SHAPE MEMORY POLYMER GOLD NANOCOMPOSITE MATERIALS FOR TRANSCATHETER CARDIOVASCULAR DEVICES by KIRAN DYAMENAHALLI B.S., University of Washington, 2008 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Bioengineering Program 2014
ii This thesis for the Doctor of Philosophy degree by Kiran Dyamenahalli has been approved for the Bioengineering Program by Dae w on Park, Chair Robin Shandas, Advisor Kristi Anseth Yiming Deng Jeff Stansbury Date: 10/1 /2014
iii Dyamenahalli, Kiran (Ph.D., Bioengineering) Shape Memory Polymer Gold Nanocomposite Materials for Trans catheter Cardiovascular Devices Thesis directed by Professor Robin Shandas ABSTRACT Modern trans catheter cardiovascular devices (TCDs) include coils and particulates for embolization of aneurysms and other vascular malformations, patches for cardiac septal defect closure, coronary and peripheral artery stents, and filters designed to cat ch blood clots. Today, the vast majority are fabricated using metal alloys, since they are normally visible using X ray based imaging modalities and possess adequate, if sub optimal, mechanical properties for their intended application. However, synthetic polymers offer a far more attractive palette of features, including reduced device costs, decreased or absent magnetic resonance and computed tomography imaging artifacts, and the ability to tune stiffness, surface interactions with blood components, and b iodegradation. A class of polymers called shape memory polymers (SMPs) is particularly promising, since they can recover almost any pre defined 3D shape following deployment from a vascular catheter. Nevertheless, while their fundamental material propertie s are very encouraging, traditional SMPs are not ideal; they are X ray invisible and still may not offer enough flexibility to span the broad range in bulk mechanical and surface properties required to fabricate ideal TCDs. To overcome these obstacles, we aim ed to engineer and evaluate a new composite SMP which incorporates gold nanoparticles (GNPs). GNPs are excellent additives for a variety of reasons. First, due to X ray s very well in the diagnostic energy range an d can make SMPs radio opaque for device visualization. Moreover, compared to traditional contrast agents like iodine, gold can be imaged at even higher X ray energies, for which bone and soft tissue absorption are minimized, improving contrast and reducing radiation dose to the patient. Second, by varying their size, concentration, and surface chemistry, the bulk properties of the resulting composite material can be tailored to a very fine degree. Third, GNPs are very well
iv characterized in the literature an d numerous methods already exist to synthesize and functionalize them for dispersion in polymer environments. Ultimately, the objective of this study was to develop a customizable GNP SMP composite material that preserves the best properties of both metals and polymers and to evaluate its utility for the design of next generation TCDs. To achieve this, a variety of synthetic challenges need ed to be addressed, such as identifying the best surface modifiers for GNPs to ensure uniform and maximum incorporation into SMPs. After a protocol was developed for reproducible production of GNP SMP composites, their thermo mechanical, radiographic, and photo thermal and degradative properties were studied. Though additional development is necessary, this research yield ed a material that is radio opaque, produces minimal MRI and CT artifacts, retains important shape recovery characteristics, and can be heated indirectly with green light. It could potentially improve patient outcomes for transcatheter therapies while reducing associated costs. The form and content of this abstract are approved I recommend its publication. Approved: Robin Shandas
v ACKNOWLEDGEMENTS I am unendingly grateful to all of the advisors, teachers, collaborators, lab mates family members, and friends who have helped me during the course of my studies. While their roles have varied, I could not have reached this point in my profes sional and personal development without them. My advisor, Dr. Robin Shandas deserves particular thanks In addition to his consistent mentorship and expertise, I appreciate his patience and the freedom he provided to explore new veins of research and interesting side projects. His experience working with industry and intellec tual property has played an important role in my training. I am also indebted to my other thesis committee members, including Drs. Kristi Anseth, Yiming Deng, Daew on Park, and Jeffrey Stansbury, for their unique perspectives and willingness to support my resea rch efforts and fellowship applications. With regard to clinical mentorship, Drs. Adel Younoszai and David Kumpe have served as my preceptors during the course of my PhD My undergraduate research mentor, Dr. James Bassuk, introduced me to research. While our work was unrelated to my PhD studies, my approach to scientific investigation was influenced strongly by the rigorous environment he established in his laboratory. I am incredibly thankful for the time they devoted to teaching. One of my goals is to me ntor students with the same enthusiasm, insight, and patience that I have seen in my own mentors. I am also grateful to t he leadership of the Medical Scientist Training Program at the University of Colorado, including Drs. Arthur Gutierrez Hartmann and Ang ie Ribera, and Jodi Cropper, for their constant support and advice. Finally, I would like to thank my lab mates, friends and family. Despite some obstacles, the last four years have been the most enjoyable exciting and formative years of my life. This is due entirely to the encouragement and support of my peers and family. My parents have always encouraged me to do my best a nd I dedicate this work to them.
vi TABLE OF CONTENTS CHAPTER I. INTRODUCTION II. GENERAL PRINCIPLES FOR CHARACTERIZATION OF SHAPE MEMORY POLYMERS FOR BIOMEDICAL APPLICATIONS A GUIDE FOR READERS III. IODINE FUNCTIONALIZED SHAPE MEMORY POLYMERS FOR EMBOLIC COILS A POINT OF COMPARISON IV. PREPARATION AND CHARACTERIZATION OF SHAPE MEMORY POLYMER GOLD NANOCOMPOSITE MATERIALS V. THERMO MECHANICAL PROPERTIES OF SHAPE MEMORY POLYMER GOLD NANOCOMPOSITE MATERIALS VI. EMBEDDED GOLD NANOPARTICLES FOR MULTI MO DALITY MEDICAL IMAGING OF S HAPE MEMORY P VII. CONCLUSIONS AND F REFERENCES APPENDIX A. PROTOCOL FOR SIZE CONTROLLED SYNTHESIS OF HYDROPHOBIC THIOL DERIVATIZED GOLD CLUSTERS USING OLEYLAMINE B. PROTOCOL FOR SIZE CONTROLLED SYNTHESIS OF THIOL DERIVATIZED GOLD CLUSTERS USING ETHYLENE GLYCOL AND POLYVINYLPYRROLIDONE
vii LIST OF TABLES TABLE 1.1 COMMON TRANSCATHETER CARDIOVASCULAR DEVICES AND THEIR 3 2.1 TECHNIQUES FOR STRUCTURAL AND CHEMICAL ANALYSIS OF SHAPE MEMORY POLYMERS 7 2.2 TECHNIQUES FOR MECHANICAL AND THERMO MECHANICAL ANALYSIS OF SHAPE MEMORY POLYMERS. 2.3 TECHNIQUES FOR ANALYSIS OF SURFACE PROPERTIES OF SHAPE MEMORY POLYMERS 2.4 TECHNIQUES FOR OPTICAL/IMAGING PROPERTIES OF SHAPE MEMORY POLYMERS 2.5 TECHNIQUES FOR BIOCOMPATIBILITY ANALYSIS OF SHAPE MEMORY POLYMERS 2.6 COMPARISON OF TRANSMISSION AND SCANNING ELECTRON MICROSCOPY FOR THE ANALYSIS OF SHAPE MEMORY POLYMERS 3.1 POTENTIAL ADVAN TAGES OF SMPS, WHEN COMPARED WITH METALS, FOR THE DESIGN OF EMBOLIC COILS 3.2 POLYMER RADIOPACITY WAS CALCULATED BY SUBTRA CTING THE BACKGROUND INTENSITY VALUE FROM EACH SAMP LE INTENSITY VALUE, THEN NORMALIZING TO THE VALUE FOR STAI NLESS STEEL. ISMP AN D SMP VALUES WERE MULTIPLI ED BY A FACTOR TO AC COUNT FOR DIFFERENCES IN SAMPLE THICKNESS 4.1 RAW DATA FOR XPS SURVEY SCAN 4.2 RAW DATA FOR XPS MULTIPLEX SCAN 6.1 RELATIVE NANOCOMPOSI TE RADIOPACITY WAS C ALCULATED BY SUBTRACTING THE BACK GROUND INTENSITY VAL UE FROM EACH SAMPLE INTENSITY VALUE, THE N NORMALIZING TO THE VALUE FOR STAINLESS STEEL 7.1 MODIFIABLE NANOPARTICLE AND POLYMER VARIABLES AND THEIR
viii LIST OF FIGURES FIGURE 2.1 MULTIPLE REFLECTION, ATTENUAT ED TOTAL REFLECTANCE (ATR) SYSTEM FOR USE WITH FTIR SP ECTROMETERS. IN THE ATR ACCESSORY, INFRA RED LIGHT PASSES THROUGH A CRYSTAL EL EMENT HAVING A REFRA CTIVE INDEX GREATER THAN T HAT OF THE SAMPLE, S O THAT IT REFLECTS A T LEAST ONCE OFF THE S AMPLE SURFACE AND PR ODUCES AN EVANESCENT WAVE THAT TYPICALLY PENETRATES BETWEEN 0 .5 AND 2 M INTO THE SAMPLE. THE BEAM IS COLLE CTED BY A DETECTOR A S IT EXITS THE CRYSTAL 2.2 IN SIZE EXCLUSION CH ROMATOGRAPHY, THE PR INCIPLE UNDERLYING G EL PERMEATION CHROMATOG RAPHY, POLYMERS IN A SOLVENT ARE PUMPED THROUGH A COLUMN PAC KED WITH MICROPOROUS BEADS. LARGER MOLECULES ARE UNABLE TO ENTER THE MAJORIT Y OF PORES AND THEREFORE PASS THROU GH THE COLUMN RELATIVEL Y RAPIDLY. SMALLER MOLECULES ARE ABLE T O ENTER MANY MORE PO RES AND MUST TRAVEL A GREATER DISTANCE, RE SULTING IN SLOWER PA SSAGE THROUGH THE COLUMN. THIS ELUTION PROFILE IS MEASURED BY DETECTORS AT THE OUTLET OF THE COLUMN 2.3 A. THE AMPLITUDE OF THE DMA SIGNAL AND P HASE ANGLE ARE USED TO STORAGE AND LOSS MOD ULI, VISCOSITY, AND OTHER IMPORTANT MATE RIAL PROPERTIES. PHA SE LAG BETWEEN STRES S ), STRAIN ( DMA TEMPERATURE SCAN OF A POLYMER, SHOWIN G DECREASE IN STORAGE MODULUS AS T EMPERATURE INCREASES DUE TO INCREASED MOLECULAR MOTION (BO ND BENDING AND STRET CHING, SIDE CHAIN MOTION ETC.) AND FRE E VOLUME. 2.4 TYPICAL DSC CURVE OF CROSS LINKED POLYMER SHOWI NG THERMAL TRANSITIONS. 2.5 DSC INSTRUMENT SCHEM ATIC. ALUMEL WIRE ME ASURES SAMPLE TEMPERATURE (T). CHR T 2.6 SAMPLE GEOMETRY AND VISCOSITY GOV ERN THE SELECTION OF RHEOLOGY FI XTURES 2.7 CONTACT ANGLE IS DEP ENDENT ON THE FORCES OF INTERACTION AMONG MOLECULES OF THE LIQ UID DROPLET AND BETW EEN THOSE OF THE LIQ UID DROPLET, SOLID SAMPL E SURFACE, AND AIR. IT REFLECTS THE CO NTRIBUTION OF POLAR, DISPERSIVE, AND OTHE R FORCES TO THE TOTAL SURFACE FREE E NERGY AND CAN PREDIC INTERACTIONS WITH CE LLS, PROTEINS, AND S MALL MOLECULES
ix 2.8 A) AMPLITUDE CONTRAS T IN POLYMER TEM ARI SES WHEN AN OBJECTIV E APERTURE IS USED TO SELECTIVELY BLOCK ELECTRONS SCAT TERED AT HIGH ANGLES FROM CRY STALLINE SEGMENTS. B ) VARIATION IN THE P HASE OR ENERGY OF ELECTRO NS DEPARTING THE SAM PLE CAN PROVIDE CONTRAST WITHIN LARG ELY AMORPHOUS POLYME RS HAVING ONLY MINOR SPATIAL VARIATIONS I N DEN SITY 2.9 DESIGN CONSIDERATION S FOR IMAGING PHANTO MS INTENDED TO ASSES S SMP BIOMEDICAL DEVIC ES 3.1 (A) ELASTIC MODULUS CURVES FROM DYNAMIC MECHANICAL ANALYSIS OF CONTROL SMP AND ISMP THIN FILMS. BELOW THE GLA SS TRANSITION TEMPERATURE (T G ), SMPS ARE IN THE G LASSY STATE; ABOVE T HE T G THEY ARE IN THE RUBBERY S TATE. THE CONTROL SM P HAD A GLASS TRANSITION TEMPERATURE OF 16.6 C, A GLASSY ELASTIC MODULUS OF 168.8 MPA AND A RUBBERY MODULUS OF 24.2 MPA. THE ISMP H AD A GLASS TRANSITION TEMPERATURE OF 24.2 C, A GLASSY ELASTIC MODULUS OF 347.2 MPA AND A RUBBERY MODULUS OF 29.3 MPA. (B) SHAPE RECOVERY OF SMP COIL AFTER BEING STRAIGHT ENED, COOLED TO 0C, AND THEN PLACED IN 3 7C WATER BATH (POLYMER DYED WITH BLUE INK T O AID VISIBILITY) 3.2 (A) FLUOROSCOPIC IMA GE OF ISMP, CONTROL SMP AND METAL COIL/T HIN FILM SAMPLES. DOTTED WHITE BOXES I NDICATE LOCATION OF RADIOLUCENT CONTROL SMP COIL AND STRIP. POLYMER STRIPS WERE 0.87 MM THICK, WHILE STAINLESS STEEL STRI P WAS 1.03 MM THICK, REQUIRING NORMALIZAT ION OF RADIOPACITY V ALUES. ACQUISITION PARAMETERS: 62 KVP, 80 MA, 2 SEC E XPOSURE. (B) AREA AVERAGED INTENSITY VALUES (PR IOR TO NORMALIZATION ) FOR 0.87 MM THICK ISMP, 0.87 MM THICK SMP AN D 1.03 MM THICK STAI NLESS STEEL STRIPS. BACKGROUND VALUES FO R 6 INCHES OF NORMAL SALINE ARE ALSO SHOWN 3.3 (A) TRANSVERSE CT SL ICE SHOWING SIGNIFIC ANT BEAM HARDENING ARTIFACT FROM NITINO L AND COOK STAINLESS STEEL COILS IN A NOR MAL SALINE ENVIRONMENT. COIL DIMENSIONS: 0.7 1 MM WIRE, 52 MM LEN GTH, 12 MM CURL. IMAGE ACQUI SITION PARAMETERS: 1 20 KV, 300 MA, 3 MM SLI CE. (B) MINIMAL BEAM HAR DENING ARTIFACT FROM 5.2 AND 26 CM ISMP C OILS. YELLOW BOX INDICATES LOCATION OF CONTROL SMP COIL, WHICH IS N OT VISIBLE. COIL DIMENS IONS: 0.71 MM WIRE, 52 MM LENGTH, 12 MM CURL 3.4 THREE DIMENSIO NAL RECONSTRUCTION O F 5.2 AND 26 CM ISMP COILS (A) AS WELL AS STAINLESS STEEL AND NITINOL CO ILS (B) FOLLOWING MA NUAL SEGMENTATION IN ITK SNAP AND POST PROCESSING IN MESHLA B, TO REVEAL A SURFACE MES H OF TRIANGULAR ELEM ENTS. IN LIGHT OF TH E 3 MM SLICE THICKN ESS, WHICH WAS LARGE RELATIVE TO THE DIME NSIONS OF THE COIL, THE QUA LITY OF THE RESULTIN G SURFACE MESH IS REASONABLE. (C) CT R ECONSTRUCTION SHOWIN G COIL PACK IN OVINE FEMORAL ARTERY
x 3.5 COMMON MRI PULSE SEQ UENCES REVEAL SIGNIF ICANT MAGNETIC SUSCEPTIBILITY ARTIF ACTS FROM NITINOL AN D STAINLESS STEEL CO ILS, WHILE UNMODIFIED AND IODINATED SHAPE MEMO RY POLYMER COILS APPEAR TO BE ARTIFAC T FREE. ORIENTATION OF SAMPLES IN SUBFIGURE A APPLIES TO B AND C. (A) T1 WEIGHTED SE SEQUENCE IMAGE ACQUISITION PARAMETE RS: TE 8.1 MS, TR 80 0 MS, FLIP ANGLE 90 SLICE THICKNESS 5 MM, ETL 2. (B) T2 WEIGHTED FSE SEQUENC E. IMAGE ACQUISITION PARAMETE RS: TE 81 MS, TR 300 0 MS, FLIP ANGLE 90 SLICE THICKNESS 5 MM, ET L 16. (C) SPIN DENSI TY (PROTON) WEIGHTED GRE SEQUENCE. IMAGE ACQUISITION PARAMETE RS: TE 1.5 MS, TR 25 0 MS, FLIP ANGLE 25, SLICE THI CKNESS 5 MM, ETL 1 4.1 DRY SYNTHESIS OF POL YDISPERSE METAL NANO PARTICLES CAN BE ACHIEVED USING HIGH ENERGY BALL MILLING EQUIPME NT. IN A BALL MILL, MULTIPLE GRINDING BO WLS ARE ROTATED ON I NDEPENDENT PLATFORM, WHILE THE ENTIRE ASS EMBLY IS ROTATED IN THE OPPOSITE DIRECTI ON. GAL FORCES ALTERNATE LY ADD AND SUBTRACT, CA USING T HE GRINDING BALLS TO ROLL AROUND THE BOWLS AND BE THROWN TO THE OPPOSITE SIDE IMPACTING THE SUBSTRATE PARTICLES AT HIGH SPEED 4.2 PROCESSES FOR GAS AN D LIQUID PHASE SYNTH ESIS OF METAL NANOPARTICLES (S, L AND G = SOLID, LIQUI D AND GAS PHASE, RESPECTIVELY) 4.3 THE EFFECT OF REDUCI NG CONDITIONS ON GNP SIZE. STRONGER REDUCING AGENTS RESU LT IN FASTER NUCLEAT ION AND A GREATER NUMBER OF INITIAL SE ED PARTICLES. THIS U LTIMATELY LEADS TO SMALLER GNPS 4.4 COVALENT BONDING OF BIOCONJUGATED THIOLS AND DISULFIDES TO GOLD SURFACES 4.5 REACTION SCHEMATIC F OR THE PREPARATION O F HYDROPHOBIC THIOL DERIVATIZED GOLD NAN OPARTICLES. REACTION 1 SHOWS THE REDUCTIO N AN D CAPPING OF TETRACH LOROAURATE TRIHYDRAT E BY ETHYLENE GLYCOL AND POLYVINYL PYRROLIDONE (PVP). R EACTION 2 SHOWS THE EXCHANGE OF PVP FOR DEODECANETHIOL (DDT) 4.6 SAMPLES FOR XPS ANAL YSIS WERE PREPARED B Y SERIALLY DEPOSITIN G A DILUTE SOLUTION OF SURFACE MODIFIED GNPS IN HEX ANE ONTO 1 CM 2 POLISHED SILICON WAF ERS. A TOTAL OF 0.5 MG WAS DEPOSITED ON EACH WAFER 4.7 X RAY PHOTOELECTRON SP ECTRUM OF DODECANETH IOL FUNCTIONALIZED GOLD NANOPARTICLE SU RFACE (LOW RESOLUTI ON SURVEY SCAN), SHOWING CARBON 1S, S ULFUR 2P AND G 4.8 X RAY PHOTOELECTRON SP ECTRUM OF DDT FUNCTIONALIZED GOLD NANOPARTICLE SURFACE (HIGH RESOLUTION MULTIPLEX SCAN). SULFUR 2P, GOLD 4S, AND CAR BON 1S PEAKS ARE SHO WN. THE LOCATION OF THE
xi SULFUR 2P PEAK IS CO NSISTENT WITH THE BI NDING ENERGY OF GOLD BOUND SULFUR SPECIES. THE EXPECTE D SPIN ORBIT SPLITTING BETW EEN THE SULFUR 2P1/2 AND 2P3/2 PEAKS IS EVIDE NT. RAW DATA WAS ANALYZED USING CASAX PS SOFTWARE 78 4.9 CHARACTERIZATION OF DDT FUNCTIONALIZED GOLD NANOPARTICLES: UV VIS SPECTRUM (A) AND DYNAMIC LIGHT SCATTERING DAT A (B) OF NANOPARTICLES INDICA TING AVERAGE PARTICL E SIZES OF APPROXIMATELY 12 AND 14 NM, RESPECTIVELY; TRANSMISSION ELECTRO N MICROGRAPHS OF DDT FUNCTIONALIZED GOLD NANOPARTICLES SHOW EXCELLENT DISPERSION IN PURELY HYDROPHOBI C ENVIRONMENT S LIKE HEXANE (C) AND MODER ATE CLUSTERING WHEN EMBEDDED IN POLYMERIZED SHAPE ME MORY POLYMER AT 1 WT % (D, E) 4.10 TRANSMISSION ELECTRO N MICROSCOPY IMAGES OF DDT FUNCTIONALIZED GNPS SUSPENDED IN HE XANE AND DEPOSITED O N GLASS SLIDES. PARTICLES ARE DO NOT DISPLAY AGGRE GATION IN LARGE CLUSTERS 4.11 CONTACT ANGLE OF A L AYER OF DDT FUNCTIONALIZED GNPS DEPOSITED ON SILICON WAFERS AL ONG WITH PLAIN SMP F ILMS, WITH WATER AND DIIODOMETHANE (A NON POLAR LIQUID). SURFA CE FREE ENERGY WAS CALCULATED USING THE 4.12 CHEMICAL STRUCTURE O F MONOMERS USED IN T HIS STUDY: (A) TERT BUTYL ACRYLATE (TBA); (B) POLY(ETHYLENE GLYCOL ) DIMETHACRYLATE (PEGDMA) 550 MN 4.13 INFRARED SPECTRA OF AN UNREAC TED ACRYLATE MONOMER FILM AND TWO POLYMERIZED NANO COMPOSITE MATERIALS CONTAINING 0 AND 1 WT% GOLD NANOPARTICL ES. THE ABSENCE OF A VINYL GROUP ABSORBANCE PEAK AT 9 00 CM 1 SHOWS ADEQUATE BOND CONVERSION INDEPENDENT OF GNP I NCORPORATION 4.14 INFRARED SPECTRA OF AN UNREACTED ACRYLAT E MONOMER FILM AND TWO POLYMERIZED NANO COMPOSITE MATERIALS CONTAINING 0 AND 1 WT% GOLD NANOPARTICL ES. A BROAD AND WEAK ER ABSORPTION PEAK I S EXPECTED AT 880 900 CM 1 REPRESENTING GEM DI SUBST ITUTED ALKENES (AS FOUND IN DI METHACRYLATES SUCH A S PEGDMA) 4.15 TRANSMISSION ELECTRO N MICROSCOPY IMAGES OF DDT FUNCTIONALIZED GNPS EMBEDDED, AT VA RYING CONCENTRATIONS IN A (METH)ACRYLAT E SMP CONSISTING OF 80 % (W/W) TERT BUTYL ACRYLATE AND 2 0% ( W/W) POLY(ETHYLENE GLYCOL ) DIMETHACRYLATE, MN 550. SUBFIGURES A THROUGH H CORRESPOND TO GNP CONCENTRATION S OF 0.1, 0.2, 0.5, 1, 1.5, 2, 2.5 AND 3 WT %. AS THE GNP INCORP ORATION IS INCREASED CLUSTER SIZE ALSO IN CREASES, WITH A DRAM ATIC INCREASE IN PAR TICLE AGGREGATION SEEN BET WEEN 2.5 AND 3 WT% 4.16 LARGE FIELD TRANSMISSION E LECTRON MICROSCOPY I MAGES OF DDT FUNCTIONALIZED GNPS EMBEDDED IN A (METH) ACRYLATE SMP
xii CONSISTING OF 80% (W /W) TERT BUTYL ACRYLATE AND 2 0% (W/W) POLY(ETHYLENE GLYCOL ) DIME THACRYLATE, MN 550. THE GNPS FORMED INCREASINGLY LARGE C LUSTERS AS THE GNP C ONTENT IN THE SMP IS INCREASED, BUT THE C LUSTERS THEMSELVES W ERE WELL SEPARATED. (A) 2.5 WT% GNPS. (B ) 3 WT% GNPS 5.1 PROCEDURE USED TO EV ALUATE SHAPE RECOVERY OF SMP GOLD NANOCOMPOSITE MATERI ALS .. 5.2 THERMO MECHANICAL PROPERTIE S OF NANOCOMPOSITE M ATERIALS AS MEASURED BY DYNAMIC MECHANICAL ANALYSIS: (A) GLASS TRANSITION TEMPERATURE AND TRAN SITION WIDTH; (B) GL ASSY MODULUS; (C) RU BBERY MODULUS; (D) F REE STRAIN RECOVERY AND STRAIN FIXITY; ( E) SHAPE RECOVERY SHARP NESS 6.1 NATIONAL INSTITUTE O F STANDARDS AND TECH NOLOGY ELEMENTAL MAS S ATTENUATION DATA FOR GOLD, IODINE, AND SO FT TISSUE. THE HIGHE R K ELECTRON SHE K RED TO TRADITIONAL CONTRAST AGENTS MAY ALLOW GNP EMBEDDED BIOMATERIALS TO BE I MAGED AT HIGHER X RAY ENERGIES, FOR WH ICH THE RADIATION DOSE T O HUMAN TISSUE IS MI NIMIZED 101 6.2 (A) SOLIDWORKS ILLUS TRATION O F WATERTIGHT ACRYLIC PHANTOM. (B) COMPLETED PHANTOM. ( C) CLOSE UP VIEW OF 0.1 MM MO NOFILAMENT NYLON WIRE USED TO S USPEND SAMPLES FOR I MAGING 106 6.3 FLUOROSCOPIC IMAGES OF 0.5 ML SMP GOLD NANOCOMPOSITE S AMPLES CONTAINED IN POLYPRO PYLENE CENTRIFUGE T UBES AND SUSPENDED I N A NORMAL SALINE FILLED ACRYLIC PHANT OM. COLLOIDAL GOLD C ONTENT RANGED FROM 0 TO 3% BY WEIGHT, AS SHOWN IN SUBFIGURE A. 4 X 15 MM (D X L) STAINLESS ST EEL ROD AND BUNDLED 22 GAUGE NITINOL WIR E OF EQUIVALENT VOLUME WE RE USED AS POSITIVE CONTROLS. THE 100 PIXEL REGIONS OF INTEREST DEFINED FOR DENSITOM ETRY ANALYSIS (TABLE 6.1) ARE SHOWN IN SUBFIGU RE B. THE ORIENTATIO N/LABELING OF SAMPLE S IN SUBFIGURE A ALSO APP LIES TO B, C, AND D. ACQUISITION PARAMETE RS FOR EACH IMAGE ARE A S FOLLOWS: (A) 60 KVP, 120 MA, 1.4 MGY ; (B) 81 KVP, 118 MA, 0.6 MGY; (C) 102 KVP, 79 MA, 0.3 MGY; (D) 125 KVP, 40 0.1 MGY 6.4 TRANSVERSE CT SLICES OF NANOCOMPOSITE AND METAL SAMPLES IN A 0.9% (W/V) SALINE EN VIRONMENT. (A) MINOR DISTOR TIONS FROM SMP GOLD NANOCOMPOSITE M ATERIALS WITH UP TO 0.5% (W/W) GOLD NANOPARTICLE INCORPO RATION. ACQUISITION PARAMETERS: 120 KVP, 316 MA. (B) MODERATE DIS TORTIONS FROM SMP GOLD NANOCOMPOSITE MATERIALS WITH 1 3% (W/V) GOLD NANOPA RTICLE INCORPORATION ACQU ISITION PARAMETERS: 120 KVP, 251 MA. (C) SIGNIFICANT BEAM HARDENING AND PHOTON STARVATION ARTIFACTS FROM STAINLESS STEEL AND NITINOL SA MPLES
xiii 6.5 AGING OF NANOCOMPOSI TE AND METAL SAMPLES IN A 0.9% (W/V) SALINE EN VIRONMENT, UTILIZING LOWER ACCELERATING VOLTAGE. (A,B) NO VI SIBLE DISTORTIONS FR OM SMP GOLD NANOCOMPOSITE MATERI ALS WITH UP TO 3% (W /W) GOLD NANOPARTICL E INCORPORATION. ACQUI SITION PARAMETERS: S UBFIGURE A 80 KVP, 499 MA; SUBFIGURE B 80 KVP, 404 MA. (C) SIGNIFICANT BEAM HAR DENING AND PHOTON STARVATION ARTIFACTS FROM STAINLESS STEEL AND NITINOL SAMPLES REMAIN. ACQUISITION PARAMETERS: 80 KVP, 516 MA. RESULTS FROM LOW KVP AND NOR MAL (FIGURE 4) SCANS ARE COMPARABLE FOR METAL SAMPLES 6.6 HIGH SLICE THICKNESS T2 WEIGHTED FSE MRI IMA GE OF 0.5 ML SMP GOLD NANOCOMPOSITE MATERI ALS AND NITINOL WIRE BUNDLE CONTAINED IN POLYPROPYLENE CENTRI FUGE TUBES AND SUSPE NDED AT THE CENTER F 0.9% (W/V) S ALINE + 5 MM GADOLIN IUM CHELATE CONTRAST. IMAGE ACQU ISITION PARAMETERS TE: 64.896 MS, TR: 3 000 MS, FLIP ANGLE: 90, SLI CE THICKNESS: 40 MM, ETL: 8, PIXEL BANDWI DTH: 244.141, AVERAGES: 1 NANOCOMPOSITE SAMP LES WERE ARTIFACT FR EE AND ARE VISUALIZED T HROUGH NEGATIVE CONTRAST, W HILE THE NITINOL SAMPLE GENERATED SIG NIFICANT MAGNETIC SU SCEPTIBILITY ARTIFACT 6.7 LOW SLICE THICKNESS MR IMAGES OF SMP GOLD NANOCOMPOSITE MATERIALS, CONTAININ G VARYING CONCENTRAT IONS OF COLLOIDAL G OLD, AND BUNDLED NIT INOL WIRE SAMPLE. AL L SAMPLES WERE CONTAINED IN POLYPRO PYLENE CENTRIFUGE TU BES AND SUSPENDED AT THE CENTER OF A 6X6X /V) SALINE + 5 MM GADOLINIUM CHELATE C ONTRAST. COMMON MRI PULSE SEQUENCES REVEAL SIGNIFICANT M AGNETI C SUSCEPTIBILITY ART IFACTS FROM NITINOL, WHILE SMP GOLD NANOCOMPOSITE S AMPLES APPEAR TO BE ARTIFACT FREE. THE ORIENTATIO N/LABELING OF SAMPLE S IN SUBFIGURE A ALSO APPLIES TO SUBF IGURES B F. (A) T1 WEIGHTED SE SEQUENCE IMAGE ACQUISITION PARAMETE RS TE: 11 MS, TR: 700 MS, FLIP ANGLE: 90, SLICE THICKNESS: 5 MM, ETL : 1, PIXEL BANDWIDTH : 195.312, AVERAGES: 1. (B) T2 WEIGHTED FSE SEQUENC E. IMAGE ACQUISITION PARAMETERS TE: 64.896 MS, TR: 3000 MS, FLI P ANGLE: 90, SLICE THICKNESS: 5 MM, ETL : 8, PIXEL BANDWIDTH: 195.312, AVERAGES: 1 (C) SPIN DENSITY ( PROTON) WEIGHTED GRE SEQUENCE. IMAGE ACQUISITION PARAMETE RS TE: 11 MS, TR: 2000 MS, FLIP ANGLE: 90, SLICE THICKNESS: 5 M M, ETL: 1, PIXEL BAN DWIDTH: 195.312, AVERAGES: 1 116 7.1 SHAPE RECOVERY O F SMP GNP COMPOSITE STRIP CONTAINING 0.5 WT% GNPS AFTER TEMPORARY DEFO RMATION (ROLLING) UP ON ILLUMINATION BY 1 00 MW, 532 NM SOLID STATE LASER ELAPSED TIME ~20 SEC ....... .123
xiv ABBREVIATIONS AFM ATOMIC FORCE MICROSC OPY AIBN AZOBIS(2 METHYLPROPIONITRILE) ASTM AMERICAN SOCIETY FOR TESTING AND MATERIAL S ATR FTIR ATTENUATED TOTAL REF LECTANCE FTIR AUC AREA UNDER THE CURVE CT COMPUTED TOMOGRAPHY CTAB CETYL TRIMETHYLAMMON IUM BROMIDE DICOM DIGITAL IMAGING AND COMMUNICATIONS IN ME DICINE DLS DYNAMIC LIGHT SCATTE RING DMA DYNAMIC MECHANICAL A NALYSIS DSC DIFFERENTIAL SCANNIN G CALORIMETRY DTEM DYNAMIC IN SITU TRAN SMISSION ELECTRON MI CROSCOPY DTT DODECANETHIOL EG ETHYLENE GLYCOL E G GLASSY MODULUS E R RUBBERY MODULUS ETL ECHO TRAIN LENG TH FDA (UNITED STATES) FOOD AND DRUG ADMINISTRAT ION FSE FAST SPIN ECHO FTIR FOURIER TRANSFORM INFRARED ( SPECTROSCOPY) FWHM FULL WIDTH AT HALF MAXIMUM GDC GUGLIELMI DETACHABLE COIL GNP GOLD NANOPARTICLE GPC GAS PERMEATION CHROMATOG RAPHY GRE GRADIENT RECALLED EC HO HPLC HIGH PRESSURE LIQUID CHROMATOGRAPHY ISO INTERNATIONAL ORGANI ZATION FOR STANDARDI ZATION KVP PEAK KILOVOLTAGE MA MILLIAMPERES MRI MAGNETIC RESONANCE I MAGING MS MAS SPECTROMETRY MW MOLECULAR WEIGHT NMR NUCLEAR MAGNERIC RES ONANCE (SPECTROSCOPY ) PCL POLYCAPROLACTONE PDI POLYDISPERSITY INDEX PEGDMA POLY(ETHYLENE GLYCOL ) DIMETHACRYLATE PMMA POLY(METHYL METHACRY LATE) PTFE POLYTETRAFLUOROETHYL ENE PVP POLYVINYLPYRROLIDONE SE SPIN ECHO SEM SCANNING ELECTRON MI C ROSCOPY i SMP IODINATED SHAPE MEMO RY POLYMER SMP SHAPE MEMORY POLYMER STA SIMULTANEOUS THERMAL ANALYSIS t BA TERT BUTYL ACRYLATE TCD TRANSCATHETER CARDIO VASCULAR DEVICE TE ECHO TIME TEM TRANSMISSION ELECTRO N MICROSCOPY T g GLASS TRANSITION TEMPERATU RE TGA THERMOGRAVIMETRIC AN ALYSIS TIA 2 ACRYLOYLOXYETHYL 2,3,5 TRIIODOBENZOATE TOF SIMS TIME OF FLIGHT SECONDARY ION MASS SPECTROMETRY TR REPETITION TIME
xv UV ULTRAVIOLET UV V is ULTRAVIOLET VISIBLE WAXD WIDE ANGLE X RAY SCATTERING XPS X RAY PHOTOELECTRON SPECTROSCOPY
1 CHAPTER I INTRODUCTION Percutaneous intervention with trans catheter devices is based on the principle that lesions in the cardiovascular system can be accessed and repaired from inside the heart and vascular compartment, without the need for open surgical procedures. Its applic ation has grown significantly over the last decade and trans catheter cardiovascular devices (TCDs) now include coils and particulates for embolization of vascular malformations (especially aneurysms) and arterial tumor supplies, patches and grafts for car diac septal defect closure or vascular reconstruction, coronary and peripheral artery stents, and filters designed to catch blood clots before they reach the lungs or carotid arteries (reducing the risk for pulmonary embolism or stroke, respectively). 1 2 3 Percutaneous intervention is often a good option in patients for whom customary vascular or trans thoracic cardiac surgery is contraindicated, due to heart failure, previous surgeries, or age. Moreover, the initial promises of this approach, including shortened procedure and recovery times, reduced costs and repeat procedure rates, and improved patient outcomes, have been achieved in many cases. 4 However, the adoption of TCDs has been slowed by recent reports which fail to show equivalent or superior clinical outcomes compared to traditional surgical repairs, narrow indication windows for certain devices, is sues with biocompatibility, and related post procedure complications. Many of these complications, such as recanalization and bleeding of coiled aneurysms or stent associated thrombus formation, are caused by failures in either the design or composition of TCDs. 5 Accordingly, their resolution requires us t o solve fundamental materials science problems. Conventional materials used in most TCDs include platinum, stainless steel, and alloys of titanium, such as Nitinol. Metals are used because they are durable and generally visible using X ray based imaging mo dalities. However, their mechanical properties, such as flexural stiffness and compressibility, are only tunable within a very narrow range and their surface properties, which determine their tendency to incite an inflammatory response or form blood clots, are largely static. Additionally, metals are unable to elute drugs without a polymer coating or biodegrade
2 once their function is complete, and they distort computed tomography (CT) and magnetic resonance imaging (MRI) scans through significant artifact g eneration. In some cases, placement of ferromagnetic metal TCDs is a contraindication to MRI scans due to its use of high magnetic field strengths. Finally, in terms of both material and processing costs, metals are very expensive to manufacture, transferr ing higher costs to patients and providers. 6 The shift toward percutaneous interventions and importance of material properties in TCD pe rformance is perhaps best illustrated by a specific example; here, we take the case of embolic coils for aneurysm repair a bit further. Aneurysms are pathologically weakened and dilated sections of blood vessels that are at increased risk of rupture. In th e cerebral vasculature, rupture of an aneurysm can lead to catastrophic hemorrhagic stroke. One option for early intervention is a major neurosurgical procedure involving a craniotomy and placement of a clip at the neck of the malformation. However, this a pproach has lost ground to minimally invasive embolization techniques, which involve trans catheter delivery of small coils into the aneurysm, producing an effective seal through physical packing, hemostasis, thrombosis, and eventually neointimal formation To date, embolic coils have primarily been fabricated using stainless steel or platinum. Although such coils are well accepted clinically, they are limited by their poor capacity for shape memory, poor resistance to kinking, and relatively high stiffness all of which prevent optimal packing of the aneurysm. Furthermore, CT and MRI artifacts generated by metal coils prevent accurate visualization of proximal anatomy. Clinicians are obligated to use fluoroscopy for follow up evaluation, increasing radiatio n dose to the patient. Though it is simplest to view the potential benefits of polymers over metals through a case study of one device, the true value of polymers lies in their flexibility. With that in mind, the approach of this dissertation was not to fo cus on a single device but rather all TCDs, w ithout bro adening the experimental scope. With this approach, the range in attainable material properties, rather than more rigid pre defined design aims, would suggest specific devices that are more promising than others. Table 1.1 outlines common TCDs used in modern medical practice and their ideal material properties.
3 Table 1.1 : Common transcatheter cardiovascular devices and their idealized material properties. TCD Ideal material properties Embolic coil Low stiffness Rapid deployment High X ray visibility Low cost Inferior vena cava filter High recovery force X ray visibility Low cost Amplatzer occlusion device Large recoverable deformation X ray visibility Low cost Stent Potential for biodegradation X ray visibility Low cost In view of the drawbacks of metals, the prospect of fully polymeric TCDs have sparked much interest. Synthetic polymers offer a far more attractive palette of features, including reduced device costs, decreased or absent MRI and CT imaging artifacts, and t he ability to tune stiffness, surface interactions with blood components, biodegradation, and drug elution. Among polymers, shape memory polymers (SMPs) have highly desirable properties for catheter based storage and release. These materials can recover al most any pre determined shape of very low stiffness after being heated above a tunable glass transition temperature (T g ) 1 7 8 Repeated strain recoveries of
4 several hundred percent are possible over multiple cycles. 9 10 This feature could allow, for instance, emboli c polymer devices to recover a helical coil conformation upon release at body temperature, after being stored in a catheter in a compacted or elongated form. 11 12 Even so, SMPs do have drawbacks. For instance, native SMPs st ill do not offer enough flexibility to span the broad range in bulk mechanical properties required to fabricate ideal TCDs. Moreover, in any TCD application, accurate placement is critical to device performance and safety, and even with the advent of real time/4D MRI, X ray based imaging modalities are almost always employed in this capacity. 13 14 Polymers are largely radiolucent, so heavy element fillers are often used to absorb and scatter X ray s. 15 16 For instance, barium sulfate, zirconium oxide and tantalum have been used in the orthopedic field for bone cement, 17 18 19 tantalum filled SMPs have been evaluated in the research setting as embolic coil materials, 11 and iodinated monomers have been incorporated in to denture base resins. 20 However, the addition of these fillers tend to disturb th e original SMP characteristics. To overcome these issues, our laboratory has developed a special interest in gold nanopa rticles (GNPs) for generating radio opaque and mechanically optimal SMP nanocomposite materials. GNPs are advantageous for a number of reasons. First, compared to iodine, barium, tantalum and zirconium, gold offers a higher atomic weight (~197) 21 and a superior ability to attenuate X ray s at m ost energy levels in the diagnostic range. 22 shell binding energy, it can be imaged at even higher X ray energies (> 80 keV) for which bone and soft tissue absorption are minimized, improving contrast and reducing ionizing ra diation dose to the patient. 23 Polymer gold nanocomposites have primarily found use in optical app lications, such as lenses, filters and light emitting diodes, 24 25 but in colloid form, functionalized GNPs have been evaluated as X ray contrast agents for vasculature and micro damaged bone, with favorable results. 23 26 27 Second, by varying their size, aspect ratio, concentration, and surface chemistry, the bulk properties of the resulting composite material can be tailored to a very fi ne degree. Third, GNPs are very well characterized in the literature are well known for the ease and flexibility of their synthesis, excellent size and shape control, their long term stability in a variety of solvents, and amenity to surface modification w ith thiols, amines and phosphines for dispersion in polymer
5 environments. 28 29 30 31 Fourth, GNPs can confer entirely new properties to SMPs. Native polymers are electrical and thermal insulators, but GNPs may allow these properties to be controlled in a concentration dependent manner. Likewise, GNPs can dissipate visible light as heat. Zhang et al. recently harnessed the unique surface plasmon resonance enhanced scattering and absorptio n of green light to trigger shape changes in SMPs. 32 In short the objective of this study was to develop a novel, customizable, and thermally responsive gold SMP nanoc omposite material for the design of next generation TCDs. While additional investigation is needed, this material preserve s many of the best features of both metals and polymers, while adding properties that could allow for new modes of device delivery and use. In this study, characterization of the composite material focused primarily on (1) imaging (ie. enhancing X ray contrast, while minimizing CT and MR artifacts); (2) thermomechanical properties (ie. preserving and fine tuning temperature dependent mec hanical properties of the original SMP); and (3) photo thermal properties (ie. using laser light to heat the material and trigger thermal transitions without raising the ambient temperature). While biocompatibility evaluation (cytotoxicity, residual materi al analysis, clotting tendency etc.) is a mainstay of new device development, it is not the focus of this study. However, in light of the significant precedent for the use of (meth)acrylate polymers in intravascular devices, the insoluble nature of cross l inked polymer systems, and the inertness of passivated gold surfaces, it is likely that the composite material would perform well. Before presenting the SMP GNP composite, results from the evaluation of a new iodinated SMP system, with similar objectives, are shown in order to provide a point of comparison. Importantly, investigations into the last area of focus, pho to thermal properties, are still early, but show promising results and interesting implications for the development of photo thermal release me chanisms. Consequent ly, these initial results are presented in the Future Directions chapter.
6 CHAPTER II GENERAL PRINCIPLES FOR CHARACTERIZATION OF SHAPE MEMORY POLYMERS FOR BIOMEDICAL APPLICATIONS A GUIDE FOR READERS [Adapted from the following published book chapter: Dyamenahalli K, Famili A, Shandas R. Woodhead, 2014.] Introduction Potential uses of shape memory polymers (SMPs) and their composites in the medical field are numerous and include endovascular devices, such as stents and embolic coils, intra abdominal patch repair of inguinal hernias or pelvic organ prolapse, intra ocular lenses, and drug delivery devices. The goal of this chapter is to familiarize the reader with modern laboratory tools available to study properties of SMPs in view of their proposed function. While some of the techniques described he rein can be generalized to the characterization of traditional polymers or SMPs designed for industrial uses, most were selected specifically for their relevance to SMPs intended for biomedical applications. In such roles, the utility of SMPs is most often found in the mechanical properties they display following a thermal transition, typically occurring near body temperature. Primary among these properties are the glass transition temperature, glassy and rubbery moduli, strain recovery and fixity rates, an d recovery stress. That said, this chapter encompasses a much wider range of techniques, from micro structural characterization to measurement of oxidative stability and surface fouling. In addition to discussing theoretical concepts and practical consider ations associated with each SMP characterization technique, we attempt to lead the reader through the process of identifying the most relevant properties to evaluate and the comparative advantages of each technique, in an application specific manner. Accor dingly, we begin with a series of tables (1 5), which score the ability of various material characterization techniques to evaluate SMP properties. Guidance for further reading, a listing of important American Society for Testing and Materials ( ASTM ) Unit ed States Pharmacopoeia
7 (USP), and International Organization for Standardization ( ISO ) reference standards, and additional res ources have also been provided. Table 2.1: Techniques for structural and chemical analysis of shape memory polymers. UV Vis spectroscopy FTIR spectroscopy GPC WAXD NMR spectroscopy MS Molecular weight and polydispersity ** ** Functional groups ** ** Chemical composition ** ** Chain conformations * ** ** Transitions ** ** Residual monomers ** ** ** ** End groups ** ** Degree of crystallinity ** ** = strong technique for acquiring this information, = weak technique for acquiring this information, = technique inadequate for acquiring this information, UV Vis = ultraviolet visible FTIR = Fourier transform infrared, GPC = gel permeation chromatography, WAXD = wide angle X ray diffraction, NMR = nuclear magnetic resonance, MS = mass spectrometry, DSC = differential scanning calorimetry
8 Table 2 .2: Techniques for mechanical and thermo mechanical analysis of shape memory polymers. Tensometry DMA DSC TGA Rheology Isothermal elastic deformation properties ** Flexural and shear moduli ** ** Complex, storage and loss moduli ** Shape recovery parameters ** Creep and stress relaxation ** ** Thermal decomposition behavior ** T g and T m ** ** Sub T g thermal transitions ** * Viscosity ** ** ** ** = strong technique for acquiring this information, = weak technique for acquiring this information, = technique inadequate for acquiring this information, DMA = dynamic mechanical analysis, DSC = differential scanning calorimetry, TGA = thermogravim etric analysis Table 2.3: Techniques for analysis of surface properties of shape memory polymers. AFM Contact angle measurements XPS TOF SIMS Surface chemistry ** ** ** Surface wetting and free energy ** Surface functionalization ** ** ** Surface topology ** Surface mechanical properties ** ** = strong technique for acquiring this information, = weak technique for acquiring this information, = technique inadequate for acquiring this information, AFM = atomic force microscopy, XPS = X ray photoelectron spectroscopy, TOF SIMS = time of flight secondary ion mass spectrometry
9 Table 2.4: Techniques for optical/imaging properties of shape memory polymers. SEM TEM Phantom based X ray imaging Phantom based MR imaging UV Vis spectroscopy FTIR spectroscopy Ellipsometry Near visible absorbance / transmittance spectra ** ** Refraction, aberration, and birefringence ** Radio opacity ** MR signal and artifacts ** Porosity, density and free space ** ** = strong technique for acquiring this information, = weak technique for acquiri ng this information, = technique inadequate for acquiring this information, SEM = scanning electron microscopy, TEM = transmission electron microscopy, UV Vis = ultraviolet visible, FTIR = Fourier transform infrared Table 2.5: Techniques for biocompatibility analysis of shape memory polymers. In vitro cytotoxicity In vitro functional testing In vivo sensitization In vivo irritation In vivo functional testing Material stability * * ** Cytotoxicity ** ** ** Pyrogenicity ** ** Cellular functional response ** Physiological functional response ** Protein adsorption ** ** ** = strong technique for acquiring this information, = weak technique for acquiring this information, = technique inadequate for acquiring this information
10 Structural and Chemical Characterization Extensive structural and chemical characterization is necessary to understand the nature of SMPs and predict their behavior. While a precise composition is typically targeted at the start of any synthesis procedure, it must be verified experimentally using multiple specific methods. The combined output of these analyses provides insight into the chemical structure of the synthesized SMP, with each providing insight into one or more desired polymer prop erties, as outlined in Table 2.1 Ultraviolet Visible Spectroscopy Ultraviolet visible (UV Vis) spectroscopy is a technique most commonly applied to quantitative measurement of solutions, specifically for determining concentrations of known solutes. However, its application to solid state samples can also provide valuabl e insights. Some of the key advantages of UV Vis spectroscopy are its relative ease of use, cost effectiveness and sampling speed. Hence, it can be a good starting point for proof of concept work or to guide early experimentation. The physical principles u nderlying this method are straightforward, making the instrumentation simple and robust. Light of known wavelength and intensity is directed at the sample and its final intensity, after passing through, is measured by a detector. By comparing the incident radiation ( I O ) and the transmitted radiation ( I ), the amount of light absorbed by the sample at that particular wavelength can be easily calculated. Using the Beer Lambert law, this absorption can be used to measure c oncentrations of known solutes: In this equation, A the molar absorbtivity (L mol 1 cm 1 ), c is the concentration of the dissolved solute (mol L 1 ), and L is the path length (cm). UV Vis spectroscopy is relatively weak at identifying compounds; it is much more useful in quantitative assessments. In SMP
11 characterization, UV Vis spectroscopy is an especially effective method for measuring basic optical charac teristics of a sample and the concentrations of its known extractables or degradation products. Optical characteristics measured in the solid phase will be of particular interest for ocular device applications, where UV blocking properties must be characte rized and quantified and optical clarity in the visible region is critical 33 In addition, knowledge of the absorption spectrum is important for light activated SMP systems, depending on the wavelength of light in question 34 Extraction studies, used to quantify the ability of unreacted monomers or other additives (e.g. plasticizers or stabilizers) to diffuse out of the SMP network, often emp loy UV Vis spectroscopy to measure the concentration of these compounds in the extraction medium. However, more sensitive methods, such as high performance liquid chromatography (HPLC), can also be used 35 Fourier Transform Infrared Spectroscopy Fourier Transform Infrared Spectroscopy (FTIR) is an invaluable analytical tool, especially when assessing bond conversion during polymerization. Like UV Vis spectroscopy, FTIR relies on transmission of radiation through a sample. However, moving from the UV and visible to the infrared regions of the electromagnetic spectrum allows for detection of molecular vibrational states, which provides information on chemical structure and bonding. For a molecule to absorb energy in the infrared energy range, it must posses s a mode of vibration that causes a shift in dipole moment. This dipole moment can be caused either by permanent dipoles, which occur when two atoms in a molecular have significantly different electronegativity or asymmetrical stretching or bending in a mo lecule that has no permanent dipole. The constituent atoms responsible for dipole moments will influence the energy required to vibrate or stretch these bonds, allowing infrared absorption peaks to be linked with chemical structure. These spectra act as fi ngerprints for particular types of bonding, allowing for their identificati on and relative quantification.
12 In SMP applications, FTIR can provide a vast array of data. In addition to information about general bonding and structure within the polymer network FTIR can identify and quantify functional groups. This is particularly advantageous when functional groups are either involved in the shape memory transition or are to be used for subsequent tethering/modification of the polymer, such as with a biomacrom olecule. Perhaps the most valuable function of FTIR is its ability to monitor polymerization specific parameters, such as degree of conversion. By tracking the evolution of absorbance peaks associated with functional groups formed or consumed during polyme rization (e.g. vinyl groups), the polymerization process can be tracked in real time. Reactions can be optimized by measuring conversion as a function of time, temperature, initiator concentration and other parameters 36 For many polymer systems, attenuated total reflectance FTIR (ATR FTIR) can be used to extend FTIR analytical principles to polymers in solution or to isolate surface properties of bulk polymers. For example, it may be used to assess the effects of storage or sterilization methods on SMP surface chemistry, including oxidation and other degradation processes. Additionally, ATR (attenuated total refl ection) FTIR analysis can be used to identify protein surface adsorption, with the additional benefit of providing structural information about the protein after adsorption. Conveniently, most modern FTIR spectrometers can be converted simply by mounting an ATR accessory to the sample compartment. In the ATR accessory, infrared light passes through a crystal element having a refractive index greater than that of the sample, so that it reflects at least once off the sample surface and produces an evanescent wave that typically penetrates between 0.5 and 2 m into the sample. The beam is collected by a detector as it exits the crystal. Th is scheme is shown in figure 2. 1.
13 Figure 2.1: Multiple reflection, attenuated total reflectance (ATR) system for use with FTIR spectrometers. In the ATR accessory, infrared light passes through a crystal element having a refractive index greater than that of the sample, so that it reflects at least once off the sample su rface and produces an evanescent wave that typically penetrates between 0.5 and 2 m into the sample. The beam is collected by a de tector as it exits the crystal. Gel Permeation Chromatography Gel Permeation Chromatography (GPC) is by far the most common choice in polymer characterization for the evaluation of molecular weight (MW) and polydispersity index (PDI), which measures the distribution or heterogeneity in molecular mass in a polymer sample. Both are critical parameters; most mechanical properties vary considerably with MW and PDI plays a key part in determining the sh arpness of thermal transitions. GPC is based on the principles of size exclusion chromatography, by which polymers are separated based on their hydrodynamic radius, as illustrated in F igure 2. A column is packed with microporous polystyrene beads with carefully controlled pore sizing. As the polymer solution permeates the column, molecules will either pass primarily through interstitial spaces (for molecules too large to enter the pores ) and elute rapidly from the column, or will be transported into and between the beads (for molecules small enough to enter the pores), eluting slowly due to the increased path length. The amount of polymer exiting the column as a function of time is measu ring by an appropriate detector, generally some combination of refractive index, viscosity and/or light scattering 37 Sample (liquid or solid) ATR crystal Infrared beam (from source) Infrared beam ( to detector)
14 Figure 2 .2: In size exclusion chromatography, t he principle underlying gel permeation chromatography, polymers in a solvent are pumped through a column packed with microporous beads. Larger molecules are unable to enter the majority of pores and therefore pass through the column relatively rapidly. Sma ller molecules are able to enter many more pores and must travel a greater distance, resulting in slower passage through the column. This elution profile is measured by detecto rs at the outlet of the column. The application of GPC to SMP systems composed o f linear or branched polymers is readily understandable. Given the variety of these systems that are formed from co polymerization of homo polymers, GPC plays a critical role in characterizing both the homo polymers and the resulting co polymer. That said, many SMP applications use cross linked systems and the utility of GPC may not be as apparent. GPC is still an important tool in characterizing these polymers, especially when measuring the MW and PDI of the pre cursors used in their synthesis 38 For instance, a recent study using a semi crystalline polycaprolactone ( PCL ) based SMP found that the MW of PCL segments could be used to control the shape memory behavior of the network, in particular the strain recovery rate, due to its dependence on c ross link density 39 Nuclear Magnetic Resonance Spectroscopy Understanding nuclear magnetic resonance (NMR) spectroscopy theory requires a complex discussion with more space than can be allotted here. Since theoretical and mechanistic discussions are not the focus of this chapter, readers are instead directed to a number of other works with thorough discussions in this regard 40 42 We provide brief details be low. To detector(s) Inject polymer in solvent with steady flow rate A polymer with a small hydrodynamic radius is entrapped in many pores resulting in a long path length and time to elute from the column. These smaller molecules will reach the detector last. A polymer with a large hydrodynamic radius is too large to enter most pores resulting in a short path length and time to elute from the column. These larger molecules will reach the detector first. Porous beads Column
15 Given the insolubility of most SMP systems, solid state NMR is generally m ore valuable for NMR characterization than the liquid state alternative. From these studies, information regarding chemical composition, network architecture, cross link density, degree of conversion, and, most importantly, shape memory transitions can be garnered at the molecular level. As first reported by Lendlein and colleagues 43 in a covalently cross lin ked system, the shape memory effect is characterized by way of double quantum excitation, which reflects the strength of dipolar coupling in the network. When elongated and fixed in the temporary state, segments between cross linking points are stretched a nd partially aligned resulting in greater dipolar coupling. After recovering the permanent shape, this alignment is lost and dipolar coupling in the network is reduced. The shift in the excitation time associated with maximum double quantum coherence refle cts this change in the network and correlates strongly with the shape memory effect. Shape memory reversibility can also be characterized by monitoring this effect before deformation and again after r ecovery of the permanent shape. Wide Angle X ray Diffrac tion Entirely crystalline polymers are rarely encountered; most polymeric systems display characteristics of both crystalline and amorphous solids. This balance between crystalline and amorphous domains in the polymer architecture underlies the shape memor y effect in several classes of SMPs and so represents an impo rtant property to characterize. When an incident X ray interacts with a material in a wide angle X ray diffraction (WAXD) instrument, the electrons begin to oscillate at the same frequency as the incoming beam. If the atoms have no regular arrangement (ie. in entirely amorphous materials), the oscillations of the electrons will destructively interfere with one another and little to no energy leaves the system. However, if there is a regular arrang ement of atoms (i.e. crystalline domains), constructive interference will result in X ray beams leaving the sample in well defined directions. When studying a polymeric sample, the amorphous regions produce very broad peaks in the diffraction spectrum and crystalline regions produce sharp peaks. The crystallinity in the sample can be
16 calculated by comparing the area under the curves in these regions, resulting in a measure of mass fraction crystallinity. WAXD has been broadly applied to characterization of SMP materials, often times coupled with di fferential scanning calorimetry measurements for a more complete picture of crystallinity. For example, Coughlin et al. used WAXD measurements to characterize the effect of cross linking on polycyclooctene (PCO) an d the resultant impact on shape memory behavior 44 They noted a monotonic decrease in the degree of crystallinity with increased cross linking, explained by the constraining of crystal growth by cross linking points. For the PCO system, they also concluded that the rate and extent of strain recovery both increase with decreasing degree of crystallinity. This mechanistic understanding gives them the ability to tailor these parameters to the given application, improving t he flexibility of their system. WAXD can determine precise crystal structures in SMP samples, aiding understanding of shape memory effects in systems where crystallinity is involved in shape recovery and storage. However, its application is somewhat limi ted by the relative difficulty in translating data to a structural determination. Mechanical and Thermo M echanical Characterization As noted earlier, the utility of SMPs intended for biomedical device applications is most often judged by the mechanical properties they display about thermal transition. Accordingly, this section encompasses some of the most ubiquitous and useful techniques for SMP characterization, including tensile/compression testing, dynamic mechanical analysis, differential scanning ca lorimetry, thermogravi metric analysis, and rheology. Mechanical Tensile and Compression Testing Tension/tensile and compression testing are fundamental to materials science and comprise some of the simplest methods available to characterize the mechanical properties of polymers.
17 Perhaps the most basic of these is the uniaxial tensile test, in which a sample is subjected to a fixed strain rate, along a single axis, until failure. The applied force and extension are measured and recorded simultaneously at regular intervals and the results take the form of a stress strain curve. Important parameters deri ved from this curve include ultimate tensile strength (peak yield (typically expressed as % of original strain), and elongation at failure. Typical material tensile systems operate at a single temperature throughout each test, with control provided by insulated chambers connected to recirculating water baths or furnaces. When testing SMPs, it is critical to take thermal transition temperatures into account and test samples at the temperature( s) seen in the end application. While uniaxial testing is sufficient for isotropic samples, anisotropic samples (e.g. polymer composites in which the discontinuous phase or additive takes the form of fibers), require the us e of biaxial systems. There are three main types of biaxial tensile testing: (1) planar biaxial tests in which two forces of independent origin are introduced along two primary axes; (2) bursting tests in which a disc shaped specimen is clamped along its edge and inflated by a gas or fluid until it bursts; and (3) cylinder tests which require a hollow, cylindrical sample that is subjected to internal p ressure and axial tension. Compression tests generate similar results, but the moduli determined in comp ression may be higher than those found in tension if the material exhibits anisotropy. In general, however, compression testing is used for materials that cannot withstand tensile loading, such as gels and extremely hydrophilic polymers. Additionally, in s ome cases, polymers may fail in a brittle manner under tension, while exhibiting significant ductility under compression. These differences may reflect variations in the applied stress fields during testing, but are more often due to variations in molecula r and small scale responses inherent to the polymer specimen. 45
18 Dynamic Mechanical Analysis Dynamic mechanical analysis (DMA) or dynamic mechanical spectroscopy is a powerful and commonly used tool to study the viscoelastic behavior of polymers. In simple terms, it involves stress an DMA provides a modulus value for each cycle of the sinusoidal stress allowing the investigator to sweep across a range of temperatures and frequencies or shear rates. Temperature and frequency sweeps are the most common operational modes for DMA. It is also important to note that the modulus obtained in DMA may be differe linear region of a classical stress strain curve. DMA provides complex, elastic/storage, and ability to store and return energy, while the loss modulus reveals its propensity for viscous energy loss. Along with moduli and viscosity, DMA can measure thermal transitions in polymers, including the glass transition temperature (T g ), which separates the glassy and rubbery regimes of semi crystalline polymers and is critical in the characterization of SMPs. In a typical plot of storage modulus vs. temperature, sub T g transitions, that reflect molecular vibrations on the order of bond bending and stretching and side group mot ion, can also often be seen. They are less prominent than the T g since they are associated with comparatively small changes in modulus. Figure 3 illustrates these transitions and describes some of the important parameters measured by DMA. It also shows ho w DMA data can be used to calculate free strain recovery and fi xity rates and recovery stress.
19 Figure 2.3: (A) The amplitude of the dynamic mechanical analysis ( DMA ) signal and phase angle properties. Phase lag between stres (B) Prototypical DMA temperature scan of a polymer, showing decrease in storage modulus as temperature increases due to increased molecular motion (bond bending and stretching, side chain motion etc. ) and free volume. DMA instrumentation consists of sample clamps, which secure the test sample, a linear drive motor, which provides load for the applied force, a displacement sensor, and a temperature controlled chamber (furnace and liquid nitrogen line). Sample and clamp geometry are highly dependent on the desired testing mode. Differential Scanning Calorimetry The changes in physical and chemical properties underlying temperature or time dependent polymer transitions are accompanied by changes in enthalpy. As its name suggests, differential scanning calorimetry (DSC) can measure these changes in heat flow in an experimental sample in reference to a known sample. The observed thermal transitions are very useful for assessing A B
20 uniformity between batches of the same material or determining the thermal properties of new materials. DSC can measure glass transitions, m elting and boiling points, crystallization time and temperature, percent crystallinity, heats of fusion and other reactions, reaction kinetics, specific heat, oxidative/thermal stability, rate and degree of polymerization, and sample purity. That said, DSC is not ideally suited to assess all of these properties and other techniques may be superior. For instance, observed thermal transitions in an unknown polymer can be compared to those of common polymers in published reference databases, but they do not un iquely identify composition. FTIR is far more effective for determining polymer composition. Similarly, melting points, which reflect molecular weight and thermal processing history, are sometimes available from standard compilations and DSC can reveal pol ymer degradation if the measured melting point is lower than expected. However, structural/chemical decomposition is best measured using thermogravimetric analysis DSC is comparatively useful for measuring melting temperature (T m ) and T g but sub T g therm al transitions are usually too weak or broad to be detected by DSC and require DMA. A typical DSC curve of heat flow as a function of temperature is shown in figure 4. More recently, investigators have used DSC to measure thermal conductivity in polymers 46 and to predict drug solubility in polymers with the fox equation. 47 Figure 2.4: Typical differential scanning calorimetry curve of cross linked polymer showing thermal transiti ons. There are several variants of DSC, but differential thermal analysis, in which temperature and temperature differences are measured in association with time and temperature dependent material transitions, is the most common. In differential photocalorimetry, the sample is exposed Temperature Exothermic heat flow Glass transition Crystallization Melting Cross linking Decomposition/ oxidation
21 to UV Vis radiation. When running a DSC experiment, investigators must optimize both sample and instrument parameters. Polymer samples should be as thin as possible and cover as much of the plate as possible. Just as with DMA, careful control of heating and cooling rates is critical. Figure 5 displays a simpl ified DSC instrument schematic. Figure 2.5: Differential scanning calorimeter instrument schematic. Alumel wire measures sample temperature Thermogravimetric Analysis Thermogravimetric analysis ( TGA ) measures the rate and amount of mass change in a material as a function of temperature or time in a controlled atmosphere. It can determine selected characteristics of materials that exhibit mass loss or gain as a result of decomposition, oxidation, dehyd ration, or loss of other volatile components. These characteristics include composition (of multi component polymer systems), thermal and oxidative stability, decomposition kinetics, and moisture or volatile component content. TGA instruments rely on preci se measurements of three variables: temperature, temperature change, and mass change. Accordingly, they consist of a precise balance and a programmable furnace that allows for either a constant heating rate (more common) or variable heating, in order Gas vent Gas inlet Heat flux plate Furnace Sample plate Reference plate Alumel wire Chromel wire Chromel disc
22 to ac hieve a constant mass loss rate (useful for studying specific reaction kinetics). The experimental material sample is placed on a balance pan inside the furnace. If a reference sample is used, it must be placed on a separate balance in a difference chamber The furnace chamber also allows the atmosphere surrounding the sample to be purged with an inert gas, such as nitrogen or helium, to prevent oxidation. As the sample is heated, its components decompose and the weight percentage corresponding to each resu lting mass change can be measured. Some modern TGA instruments, which use quartz crystal microbalances, can measure mass changes on the order of micrograms (versus milligrams for conventional TGA). For structure determination, thermogravimetric analyzers c an be coupled to an FTIR or mass spectrometer, and temperatures of up to 2000C may be used for gas phase analysis. Rheology The field of rheology is concerned with the viscoelastic behavior of matter, particularly its tendency to flow and respond plastica lly to an applied force, rather than elastically. Viscoelasticity can be related directly to the configuration of individual polymer molecules, network junctions, and entanglements, all of which can change in response to an applied force. Local rearrangeme nts occur much more rapidly than larger segmental motions, so the stress response is relatively long and continuous. This effect is lessened in low molecular weight polymers, which display near Newtonian behavior (viscosity independent of applied shear str ess), due to a lack of significant entanglements and the absence of chain stretching, rotation, or other relaxation processes. High molecular weight polymers exhibit shear thinning in which viscosity decreases a s a function of the shear rate. Rheometers a re most often used to study liquids, but soft solids, such as polymers, are also routinely studied. As such, they can be used to monitor polymerization, like FTIR, and gel formation by measuring associated changes in viscosity. More generally, rheology ena bles the translation of measured force, pressure, torque, and angular velocity, into stress and strain. These can, in turn, be used to calculate other rheological properties. A typical set up establishes
23 contact between the sample and two surfaces ( slip su rfaces ), at least one of which is attached to a drive motor. Characterization of both steady and oscillatory shear properties can be made. Several fixture geometries have been developed to accommodate different sample geometries and viscosities. The four m ost common ge ometries are shown in figure 2. 6. Figure 2.6: Sample geometry and viscosity govern the selection of rheology fi xtures. Concentric cylinders are best for very low viscosity systems, including polymer suspensions in solvents or colloidal mixtures that are prone to settling. Note that for systems with extremely low viscosities, the contact area should be maximized so that an adequate force is seen at the transducer. The cone and plate is a particularly versatile set up for characterization in the non linear viscoelastic region and is the most common fixture used in rheology. It has the benefits of constant shear rate along the flow axis and low sample volume, but is quite sensitive to plate separation and cone angle (typically ~ 1). It can be used with suspensions or emulsions, but the suspended particles must be no larger than 1/10 th the truncation gap. In addition, care should be taken to avoid shear heating. Parallel plate fixtures are simple to use with respect to sample preparation and loading. They are appropriate for samples ranging from low viscosity liquids to soft solids, but produce non uniform velocity fields requiring complex integration over the sample Concentric cylinders Cone and plate Parallel plate Torsion rectangular Low to medium viscosity liquids Low to high viscosity liquids Low viscosity liquids to soft solids Soft to rigid solids Polymer
24 diameter. For both the cone and plate and parallel plate fixtures, instabilities in the shear field can cause ejections of the fluid at high rates. The last fixture shown in figure 2.6 is adequate for solid samples, but most instruments are inaccurate at high strains. Surface Characterization The success or failure of a polymers employed in biomedical device applications is often determined by their surface properties. It is well known that protein surface fouling and cell can be defined qualitatively as the zone where the structure and composition differ from that of the bulk materi al, as influenced by the interface between that material and the surrounding medium (i.e. air, water, blood, etc.). In the case of polymers, this zone can be significantly deeper than other materials, extending anywhere from 10 to 100 nm from the interface Because of the heightened reactivity of this interfacial zone and its critical role in determining the response of biological systems to the material, its chemical makeup and structure must be thoroughly understood. Heightened reactivity of this interfac e zone and its critical role in d etermining anywhere Atomic Force Microscopy relevant to biological systems. Atomic force microscopy (AFM) accomplishes this task by providing a map of surface topography with nanoscale resolution. In fact, modern AFM instruments often have sub nanometer resolution. They essentially consist of a sharp tip/probe mounted to a cantilever. When the trip is brought in close proximity with a surface, forces of interaction between the tip and surface result in a deflection of the cantilever. This deflection is registered through changes in voltage (when piezoelectric elements are incorporated into the cantilever), laser beam deflection, or optical interferometry. Interaction forces include the mechanical contact force, van der Waals forces, capillary forces, electrostatic forces, chemical bonding, and magnetic forces, among others.
25 AFM has advanced greatly since its commercial introducti on in the late 1980s, resulting in the application of scanning probe microscopy to provide an array of surface information including surface potential, thermal conductivity, chemical interaction strength and biorecognition of specific biomolecules. Togethe r, these variants can provide an impressively complete collection of the to infer morphology and molecular structure of different regions in the network, includi ng the existence and evolution of crystallites 48 However, AFM does have several disadvantages, many of which are specific to soft materials like SMPs. For instance, the area and depth of an AFM scan is limited, meaning it is unlikely that the entire surface of a sample could be imaged. Inste ad, a small section must be selected to represent the entire surface. In addition, soft materials can be influenced by the small forces applied by the cantilever tip, causing sampling artifacts. However, such concerns have been minimized by the recent appl ication of real time correction algorithms suc h as feature oriented scanning. Contact Angle Measurements A drop of water sitting on a surface can be a surprisingly powerful tool in characterizing interfacial energies. Determining the angle formed between t he surface and droplet is a balance between the cohesive force attracting the water molecules to each other and the adhesive force, which is driven by attraction between water molecules and the molecules comprising the surface of the material. For instance a polar, hydrophilic surface will form a greater angle of contact with a water droplet than a non polar, hydrophobic surface (figure 2. 7). Surface free energy, which is correlated with contact angle, has been shown to be a very useful and easily obtained parameter showed that adhesion of human umbilical vein endothelial cells and HeLA cells onto self assembled monolayers of alkanethiols was strongly dependent on the wettability of the surface, as determined by contact angle measurements 49 Surface free energy can be partitioned into individual components, each of which represents a contribution from a specific type of interfacial interaction. The primary in teraction forces include London dispersion forces, polar forces,
26 hydrogen bonding, induction, and acid base interactions. There are several methods to calculate surface free energy and its components, depending on the material being studied; readers are re ferred to excellent reviews on the subject, such as the work by Zenkiewicz. 50 Note that liquids other than water are often used when estimating non polar contributions to surface energy. A dditionally, readers should keep in mind that contact angle measurements are notoriously sensitive, with slight differences in sample preparation capable of significantly altering experimental results. Concerns include surface roughness, surface and liquid purity, and absorption of the liquid into the polymer. The sessile drop experiment is the most sensitive to these factors, but more sophisticated instruments exist that instead measure advancing and receding contact angles or perform capillary based measu rements, which impr ove accuracy and repeatability. Figure 2.7: Contact angle is dependent on the forces of interaction among molecules of the liquid droplet and between those of the liquid droplet, solid sample surface, and air. It reflects the contribution of polar, dispersive, and other forces to the total surface free energy and can predict proteins, and small molecules. X R ay Photoelectron Spectroscopy Based on the photoelectric effect, X ray photoelectron spectroscopy (XPS) involves focusing X rays onto a sample and subsequen tly detecting the kinetic energy and number of electrons that are ejected. Interaction of X ray photons with atoms at the surface of the material cause inner shell electrons to be liberated. Since the energies of the incident X ray photons are known, the b inding energy of the liberated electrons can be calculated based on their kinetic energy. Binding energies simply represent the difference in energy between the neutral and ionized atoms. They are characteristic of the atoms from which the electrons were l iberated and their environment, 1 2 Contact angles 1 2 Polymer sample Water droplet Increasing surface hydrophobicity
27 allowing determination of the atomic percentages of the chemical elements present at the surface of the material and their chemical and electronic states. Since only electrons near the surface of the material are ejected wit h enough kinetic energy to escape the sample, XPS limits the information returned about the sample to the region within 1 12 nm of the surface. Compared with many other surface chara cterization techniques (e.g. scanning electron microscopy ), the lack of sa mple preparation and non destructive nature of the analysis are major advantages of XPS. Due to the surface localization and sensitivity of XPS, its application to SMP characterization can provide valuable information about surface composition, contaminati on and oxidation state, as well as characterization of surface modifications. For example, after immobilization of a biomolecule on the surface of a SMP device, XPS could be used to a) confirm covalent linking of the molecule to the SMP surface; b) provide a stoichiometric analysis of biomolecule content; and c) permit an estimation of the thickness of the surface brush. XPS likely provides the most expansive set of information of any surface characterization technique relevant to SMPs; however, the rarity and high cost of XPS instruments and the reliance on experienc ed operators limit its utility. If more detailed surface characterization is required than can be provided by XPS, time of flight secondary ion mass spectrometry (TOF SIMS) analysis may be used. This technique can provide not only the mass spectra of the top 1 2 atomic layers of the material, but also detailed surface maps and depth profiles. While TOF SIMS is not as common as XPS in organic applications, its use in SMP applications may increase as composite materials, incorporating inorganic fillers, are more heavily investigated. Imaging B ased Characterization Electron M icroscopy Transmission and scanning electron microscopy (TEM and SEM) have long been used for structural characterization of polymers. In lieu of light and optical glass lenses, electron
28 microscopes employ electrons (having a much lower wavelength) and electromagnet ic lenses, allowing them to resolve structures well under 1 nm. When deciding whether to use TEM or SEM, one must consider material processing restrictions as well as the desired resolution. Table 2. 6 summarizes these considerations. With either modality, the user must balance image quality, contrast, and radiation dama ge, as with all soft materials. Table 2. 6 Comparison of transmission a nd scanning electron microscopy for the analysis of shape memory polymers. TEM SEM Sample preparation Ultrathin section Thick, conductive Peak resolution ~ 0.5 ~ 0.5 nm Beam intensity/voltage High (~ 100 keV) Low medium Final image Interior (2D) Surface or interior via fracture (3D) Transmission Elec tron Microscopy Like biological specimens, synthetic polymers consist primarily of low atomic number elements, whose elastic interactions with high energy electrons are fairly weak. There is almost no absorption of electrons directed at samples thin enough for TEM. Generating contrast in polymer TEM can thus be difficult. However, inelastic interactions can be strong. This allows for powerful spectroscopic techniques, but can also lead to radiation damage, a constraint that affects polymers far more than inorganic samples 51 Radiation damage is a particular concern with TEM, as it em ploys high voltages to accelerate the electrons (40 400 keV). Even so, several well established techniques can be used to image polymers, including bright and dark field imaging, high resolution imaging (to investigate molecular level structures in crys talline polymer solids), electron diffract ion, and analytical microscopy. SMPs inherently contain spatial variations in density, crystallinity, and crystal orientation, and it is often important to characterize the distribution and size of crystalline and amorphous regions.
29 Crystallinity generates Bragg diffraction. In dark field imaging, the aperture of the objective lens is set to block the primary beam and allow scattered electrons to pass. This produces an image containing bright crystallites and dark a morphous surroundings. Amorphous regions do not exhibit appreciable spatial variations in density, leading to forward scattering. However, they can still produce variations in the energy or phase of incident electrons, providing additional sources of contr ast (electron energy loss spectroscopy and phase contrast, respectively). Figure 2.8 illustrates the difference between amplitude and phase contrast. Composite SMPs containing heavy elements are straightforward to image using standard TEM. They produce significant Rutherford scattering, in which electrons are scattered at high angles and can be blocked by an objective aperture to generate image contrast, with the heavy discontinuous phase appearing dark. Modern and comprehensive works have been published covering contrast mechanisms, resolution, and specimen preparation in polymer TEM 48,52 Special attention should be paid to positive and negative staining met hods; for instance, compounds containing double bonds can be selectively stained with osmium tetroxide (OsO 4 ). It is often used to stain one phase of block copolymers, revealing their microstructure.
30 Figure 2. 8 : ( A) Amplitude contrast in polymer TEM arises when an objective aperture is used to selectively block electrons scattered at high angles from crystalline segments. ( B) Variation in the phase or energy of electrons departing the sample can provide contrast within largely amorphous polymers having only minor spatial variations in density. Sc anning Electron Microscopy SEM is generally used to examine the surface topography of polymers, though internal structures can be imaged by introducing fractures. The image resolution of SEM is typically an order of magnitude poorer than TEM, but it has excellent depth of field, allowing it to produce accurate representations of a polym dimensional shape. In addition, since SEM does not rely on transmission, samples up to several cen timeters in size can be imaged. Image contrast in SEM can be generated through several mechanisms. Surface tilt contrast arises due to the angle be tween the incident electron beam and the sample surface, which makes the surface features normal to the beam appear darkest. Shadow and diffusion contrast causes edges and protrusions on the surface to appear bright. Material contrast results from spatial arily on the constituent atoms. Polymer sample e e A B Objective lens Aperture Image Amplitude contrast Phase contrast
31 Sample preparation for SEM is generally much simpler than for TEM. The sample is simply placed in the specimen holder with an electroconductive adhesive. However, with polymers and other insulating specimens, care must be taken to prevent surface charging from the incident electron beam as this can create severe imaging artifacts. Ultra thin heavy metal coatings such as gold, platinum, or their palladium al loys are often applied to prevent surface charging, though it is sometimes adequate to simply r educe the accelerating voltage. Phantom B ased Medical I maging When characterizing SMPs for biomedical device applications, it is often necessary to assess signal and artifacts generated from radiography and magnetic resonance based imaging modalities. These modalities, which include digit al X ray, fluoroscopy, CT MRI are often used to confirm proper device placement in the body. Initial assessment is most often p erformed in artificial imaging phantoms which are designed to simulate the imaging characteristics of the target organ or delivery site (e.g. vascular compartment, brain parenchyma etc.). Due to the abundance of water in the body, the simplest phantoms of ten consist of an acrylic polymer (or equivalent) container filled with water or saline, in which the specimen of interest is suspended. However, more purposeful phantoms can be constructed, as discussed below. General considerations when designing imaging p hantoms are shown in figure 2.9
32 Figure 2. 9 : Design considerations for imaging phantoms intended to assess SMP biomedical devices. Digital X R ay, Computed Tomography, and Fluoroscopy Polymers are made primarily of low atomic number elements, including carbon, hydrogen, oxygen and nitrogen, which are poor attenuators of photons. Developing radio opaque SMPs typically requires the addition of heavy element fillers. 15 16 For instance, barium sulfate, zirconium oxide and tantalum have been used in the orthopedic field for bone cement, 17 18 19 tantalum filled SMPs have been evaluated as embolic coil materials, 11 and iodinated mono mers have been incorporated into denture base resins. 20 Methods exist too for a priori design of radiographic phantoms having specific elemental compositions and physical densities 53 but it is most often sufficient to use normal saline (0.90% w/v NaCl) as the scattering medium. Magnetic Resonance Imaging The goal of a MRI phantom is to surround the specimen of interest in an artificial environment that mirrors the T1 and T2 relaxation properties and conductivity of tissue surrounding the putative device delivery site. Since MRI signal arises primarily from the relaxation of oriented protons found in water molecules, a simple MRI phantom can be designed u sing water. However, a X ray modalities Both MRI Simple vs. biomimetic form Minimize exogenous material near sample Liquid vs. solid phantom Optical clarity for direct visualization Avoid materials known to generate artifacts Mimic X ray attenuation/scattering properties of tissue surrounding delivery site Mimic T1 and T2 relaxation properties of tissue surrounding delivery site If using water, T1 relaxation time must be lowered (e.g. with gadolinium) Avoid materials suscep tible to induction
33 paramagnetic contrast agent such as gadolinium must be added (at approximately 5 10 mmol) to reduce the T1 relaxation time to useable levels. It is often useful to study the geometry and properties of accreditation phantoms used to v alidate clinical scanners. For instance, the American College of Radiology accreditation phantom uses a solution of 10 mmol nickel chloride and 45 mmol sodium chloride to simulate biological conductivity and includes a contrast vial containing 20 mmol nick el chloride and 15 mmol sodium chloride, allowing for a difference in the T1 and T2 relaxation times. Due to the strong magnetic fields used in MRI, care must be taken not to include ferromagnetic materials or conductive materials susceptible to torque or current generation thro Biological Testing Given the immense variety and complexity of biological responses to foreign bodies, characterizing the response of biological systems to implanted materials can be a daunting task. Because of this, the field has strived to standardize test procedures to en sure repeatability of experiments and facilitate direct comparisons between materials. However, these efforts have only been moderately successful; wide variation still exists. The Sources of Further Information section includes standards for biological te sting, many of which provide detailed experimental protocols. Determining which protocol is appropriate for a given material will depend heavily on the nature of the material and the clinical applic ation for which it is designed. In Vitro T esting Cell or tissue culture based testing can be used to gather a variety of information on how certain biological systems, in a controlled environment, will react to a material. For a typical SMP application, three major categories of testing are used: cytoto xicity testing, functional testing, and hemocompatibility testing. Cytotoxicity testing is one of the most crucial steps in characterizing a SMP for biomedical applications and should be performed as early as realistically possible in the material developm ent process. While it does not provide a complete picture, in vitro testing can still provide valuable insight that can guide subsequent product development.
34 The purpose of in vitro cytotoxicity testing is to determine whether exposure to a material, or su bstances extracted from that material, is toxic at the cellular level. However, because toxicity involves multiple mechanisms, there are various methods by which its effects can be assessed, either quantitatively (cell count or cellular morphology) or qual itatively (metabolic activity, membrane integrity, necrosis, or apoptosis). In addition, there are three methods by which cells can be exposed to the material of interest: direct contact, agar diffusion or extraction. In direct contact tests, cells are gro wn to approximately 80% confluence in a culture dish and, concomitant with a change in the culture medium, a coupon of the sample material is placed in the culture dish. After a period of incubation, the sample is removed and the cells are fixed and histoc hemically stained for microscopy analysis. In the agar diffusion test, cells are grown to near confluence in a culture dish and the medium is replaced with fresh medium containing 2% agar and a vital stain. After the agar solidifies, the sample material is placed on top of the agar gel and the plate is incubated, after which live cells can be identified. In the extraction test, a sample of the material is incubated in medium for a period of time, after which the material is removed and the medium is added t o nearly confluent cells. This test is the most flexible assay that can be used for quantitative assessment of cytotoxicity, allowing evaluation of specific toxicity mechanisms. However, the direct contact and agar diffusion tests may be considered better analogues to the clini cal implementation of a device. In vitro testing can also be used to measure functional responses at the cellular level. These tests take characterization beyond the simple question of toxicity and attempt to determine how cells will functionally respond to implantation of the sample material, whether such a response is desirable or not. This level of testing is especially important for polymers that are bio functionalized, or otherwise modified with the intention of eliciting a specif ic cellular/physiological response. For example, in an application where tissue ingrowth into a SMP network is crucial, functional testing might entail seeding cells with a sample of the SMP, followed by electron microscopy and histological analysis of sam ple cross sections to determine the extent of cellular penetration 54 While in vitro functional testing can provide a rapid and inexpensive means of
35 optimizing polymer parameters, only in vivo testing can give a complete picture of how the material will behave after implantation. Hemocompatibility testing is a critical step for devices that will come into contact with blood. As blood is a complex system, materials can interact with it in a variety of ways, many of which can be deleterious to the clinical application. Several tests are outlined by ISO 10993 4 as first steps in establishing the hemocompatibility of a material. These tests are designed to assess the potential for a material to cause adverse effects upon contact with blood, whether at the site of the device itself or do wnstream of it. Accordingly, the expected mode of interaction between the material and blood must be established first. The aforementioned ISO standard breaks this into three categories: external communicating device blood path indirect (a device that con tacts the blood path at one point as a conduit into the vascular system); external communicating device circulating blood (a device that contacts circulating blood); and implant device (a device principally contacting blood). Based on the category, severa l Level 1 (primary) and Level 2 (optional) tests are outlined to evaluate potential adverse interactions. Primarily, these tests assess responses related to thrombosis, coagulation, platelets, hematology, and immunology. While some of these assessments can be performed by in vitro tests (e.g. immunological testing), most require in vivo or ex vivo work. Based on the results of the Level 1 tests, Level 2 testing may be required if specific interactions need additional investigation. In Vivo T esting While use ful for modelling biological systems, in vitro testing cannot begin to duplicate all the complexities of a living system. As such, in vivo testing is a necessary step in the development of any biomaterial. In vivo assessments are meant to ensure a) that a given material is not harmful to the patient and b) that it perform its intended role. As with in vitro studies, several organizations have published standardized protocols to be followed when performing in vivo experimentation. The reader is directed to t he Sources of Further Information section for these references.
36 The appropriate in vivo testing protocol to be implemented will depend on the specific application of the SMP device, including the duration and mode of contact. Contact duration is specified as the modes of contact are categorized as surface device, external communicating device, or implant device. While a multitude of testing regimens exist depending on the combination of duration and mode of contact, three tests are generally recommended by ISO 10993 for all medical devices and will be briefly described here: cytotoxicity (as described in the In Vitro Testing section), sensitization, and irritation. S ensitization studies highlight immunologic mechanisms influencing host reactions to an implanted device, with an emphasis on local over systemic responses. Sensitization studies may be performed using individual chemicals used to synthesize the test materi al (to isolate specific responses), the entire material itself, or extracts from the material. Typical test methods include the Guinea Pig Maximization Test, the Closed Patch Test and the Murine Lymph Node Assay. Irritation studies estimate the potential f or an implanted material to cause dermal irritation, generally by applying the material or its extracts to a dermal, mucosal or ocular surface of young adult albino rabbits. For devices that will be externally communicating or have internal contact, intrad ermal injections of material extracts are recommended. In a typical SMP application, where extracts will likely contain hydrophilic and lipophilic components, it is important to include testing with solvents for both classes of chemicals. For devices that will have only external communication with intact or abraded skin, the test material or its extractions should be applied directly to both skin types. In both of these ar formation, edema, or other observed reactions. Repeated exposure studies may also be necessary. Example Applications In this section, we will approach three distinct SMP biomedical device applications from the perspective of an investigator. Each case study will lead the reader through the process of
37 identifying important application specific material properties and the techniques useful in characterizing them. Case Study 1: Shape Memory Polymer Vascular S tent Vascular stents are mesh tubes used to keep blood vessel open following the repair of a stenosis secondary to a pathological biological process. For instance, ather osclerotic lesions in the coronary arteries of the heart, which restrict blood flow to the myocardium, can be treated by balloon dilation (angioplasty) of the constricted region followed by placement of a stent to maintain patency in the long term. Stents are typically made of metals, with or without a polymer coating designed for elution of anti fibrotic drugs and prevention of encapsulation. However, fully polymeric stents have gained United States Food and Drug Administration ( FDA ) approval and SMPs appe ar to be well suited for percutaneous vascular devices as they can recover a complex 3D conformation after prolonged storage i n a catheter under deformation. Comparison of a group of candidate SMPs for vascular stent applications would require extensive testing. First, the mechanical and thermo mechanical properties would have to be assessed. Simple uniaxial tensile testing in a temperature controlled chamber would determine whether the SMP stent could withstand typical peak stresses seen during device de ployment or retraction. It would also be important to determine how the SMP stent transitions from room to body temperature. While DSC could accurately determine thermal transition temperatures, DMA may be more appropriate, since can also provide the modul us and recovery stress over that same range. Recovery stress may be particularly important as this would help to determine whether the deployed device exerts sufficient force on the arterial wall to maintain patency following angioplasty. Compositional and structural characterization is a ubiquitous requirement in polymeric biomaterial development and includes analysis of molecular weight and weight distribution, the degree of branching and crystallinity, and component concentrations. However, it has unique importance when designing materials intended for delivery to the vascular compartment. FDA guidelines
38 require materials in contact with blood to undergo testing for cytotoxicity ( in vitro ), hemolysis, complement activation, cell adhesion, protein adsorpti on, thrombogenicity (whole blood clotting time, platelet and fibrinogen turnover etc.) and pyrogenicity, among others. Particular attention should be paid to chemical composition and leaching, which reflects the monomeric composition, impurities introduced by the fabrication process, and the degree of conversion. Variables such as protein and cell adsorption can be tested directly using human serum proteins or whole blood. However, candidate polymers can be pre screened for their tendency to interact with p roteins or cells using contact angle measurements. It is well known that hydrophilic/ionic surfaces tend to increase cell adhesion, while protein fouling can be driven by hydrophobic interactions between the substrate and prot ein core or ionic interactions Case Study 2: Growth Factor Modified Neurovascular Shape Memory Polymer D evice A SMP device has been designed for a neurovascular application, in which the device is meant to promote neuronal regeneration at the site of implantation. To promote regenera tion, the device has been surface modifie d with neuronal growth factors. As a first step, the raw material itself should be tested for in vitro biocompatibility, followed by testing of the material after surface modification. To confirm immobilization of t he growth factor on state NMR, AFM or XPS could be used, with NMR or AFM providing a quicker analysis, but XPS providing better accuracy 55 Functional in vitro studies should follow to demonstrate the potential of the device to promote neuronal outgrowth and the added benefit of the immobilized growth factors. These studies should also identify the optimal concentration of tethered growth factor for maximized re sponse. For example, primary hippocampal neurons isolated from rats could be cultured on surfaces of the SMP material itself, modified SMP, and poly L lysine coated surfaces (positive control) 56 After a period of incubation, cells would undergo morphological assessment in addition to neurofilament antibody staining to qualitat ively assess neurite outgrowth.
39 Testing of shape memory and mechanical properties would follow tradit ional characterization methods, including DMA or DSC for thermo mechanical testing and solid state NMR and FTIR for chemical characterization. In this case, chemical characterization might focus on ensuring the presence of functional groups amenable to imm obilization of the growth factor on the surface of the material. Case Study 3: Subcutaneous Shape Memory Polymer Drug Delivery D evice In this case study, a SMP device has been developed with the intention of being administered subcutaneously, in a minimall y invasive fashion. Once implanted, the device will deploy to its permanent shape, opening up macro will be loaded with enough of the drug to sustain several months of delivery at therapeutic leve ls. Besides universal characterization steps, including thermo mechanical testing and chemical characterization, biocompatibility testing and examinations of the kinetics of drug release and incorporation will be pivotal for this device. Cytotoxicity testi ng should begin with in vitro exposure of a relevant cell line (e.g. L929, fibroblasts isolated from the subcutaneous tissue of mice) to extractions from the material, as outlined in ISO 10993, followed by an MTT ( 3 ( 4 5 dimethylthiazol 2yl ) 2,5 diphenyltetrazolium bromide ) assay to determine whether metabolic activity is affected. Finding no cytotoxic effects, biocompatibility testing should move to in vivo studies for sensitization and irritation with the protocol extending at least eight we eks. This eight week timeframe will allow for evaluation of a) the onset and intensity of acute inflammation, b) the time required for acute inflammation to subside and c) any indication of chronic inflammation, which generally take s about eight weeks to p resent. The combination device/drug side of the system will also need careful examination to ensure the mechanisms and kinetics of drug release are thoroughly understood. Considerations include drug stability, kinetics of drug release from the system, unif ormity of drug distribution, and the physical state of the drug within the network (crystalline or solid dispersion). The first two considerations are best assessed by in vitro release testing, in which the device is placed in a release medium
40 (e.g. phosph ate buffered saline pH 7.4) at physiologically relevant conditions and samples of the release medium are taken periodically. HPLC, MS, or UV Vis spectroscopy can be used to determine the amount of drug released at a given time and the fraction of that dru g that is pharmaceutically active. Drug stability is especially critical for delivery of large molecule drugs such as proteins or peptides; however, more sensitive methods such as enzyme linked immunosorbent assay s are generally needed to make these assess ments. Drug distribution and physical state can be assessed by XPS or AFM if it can be assumed that conditions at the surface are indicative of the entire network. Otherwise, more advanced techniques such as confocal Raman spectroscopy may be employed 57,58 Future Trends and Conclusions Shape memory behavior in polymers intended for biomedical applications has been studied extensively since the 1970s. However, advances in existing characterization techniques and development of new methods will continue into the foreseeable future. Adoptio n of new methods has been driven in part by the emergence of functionalized SMPs, including nanocomposites and surface modified SMPs coated with bio antigens/receptors. With these new materials, the study of interfaces (between the continuous and discontin uous phases of a composite or immobilized surface biomolecules and foreign surfaces) takes center stage. Accordingly, surface analysis techniques like XPS, which determines the identity and oxidation state of elements within 10 nm are being seen more commonly in the literature. Another interesting trend is the combination of existing analytical instruments. For instance, simultaneous thermal analysis (STA) is the application of DSC and TGA to a sample in a single instrument. This is possible because the environmental conditions of the test (atmospheric composition, gas flow rate, heating rate etc.) are identical for both signals. The results are often enhanced further by coupling the STA machine to an FTIR instrument or mass spectrom eter. Advances in imaging techniques, such as dynamic in situ TEM (DTEM), may also provide important data on the micro structural changes in SMPs during thermal transitions. DTEM allows for real time atomic level observations of changes in materials in res ponse to external stimuli, such as heat, stress, light, or
41 processes occurring durin g SMP device synthesis and use. Computational modeling and simulation may also play a key role in the future of SMP characterization, particularly in initial feasibility and proof of concept studies. Commercial software packages such as SolidWorks and ANSYS are routinely used to model the deformation and internal stresses exp ected in biomaterials, once deployed in the body, and could one day be used to model more complex behavior. For instance, an investigator could conceivably simulate the trans catheter delivery of a SMP stent into a coronary artery and predict the likelihoo d of stent displacement, vessel rupture, or the time to complete encapsulation. Such simulations would require information about the device geometry, mechanical properties of the SMP and vessel wall, arterial blood flow, and even protein/cell adsorption ki netics. Clearly, computational simulations are also well suited for evaluation of drug release from SMPs and may provide a rapid means of testing the effects of polymer composition, molecular weight, and other vari ables on drug release kinetics. This chapter provided an overview of characterization methods relevant to the development and implementation of SMP biomaterials, with an emphasis on human medical device applications. The task of assembling a comprehensive material description is now more auto mated, faster, and more accurate than it has ever been. Even so, the investigator must always bear in mind that the identification of important material properties and the methods best suited to evaluate them are highly application specific and that SMP va lidation should be a constant process of feedback and refineme nt, rather than a single event. Source of Further Information ISO 10993: Biological Evaluation of Medical Devices Test Methods for Biological Safety Evaluation of Medical Devices, Assessment of Medical Device, Notice 36 (Pharmaceutical and Food Safety Bureau, Ministry of Health, Labour and Welfare, Japan)
42 USP <87>: Biological Reactivity Tests In Vivo USP <88>: Biological Reactivity Tests In Vitro ASTM F748 06: Standard Practice for Selecting Gen eric Biological Test Methods for Materials and Devices ASTM D3418/ISO 11357: Transition Temperatures and Enthalpies of Fusion and Crystallization of Polymers by Differential Scanning Calorimetry ASTM D4065: Dynamic Mechanical Properties ASTM D638/ISO 527: Tensile Properties of Plastics ASTM D695/ISO 604: Compressive Properties of Rigid Plastics Menard, K.P., 1999. Dynamic Mechanical Analysis: A Practical Introduction, Boca Raton, FL: CRC Press.
43 CHAPTER III IODINE FUNCTIONALIZED SHAPE MEMORY POLYMERS FOR EMBOLIC COILS A POINT OF CO MPARISON Introduction The goal of this chapter is to familiarize the reader with a shape memory polymer chemistry we have studied that h as shown significant success with regard to radiopacity and preservation of de sirable thermo mechanical properties. It takes a different approach than the SMP GNP composite and involves covalently incorporating iodinated chemical groups into the backbone of the SMP. It will act as a point of comparison for the gold nanocomposite des cribed later. Metallic coils have been used for endovascular arterial embolization since 1975, 59 though their use in the treatment of intracranial aneurysms has only been reported starting in the late 1980s. 60 63 Initial designs evolved with the development of helical, platinum based Guglielmi detachable coils (GDCs) in 1990 64 and subsequent FDA approval in 1995. The GDC spawned a generation of coils with varying softness, caliber, 3D form, and controlled detachment mechanisms. Since then, manufacturers hav e introduced supplementary tools such as balloons and stents to aid in intracranial deployment 65 68 and clinical studies have helped to widen the indication for endovascular embolization, to include treatment of arteriovenous malformations, gastrointestinal bleeding, primary post partum hemorrhage, malignant hypertension secondary to renal fai lure, and lesions in the kidneys, liver and uterus. Moreover, polymeric coatings have been developed with the goal of enhancing endothelial tissue reaction and accelerating thrombus and fibrocellular tissue organization, or improving volumetric filling of vascular defects. Respective co co acrylic acid) hydrogel. However, these de signs have demonstrated limited to no clinical benefit over bare platinum coils, with regard to common complications and long term clinical outcomes. 69 74
44 It is fair to say that the fundamental design of coils and the materials they employ have not changed significantly since the introduction of the GDC. Modern coils are still primarily fabricated using platinum, platinum iridium alloys, stainless steel, tungsten, or titanium alloys. 75 This reflects a growing body of clinical data surrounding their use and the fact that metals are generally durable and radiopaque. However, their mechanical properties, such as flexural stiffness and compressibility are only tunable within a narrow range 76,77 and their surface properties, which determine their te ndency to corrode, incite an inflammatory response, or form a stable thrombus, are relatively static without deposition of exogenous materials. 78 80 Additionally, metals are unable to elute drugs without a polymer coating or biodegrade once their intende d function is complete. They also distort CT and magnetic resonance imaging MRI scans through significant artifact generation, preventing accurate visualization of proximal anatomy. 81,82 In certain cases, the presence of ferromagnetic metal TCDs is a contraindication to MRI scans, particularly in the high magnetic fields generated by modern inst ruments. 83 Today, artifacts and contraindications are accepted as inevitable constraints of working with metal devices. Finally, metals are very expensive to manufacture, t ransferring higher costs to patients and providers. In addition to high raw material costs, metal coils require relatively complex wire forming fabrication techniques. In contrast, polymers rely on mature and flexible fabrication techniques, such as moldin g and extrusion, which allow for highly customized device geometries with significantly reduced material and processing costs. 84 Given the above issues, the prospect of fully polymeric embolic devices has sparked much interest. These devices offer a far more attractive palette of features, including decreased or absent MRI and CT imaging artifacts, and the ability to tune stiffness, surface interactions with blood components, biodegradation, and drug elution with relative ease. Among polymers, shape memory polymers (SMPs) have high ly desirable properties for catheter based storage and release. 3 SMPs are stimuli responsive mat erials; they can recover almost any pre determined shape after being heated above a tunable glass transition temperature (T g ). 8,85 In fact, repeated strain recoveries of several hundred percent are possible over multiple cycles. 9 This feature allows embolic SMP devices to recover a helical coil or other 3D conformation upon release at
45 body temperature, after being stored i n a catheter in a compacted or elongated form. The thermally induced shape memory effect may also allow coils to self pack within aneurysms and resist permanent deformation associated with long term exposure to high blood pressures. Moreover, at temperatur es above the T g SMPs undergo a reduction in stiffness of several orders of magnitude, resulting in elastic moduli below 30 MPa. 8,86 In comparison, platinum and stainless steel have elastic moduli of roughly 168 and 180 GPa, respectively. 76 It is expected that lower stiffness materials will allow for higher coil packing densities. Table 3.1 summarizes the potential advantages of SMPs over metals in the design of embo lic coils. Tab le 3.1: Potential advantages of SMPs, when compared with metals, for the design of embolic coils SMP properties Practical coil features Clinical correlate s Highly tunable modulus Softer coils at physiologic temperature Lower vessel/aneurysm rupture rate during procedure and reduced Higher packing density Diamagnetic Reduced or absent MRI artifacts Subsequent MR imaging of nearby vascular or parenchymal structures is possible Allows for MR guided intervention Tunable bulk stability Controlled biodegradation Elution of pro thrombotic and/or neointimal proliferative factors Rapid tissue response and permanent aneurysm occlusion Thermally induced shape recovery Self packing ability Pre programmed 3D shape Higher packing density in lesion Enhanced immediate physical occlusion and hemostasis Resistance to permanent elastic deformation Reduced long term coil compaction lower re bleeding rates Fewer repeat procedures Low raw material and processing costs -Low device cost
46 While SMPs have many favorable properties, they also have drawbacks which must be addressed before they can be implemented clinically. The most significant of these is a lack of radiopacity. In any trans catheter device application, accurate placement is c ritical to device performance and safety, and even with the advent of real time/4D MRI, fluoroscopy continues to be the clinical standard for guiding embolic device deployment. Polymers are largely radiolucent, so heavy element fillers are often used to ab sorb and scatter X ray s. For instance, barium sulfate, zirconium oxide and tantalum have been used in the orthopedic field for bone cement 18,19 and tantalum filled SMPs have been evaluated in the research setting as embolic devices. 11 However, the addition of these fillers tends to disturb the origina l, favorable SMP characteristics. To overcome this, we have developed a special interest in SMPs that incorporate monomers tethered with iodine substituted aromatic groups. This method promises to maximize the amount of opacifier that can be incorporated, recovery characteristics and avoiding leach ing or biocompatibility issues. Simply put, the objective of this study was to develop a novel, customizable, and thermally responsive SMP material for the design of next generation embolic coils. Although there are commercially available embolic coils that incorporate polymers in the form of hydrogel coatings or pro thrombotic polyethylene fibers, these still require metal cores for mechanical integrity and radiopacity. To our knowledge, iSMP based vascular plugs are the first fully polymeric devices that have been developed and approved for clinical embolization use. We report here on the results of material characterization, with an emphasis on mechanical properties and imaging characteristics.
47 Materials and Methods Sample P reparation Polymer Coil and Film S amples Shape memory polymer (SMP) samples used in this study were generally prepared by casting a mixture of an acrylic monomer, crosslinking acrylic oligomer, and photoinitiator into glass or polytetrafluoroethylene (PTFE) molds. Specifically, iodinated SMPs (iSMPs) contained 75% (w/w) 2 acryloyloxyethyl 2,3,5 triiodobenzoate monomer (TIA, EndoShape Inc., Boulder, CO) and 25% (w/w) poly(ethylen e glycol) based oligomer. Note that TIA contains iodine substituted benzyl groups and accounts for the radiopacity of iSMP samples. Negative control polymers were synthesized by substituting tert Butyl acrylate (tBA, Sigma Aldrich, St. Louis, MO) for TIA, in the above mixture. Thin film samples of the polymer were prepared in chambers consisting of siliconized glass microscope slides separated by 1 mm rubber gaskets. Thin walled 22 gauge PTFE tubing wrapped around a 12 mm mandrel was used to mold coil sampl es, resulting in coils with a 0.028" wire diameter and a 12 mm diameter curl. Polymerization was achieved through exposure of t he cast mixture to UV light at 10 mW/cm 2 for 30 min. Following polymerization, SMP samples were removed from their molds and incu bated at 90C for 1 hr in order to volatilize unreacted components. Prior to imaging, polymer coil samples were cut into 5.2 or 26 cm wire lengths. Metal C oil an d Film S amples Nitinol coils and stainless steel coils and films were used for comparison to corresponding polymer samples during all imaging experiments. Nitinol coils were prepared using polished 22 gauge wire (Nitinol Devices & Components, Inc., Fremont, CA) wrapped a round a 4 mm diameter steel rod and annealed at 520C for 1 hour in an air oven. Stainless steel coils were acquired from a commercial source (Cook Medical, Bloomington, IN). All metal coil samples were cut to 5.2
48 cm wire lengths prior to imaging. All coil s had a final wire diameter of 0.028" with a 12 mm diameter curl. Stainless steel film samples were c ut from 1.03 mm 316 shim stock. Thermo Mechanical E valuation Thermo mechanical analysis was performed using a TA Instruments Q800 dynamic mechanical analyz er. 5 x 35 x 1 mm samples were prepared from both the radiopaque and control SMP sheets. Samples were loaded between uniaxial thin film clamps for thermal scans, which ran from 0 to 100C at a rate of 3C/min, a frequency of 1 Hz, and a strain of 0.05% in tension. Glassy and rubbery elastic moduli (E g and E r respectively) were reported. Glass transition temperatures (T g ) were assigned at the tan delta curve maxima. Imaging E valuation Imaging Phantom D esign A simple phantom was designed and fabricated to su spend coil and thin film material samples at the center of a 6 inch column of fluid using minimal support material. It consisted of an open acrylic polymer box with three anchor points along each interior wall at mid level. A 0.1 mm diameter nylon monofila ment wire was threaded through each anchor point, producing a grid for sample suspens ion. Image A cquisition Fluoroscopic and multi slice helical CT imaging was performed using a Swissray ddR Trauma System fluoroscope and Siemens Somatom Sensation 40 scann er, respectively. For both of these modalities, the imaging phantom was filled with normal saline (0.9% w/v NaCl in deionized water). Fluoroscopy images were acquired at 62 kVp and 80 mA, using a 2 sec exposure. CT images were acquired at 120 kV and 300 mA using a 3 mm slice thickness. MR images were acquired using a Signa HDxt 3.0 T scanner and the phantom was filled with normal saline
49 containing 5 mM gadolinium chelate contrast (Omniscan, Amersham Health, Princeton, NJ). The paramagnetic contrast was use d to reduce the T1 relaxat ion time to appropriate levels. Image P rocessing All Digital Imaging and Communications in Medicine ( DICOM ) images were viewed and exported to other formats using MicroDicom software. No further image processing was performed prior to analysis. However, CT data was used to reconstruct geometric surface meshes for 5.2 and 26 cm iSMP coils. Briefly, a manual segm ent ation was performed with ITK SNAP software using a single pixel brush tool. Post processing/mesh refinement was performed using MeshLab software and included minimal Gaussian filtering an d removal of spurious elements. Results Thermo Mechanical E valuati on Thin film samples of unmodified and iodinated SMPs were prepared in order to study each both highly temperature dependent, key thermo mechanical properties that are relevant to device design were identified. These included the elastic modulus and T g Elastic modulus is a quantity energy losses that occur during deform ation (contributing to the viscous modulus) are ignored, it rubbery and glassy states. While the qualitative definition of T g can vary, it is best seen as dema rcating a physical transformation in the disordered, amorphous region of polymers. As the temperature of the polymer is lowered below the T g glass transition involves a completely continuous decrease in viscosity. As a result, the T g of SMPs provides information about the physical state a device will take at body temperature. In this study, it was defined quantitatively as the temperature corr Tan( ) is simply
50 the ratio of the viscous modulus to the elastic modulus and so represents the balance between energy loss and storage in a mat erial at any given temperature. The T g and elastic modulus of SMP samp les were measured using a dynamic mechanical analyzer, which uses a furnace and oscillatory drive motor to sweep the temperature of the samples between 0 and 100C, while straining the samples at a fixed frequency. It provides a semi continuous reading of modulus as a function of temperature. Th is data is displayed in figure 3.1 The iodinated SMP demonstrated a T g of 24.2 C, approximately 7.6 C above that of the unmodified SMP. Critically, however, the T g was still well below body temperature, indicating that iSMP coils should be in the rubbery state when deployed in the body. The elastic modulus of the iSMP material was 29.3 MPa in the rubbery state, which represents a decrease in stiffness of over 1000 fold when compared to typical grades of stainless s teel. Figure 3.1: (A) Elastic modulus curves from dynamic mechanical analysis of control SMP and iSMP thin films. Below the glass transition temperature (T g ), SMPs are in the glassy state; above the T g they are in the rubbery state. The control SMP had a glass transition temperature of 16.6 C, a glassy elastic modulus of 168.8 MPa, and a rubbery modulus of 24.2 MPa. The iSMP had a glass transition temperature of 24.2 C, a glassy elastic modulus of 34 7.2 MPa, and a rubbery modulus of 29.3 MPa. (B) Shape recovery of SMP coil after being straightened, cooled to 0C, and then placed in 37C water bath (polymer dyed wi th blue ink to aid visibility). 10 100 1000 0 5 11 17 22 29 35 41 48 54 60 66 72 Elastic Modulus (MPa) Temperature ( C) Control iSMP A B region region
51 Fluoro scopic I maging In order to evaluate the effect of iodine incorporation on SMP radiopacity, short exposure fluoroscopic images of iSMP thin film and coil samples were acquired, alongside unmodified SMP, stainless steel and Nitinol samples of comparable geometry. All Nitinol samples and stainless steel film samples were prepared in the laboratory, while the stainless steel coil samples column of normal saline (0.90% w/v) and the results, without post pro cessing, are shown in figure 3.2 Figure 3.2: (A) Fluoroscopic image of iSMP, control SMP and metal coil/thin film samples. Dotted white boxes indicate location of radiolucent control SMP coil and strip. Polymer strips were 0.87 mm thick, while stainless steel strip was 1.03 mm thick, requiring normaliza tion of radiopacity values. Acquisition parameters: 62 kVp, 80 mA, 2 sec exposure. (B) Area averaged intensity values (prior to normalization) for 0.87 mm thick iSMP, 0.87 mm thick SMP and 1.03 mm thick stainless steel strips. Background values for 6 inche s o f normal saline are also shown. As expected, the non functionalized SMP samples were completely radiolucent, while the iSMP samples were readily visible. Qualitatively, the iSMP coil samples were visualized as easily as the stainless steel and nitinol coils. However, quantitative densito metry analysis of iSMP and stainless steel film samples revealed that the iSMP material does not match the radiopacity of stainless steel, displaying a normalized radiopacity of 46 .7% (table 3.2 ). Mean: 2474.7 Std. Dev: 101.5 5.2 cm Nitinol coil 5.2 cm Cook stainless steel coil 26 cm iSMP coil 5.2 cm iSMP coil 0.87 mm thick iSMP strip 0.87 mm thick SMP strip 1.03 mm thick stainless steel strip 5.2 cm SMP coil Mean: 2189.5 Std. Dev: 97.8 Mean: 3032.6 Std. Dev: 91.0 Mean: 2108.8 Std Dev: 98.7 A B
52 Table 3.2: Polymer radiopacity was calculated by subtractin g the background intensity value from each sample intensity value, then normalizing to the value for stainless steel. iSMP and SMP values were multiplied by a factor to account for differences in sample thickness. Sample Normalized Radiopacity Stainless steel 1.000 iSMP 0.467 SMP 0.103 Computed Tomography It is well known that implanted metal devices can generate significant artifacts during CT imaging and obscure proximal anatomical structures. These artifacts often manifest as streaking across the image and can be due to a combination of beam hardening, photon starvation, undersampling, motion, partial volume averaging, and Compton scatter. Streak artifacts from intracranial embolic coils and aneurysm clips are routinely observed. 87 Accordingly, it is important to characterize potential artifact generation from new embolization materials. Transverse CT images of both SMP and metal coil samples were acquired using a helical multi slice scanner and the slices showing maximum distortion were selected for display in figure 3 .3 The same imaging phantom used during fluoros copic evalu ation was used here.
53 Figure 3.3: (A) Transverse CT slice showing significant beam hardening artifact from Nitinol and Cook stainless steel coils in a normal saline environment. Coil dimensions: 0.71 mm wire, 52 mm length, 12 mm curl. Image acquisition parameters: 120 kV, 300 mA, 3 mm slice. (B) Minimal beam hardening artifact from 5.2 and 26 cm iSMP coils. Yellow box indicates location of control SMP coil, which is not visible. Coil dimensions: 0.71 mm wire, 5 2 mm length, 12 mm curl. Figure 3 .3 A shows the axial slice passing directly though the two individual metal coils. Both Nitinol and stainless steel generated abundant streaking artifacts along the long axis of the coils. Shadows were cast across the full artifacts will be even more significant for reconstructions along the z axis (parallel to the gantry) due to the conventional use of anisotropic voxels having a lower out of plane resolution, particularly for older scanners. Figure 3 .3 B shows a corresponding slice passing throug h the polymer coil samples. The non functionalized SMP coil appeared to be completely radiolucent. The iSMP coils were readily visible, but generated minimal streaking artifacts, likely due to reduced beam hardening and photon starvation effects. It is cle ar that even 5 packed iSMP coils produce less shadowing than a single s tainless steel or Nitinol coil. Developments in software and computational power have made segmentation and 3D reconstruction of tomographic medical imaging datasets much more common. In order to assess the ability to reconstruct iSMP coils from CT datasets, a surface mesh of the coils shown in figure 3B was developed using ITK SNAP and MeshLab software. In this case, manual segmentation 5.2 cm Cook stainless steel 5.2 cm Nitinol 26 cm iSMP 5.2 cm iSMP 26 cm SMP A B
54 was used. For comparison, a packed multi strand e mbolic device delivered into the femoral artery of a sheep, from a separate study, was also reconstructed from an automatic segmentation. These results are displayed in figure 3.4 Magnetic Resonance Imaging As with CT imaging, artifact generation from implanted metal devices during MRI scans is pervasive and its causes are numerous. Figure 5 shows T1 T2 and proton density weighted MR images of a gadolinium doped saline phantom containing polymer and metal coil samples. Functionalization with iodinated monomers does not appear to affect the signal or generate distortions from SMP coils in any of the images. Nitinol coils produced moderate distortion, while stainless steel coils produced significant distortio n in all three sequences. Moderate ringing artifact seen with both metal coils is associated with sharp boundaries in the imaging field. However, the majority of the observed artifact is due to the ferromagnetism of iron and nickel in the stainless steel a nd Nitinol coils, respectively. Materials like these, with significant magnetic Figure 3.4 Three dimensional reconstruction of 5.2 and 26 cm iSMP coils (A) as well as stainless steel and Nitinol coils (B) following manual segmentation in ITK SNAP and post processing in MeshLab, to reveal a surface mesh of triangular elements. In light of the 3 mm slice thickness, which was large relative to the dimension s of the coil, the quality of the resulting surface mesh is reasonable. (C) CT reconstruction showing coil pack in ovine femoral artery. B 2.6 cm Nitinol 2.6 cm stainless steel 2.6 cm iSMP 26 cm iSMP A
55 susceptibility, distort the linear magnetic field gradients and generate large signal voids. That said, artifact size from the Nitinol coil was significantly less than that see n with the stainless steel coil, supporting the findings of Meyer et al. 88 Interestingly, though coils were visible in all ca ses due to negative contrast/signal void, they were best visualized by the proton density weighted gradient recalled echo (GRE) sequence. GRE sequences are known to be more sensitive to susceptibility artifacts than the spin echo (SE) and fast spin echo (F SE) sequences shown here. Even when using this sequence, however, no dist ortion was seen from SMP coils. 5.2 cm Nitinol 5.2 cm iSMP 5.2 cm SMP 5.2 cm Cook stainless steel 26 cm iSMP A B C Figure 3.5: Common MRI pulse sequences reveal significant magnetic susceptibility artifacts from Nitinol and stainless steel coils, while unmodified and iodinated shape memory polymer coils appear to be artifact free. Orientation of samples in subfigure A applies to B and C. (A) T1 weighted SE sequence. Image acquisition parameters: echo time ( TE ) 8.1 ms, repetition time ( TR ) 800 ms, flip angle 90, slice thickness 5 mm, echo train length ( ETL ) 2. (B) T2 weighted FSE sequence. Image acquisition parameters: TE 81 ms, TR 3000 ms, flip angle 90, slice thickness 5 mm, ETL 16. (C) Spin density (proton) weighted GRE sequence. Image acquisition parameters: TE 1.5 ms, TR 250 ms, flip angle 25, slice thicknes s 5 mm, ETL 1.
56 With regard to torque, displacement, and heating during MR imaging, only the stainless steel coil the imaging phantom, it spun o n the nylon wire to allow for alignment of the induced and permanent magnetic fields. It follows that motion artifacts may also have contributed to the curren ts may be generated when they are moved through a static magnetic field or when the field changes rapidly during imaging. No significant heating of any coil, secondary to current induction, was detected before or directly following imaging. Discussion Some of the most important considerations for next generation embolic coil designs, beyond imaging characteristics. In this study, we evaluated these properties fo r a recently approved iodinated shape memory polymer vascular plug. Preliminary findings indicate that this material may be very suitable for neuro vascular and peripheral vascular embolization. Thermo Mechanical E valuation Thermo mechanical analysis demon strated that at body temperature, the material has a modulus of only 34.7 MPa, more than three orders of magnitude lower than that of platinum, stainless steel, or titanium alloys like Nitinol. This comparatively dramatic reduction in stiffness should yiel d coils with a significantly enhanced packing ability. Furthermore, shape memory should help them resist long term compaction and associated re canalization, delayed rupture, and aneurysm enlargement, secondary to the water hammer effect. 89 Moreover, because of their inherently low elastic modulus, SMP coils do not require complex geometries, such as secondary coiling, that are used by m etal devices. Fluoroscopic I maging
57 Imaging artifacts from implanted devices, regardless of the cause, have become accepted as the status quo and an inevitable side effect of using materials with excellent radiopacity. However, there has been a gradual real ization that polymers and polymer composites may provide adequate radiopacity without generating artifacts during tomographic imaging studies. This has renewed interest in novel material development for embolic devices. The iSMP studied here demonstrated a radiopacity almost half (46.7%) that of stainless steel when imaged near the k edge of iodine. Coil samples were readily visible in phantom based imaging. That said, there may be room to increase the device visibility further by incorporating iodinated mo nomers at higher concentrations without altering the shape recovery characteristics of the material or introducing artifacts Computed Tomography lesser ex tent, the iSMP coils) during CT imaging are largely due to beam hardening and photon starvation. Beam hardening is the process through which the average energy of the polyenergetic incident X ray beam is increased as it passes through radiodense samples. This happens because low energy X rays are disproportionately absorbed compared to high energy X rays. The harde ned beam is subsequently attenuated to a lesser extent, so the detector reads a higher signal than would otherwise be expected. Ltourneau Guillon et al. showed this effect for nitinol stents with distal tantalum markers. 90 They found that detection of stenoses was not possible with CT angiography when the stents were perpendicular to the z axis because of streak artifacts induce d by the tantalum bands. Photon starvation is often seen with metal implants because the density of the metal is beyond the normal range that can be handled by the hardware, resulting in incomplete profiles. Moreover, due to the asymmetric shape of devices being imaged, incident beams from different directions are blocked to different extents, which can confuse the reconstruction algorithm and cause artifacts that appear as dark streaks (shadows) along the long axis of the device. Accordingly, while CT arti facts can be attenuated through filtration or by modifying the tube current, slice thickness etc., they are largely unavoidable when imaging pure
58 metal devices. One of the greatest benefits of polymeric coils, then, is the ability to vary the amount of opa cifier they contain. In other words, they allow investigators to select a level of opacifier (in this case iodinated monomers) that best balances radiop acity with artifact generation. In general, artifacts are especially problematic for the C arm cone beam CT systems typically used in interventional procedures because of increased motion artifacts and scatter. 91 A traditional multi slice helical CT scanner was used in this study, so the artifacts show n here may be underestimations. Magnetic Resonance Imaging When assessing MRI signal and artifact generation, three common pulse sequences were used. These included a T1 weighted SE sequence, a T2 weighted FSE sequence, and a proton density weighted GRE sequence. In all cases, the stainless steel and Nitinol coils generated large ringing and susceptibility artifacts, while the SMP c oils were artifact free. SMP coils are visible through negative contrast, though minor differences between sequences were seen. In particular, the best apparent contrast was seen with the proton density weighted image. A single magnetic field strength (3T) was used in this study, since prior studies have shown that it has minimal to no effect on the appearance of artifacts. 92 I t is known that the orientation relative to the main magnetic field (B 0 ) and readout gradient does influence the appearance of artifacts for stents; 90 however, a uniform orientation was chosen in this study for simplicity. Visualizing polymers in MRI studies and explaining their appearance is a fairly complex process and necessitates an examination of fundamental MR physics. Nevertheless, it will likely be important in the future given the advent of time resolved 3D MRI for guided placement and retrieval of endovascular devices. 93,94 T1 relaxation is caused by the release of energy from individual excited spins/protons into th e surrounding molecular lattice. The greater the overlap between the vibrational frequency of the molecules in the surrounding lattice and the Larmor frequency ( B 0 dependent precessional frequency), the faster this energy dissipation occurs and the shorter the T1 relaxation time. In T1 weighted images, materials with short T1 will appear
59 bright since the longitudinal magnetization M Z will have recovered to a higher percentage of the original/equilibrium value. Polymers contain many bonded hydrogen atoms and their fundamental vibrational frequencies depend on the size of the monomers and composition of the polymer chains. In comparison to very large biological molecules, polymer components occupy higher frequency ranges; conversely, when compared to very smal l biological molecules in mobile aqueous environments, polymer components occupy a lower frequency range. Thus, with respect to the Larmor frequency for protons, polymer components likely do not have considerable overlap. This leads to slow spin lattice re laxation, a long T1, and a dark appearance on T1 weig hted images. In contrast, T2 decay is not a form of energy dissipation. Rather, it represents the loss of coherence (phase match) between individual protons that was originally achieved by a 90 radiofre quency pulse. Inhomogeneities in the local magnetic field (intrinsic to the sample being imaged in the case of T2 and extrinsic in the case of T2*) accelerate the dephasing process and result in a shorter T2. In T2 weighted images, materials/tissues with l ong T2 appear bright. This is be cause transverse magnetization M XY after excitation, is a decaying function, so materials with a long T2 would still have retained a significant portion of the transverse magnetization at the time of signal acquisition. Mol ecular size, composition and physical state can affect T2, since molecular motion can reduce or cancel the intrinsic micromagnetic inhomogeneities (averaging effect). The extensive cross linking of most polymers, such as the SMPs evaluated in this study, l ikely leads to considerable constraint on molecular motion. This prevents cancelation of any magnetic inhomogeneities that may be present from other sources. It leads to faster transverse dephasing, shorter T2, and, as with T1 weighting, a dark appearance on the final image. Traditionally, one significant disadvantage of GRE sequences, such as the one used to generate the proton density weighted image in figure 5, is the loss of signal from static magnetic field inhomogeneity. This occurs to a lesser extent and for a different reason with SE and FSE sequences. Consequently, magnetic susceptibility artifacts are more pronounced for GRE sequences. This can be seen in figure 5c, which shows greater distortions from the stainless steel
60 and Nitinol coils. However polymers are not prone to this effect, allowing GRE sequences to be used without these drawbacks. Interestingly, Bartels et al. demonstrated that artifacts from metal stents are generally minimized when short echo times ( TE s) are used; 95 however, the utility of th is technique is likely limited. There appears to be renewed debate in the literature concerning the efficacy and safety of endovascular embolization in comparison to surgical alternatives, particularly with respect to aneurysm repair. Given the relative infancy of coil embolization technologies, we believe that the favorable mechanical and imaging properties of fully polymeric coils, such as the iSMP o cclusion devices evaluated in this study, may point to the potential for improved patient outcomes during trans catheter embolization procedures.
61 CHAPTER IV PREPARATION AND CHARACTERIZATION OF SHAPE MEMORY POLYMER GOLD NANOCOMPOSITE MATERIALS Introduction Top Down and Bottom Up Techniques for Metal Nanoparticle S ynthesis M their intended application and available equipment. Top down techniques involve me chanical attrition of larger elemental or compound fragments. They often employ ball mills, in which metal fragments are trapped between highly kinetic balls and the inner surface of a rotating cylinder and undergo deformation, cold welding, and fragmentat ion (figure 4. 1 ) 96 99 T his technique can be used to synthesize metal particles, oxides, a nd metal/non metal composite particles. Figure 4. 1 : Dry synthesis of polydisperse metal nanoparticles can be achieved using high energy ball milling equipment. In a ball mill, multiple grinding bowls are r otated on independent platform, while the entire assembly is rotated in the opposite direction. During centrifugal forces alternately add and subtract causing the grinding balls to roll around the bowls and be thrown to the opposite side, impacting the substrate particles at high speed Rotation of bowl Grinding medium Metal fragments Rotation of assembly
62 Mechanical milling does not require high temperature treatment and typic ally produces excellent yields, but also leads to broad particle size distributions (10 1000 nm), and varied/ irregular geom etry. Impurities are also a problem, even with the use of tungsten carbide milling components and iner t atmosphere and/or high vacuum processes Moreover, milling typically generates an amorphous powder, which requires subsequent partial recrystallization before it can be consolidated into nanostructured materials Due to these significant limitations, bot tom up processes, such as pyrolysis, inert gas condensation, sol gel fabrication micellar structured media (microemulsion) and solvo thermal reactions dominate in laboratory settings. Unlike milling techniques, gas phase reactions (p yrolysis and gas condensation ) produce particles with excellent monodispersity, but often require a solid surface to support grain growth and are generally considered low yield processes (up to 100 mg/hour). This is in part because they often require a solid surface to sup port particle growth adding a rate limiting transport/diffusion term to equations governing particle nucleation and growth. Liquid phase techniques (sol gel fabrication, microemulsions, and solvo thermal reactions) offer several benefits. Like gas phase t echniques, they produce particles with a narrow size range and allow for controlled layering of particles on solid surfaces. They are also amenable to scale up with standard laboratory equipment or on industrial scales, with reactors that allow for contin uous particle generation and solvent recycling Figure 4. 2 compares the steps involved in gas and liquid phase nanoparticle fabrication.
63 Figure 4. 2 : Processes for gas and liquid phase synthesis of metal nanoparticles (S, L and G = solid, liquid and gas phase, respectively). Liquid phase, solvo thermal techniques were selected for the purposes of this study due to the combination of excellent monodispersity and adequate yield that they provide. This category also encompasses, by a wide margin, the best characterized approaches for co lloidal gold synthesis in the literature. Specific In Situ and Ex Situ Techniques for Solvo Thermal Gold Nanoparticle S ynthesis When the nanoparticle in question is intended to be part of a composite system, investigators have the option of u sing either an in situ or ex situ approach for nanoparticle formation and incorporation In the case of bulk polymer GNP composites, wherein the GNPs form a distinct, discontinuous phase, in situ particle formation involves reducing the gold precursor in a pre formed polymer matrix or liquid m onomer mixture prior to polymerization. The e x situ approach requires the investigator to generate the particles separately, add them to the liquid mon omer mixtur e, and disperse them by adding energy to the system (e.g. sonication ). The ex situ method was chosen for this study because it simplifies the removal of unwanted reactants and reaction byproducts simplifies GNP surface modification and ligand exchange, and allows nanoparticle properties to be studied independently (outsid e the context of the surrounding polymer network) Gas phase synthesis Liquid phase synthesis Precursor (S, L, or G) Precursor (G) Intermediate (G) Seed particles (S) Precursor (S or L) Intermediate (S or L) Seed particles (S) Nanoparticles (S) Evaporation Gas phase reaction Gas solid surface reaction Particle growth Liquid phase reaction Liquid solid surface reaction Particle growth Liquid solid surface reaction Nucleation/ condensation Super saturation Nucleation/ condensation
64 In the past, ex situ GNP synthesis in the laboratory fell into one of two categories. The first w as the citrate method which was first described in 1857 by Faraday 100 (who called the product and later refined by Frens 101 and others The second was the two phase method introduced by Wilcoxon et al. 102 and later refined by Brust et al. 103 Even though it was not used in this study, it is useful to describe the citrate method in more detail due to its simplicity and because it helps to convey certain universal principles of GNP preparation. In essence, it involves reacting a gold salt with a reducing agent (citrate), at elevated temperature, in a solution containing a complexing component, such as a polymer. 104 This complexing component is necessary in order to stabilize the nanoparticles and regulate their average size. A surfactant may also be used for this purpose. For example, a simple protocol involves the reduction of hydrogen tetrachloroaurate(III), HAuCl 4 with trisodium citrate dihydrate, Na 3 C 6 H 5 O 7 2H 2 O, in an aqueous solution. 105 This reaction only proceeds at elevated temperatures (typically > 60C) T he citrate donates electrons to the gold ions, binding to the surface and producing gold metal T he negative charges of the citrate molecules help to separate and stabilize the gold nanoparticles through electrostatic repulsion. Note that in this scenario, citrate acts as both the reducing agent and complexing componen t W hile the citrate method produces monodisperse GNPs between 2 and 100 nm in diameter, it suffers from a low concentration of GNPs in the resulting solution and the re striction to water as the solvent. As a result, derivitization with hydrophobic surface groups is not possible While the B rust method allows for the use of hydrophobic thiols as surface ligands it usually results in GNPs with greater polydi spersity Thankfully newer methods have been developed which allow for the synthesis of r elatively large amounts of monodisperse GNPs (> 100 mg per reaction ) and the use of hydrophobic surface ligands. One such method involves the reduction and capping of tetrachloroaurate with ethylene glycol and polyvinylpyrrolidone respectively While others were explored, this method was ultimately ex ploited to produce all GNPs used in this study and is described in detail later in this chapter
65 Control of Gold Nanoparticle S ize Aspect R atio, and Surface F unctionality The choice of reducing agent can signific antly affect the size of GNPs. For instance, tannic acid is a much stronger reducing agent than citrate. The ratio of tannic acid and citrate can then be modified to control GNP size in aqueous preparations where higher initial tannic acid concentrations yield smaller particles. 106 In general, more aggressive reducing agents are required for smaller particle sizes. Higher reducing agent concentrations have a similar effect. This scheme is summarized in figure 4.3 Figure 4.3: The effect of reducing conditions on GNP size Stronger redu cing agents result in faster nucleation and a greater number of initial seed particles. This ultimately leads to smaller GNPs. Other factors can also influence final GNP size, including the ratio of capping agent to the gold precursor and the reaction temperature and time. Higher capping agent to gold precursor ratios, higher reaction temperatures and shorter reaction times all tend to result in smaller particles. Once formed GNPs are stable for several months in aqueous media. However, stability can be improved by performing a ligand exchange, in which citrate or similar capping agent is exchang ed Particle growth Slow nucleation (fewer seed particles) Fast nucleation (more seed particles) Stronger reducing agent Larger GNPs Smaller GNPs higher [reducing agent]
66 for a ligand with greater affinity for the gold surface. Usually, this requires molecules terminated in thiol s amine s, or phosphine s 28 Such ligands often provid e new functionality to the GNPs or may serve as platforms for attachment of additional molecular payloads such as oligonucleotides. 107 108 109 The attractive interaction be tween the head group of a ligand and a nanoparticle surface can be described as chemisorption hydropho bic interaction, or electrostatic interaction. Gold thiol interactions are probably the most well known inorganic nanoparticle ligand interactions. They are readily exploited during GNP synthesis and have been used to tether biomolecules including antibodies and thiolated peptides. 110 Th iol groups are known to have the highest affinity to noble metal surfaces, particularly gold 111 This binding is described as either chemisorption or covalent bonding in the literature. In fact, the true nature and process underlying the interaction is still subject to research and discussion. A more in depth discussi on of the sulfur gold bond and the nature of thiolates is included in the next section. Over time, GNPs may aggregate and lose their dispersed colloidal properties, a phenomenon which is signaled by a shift in solution color from red to blue. A variety of additional wet and dry synthetic procedures have been developed. 106 In terms of morphology, GNPs can be divided into spheres and rods. Rod shaped GNPs are less commonly produ ced, but have unique properties and applications. They are defined grossly by their aspect ratio. 112 Gold nanorods are sometimes formed through reduction in rod shaped micelles, a process which is convenient for large scale production, but difficult with regard to controlling morphology. Alternatively, solid rod shaped templates, such as anodized porous alumina, can be used. 113 Spherical GNPs, often with positively or negatively charged surface groups, can also be used as seed agents for generating rod shaped particles, through a process called nucleation. 113 Interestingly, Gole et al. observed that a seed mediated synthesis resulted in a small number of additional shapes including triangles and hexagons in each batch of gold nanorods. 113 They also observed a dependence of the aspect ratio of the nanorod on the size of the seed particle. As with spherical GNPs, a variety of seed materials and synthetic parameters have been developed during the past decade. 114 115 116 117
67 The G old Sulfur B ond While gold is very inert in isolation it has rich and complex ligand chemistry. The gold thiolate bond in particular is subject to considerable investigation. This is because of its strength, flexibility, and utility in areas as far ranging a s drug delivery, reaction catalysis, and functionalization of surfaces for molecular sensing/recognition. Even so, the exact nature of the covalent interaction at the gold sulfur interface is not completely understood Since it is fundamental to the prepar ation of thiolate protected gold nanoparticles in this study, it is useful briefly review the current understanding of this bond and its properties Thiols are quite acidic and it is easy to recognize th at a surrounding aqueous medium could act as a base and scavenge protons. The consensus in the literature follows this conventional understanding. It is accepted that covalent interaction at the gold sulfur interface requires the formation of gold thiolate bon ds. 118 The sulfhydryl group is deprotonated, creating a thiyl radical (RS) that readily reacts with the gold surface. Experimental and theoretical studies have shown that, b ecause of the extremely high thiolate gold bond strength, the interaction between sulfur and gold can significantly al ter proximal gold gold bonding. 118 The bond is on the order of 200 kJ/mol, rivaling the strength of the gold gold bond ( 221.3 kJ/mol) and is considered to be nearly irreversible. 110 Disulfides also readily bind to gold surfaces (figure 4. 4 ). 119
68 Figure 4. 4 : Covalent bonding of bioconjugated thiols and disulfides to gold surfaces. Adapted from Synthetic Remarks [ 120 ]. oxidation states when bound ( I to +V). With regard to thiolate protected gold nanoparticles/nanoclusters alone, the structures of several highly stable compounds in the range of 1 3 nm have been elucidated. These include Au 20 (SR) 16 Au 25 (SR) 18 Au 38 (SR) 24 Au 40 (SR) 24 Au 68 (SR) 34 Au 102 (SR) 44 and additional compounds having approxim ately 144 Au atoms and 60 thiolates. 118 Now that the exact structure o f several of these compounds has been determined using X ray crystallography, future theoretical studies shoul d be able to clarify the nature of geometric and electronic factors underlying the remarkable stability of these compounds. Informative models have already been developed. In a marked departure from prior models, which considered an atomically smooth Au S interface and compact Au cores, Au atoms at the center of the particle and in the thiolate layer a re now considered to be in two distinct chemica l states 121 It is now all but established that the gold sulfur interfaces in thiolate protected gold nanoclusters consist of oligomeric RS(AuSR) n units. Characterization of Gold Nanoparticles As in all nanomaterials, size and morphology stro ngly influence physical properties of GNPs. Techniques used to characterize GNPs reflect their 1 100nm size range and include scanning S Au Au Au Au Au Au Au Au Au Au Au Au Au Au Au Au Au Au Au Au SH Molecular payload S Molecular payload SH Molecular payload S Molecular payload S S Molecular payload SH or Thiols Disulfides
69 electron microscopy, transmission electron microscopy, atomic force microscopy, Fourier transform infrared spectrosc opy, UV vis absorption spectroscopy, nuclear magnetic resonance spectroscopy, dynamic light scattering, X ray photo electron spectroscopy, X ray diffraction, and dual polarization interferometry. Tracking of nanoparticle Brownian motion through a variety of techniques also allows for size determination. 122 Certainly, UV vis spectroscopy is the most widely used method for characterizing the physical features and optical properties of GNPs, since the electromagnetic absorption bands are strongly related to their diameter and aspect ratio. 106 Rationale for Selecting Composite S ystem Shape Memory Polymer The Continuous Phase It is useful now to take a step back and discuss the reasons why a composite system consisting of an acrylate/methacrylate SMP and gold nanoparticles was chosen for this study instead of alternative SMPs and additives. Some examples of polymers which can exhibit the shape memory effect include (meth)acrylates, polyurethanes, polystyrenes, and polyvinylchloride (Meth)acrylate systems (vinyl monomers ) have been studied extensively in our laboratory in the c ontext of implanted biomaterials, particularly for vascular applications. Their biological inertness stems from the fact that the carbon carbon bonds that form their backbone are stable towards oxidative stress, hydrolysis and high temperatures. In additio n (meth)acrylates are fairly non selective in their reactivity, which allows for co polymerization with almost any vinyl monomer. Finally, subtle differences in the reactivity and steric hindrance of acrylate and methacrylate monomers means that end polym er properties can be tuned by varying their relative ratios There is significant precedent for the development of (meth)acrylate SMPs. These systems include tert B utyl acrylate/diethyleneglycol diacrylate/poly(ethyle ne glycol) dimethacrylates oligo(3 cap rolactone) dime thacrylate/n butyl acrylate, and pol yurethane based acrylic systems. Most formulations used in our laboratory are co polymers consisting of a monofunctional acryl ate
70 or methacrylate, such as tert Butyl acrylate, and a difunctional cross linking methacrylate, such as poly(ethylene glycol) dimethacrylate. In order to vary the glass transition temperature, substitutions for the monof unctional acrylate can be made, such as n Butyl acrylate or isobornyl acrylate. The molecular weight of the cross linker can also be modified, with corresponding effects on mechanical properties (M n of 550, 750, and 1000 are readily available) Important limitations of the (meth)acrylate based SMP systems include the formation of a hetero geneous polymer network a relatively broad glassy to rubbery transition, and the inhibition of polymerization by oxygen. Oxygen inhibition results from the fact that the ground state of mo lecular oxygen has a diradical nature, so it has a high reactivity towards radical species. Oxygen thus scavenges effectively propagating radicals from cleaved initiator species, forming peroxy radicals with low reactivity. Thiol ene (e.g. thiol (meth) acrylate ) systems were also considered for this study. They go a long way in addressing the major drawbacks of traditional (meth)acrylate polymer systems and have been shown to form excellent SMPs. 10 Their advantages include the formation of homogeneous networks, insensitivity to oxygen inhibition, low volume shrinkage/shrinkage stress, and rapid polymerization. Unlike purely (meth)acrylate systems, which proceed v ia free radical chain polymerization, thiol ene polymerization proceeds via a free radical step growth mechanism. Once formed a thiyl radical adds across a vinyl functional group to generate a carbon centered radical, which subsequently undergoes chain tra nsfer to a thiol group (re generating the thiyl radical). The gradual molecular weight evolution associated with step growth polymerization results in more homogeneous polymer networks and narrower glassy to rubbery transitions. However, despite their many advantages, the thioether bond formed in thiol ene polymers is very susceptible to cleavage in oxidative and hydrolytic environments, making them fundamentally un suitable for intravascular device applications.
71 Inorganic Additive The Discontinuous Phase Considerable forethought also went into the selection of the appropriate additive(s) for the SMP Along with the actual elemental composition, the size scale (atomic/molecular, nano or macro ), and method of incorporation (solution, suspension, or coati ng) had to be considered. Since one of the primary goals was to confer radiopacity, the additive had to be composed, at least partially, of metals (Ba Bi, W, Au, Ag, Ti, Ta Sn Gd, Fe etc.) or heavier halides like iodine and bromine. The ideal additive would provide sufficient X ray contrast without introducing artifacts during MR or CT imaging, have minimal adverse effect on shape memory properties, and possess unique properties that might expand the ability to process or control the behavior of the composite material. GNPs were ultimately selected as the additive for many reasons. The first benefit is that the ability of gold to absorb and scatter incid ent X rays is greater than that of other common contrast agents such as iodine, bari um, tantalum and zirconium. In comparison to these elements gold has a higher atomic weight (~197) 21 When considered a cross a broad energy range, heavier elements have a correspondingly greater average mass attenuation coefficient This is in part because of the heavier nuclei and also because of the greater number of electron shells that surround them. As X ray energy is increased, absorption generally decreases. However, local absorption peaks appear at cer tain energies, corresponding to electron shell binding energies. In the diagnostic X ray energy range, it is the innermost K shell that is exploited In other words, the X ray source can be tuned to emit X rays with an average energy just above the K edge of a contrast agent (where there is a local maximum), increasing contrast between the actual contrast agent and surrounding tissue. Due shell binding energy, it can be imaged at even higher X ray energies (> 80 keV) for which bone and soft tissue absorption are minimized, improving contrast and reducing ionizing radiation dose to the patient. 23 The second benefit of GNPs is that there is precedent for their use in polymer composites and as radiopacifiers (though separately). Polymer gold nanocomposites have found use in optical applications, such as lenses, fi lters and light emitting di odes. 24 25 I n colloid al form, surface functionalized GNPs have
72 been evaluated as X ray contrast agents for vasculature and micro damaged bone, with favorable results. 23 26 27 The third benefit is the flexibility with which GNPs are synthesized and modified. GNPs are well characterized in the literature and are well known for the ease and flexibility of their synthesis, excellent size and s hape control, their long term stability in a variety of solvents, and amenity to surface modification with thiols, amines and phosphines for dispersion in polymer environments. 28 29 30 31 Fou rth, gold is a diamagnetic element and should not alter local magnetic fields during MR imaging, which is a common source of artifacts from exogenous implanted biomaterials. Fifth GNPs have special properties that could change the way SMPs are processed a nd manipulated with r espect to thermal transitions. For instance, u nmodified polymers are electrical and thermal insulators, but GNPs may allow these properties to be enhanced in a concentration depend ent manner. Likewise, GNPs display a surface plasmon resonance enhanced absorption peak near the wavelength of visible green light. By efficiently dissipating visible light as heat, thermal transitions in the polymer may be manipulated indirectly, in a spatially controlled manner. Finally, GNPs have been sho wn to be re latively biocompatible and safe. GNP Synthesis and C haracterization In this study, GNP preparation followed a modified version of the protocol developed by Carotenuto and Nicolais. 123 Tetrachloroaurate trihydrate (HAuCl 4 3H 2 O), ethylene glycol (EG), polyvinylpyrrolidone (PVP, MW n 10,000), and dodecanethiol (DDT), were selected as the gold salt, reducing agent, capping agent, and final surface ligand, respectively. The reaction setup consisted of a 250 ml round bottom boiling flask, equipped with a condenser and incubated in a temp erature controlled paraffin oil bath. Before starting, all glassware to come in contact with GNPs was incubated with aqua regia (1:3 volumetric mixture of nitric and hydrochloric acids), in order to dissolve remnant colloidal gold and other trace metals, r insed in distilled water, and baked at 100C for 3 hours. In the first step, a solution of 200 mg/ml HAuCl 4 3H 2 O in EG was rapidly injected into 100 ml of 30% (w/v) PVP in EG stabilized at 90C in the boiling flask. Note that preparation of the PVP/EG solu tion required heat and stirring over several hours and that
73 glass spatulas were used to handle HAuCl 4 3H 2 O, due to its corrosive nature. This reaction proceeded for 10 minutes and was characterized by a change in the solution color from yellow, to clear an d, finally, deep red/purple. The reaction mixture was then cast into 300 ml of chilled acetone and sonicated for 2 minutes in order to terminate the reaction and separate the nanoparticles. A gold PVP pellet was collected by centrifugation at 1000 x g for 15 min and washed with chilled acetone. The centrifugation and washing steps were repeated three more times and the product was at room temperature The pellet was subsequently dissolved into a 100 ml solution of 50 mM DDT i n ethanol and incubated at room temperature, with agitation, for 1 hour, facilitating surface ligand exchange. Once again, the product was collected by three cycles of centrifugation and washing with chilled ethanol. The final DDT functionalized gold nanop article pellet was dried under vacuum and stored at room temperature. Figure 4.5 illustrates the nanoparticle formation and ligand exchange reactions. Figure 4.5 : Reaction schematic for the preparation of hydrophobic thiol derivatized gold nanoparticles. Reaction 1 shows the reduction and capping of tetrachloroaurate trihydrate by ethylene glycol and polyvinylpyrrolidone (PVP). Reaction 2 shows the exchange of PVP for deodecanethiol (DDT). Early on in this study, a number of other methods for high yield GNP preparation were evaluated. One such method involved the reduction and capping of GNPs with oleylamine in toluene. It was HAuCl 4 3 H 2 O + ethylene glycol + PVP Dodecanethiol (Reaction 1) (Reaction 2) DDT DDT Au + PVP PVP Au
74 based on a m odified version of the protocol developed by H. Hiramatsu 124 However, this method proved to be highly variable and unpredictable, perhaps due to the fact that oleylamine is only available at 70% (v/v) and exhibits a greater lot to lot variation in activity. Exact protocols for GNP prepa ration using both the EG/PVP and oleylami ne methods can be found in the A ppendix. XPS survey and multiplex scans were used to identify gold thiolates at the surface of GNPs as an indirect confirmation of successful ligand exchange with DDT. To prepare the sample for analysis 0.5 mg of surface modified GNPs was dispersed in 2 ml of hexane using sonication. Drops of this dilute solution were s erially deposited onto the surfa ce of a 1 cm 2 polished silicon wafer and allowed to dry after each application (figure 4. 6 ). Figure 4. 6 : Samples for X ray photoelectron spectroscopy analysis were prepared by serially depositing a dilute solution of surface modified GNPs in hexane onto 1 cm 2 polished silicon wafers. A total of 0.5 mg was deposited on each wafer. In general, peaks are assigned by checking the position and relative intensities of two or more peaks (photoemission and Auger lines) for an element. Checking the spin orbital splitting and relative area unde r p, d, and f peaks is also helpful. Modern software tools simplify identification by automatically comparing unknown compound spectra to spectra for individual elements (e.g. Scofield element library). In this study, CasaXPS software was used to analyze a ll XPS spectra. Selected peaks from the initial survey scan representing electron binding energies of atoms within approximately 5 nm of the sample s u rface, are labeled in figure 4.7 Once the survey scan revealed the approximate location of cert ain peaks of interest (Sulfur 2p Gold 4f, and Carbon 1s ), Bare silicon wafer Deposited GNPs
75 a multiple x scan was performed (figure 4.8 ). As expected, sulfur was present in thiolate form, bound to gold. This was determined by comparing the observed binding energy corresponding to the Sulfur 2P pea k to those found in the literature for gold thiolates. 125 S pin orbit splitting between the sulfur 2p1/2 and 2p3/2 peaks (it is expected for all p, d and f sublevels) was obs erved Figure 4.7 : X ray photoelectron spectrum of dodecanethiol functionalized gold nano particle surfa ce (low resolution survey scan), showing carbon 1s, sulfur 2p and gold 4f peaks. Table 4.1: Raw data for X ray photoelectron spectroscopy survey scan. Name Position FWHM AUC C 1s 284.4 2.115 6987.0 S 2p 162.8 2.472 762.8 Au 4f 83.6 1.842 77923.9 Au 4f7/2 83.5 1.793 40331.7 Au 4f5/2 87.1 2.061 36016.5 FWHM: Full width at half maximum AUC: Area under the curve 0 5 10 15 20 25 1100 990 880 770 660 550 440 330 220 110 0 Counts per second (x103) Binding energy (eV) Au 4f C 1s S 2p
78 Figure 4.8 X ray photoelectron spectrum of DDT functionalized gold nanoparticle surface (high resolution multiplex scan) Sulfur 2p, gold 4s, and carbon 1s peaks are shown. The l ocation of the sulfur 2p peak is consistent with the binding energy of gold bound sulfur species. The expected spin orbit splitting between the sulfur 2p1/2 and 2p3/2 peaks is evident. Raw data was analyzed using CasaXP S software.
79 Table 4.2 Raw data for X ray photoelectron spectroscopy multiplex scan. Name Position FWHM AUC % Conc. S 2p3/2 161.52 1.15 110.78 2.62 S 2p1/2 162.75 1.103 62.43 2.89 Au 4f7/2 83.52 1.018 9572.85 26.27 Au 4f5/2 87.2 0 1.043 7451.66 25.98 C 1s fit 284.46 1.345 1606.02 42.23 FWHM: Full width at half maximum AUC: Area under the curve Nanoparticle size is best assessed through TEM or SEM or using spectroscopic techniques such as dynamic light scattering (DLS) and UV Vis spectroscopy. Spectroscopic techniques rely on the application of the Mie and Gans theories for spherical particles and are prone to minor inaccuracies. 126 127 However, they are used routinely and provide rapid results. Specifically, Haiss et al. 126 determined that the size of spherical GNPs can be determined from the ratio of absorbances at the surface plasmon resonance peak and at 450 nm: Here, d is the particle diameter in nm, Aspr is the absorbance of GNps at the surface plasmon resonance peak, A450 is the absorbance at 450 nm, and B 1 and B 2 are parameters derived from a linear fit of a plot of the theoretical value of Aspr/A450 as a function of ln(d). B 1 and B 2 were found to be 3.00 and 2.20, respectively, resulting in an average error of approximately 11% for particles in the size range from 5 to 80 nm. DLS also helps to quantify monodispersity, as it determines the relative abundance of particles or clusters o f varying diameter. The results of UV Vis spectroscopy and DLS analysis of DDT functionalized GNPs are displayed in figure 4.9
80 Figure 4.9 : Characterization of dodecanethiol funct ionalized gold nanoparticles: ultraviolet v is ible spectrum (A) and dynamic light scattering (DLS) data (B) of nanoparticles indicating average particle sizes of approximat ely 12 and 14 nm, respectively. Volume weighting was used generate the DLS spectrum. Before attempting to embed the GNPs in the polymer, they were suspended in a dilute hexane solution and deposited onto glass slides for TEM analysis. Figure 4.10 reveals a relatively monodisperse group of particles with sizes between 5 and 15 nm in diameter. As a rule of thumb, a DLS spectra can only be used to justi fy the absolute label monodisperse when particle diameters fall within 20% of the mean ( in this case, ~ 11.2 16.8). However, particle aggregates have been known to skew the results. When assessed by TEM, i n a purely hydrophobic environment ( hexane ) the pa rticles do not show any significant tendency to aggregate, though strings or small clusters of 2 to 5 particles were observed. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 350 400 450 500 550 600 650 700 750 Absorbance Wavelength (nm) 526 nm surface plasmon resonance absorption peak 0 10 20 30 40 50 60 70 80 90 100 0 5 10 15 20 25 30 35 40 Relative abundance Diameter (nm)
81 Figure 4.10 : Transmission electron microscopy im ages of DDT functionalized GNPs suspended in hexane and deposited on glass slides. 100 (A) nd 50 nm (B) size bars are shown. Particles do not display aggregation in large clusters. The final method used to evaluate the GNPs alone was contact angle analysis. The goal was to determine how closely matched the GNP and SMP surfaces were with respect to surface polar and non polar (dispersive) forces and, ultimately, hydrophobicity. Essentially, the static contact angle of a layer of DDT functionalized GNPs deposit ed on silicon wafers along with plain SMP films, with water and diiodomethane (a non polar liquid), was assessed. Using this information, 50 The results are presented in figure 4.11. A B
82 Figure 4.11: Contact angle of a layer of DDT functionalized GNPs deposited on silicon wafers along with plain SMP films, with water and diiodomethane (a non polar liquid). Surface free The lower contact angle of diiodomethane with the GNP surface reflects the larger contribution from london dispersion forces in the interaction between the solid and liquid, when compared with t he SMP surface. However, the calculated surface energy demonstrates that in both materials, that interaction forces not generated by fixed dipoles predominate in both materials. The wettability, as measured by the water contact angle, is generally equivale nt. This indicates that one can conclude that the GNP and SMP surfaces are broadly similar and have a good chance of interacting stably at their respecti ve surfaces. One limitation of this testing method is the fact that the GNP layer and SMP did not have the same surface topography. The SMP was polymerized against glass microscope slides, resulting in sub nanometer topographical features. On the other han d, the gold surface has topographical feature sizes on the order of the nanoparticles themselves. As a result, it must be noted that this analysis makes the assumption that topographical changes on a sub 20 nm scale do not significantly alter macroscopic c ontact angle measurements. 0 20 40 60 80 100 120 Water Diiodomethane Contact angle GNP surface SMP surface 0 5 10 15 20 25 30 35 40 45 50 Water Diiodomethane Surface free energy (mJ/m^2) GNP surface SMP surface
83 Shape Memory Polymer S ynthesis an d Composite C haracterization Plain SMP and SMP GNP composite samples used in this study were generally prepared by casting a mixture of a monofunctional acrylic monomer, difunctional acrylic oligomer, thermal initiator, and isolated GNPs, into either custom chambers consisting of sili conized glass microscope slides separated by 1 mm silicon rubber gaskets or polypropylene centrifuge tubes. Specifically, SMPs consisted of 75% (w/w) tert Butyl acrylate monomer (tBA, Sigma Aldrich, St. Louis, MO) and 25% (w/w) poly(ethylene glycol) dimeth acrylate (PEGDMA, 550 M n Sigma Aldrich, St. Louis, MO). Note that tBA forms the backbone of the polymer, while PEGDMA serves as the cross linker. Dispersion of the GNPs and polymerization were achieved by sonicating the mixtures in an ice water bath for 2 hours, after which the temperature of the bath was ramped up to 70C. The time required for polymerization following the temperature ramp depended on the Azobis( 2 methylpropionitrile) (AIBN, Sigma Aldrich, St. Louis, MO) and reached approximately 100% conversion, as measured by Fourier transform infrared spectroscopy, after 5 hours. GNP content in the SMPs varied over a wide range: 0, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10, 15, 25, and 30% (w/w). The composite samples intended for thermo mechanical evaluation consisted of 5 x 35 x 1 mm coupons. The composite samples prepared for imaging analysis were left in polypropylene centrifuge tubes. All of these samples h ad a final volume of 0.5 ml, neglecting minor shrinkage stresses during polymerization and t he remaining volume was filled with 0.9% (w/v) saline. Figure 4. 12 displays the structure of acrylate and methacrylate monomers used to synthesize the shape memory co p olymer. Figure 4.12 : Chemical structure of monomers used in this study : (A) tert Butyl acrylate (tBA); (B) poly(ethylene glycol) dimethacrylate (PEGDMA) 550 Mn. A B
84 When attempting to assess the effect of an additive on polymer properties, it is important to make sure that any observed changes are due to the nature of the additiv e and the interface between phases, rather than any effect the additive might have directly on polymerization One way to do this is to examine the FTIR absorption spectru m for the samples. These spectra are routinely used to track the disappearance of absorption peaks corresponding to functional groups (e.g. vinyl groups) that should be consumed during polymerization. Briefly, FTIR spectra should be acquired for the SMP both with and without GNPs. These spectra should be compared to that of the liquid monomer/GNP mixture. In this case, we are interested in monosubstituted alkenes, which typically display strong peaks at 900 and 990 cm 1 Since the 990 cm 1 peak can be obs cured by other chemical groups, particular attention was paid to the 900 cm 1 peak. Figure 4.13 displays these spectra. A broad and weaker absorption peak is also expected at 880 900 cm 1 representing gem di substituted alkenes (as found in di methacrylat es such as PEGDMA). Thi s region is shown in figure 4.14
85 Figure 4.13 : Infrared spectra of an unreacted acrylate monomer film and two polymerized nanocomposite materials containing 0 and 1 wt% gold nanoparticles. The absence of a vinyl group absorbance peak at 900 cm 1 shows adequate bond convers ion independent of GNP incorporation. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Absorbance Wavenumber (cm 1 ) 0 wt% Polymer 1 wt% Polymer Monomer mixture 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Absorbance Wavelength (cm 1 ) tBA monomer PEGDMA550 monomer
86 Figure 4. 14 : Infrared spectra of an unreacted acrylate monomer film and two polymerized nanocomposite materials containing 0 and 1 wt% gold nanoparticles. A broad and weaker absorption peak is expected at 880 900 cm 1 representing gem di substituted alkenes (as found in di methacrylates such as PEGDMA). A simple method to calculate conversion is to use an internal reference peak (such as that corresponding to the carbonyl group, C=O). The function used to calculate vinyl bond conversion, normalized to the unaltered carbon oxygen bond in carbonyl groups is: ( 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 830 842 854 866 878 890 902 914 926 938 950 Absorbance Wavenumber (cm 1 ) 0 wt% Polymer 1 wt% Polymer Monomer mixture 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 830 842 854 866 878 890 902 914 926 938 950 Absorbance Wavelength (cm 1 ) tBA monomer PEGDMA550 monomer
87 Using this method, and assuming the vinyl carbon carbon double bond conversion in tBA is representative of the whole polymer, the approximate degree of conversion is 78.5% and 77.0% for the SMPs with 0 and 1 wt% GNP, respectively. This suggests that the pr esence of GNPs does not significantly influence vinyl bond conversion. Once incorporated into the polymerized and annealed SMP, GNP monodispersity and aggregation can be accurately assessed with electron microscopy techniques. It was expected that smaller GNPs would reduce the entropic penalty associated with their incorporation into the continuous phase of the polymer and thereby disperse more readily. It was also expected that miscibility of the GNPs with the SMP may be optimized by matching the hydrophob icity of chemical groups on the gold surface to that of the constituent monomers. That said, the polymer network of interest in this study has a mixed hydrophobic/hydrophilic nature. Though the tBA monomer is highly hydrophobic (due to the bulky tertiary b utyl group) and occupies 80% of the polymer by mass, PEGDMA is hydrophilic. Since the GNP surface brush is composed entirely of short chain alkanes, it was expected to exhibit some degree of phase separation/aggregation. F igure 4.15 demonstrates this principle by displaying representative TEM images for composite samples ranging from 0.1 to 3 wt% GNPs.
88 A B C D E F
89 Figure 4.15 : Transmission electron microscopy images of dodecanethiol functionalized GNPs embedded, at varying concentrations, in a (meth)acrylate shape memory polymer consisting of 80% (w/w) tert Butyl acrylate and 20% (w/w) poly(ethylene glycol) dimethacrylate Mn 550. Subfigures (A H) correspond to gold nanoparticle (GNP) concentrations of 0.1, 0.2, 0.5, 1, 1.5, 2, 2.5 and 3 wt%. As the GNP incorporation is increased, cluster size also increases, with a dramatic increase in particle aggregation seen between 2.5 and 3 wt%. Size bars in subfigures (A G) are 50 nm. Size bar in subfigure H is 0.2 m. As the GNP concentration was raised, the particle cluster size increased steadily. A disproportionately large increase in GNP cluster size was seen between 2.5 and 3 wt%. However, as shown in figure 4.16 the clusters themselv es were fairly well separated. G H
90 Figure 4.16 : Large field transmission electron microscopy images of dodecanethiol functionalized gold nanoparticles (GNPs) embedded in a (meth)acrylate shape memory polymer ( SMP ) consisting of 80% (w/w) tert Butyl acrylate and 20% (w/w) poly(ethylene glycol) dimethacrylate, Mn 550. The GNPs formed increasingly large clusters as the GNP content in the SMP is increased, but the clusters themselves were well separated. (A) 2.5 wt% GN Ps. (B) 3 wt% GNPs. Visualization of the surface of cryosectioned nanocomposite samples was attempted using both traditional SEM instruments in low vacuum mode as well as field emission SEM instruments. However, due to the small particle size and artifact s introduced by surface charging, satisfactory images were not obtained. Biocompatibility of Gold Nanoparticles The relative biocompatibility of GNPs is widely cited in the literature. 128 129 130 131 In general, most though the distinction is often ignored. Many groups have studied the biological effects of GNP exposure, realizing that their unique size and shape may enhance accumulation in specific cellula r compartments or organs. Even so, the interactions of GNPs with specific cellular targets are still poorly understood. As an example, Bhattacharya et al. observed G1 phase cell cycle arrest in a GNP treated multiple myeloma cell line. 132 Western blot analysis revealed GNP concentration dependent upregulation of specific cell cycle regulator proteins p21 and p27 (ie. tumor suppressor genes). No such inhibitory effect was seen in normal peripheral blood A B
91 mononuclear cells. The complex mechan isms behind this cell specific anti proliferative effect are yet unknown, but their elucidation will certainly reveal new uses for GNPs in nanotherapeutics. Shukla et al. utilized a variety of microsc opy tools, including atomic force microscopy confocal l aser scanning microscopy, and transmission electron microscopy to study the cytotoxic and immunogenic effects of GNPs on RAW264.7 macrophages. 133 As a function of exposure to various concentrations of colloidal gold, they measured overall cell survival, secretion of pro inflammatory cytokines such as TNF nitrite species. Their primary fi nding was that unmodified GNPs are inert and non toxic to cultured macrophages over a 48 hour time period. Moreover, GNPs did not illicit secretion of pro inflammatory cytokines and actually inhibited the production of reactive oxygen and nitrite species k nown to cause oxidative damage in cells and trigger apoptosis. However, some questions were left unanswered. The authors observed accumulation of GNPs in perinuclear lysozomes, but possible mechanisms for their exo cytosis and clearance are unknown. Unlike proteins, it is unlikely that lysozomal enzymes will digest GNPs. As a result, bioaccumulation could affect long term biocompatibility. Pan et al. evaluated the potential cytotoxic effects of triphenylphosphine stabilized GNPs in a wider variety of cellul ar lineages, including connective tissue fibroblasts, epithelial cells, macrophages and melanoma cells. 134 They found that cytotoxicity was highly size dependent, with the proportion of surviving cells and mode of cell death varying dramatically. Specifically, the authors examined GNPs ranging in size from 0.8 to 15 nm and found that 1.4 nm particles produced the highest toxicity in all four lineages, which represented major variations in barrier and phagocytic function. IC 50 values ranged from 30 to 56, depending on the cell population in question. In stark contrast, 15 nm GNPs were found to be non toxic at up to 60 fold higher concentrations. Moreover, the 1.4 nm particles predominantly caused cell death by necrosis within ~12 hrs, whereas particles only slightly smaller at 1.2 nm predominantly effected cell death through apoptosis. This finding suggests varia tion in uptake kinetics and/or specificity for membrane bound or intracellular targets for GNPs with only small differences in size.
92 The biocompatibility of GNPs is further complicated by surface functionalization. It is often the fact that the surface gro up itself causes the observed toxicity. Smith et al. correctly note that while gold itself is generally inert and non cytotoxic, c etyl trimethylammonium bromide ( CTAB ) a very common surface ligand, is quite toxic, suggesting that CTAB coated GNPs may not in fact be useful for in vivo diagnostic studies and therapy. 116 On the other hand, the work of Connor et al. suggests that free CTAB, not GNP bound CTAB is toxic to cells. This group evaluated cell survival in a K562 leukemia cell line as a function of treatment with 18 nm GNPs modified with various surface ligands, including citrate, biotin and CTAB. Using a standard preparation protocol, t hey found that both CTAB alone and CTAB modified GNPs displayed significant toxicity. However, when the CTAB modified GNPs were centrifuged and washed with de ionized water to remove unbound CTAB, the cytotoxic effect disappeared. Moreover, additional work in this area has directly shown that the toxicity of ligands like CTAB are reduced or eliminated when complexed with GNPs. 135 It is possible that GNP binding reduces toxicity by altering cellular localization and targeting. Connor et al. also discovered that the precursors to GNP production were significantly more toxic than the GNPs themselves. Specifically, the MTT assay revealed no significant cell death in K562 cells treated with citrate or biotin modified GNPs, while the gold salt (AuCl 4 ) precursor solution was over 90% toxic at a concentration of 200 M.
93 CHAPTER V THERMO MECHANICAL PROPERTIES OF SHAPE MEMORY POLYMER GOLD NANOCOMPOSITE MATERIALS Introduction T he incorporation of inorganic nanoparticles into polymers for biomedical applications has been discussed at length in the literature. Applications include drug delivery, imaging, and photo thermal or magnetic control of micro scale actuators or larger devi ces. As discussed earlier, we have identified gold as a particularly useful additive because of two special properties: excellent X ray mass attenuation coefficients in the diagnostic energy range, and surface plasmon resonance enhanced light absorption (i e. the photo thermal effect). Moreover, gold nanoparticles have numerous routes of synthesis, with flexibility with regard to control of particle size and geometry, and are amenable to surface modification with chemical groups terminated in thiols, amines, and phosphines. With regard to SMP gold nanocomposite materials specifically photo thermal control of cross linking, annealing and shape memory actuation has been explored, but little attention is paid to the effect of the gold nanoparticles on the funda mental mechanical and thermo mechanical properties of the materials. It is critical to evaluate these properties in order to identify appropriate biomedical device applications. The e ffects of an additive can be predicted based on fundamental nanocomposite principles and computational models that have been developed to predict their influence on Tg, shape recovery sharpness, and recovery stress, among other important variables 136 Depending on a destabilizing effects on a polymer network. In conventional polymer composite materials, many inorganic additives with dimensions in the micrometer range (e.g. calcium carbonate, glass beads, talc, and colloidal silica, and naturally occurring fibers like asbestos) have been used to enhance mechanical properties. These properties ca n be modified by controlling the volume fraction, shape, and size of the filler particles. A more dramatic improvement of the mechanical
94 properties can be achieved by using additives with larger aspect ratios such as glass fibers. It follows that dispersio n of particles or fibers with nanometer dimensions can have a very large impact on mechanical and thermo mechanical properties. So far, only layered silicates and carbon nanotubes have been studies extensively in this context. The mechanical properties of a nanocomposite are influenced by several parameters, including the properties of the polymer matrix, the material properties and distribution of the nanoscale additive, interfacial bonding, and processing methods. The interfaces are particularly important though understudied, because they determine the effectiveness of load transfer between the continuous and discontinuous phases of the material. As discussed earlier fabrication of homogeneous polymer nanocomposites remains a major area of scientific rese arch. One area that has received significant attention is the influence of nano scale additives on the T g of polymers. Both decreases and increases in the Tg have been reported, depending on the nature of the nanoparticle and its interaction with the polym er matrix. It is important to remember that if the addition of a nanoparticle to an amorphous polymer results in a change in the Tg, the traditional con tinuum mechanic s relationships. It is known that the Tg of a polymer will be affected when its chains are within several nanometers of a distinct phase. This is even true when the other phase is air. Investigators have found that the Tg of a polymer tends to be lower at the air polymer interface than in the bulk polymer. This is sometimes called a confinement effect. A specific experimental example was reported in which poly(2 vinyl pyridine) showed an increase in T g poly(methyl methacrylate) (PMMA) showed a decrease in T g and polystyrene showed no change with s ilica nanosphere incorporation. These differences were ascribed to surface wetting The authors ascribed the decrease in Tg seen with PMMA to poor surface wetting and the resulting free volume existing at the polymer surface interface. For cross linked polymers, other consideration s are necessary since the presence of nanoparticles could result in a change in cross link densit y when compared to the unmodified composite. This could be due to preferential
95 interactions of the cross linking agent with the nanoparticle surface or interruption of the crosslink density due to confinement effects. Materials and Methods Thermo Mechanica l Analysis of Shape Memory Polymer Gold Nanocomposite Materials Thermo mechanical analysis was performed using a TA Instruments Q800 dynamic mechanical analyzer. 5 x 35 x 1 mm samples were prepared from both the radiopaque and control SMP sheets. Samples were loaded between uniaxial thin film clamps for thermal scans, which ran from 0 to 100C at a rate of 3C/min, a frequency of 1 Hz, and a strain of 0.05% in tension. Glassy and rubbery elastic moduli (E g and E r respectively) were reported. Glass transition temperatures (T g ) were assigned at the tan delta curve maxima. Evaluating Shape Memory Properties of Shape Memory Polymer Gold Nanocomposite Materials The shape memory effect in the nanocomposite materials was evaluated according to the free strain recovery and fixity tests described in the literature. Free strain recovery was measured as a function of increasing temperature from Tg 30C to Tg + 30C on the above mentioned dynamic mechanical analyzer. Similar rectangular samples were st retched to the maximum strain at a rate of 0.5s 1 at a temperature 5C above their respective glass transitions. Samples were then cooled down to Tg 30C while maintaining the maximum strain and equilibrated for 5 minutes to completely store stresses wit hin the composites. Then, the deforming stress was removed and the samples were heated to Tg + 30C at 5C/min under a zero load condition and held at that temperature, allowing strain recovery. This scheme is presented in figure 5.1.
96 Figure 5.1: Procedure used to evaluate shape recovery of shape memory polymer gold nanocomposite materials. Results and Discussion Polymer networks fabricated using (meth)acrylate monomers have the potential to exhibit a very broad range of thermo mechanical properties, making them excellent candidates for shape memory materials in biomedical device applications. It is crucial that the thermo mechanical response of SMPs is well controlled and understood so that it can be predicted and optimiz ed based on compositional changes. In this study, important shape recovery properties were studied as a function of increasing GNP content. Nanoparticle size and aspect ratio was kept constant. Figure 5.2 summarizes changes in Tg, glass transition width, g lassy and rubbery storage moduli, free strain recovery, shape fixity, and shape recovery sharpness for a range of GNP concentrations. Overall, reinforcement with GNPs had a positive im p act on shape recovery behavior. Temperature Unloaded strain (Eu) Max strain (Em) Final strain (Ep) 10C Tg + 5C Tg + 25C 20C/min Free strain recovery 3C/min Free strain recovery Shape fixity Shape recovery sharpness
97 Figure 5.2: Thermo mechanical properties of nanocomposite materials as measured by dynamic mechanical analysis: (A) glass transition temperature and transition width ; (B) glassy modulus; (C) rubbery modulus; (D) free strain recovery and strain fixity ; (E) shape recov ery sharpness. As expected, the glass transition temperature decreases gradually as the GNP content is increased from 0 to 1 wt%. This is likely due to the d isruption of packing of chain polymers resulting in higher free volum e. The glass transition width also decreases, though this change is more likely due to the increased thermal conductivity of the composite material. This may also explain the increase in shape recovery sharpness seen at 0.5 and 1 wt%. Both glassy and rubbe ry moduli increase; this effect is only significant at higher levels of GNP incorporation (>0.5 wt%). More interestingly, though no significant change is seen in strain fixity, free strain recovery does increase, meaning that GNP reinforced (meth)acrylate SMPs are able to recover a greater percentage of induced strains. 0 20 40 60 0 0.05 0.1 0.2 0.5 1 Temperature ( C) GNP weight % Glass transition Transition width 1500 1600 1700 1800 1900 0 0.05 0.1 0.2 0.5 1 Glassy modulus (MPa) GNP weight % 1500 1600 1700 1800 1900 0 0.05 0.1 0.2 0.5 1 Glassy modulus (MPa) GNP weight % 0 2 4 6 8 0 0.05 0.1 0.2 0.5 1 Rubbery modulus (Mpa) GNP weight % 92 94 96 98 100 0 0.05 0.1 0.2 0.5 1 Percent GNP weight % Free strain recovery Strain fixity 0 2 4 6 8 0 0.05 0.1 0.2 0.5 1 Shape recovery sharpness (%/ C) GNP weight %
98 CHAPTER V I EMBEDDED GOLD NANOPARTICLES FOR MULTI MODALITY MEDICAL IMAGING OF SH APE MEMORY POLYMER BIOMATERIALS Introduction It is difficult to overstate the importance of polymers in the design of modern implanted biomaterials. They are used in applications as far ranging as orthopedic joint repair, vascular grafting, and ocular drug delivery. While their properties vary considerably, many of their roles are unique and cannot be performed by metals, ceramics, or other materials. These include applications where low density, optical transmission, drug elution, biodegradation, or active surface properties are required. Perhaps more importantly, polymer properties are modifiable over a far gre ater range than other classes of materials, through precise control of composition, molecular weight, cross linking, and processing. More recently, a heavy focus on the development of multi phase polymer composites, including the addition of particles and fibers, have extended the range of attainable bulk and s urface properties even further. In comparison to other functional classes of polymers, shape memory polymers (SMPs) are relatively new and, as of yet, have seen only limited use in biomedical device a pplications outside of the laboratory. They are stimuli responsive materials and can recover almost any pre determined shape after being heated above a tunable glass transition temperature (T g ). 8,85 In fact, repeated strain recoveries of several hundred percent are possible over multiple cycles. 9 This feature allows compact SMP devices to recover complex 3D conformations u pon release at body temperature. Naturally, SMPs lend themselves to minimally invasive procedures, in which devices are delivered into the body through narrow vascular catheters, or laparoscopic ports. The thermally induced shape memory effect may also hel p certain devices resist permanent deformation associated with long term exposure to natural forces in the body (blood pressure, joint compression etc.). Moreover, at temperatures above the T g SMPs undergo a reduction in stiffness of several orders of magnitude, resulting in elastic moduli below 30 MPa. 8,86 In
99 comparison, stainless steel has an elastic modulus of roughly 180 GPa. 76 It is thus expected that SMPs will perform well in applications requiring device delivery to sensiti ve compartments of the body, s uch as the neurovascular space. Strategies for Visualizing Polymeric Devices in the B ody While SMPs show promise for biomedical device design, their implementation requires the ability to be visualized during and after placeme nt. With the exception of some ocular devices, this necessitates some form of X r ay, MRI, or ultrasound contrast. In fact, the field of interventional radiology has largely formed around the need for accurate image guidance of minimally invasive procedures involving interventions in the vascular, pulmonary, hepato biliary, and gastro intestinal compartments. While ultrasound is useful in many circumstances, fluoroscopy has easily become the clinical standard for image guided interventional procedures and ti me resolved MRI is still many years from seeing mainstream use. This presents a problem for polymeric devices, since they are predominantly made from low atomic mass elements, such as carbon, hydrogen, and oxygen. In the past, radiopacity has been achieved in polymers by adding heavy element fillers to absorb and scatter X rays. For instance, barium sulfate, zirconium oxide and tantalum have been used in the orthopedic field for bone cement 18,19 However, these fillers are composed of large particles, on the order of tens to hundreds of microns. SMP properties are highly sensitive to additives, so other strategies for incorporating opacifiers are prefer able. One of the main concerns regarding the use of colloidal gold suspensions in polymers is that an inhomogeneous mixture may result. It has been shown that incomplete mixing of physical blends can lead to failures at the interface between the polymer an d additive, allowing for penetration of fluids and leaching of the additive. 15 However, this issue may only apply to non nanoscale particles, since the polymer matrix may incorporate smaller particles during polymerization without a significa nt entropic penalty. 24
100 Gold Nanoparticles for Multi Modality Imaging C ontrast The primary goal of this study is to investigate the potential of gold nanoparticles (GNPs) in the fabrication of radiopaque SMPs and, secondarily, to e valuate their influence on CT and MRI artifacts and ultrasound scattering. GNPs are advantageous for a number of reasons. First, compared to traditional filler elements or contrast agents like iodine, gold offers a higher atomic weight (~197) and a superior ability to attenuate X ray s at most energy levels in the diagnostic shell binding energy, it can be imaged at even higher X ray energies (> 80 keV) for which bone and soft tissue absorption are minimized, improving contrast and reducing ionizing radiati on dose to the patient (figure 6.1 ). Second, gold is diamagnetic and should not generate the magnetic susceptibility related MRI artifacts seen with other metals. Third, gold has a very high density (~19.3 g/cm 3 ), so its incorporation into SMPs may enhance ultrasound contrast through a bulk density effect. Fourth, by varying their size, aspect r atio, concentration, and surface chemistry, the bulk properties of the resulting composite material can be tailored to a very fine degree. Finally, GNPs are very well characterized in the literature are well known for the ease and flexibility of their synt hesis, excellent size and shape control, and amenity to surface modification with thiols, amines and phosphines for dispe rsion in polymer environments.
101 Figure 6.1: National Institute of Standards and Technology elemental mass attenuation data for compared to traditional contrast agents may allow GNP embedded biomaterials to be imaged at higher X ray energies, for which the radiation dos e to human tissue is minimized. To our knowledge, GNPs have not been incorporated into polymeric biomaterials for the purpose of enhancing or modifying imaging properties with respect to the fo ur major medical imaging modalities: digital X ray/fl uoroscopy, CT, MRI and ultrasound. However, they have been used in free injectable/c olloidal form in this capacity. Precedent for Use of Gold Nanoparticles in X Ray Based Imaging of Biological T issues O ne of the most promising and unique areas of GNP research involves the development of novel X ray and contrast agents. Relatively few advancements in contrast technology for 2D plain film or digital radiography, fluoroscopy, mammography or computed tomography have been translated to the clinic over the last 25 years. Even so, traditional iodine based contrast agents are limited by their relatively h igh rate of clearance (limiting imaging time), patient specific allergies, occasional renal toxicity, and poor contrast and visibility, particularly in large patients. 137 138 In theory, even non functionalized GNPs offer solutions to these dr awbacks. First, they are
102 known to be cleared more slowly than traditional iodine based contrast agents, allowing for longer imaging times. 138 139 They are biologically inert, highly biocompatible and non immunogenic. 133 Lastly, the high atomic mass of gold (~196.97 amu) allows it to absorb and scatter electromagnetic radiation in the X ray range to a greater extent than iodine (~126.90 amu). Hainfield et al. imaged mice injected with GNPs using a mammography unit. 23 They found that vessels less than 100 m were discernable and certain organs, including the kidneys, were highlighted, while others, including the liver and spleen showed low levels of retention. With regard to the biocompatibility, no histological or behavioral evidence of toxicity at 30 post injection was present. Zhang et al. investigated the use of 15 40 nm glutamic ac id functionalized GNPs as a targeted X ray contrast agent for micro damaged bone. 2 7 It is known that fluorochromes with carboxylate functional groups such as calcein target cracks in cortical bone by chelating calcium ions on the surfaces of exposed bone mineral crystals. Glutamic acid has a primary amine opposite carboxylic acid gro ups, so it was a natural candidate for a GNP functional group with this application in mind. The authors found that the functionalized GNPs had high colloidal stability and monodispersion. Localization to artificially induced surface cracks was evident by their grossly red appearance and atomic force microscopy and backscattered electron imaging revealed t he presence of individual GNPs. Simply put, the objective of this study was to study the influence of study the influence of GNP incorporation on SMP radi opacity, CT and MRI artifacts, and ultrasound scattering, in a concentration dependent manner. Materials and Methods Sample P reparation Preparation of Gold N anoparticles As with other metal nanoparticles, GNPs are synthesized chemically by reacting a metal salt precursor with a reducing agent at elevated temperature. To ensure that gold metal forms in
103 distinct particles, this reaction also requires a stabilizing component, such as a polymer, which acts as a temporary surface ligand, preventing aggre gation of particles and regulating their size. In general, higher initial gold salt concentrations, stronger or more concentrated reducing agents, higher capping ligand concentration, and shorter reaction times all yield smaller particles. Once formed, the temporary surface passivating agent can be replaced by reacting the GNPs with any molecule terminated by a chemical group with higher affinity for the gold surface. These groups include thiols, amines and phosphines. Ultimately, the GNPs take on the chara cter (e.g. hydrophobicity) of this secondary surface ligand. In this study, GNP preparation followed a modified version of the protocol developed by Carotenuto and Nicolais. 123 Tetrachloroaurate trihydrate (HAuCl 4 3H 2 O), ethylene glycol (EG), polyvinylpyrrolidone (PVP, MW n 10,000), and dodecanethiol (DDT), were selected as the gold salt, reducing agent, capping agent, and final surface ligand, respectively. The reaction setup consisted of a 250 ml round bottom boiling flask, equipped with a condenser and incubated in a temp erature controlled paraffin oil bath. Before starting, all glassware to come in contact with GNPs was incubated with aqua regia (1:3 volumetric mixture of nitric and hydrochloric acids),in order to dissolve remnant colloidal gold and other trace metals, ri nsed in distilled water, and baked at 100C for 3 hours. In the first step, a solution of 200 mg/ml HAuCl 4 3H 2 O in EG was rapidly injected into 100 ml of 30% (w/v) PVP in EG stabilized at 90C in the boiling flask. Note that preparation of the PVP/EG solut ion required heat and stirring over several hours and that glass spatulas were used to handle HAuCl 4 3H 2 O, due to its corrosive nature. This reaction proceeded for 10 minutes and was characterized by a change in the solution color from yellow, to clear and finally, deep red/purple. The reaction mixture was then cast into 300 ml of chilled acetone and sonicated for 2 minutes in order to terminate the reaction and separate the nanoparticles. A gold PVP pellet was collected by sonication at 1000 x g for 15 mi n, washed subsequently dissolved into 100 ml of 50 mM DDT in ethanol and incubated at room temperature, with agitation, for 1 hour, facilitating surface ligand exchange. Once again, the product was
104 collected by three cycles of centrifugation and washing with chilled ethanol. The final DDT functionalized gold nanoparticle pellet was dried under vacuum and stored at room temperature. Preparation of Shape Memory Polymer Gold N anocomposites Plain SMP and SMP GNP composite samples used in this study were generally prepared by casting a mixture of a monofunctional acrylic monomer, difunctional acrylic oligomer, thermal initiator, and isolated GNPs, into polypropylene centrifuge t ubes. Specifically, SMPs consisted of 75% (w/w) tert Butyl acrylate monomer (tBA, Sigma Aldrich, St. Louis, MO) and 25% (w/w) poly(ethylene glycol) dimethacrylate (PEGDMA, 550 M n Sigma Aldrich, St. Louis, MO). Note that tBA forms the backbone of the polym er, while PEGDMA serves as the cross linker. Dispersion of the GNPs and polymerization were achieved by sonicating the mixtures in an ice water bath for 2 hours, after which the temperature of the bath was ramped up to 70C. The time required for polymeriz ation following the temperature ramp depended on the type and concentration of Azobis(2 methylpropionitrile) (AIBN, Sigma Aldrich, St. Louis, MO) and reached adequate conversion as measur ed by Fourier transform infrared spectroscopy, after 5 hours. GNP content in the SMPs varied over a wide range: 0, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10, 15, 25, and 30% (w/w). All composite samples had a final volume of 0.5 ml, neglecting minor shrinkage stres ses during polymerization. The composite samples were left in polypropylene centrifuge tubes for all imaging experiments except ultrasound (where the presence of external material would obscure the results). In experiments where the centrifuge tubes were r etained, the remaining volume was filled with 0.9% (w/v) saline. Preparation of Metal Control S amples Stainless steel and Nitinol (~50:50 nickel:titanium) samples were used as controls in all imaging experiments. The stainless steel sample consisted of a 4 mm diameter 316L stainless steel rod (McMaster Carr, Elmhurst, IL). Its dimensions were selected to provide the same vertical transmission path length as the composite samples when suspended in the imaging phantom. The Nitinol sample consisted of 22 gaug e wire (Nitinol Devices & Components, Inc., Fremont,
105 CA) bundled to provide the same material path length as the composite polymer and stainless steel samples. Both metal samples were placed in saline filled centrifuge tubes to match the composites. Imagi ng E valuation Imaging Phantom D esign A simple phantom was designed and fabricated to suspend samples near the center of a 6 inch column of fluid using minimal support material. It consisted of an open acrylic polymer box with three anchor points along each interior wall at mid level. A 0.1 mm diameter nylon monofilament wire was threaded through each anchor point, producing a grid for sample suspension. Centrifuge tubes containing samples for imaging were strung along the nylon wire at regular intervals. A schematic of the imaging phantom can be seen in figure 6.2
106 Figure 6.2: (A) SolidWorks illustration of watertight acrylic phantom. (B) Completed phantom. (C) Close up view of 0.1 mm monofilament nylon wire used to suspend samples for imaging. Image A cquisition Fluoroscopic and multi slice helical CT imaging was performed using a Swissray ddR Trauma System fluoroscope and Siemens Somatom Definition Flash scanner, respectively. For both of these modalities, the imaging phantom was filled with normal saline (0.9% w /v NaCl in deionized water). Fluoroscopy images were acquired at 62 kVp and 80 mA, using a 2 sec exposure. CT images were acquired at 120 kV and 300 mA, using a 3 mm slice thickness. MR images were A B C
107 acquired using a Signa HDxt 3.0 T scanner and the phantom was filled with normal saline containing 5 mM gadolinium chelate contrast (Omniscan, Amersham Health, Princeton, NJ). The paramagnetic contrast was used to reduce the T1 relaxat ion time to appropriate levels. Image P rocessing All DICOM images were viewed a nd exported to other formats using MicroDicom software (MicroDicom DICOM Viewer, Sofia, Bulgaria). Window/level values were assigned automatically and no further image processing was performed prior to analysis. However, densitometry analysis of fluoroscop y data was performed using ImageJ software (ImageJ Image Processing and Analysis in Java, National Institutes of Health, Baltimore, MD). Additionally, CT data was used to reconstruct geometric surface meshes for 5.2 and 26 cm iSMP coils. Briefly, a manual segmentation was performed with ITK Snap software using a single pixel brush tool. Post processing/mesh refinement was performed using MeshLab software and included minimal Gaussian filtering an d removal of spurious elements. Results Fluoroscopy One aim of this study was to assess the radiopacity of SMP gold nanocomposites as a function of gold content and compare it to that of metals commonly used to fabricate implanted biomedical devices. It was not expected that any of the nanocomposite samples would app roach the radiodensity of pure metals. Rather, our purpose was to determine if the amount of colloidal gold that could be incorporated into SMPs using the method described here, provided sufficient contrast for visualization against a background mimicking results would allow for direct comparison between radiopacity and artifact generation. Briefly, nanocomposites with GNP weight percentages varying from 0 to 3% were polymerized in 1.5 ml polypropylene centrifuge tubes. Following polymerization, the remaining volume in the tubes was filled pantom. A
10 8 stainless steel rod and bundled 22 gauge Nitinol wire, suspended similarly in centrifuge tub es, were used as positive controls. Static images were acquired at varying accelerating voltages and the results are displayed i n figure 6.3 The radiopacity of each sample was quantified using densitometry analysis in ImageJ software. First, the images we re inverted and the average pixel intensity was derived from a 100 pixel region of interest for each sample. Then, after subtracting background intensity values from each sample measurement, the results were normalized to the intensity of the stainless ste el sample. T hese data are shown in Table 6.1
109 Figure 6.3: Fluoroscopic images of 0.5 ml shape memory polymer gold nanocomposite samples contained in polypropylene centrifuge tubes and suspended in a normal saline filled acrylic phantom. Colloidal gold content ranged from 0 to 3% by weight, as shown in subfigure A. 4 x 15 mm (D x L) stainless steel rod and bundle d 22 gauge Nitinol wire of equivalent volume were used as positive controls. The 100 pixel regions of interest defined for densitometry analysis (table 6. 1) are shown in subfigure B. The orientation/labeling of samples in subfigure A also applies to B, C, and D. Acquisition parameters for each image are as follows: (A) 60kVp, 120 mA, 1.4 mGy; (B) 81 kVp, 118 mA, 0.6 mGy; (C) 102 kVp, 79 mA, 0.3 mGy; (D) 125 kVp, 40, 0.1 mGy. GNP: gold nanoparticle. A B C D 0.05% 0.1% 0.2% 0.5% 1% 1.5% 2% 2.5% 3% Stainless steel Nitinol 0% (w/w) GNP 60 kVp 81 kVp 102 kVp 125 kVp
110 Table 6.1: Relative nanocomposite radiopacity was calcula ted by subtracting the background intensity value from each sample intensity value, then normalizing to the value for stainless steel. Normalized radiopacity Sample 60 kVp 81 kVp 102 kVp 125 kVp Stainless steel 1.000 1.000 1.000 1.000 Nitinol 1.000 0.960 0.892 0.986 SMP + 0 wt% GNP 0.001 0.012 0.01 0.004 SMP + 0.05 wt% GNP 0.004 0.000 0.009 0.001 SMP + 0.1 wt% GNP 0.012 0.020 0.004 0.006 SMP + 0.2 wt% GNP 0.027 0.021 0.012 0.018 SMP + 0.5 wt% GNP 0.039 0.029 0.022 0.035 SMP + 1 wt% GNP 0.075 0.053 0.024 0.060 SMP + 1.5 wt% GNP 0.146 0.092 0.042 0.098 SMP + 2 wt% GNP 0.279 0.139 0.069 0.116 SMP + 2.5 wt% GNP 0.307 0.181 0.105 0.148 SMP + 3 wt% GNP 0.421 0.233 0.150 0.164 SMP: shape memory polymer GNP: gold nanoparticle As expected, when the accelerating voltage was raised, X ray transmission through all samples and the background saline column increased significantly, resulting in lower absolute densitometry values. The normalized densitometry values for the SMP gold nanocomposite sa mples were generally higher at lower accelerating voltages. However, this likely reflects the fact that at 60 and even 81 kVp, the contrast from the metal samples was saturated and resulted in the maximum average pixel intensity being achieved (on a scale of 0 255 for 8 bit grayscale images). For instance, for SMPs containing 3% gold by weight, normalized radiopacity values of 0.150 (for 102 kVp) and 0.164 (for 125 kVp) are likely more representative. Stainless steel and Nitinol samples demonstrated broa dly similar radiopacity values.
111 Computed Tomography In contrast to the fluoroscopic imaging study, CT and MRI studies were mainly performed to determine if the nanocomposite samples generated significant artifacts. It is well known that implanted metal devices can generate significant artifacts during CT im aging and obscure proximal anatomical structures. These artifacts often manifest as streaking across the image and can be due to a combination of beam hardening, photon starvation, undersampling, motion, partial volume averaging, and Compton scatter. Accor dingly, transverse CT images of both nanocomposite and metal samples samples were acquired using a helical multi slice scanner. Two imaging regimens were selected. The first used traditional acquisition parameters and a kVp maximum distortion from both imaging regimens were selected for display in figures 6. 4 and 6. 5, respectively. Note that the same imaging phantom used during fluoroscopic evaluation was used here.
112 Figure 6.4: Transverse computed tomography slices of nanocomposite and metal samples in a 0.9% (w/v) saline environment. (A) Minor distortions from shape memory polymer (SMP) gold nanocom posite materials with up to 0.5% (w/w) gold nanoparticle incorporation. Acquisition parameters: 120 kVp, 316 mA. (B) Moderate distortions from SMP gold nanocomposite materials with 1 3% (w/v) gold nanoparticle incorporation. Acquisition parameters: 120 kVp 251 mA. (C) Significant beam hardening and photon starvation artifacts from stainless steel and Nitinol samples. GNP: gold nanoparticle. A B C 0.05% 0.1% 0.2% 0.5% 0% (w/w) GNP 1% 1.5% 2% 2.5% 3% Stainless steel Nitinol
113 As evidenced by figures 6. 4 and 6. 5, regardless of whether a high or low kVp imaging protocol was used, the stainless steel and Nitinol samples generated abundant streaking artifacts which covered the entire phantom volume. It is expected that these artifacts will be even more significant for reconstructions along the z axis (parallel to the gantry) due to the conventional use of anisotropic voxels having a lower out of plane resolution, particularly for older scanners. In comparison, while mild to moderate distortions were seen for nanoco mposite samples containing higher amounts of gold when using a traditional imaging protocol, no artifacts were seen when Figure 6.5: computed tomography imaging of nanocomposite and metal samples in a 0.9% (w/v) saline environment, utilizing lower accelerating voltage. (A,B) No visible distortions from SMP gold nanocomposite materials with up to 3% (w/w) gold nanoparticle incorporation. Acquisition parameters: subfigure A 80 kVp, 499 mA; subfigure B 80 kVp, 404 mA. (C) Significant beam hardening and photon starvation arti facts from stainless steel and Nitinol samples remain. Acquisition parameters: 80 kVp, 516 mA. Results from low kVp and normal (figure 4) scans are comparable for metal samples. GNP: gold nanoparticle. A B C 0.05% 0.1% 0.2% 0.5% 0% (w/w) GNP 1% 1.5% 2% 2.5% 3% Stainless steel Nitinol
114 the kVp was lowered. Moreover, the distortions from the 3% (w/w) GNP sample seen with the high kVp scan were not much larger than thos e generated by the 0% (w/w) sample. In short the gold content used in this study appears to be insufficient to cause significant beam hardening and photo n starvation during CT imaging. Magnetic Resonance Imaging Figures 6.6 and 6. 7 show T1 T2 and proto n density weighted MR images of a gadolinium doped saline phantom containing SMP gold nanocomposite and Nitinol samples. The stainless steel sample was not ultimately used as a control during MR imaging as it was strongly attracted by the magnet, preventin g precise placement in the phantom. Moreover, during initial scans, the artifact arising from its ferromagnetism obscured nearly the entire phantom volume and prevented visualization of other samples. Figure 6 .6 displays the resul ts from a T2 weighted FSE scan with a 40 mm slice thickness. A large slice thickness was chosen to better display the extent of any artifacts more completely along the vertical axis. Since the composite and Nitinol samples were suspended at different vertical positions, two slices were cropped and aligned to show both in the same field. Figure 6. 7 displays similar results for 5 mm slices. Taken together, the data indicate that the addition of GNPs does not affect the appearance of SMPs for any of the pulse sequences. In contrast, th e Nitinol sample produced a noticeable ringing artifact for all sequences. The distortion was associated with sharp boundaries in magnetic susceptibility. Materials like Nitinol, with significant magnetic susceptibility, distort the linear magnetic field g radients and generate large signal voids. That said, artifact size from the Nitinol coil was significantly less than that seen with the stainless steel coil, supporting the findings of Meyer et al. 88
115 Figure 6 .6: High slice thickness T2 weighted fast spin echo MRI image of 0.5 ml SMP gold nanocomposite materials and Nitinol wire bundle contained in polypropylene centrifuge tubes and contrast. Image acquisition parameters echo time: 64.896 ms, repetition time : 3000 ms, flip angle: 90, slice thickness: 40 mm, echo train length : 8, pixel bandwidth: 244.141, averages: 1. Nanocomposite samples were artifact free and are visualized through negative contrast, while the Nitinol sample generated significant magnetic susceptibility artifact. A B 0.05% 0.1% 0.2% 0.5% 0% (w/w) GNP Nitinol 1% 1.5% 2% 2.5% 3% 1% 1.5% 2% 2.5% 3% 0.05% 0.1% 0.2% 0.5% 0% (w/w) GNP Nitinol
116 Figure 6. 7. Low slice thickness magnetic resonance images of shape memory polymer (SMP) gold nanocomposite materials, containing varying concentrations of colloidal gold, and bundled Nitinol wire sample. All samples were contained in polypropylene centrifuge tubes and suspended olinium chelate contrast. Common MRI pulse sequences reveal significant magnetic susceptibility artifacts from Nitinol, while SMP gold nanocomposite samples appear to be artifact free. The orientation/labeling of samples in subfigure A also applies to subf igures B F. (A B ) T1 weighted spin echo sequence. Image acquisition parameters echo time ( TE ): 11 ms, repetition time (TR) : 700 ms, flip angle: 90, slice thickness: 5 mm, echo train length (ETL) : 1, pixel bandwidth: 195.312, averages: 1. (C D ) T2 weight ed fast spin echo sequence. Image acquisition parameters TE: 64.896 ms, TR: 3000 ms, flip angle: 90, slice thickness: 5 mm, ETL: 8, pixel bandwidth: 195.312, averages: 1. (E F ) Spin density (proton) weighted gradient recalled echo sequence. Image acquis ition parameters TE: 11 ms, TR: 2000 ms, flip angle: 90, slice thickness: 5 mm, ETL: 1, pixel b andwidth: 195.312, averages: 1. With regard to torque, displacement, and heating during MR imaging, only the stainless steel sample displayed obvious motion i C D E F
117 initial orientation in the imaging phantom, it spun on the nylon wire to allow for alignment of the induced and permanent magnetic fields. No significant heating of any coil, secondary to curre nt induction, was detected before or directly following imaging. This was measured by placing a thermocouple in the centrifuge tube before and after running each scan. Discussion Over their lon g history in the laboratory, GNP s have developed into an excell ent platform for both imaging and biomolecular delivery. Their unique core and surface properties include excellent X ray attenuation, facile modification with surface ligands through physical adsorption, tunable surface plasmon resonance and heat dissipat ion, and biological inertness. These properties have been extensively exploited in the laboratory setting, but few medical devices, treatments or diagnostic procedures that employ GNPs or their conjugates are currently approved by the FDA This reflects th e fact that several critical issues need to be addressed, such as the development of reliable and reproducible manufacturing methods, the need for further studies on the long term health effects of nanomaterial administration, and individual variations in immunogenic response to functionalized GNPs. In this study, we evaluated the imaging properties of shape memory polymers containing varying concentrations of gold nanoparticles. Preliminary findings indicate that, at least up to 3% (w/w) incorporation, GNP s may be a suitable contrast agent in polymeric biomaterials for multiple medical imaging modalities. It confers useable X ray contrast for visualization starting from 1 wt% and does not generate significant CT or MRI artifacts. Small distortions seen duri ng CT imaging can be eliminated by modifying the acquisition parameters, without altering sample visibility. They also illustrate that one of the primary benefits of polymeric biomaterials is the ability to control the concentration of contrast agent that they incorporate, allowing radiopacity to be balan ced with distortions/artifacts. Computed Tomography Imaging lesser extent, the iSMP coils) during CT imaging are largely due to beam hardening and photon
118 starvation. Beam hardening is the process through which the a verage energy of the polyenergetic incident X ray beam is increased as it passes through radiodense samples. This happens because low energy X rays are disproportionately absorbed compared to high energy X rays. The hardened beam is subsequently attenuated to a lesser extent, so the detector reads a higher signal than would otherwise be expected. Photon starvation is often seen with metal implants because the density of the metal is beyond the normal range that can be handled by the hardware, resulting in i ncomplete profiles. Moreover, due to the asymmetric shape of devices being imaged, incident beams from different directions are blocked to different extents, which can confuse the reconstruction algorithm and cause artifacts that appear as dark streaks (sh adows) along the long axis of the device. Accordingly, while CT artifacts can be attenuated through filtration or by modifying the tube current, slice thickness etc., they are largely unavoidable when imaging pure metal devices. One of the greatest benefit s of polymeric coils, then, is the ability to vary the am ount of opacifier they contain. In general, artifacts are especially problematic for the C arm cone beam CT systems typically used in interventional procedures because of increased motion artifacts and scatter. 91 A traditional multi slice helical CT scanner was used in this st udy, so the artifacts show n here may be underestimations. Magnetic Resonance Imaging Potential sources of artifacts in MR imaging are numerous, but those arising from the intrinsic material properties of the sample being imaged broadly fall under the categories of magnetic susceptibility or chemical shift artifacts. Susceptibility artifact s in MRI occur at interfaces of differing magnetic susceptibilities, such as tissue and air or between materials in different magnetic classes (diamagnetic, paramagnetic, superparamagnetic, and ferromagnetic). These differences in susceptibilities lead to a distortion in the local magnetic environment, causing dephasing of spins, with signal loss, and mismapping/artifacts. The principle behind the chemical shift artifact is that the protons from different molecules precess at slightly different frequencies.
119 This is due to to the fact that protons in different electronic environments are shielded to a lesser or greater degree from the main magne tic field. Nitinol consists approximately of equal parts nickel and titanium. Nickel is ferromagnetic, while titaniu m is paramagnetic. Since polymers are composed primarily of carbon, hydrogen, and oxygen, they are diamagnetic like water. Gold (element 79) has the electronic configuration [Xe] 6s1 4f14 5d10, so only one electron (in the 6S orbital) is unpaired. As a res ult, nanocomposite samples were visible through the same mechanism as the neat SMP negative contrast. In order to modify the MR contrast of gold nanoparticles, they may have to be combined with gadolinium (paramagnetic) or iron oxide (superparamagnetic) nanoparticles. Some work has been done in this field. Traditional MRI contrast agents include gadolinium, manganese and iron oxide. These elements are paramagnetic or ferromagnetic (super paramagnetic); they introduce external sources of magnetic field inh omogeneity, changing the relaxation properties of associated tissues and vascular compartments. 140 Certainly the relaxation enhancement effect in tissues is improved through delivery of a large quantity of the contrast agent to the target site. However, researchers have found that the limiting factors in delivering a large number of Gd 3+ ions to a desired tissue are not physical rules, but the challenges inherent in synthesizing complexed contrast compounds with appropriate biological and chemical properties. 141 Morig gi et al. theorized that nanocrystals made of noble metals, including gold and silver, would be ideal cores for loading of Gd 3+ 141 GNPs in particular are easily functionalized with thiol derivatives, which allows for coating with gadolinium chelate thiol de rivatives. This group also predicted that the known magnetic behavior of thiol covered GNPs might contribute to the relaxivity of bulk water molecules, separately from chelated gadolinium. In their study, the authors used a thiol derivative DTTA ( dipalmito ylphosphatidylethanolamine ) chelate as a protective agent for gold nanoparticles. These thiol coated GNPs were then modified with gadolinium or yttrium (diamagnetic control). The functionalized nanoconjugates showed very high relaxivities, though magnetic susceptibility measurements at high magnetic field using H 1 NMR revealed the absence of a significant magnetic contribution from the gold core. Figure 4 illustrates the use of gadolinium functionalized GNPs as an MRI contrast agent. Similarly, Sun et al. d emonstrated the utility of a noncovalent
120 method of attaching Gd3+ to gold nanorods. 142 They theorized that the noncovalent bond would increase the accessibility of Gd3+ ions to water molecules, resulting in significant longitudinal relaxivity, and that gold nanorods would present more gadolinium binding sites when compar ed to their spherical counterparts (high surface to volume ratio). In fact, this group observed longitudinal relaxivities over an order of magnitude greater than those observed for spherical GNP gadolinium chelates.
121 CHAPTER VII CONCLUSIONS & FUTURE DIRECTIONS Improving the GNP Surface Brush In this study, GNP surfaces were functionalized with dodecanethiol, which was shown both to bind strongly to the gold surface and confer significant hydrophobicity to the nanoparticle surface, allowing adeq uate dispersion in relatively hydrophobic acrylate SMP environments. However, contact angle measurements, displayed in Chapter IV, revealed that the functionalized GNP surfaces were ultimately dissimilar to those of polymerized SMPs, suggesting that there is room to improve the interaction between the continuous and discontinuous phases of the composite. This is worth investigating, as one of the major limitations of the SMP GNP composite studied here was the level of incorporable GNPs. Contact angle measur ements used water and diiodomethane droplets to distinguish the unique contributions of polar and dispersive forces to the overall surface free energy of GNPs and SMPs. Specifically, they demonstrated that the GNP surface had a higher contribution from dis persive forces, indicati ng that it is more hydrophobic than the SMP Further efforts in this area might involve passivating the GNP surface with amphip hilic molecules, which may match the surface energy profile of SMPs more closely and ultimately show impr oved solubility Almost any molecular payload (oligomer) tethered to a thiol, amine or phosphine should bind to GNP surfaces. Another option may be to introduce a mixture of oligomers (e.g. hydrophobic and hydrophilic) Specific options include other alkan ethiols, methyl PEG thiol compounds, 11 Mercaptoundecyl tetra(ethylene glycol), 1 mercapto (triethylene glycol) methyl ether, as well as variations of these ligands with fewer or more repeating units In other words, heterogeneity in ligand molecular weigh t could also be explored, with the goal of improving GNP dispersion and incorporable mass. Attention will have to be paid to the molecular weight cut off of the surface modifiers, as large molecular weight tags may strongly influence the physical propertie s of the nanoparticles and the resulting composite material. In addition, it is anticipated
122 that harvesting am phiphilic nanoparticles will be challenging, as they are solubility in a very wide array of solvents. High speed centrifugation, rather than selec tive precipitation, may be a preferable alternative. Photo Thermal Control of SMP GNP Composites One of the most unique and promising aspects of SMP GNP systems is the putative ability to indirectly heat the material by exposing it to a wavelength of monochromatic light that is efficiently absorbed the nanoparticles. This could allow for the control o f thermal transitions in the polymer (and corresponding physical and chemical properties) using light, eliminating the traditional dependence of SMPs on ambient temperature changes. In the context of TCDs, this could translate to the ability to control sha pe recovery and device release independent of body temperature Photo activation with green laser light has been shown in non acrylate SMPs (Zhang et al.), but no biomedical applications were discussed. A fiber optic catheter carrying green light could provide precise spatial and temporal control of shape recovery and device release or recovery. In addition, heat generation by GNPs upon exposure to green laser light may also allow for control of material polymerization in the presence of thermal initiato rs, which has not been shown in the literature. Preliminary, qualitative studies were undertaken to evaluate the potential for controlling glass transitions in SMP GNP composites using green laser light (figure 7.1).
123 Figure 7.1: Shape recovery o f SMP GNP composite strip containing 0.5 wt% GNPs after temporary deformation (rolling) upon illumination by 1 00 mW, 532 nm solid state laser Elapsed time ~20 sec. The e ffect shown in the above figure has also been demonstrated for relatively low GNP con tent and laser power (0 .1 wt% and 10 mW), respectively. In comparison, and as shown by UV Vis absorption spectra in Chapter IV SMPs that do not contain any GNPs are transparent to 532 nm light and do not show any heating. Further studies will model heat g eneration in SMP GNP composites and validate the models with optical surface pyrometry measurements. Ultimately, the goal of computational modeling is to identify the practical limitations of controlling device deployment and behavior with lasers. It has implications for device geometry, laser power requirements, exposure requirements, and the risk of dam age to endovascular tissue. In order to simplify the model, assumptions will have to be made. These may include assuming there is convective and radiative heat transfer at boundaries, a uniform laser beam profile, a stationary
124 (non rastering) laser source, no radiative heat transfer within the material, no dependence of the K term on temperature and pressure, no scattering, and no surface reflection of laser energy. Critically, one may also have to assume that the material itself is isotropic with respect t o thermal prope rties, a reasonable assumption. Boundary conditions will also have to be considered. Due to convective heat transfer between the composite surface and flowing blood, it may be useful to model a heat sink at all boundaries. As the thermal con ductivity properties of blood are not well studied, water may be a good substitute. Other important considerations will be whether the model should be 1D, 2D or 3D (the 2D model can be used to study processes within the plane transverse or parallel to the laser propagation direction), whether boundary conditions should be used in place of a semi infinite model, the mechanism of heat transfer at boundaries, and the practical ability to create a matching physical set up for validation. Summary of Study Limita tions and Conclusions Though specific limitations are discussed in each chapter, one of the primary and overarching limitations of this study is that it evaluates only a limited number of the modifiable nanoparticle and pol ymer variables As a result, it w ill probably underestimate the true range in final composite properties that may be achievable through similar methods. In addition, it is unlikely that all of the optimal properties for an individual TCD application will be found at a single GNP concentra tion. Instead, the results should be compared to current designs for superiority with regard to the most important material characteristics and cost. In some cases, radio opacity will have to be balanced with desired mechanical properties. Table 7.1 lists additional variables which may be considered in the future.
125 Table 7.1: Modifiable nanoparticle and polymer variables and their expected effect on final composite material properties. Variable Examples Composite property affected Nanoparticle Concentration 0 100 mg/ml range Radio opacity, MRI/CT artifact, Tg, shrinkage stress, strain recovery, rubbery and glassy moduli Size and aspect ratio 5 100 nm diameter range, spherical vs. rod shaped Optical absorption properties, concentration, monodispersity Surface modifier A lkanethiols, methyl PEG thiol compounds, 11 Mercaptoundecyl tetra(ethylene glycol), 1 mercapto (triethylene glycol) methyl ether Stability, concentration limit, monodispersity Method of nanoparticle incorporation Ex situ production and incorporation, In situ reduction of metal salts, pyrolysis of metal containing precursors Concentration limit, monodispersity Polymer Type/functionality and stoichiometry of monomers Acrylic acids, acrylic salts, methacrylates, metal chelating side groups Tg, strain recovery, rubbery and glassy moduli Cross linking monomer molecular weight and density Tg, strain recovery, rubbery and glassy moduli Polymerization method UV irradiation, thermal, redox Concentration limit, tensile strength, Tg, rubbery and glassy moduli Another limitation related to the system itself. As discussed earlier, i mportant limitations of the (meth)acrylate based SMP systems include the formation of a heterogeneous polymer network a relatively broad glassy to rubbery transition, and the inhibition of polymerization by oxygen. Thiol ene shape memory polymers may be of interest in future studies, especially since the thiol gold interaction may be harnessed to establish composite networ ks with covalently incorporated GNPs. While the nanocomposite material generated in this study has limitations, it shows promise as an alternative for pure metals in the design of low cost TCDs. The addition of GNPs to acrylate SMP networks in this study c onferred adequate radiopacity at or above 1 wt%. Critically, their addition did not introduce significant MRI or CT artifacts, a significant drawback of conventional metal devices. In light of the thermo mechanical results displayed in Chapter V, the addit ion of GNPs does not appear to disturb important shape recovery characteristics and indeed improves free strain recovery Ultimately, GNPs incorporation represents an additional variable with which polymer/composite thermo mechanical properties can be cont rolled.
126 REFERENCES 1. Sokolowski W, Metcalfe A, Hayashi S, Yahia L, Raymond J. Medical applications of shape memory polymers. Biomed Mater 2007;2(1):S23 7. doi:10.1088/1748 6041/2/1/S04. 2. Yakacki CM, Lyons MB, Rech B, Gall K, Shandas R. Cytotoxicity and thermomechanical behavior of biomedical shape memory polymer networks post sterilization. Biomed Mater 2008;3(1):015010. doi:10.1088/1748 6041/3/1/015010. 3. Small W, Singhal P, Wilson TS, Maitland DJ. Biomedical applications of thermally activated shape memory polymers. J Mater Chem 2010;20(18):3356 3366. doi:10.1039/B923717H. 4. Iribarne A, Easterwood R, Chan EYH, et al. The golden age of minimally invasive cardiothoracic surgery: current and future perspectives. Future Cardiol 2012;7(3):333 346. doi:10.2217/fca.11.23.The. 5. Tan IYL, Agid RF, Willinsky R a. Recanalization rates after endovascular coil embolization in a cohort of matched ruptured and unrup tured cerebral aneurysms. Interv Neuroradiol 2011;17(1):27 35. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3278032&tool=pmcentrez&rend ertype=abstract. 6. Hampikian J, Heaton B, Tong F, Zhang Z, Wong CP. Mechanical and radiogra phic properties of a shape memory polymer composite for intracranial aneurysm coils. Mater Sci Eng C 2006;26(8):1373 1379. 7. Baer GM, Wilson TS, Small W, et al. Thermomechanical properties, collapse pressure, and expansion of shape memory polymer neurov ascular stent prototypes. J Biomed Mater Res B Appl Biomater 2009;90(1):421 9. doi:10.1002/jbm.b.31301. 8. Yakacki CM, Shandas R, Lanning C, Rech B, Eckstein A, Gall K. Unconstrained recovery characterization of shape memory polymer networks for cardiova scular applications. Biomaterials 2007;28(14):2255 63. doi:10.1016/j.biomaterials.2007.01.030. 9. Yakacki CM, Willis S, Luders C, Gall K. Deformation Limits in Shape Memory Polymers. Adv Eng Mater 2008;10(1 2):112 119. doi:10.1002/adem.200700184. 10. N air DP, Cramer NB, Scott TF, Bowman CN, Shandas R. Photopolymerized Thiol Ene Systems as Shape Memory Polymers. Polymer (Guildf) 2010;51(19):4383 4389. doi:10.1016/j.polymer.2010.07.027. 11. Heaton B (Georgia IOT. A Shape Memory Polymer for Intracranial Aneurysm Coils: An Investigation of Mechanical and Radiographic Properties of a Tantalum Filled Shape Memory Polymer Composite. 2004. 12. Baer GM, Small W, Wilson TS, et al. Fabrication and in vitro deployment of a laser activated shape memory polymer vas cular stent. Biomed Eng Online 2007;6:43. doi:10.1186/1475 925X 6 43. 13. Spahn M. Flat detectors and their clinical applications. Eur Radiol 2005;15(9):1934 47. doi:10.1007/s00330 005 2734 9.
127 14. Ghaye B, Dondelinger RF. Imaging guided thoracic interv entions. Eur Respir J 2001:507 528. 15. Salamone J. Radiopaque Polymers. Polym Mater Encycl Q S 1996:7346 7350. 16. Moszner N, Salz U. New Developments of Polymeric Dental Composites. Prog Polym Sci 2001;26(1):535 576. 17. Behl M, Razzaq MY, Lendlein A. Multifunctional shape memory polymers. Adv Mater 2010;22(31):3388 410. doi:10.1002/adma.200904447. 18. Bohner M. Design of ceramic based cements and putties for bone graft substitution. Eur Cell Mater 2010;20:1 12. Available at: http://www.ncbi.nlm. nih.gov/pubmed/20574942. 19. Lye KW, Tideman H, Merkx M a W, Jansen J a. Bone cements and their potential use in a mandibular endoprosthesis. Tissue Eng Part B Rev 2009;15(4):485 96. doi:10.1089/ten.TEB.2009.0139. 20. Davy KW, Anseau MR, Berry C. Iodina ted methacrylate copolymers as X ray opaque denture base acrylics. J Dent 1997;25(6):499 505. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9604581. 21. Coursey JS, Schwab DJ, Tsai JJ, Dragoset RA. Atomic weights and isotopic compositions. NIST Phys M eas Lab 2010. Available at: http://www.nist.gov/pml/data/comp.cfm. Accessed November 24, 2011. 22. Hubbell JH, Seltzer SM. Tables of X ray mass attenuation coefficients and mass energy absorption coefficients. NIST Phys Meas Lab 2010. Available at: http://www.nist.gov/pml/data/xraycoef/index.cfm. Accessed November 24, 2011. 23. Hainfeld JF, Slatkin DN, Focella TM, Smilowitz HM. Gold nanoparticles: a new X ray contrast agent. Br J Radiol 2006;79:248 253. 24. Balazs AC, Emrick T, Russell TP. Nanopar ticle polymer composites: where two small worlds meet. Science 2006;314(5802):1107 10. doi:10.1126/science.1130557. 25. Park JH, Lim YT, Park OO, Kim JK, Yu J W, Kim YC. Polymer/Gold Nanoparticle Nanocomposite Light Emitting Diodes: Enhancement of Elect roluminescence Stability and Quantum Efficiency of Blue Light Emitting Polymers. Chem Mater 2004;16(4):688 692. doi:10.1021/cm0304142. 26. Alric C, Taleb J, Duc G Le, et al. Gadolinium Chelate Coated Gold Nanoparticles as Contrast Agents for Both X ray C omputed Tomography and Magnetic Resonance Imaging. J Am Chem Soc 2008;130(13):5908 5915. 27. Zhang Z, Ross RD, Roeder RK. Preparation of functionalized gold nanoparticles as a targeted X ray contrast agent for damaged bone tissue. Nanoscale 2010;2(4):58 2 6. doi:10.1039/b9nr00317g. 28. Giljohann D a, Seferos DS, Daniel WL, Massich MD, Patel PC, Mirkin C a. Gold nanoparticles for biology and medicine. Angew Chem Int Ed Engl 2010;49(19):3280 94. doi:10.1002/anie.200904359.
128 29. Pissuwan D, Niidome T, Cort ie MB. The forthcoming applications of gold nanoparticles in drug and gene delivery systems. J Control Release 2009;149(1):65 71. doi:10.1016/j.jconrel.2009.12.006. 30. Rotello VM. Drug and Gene Delivery using Gold Nanoparticles. Drug Deliv 2007:40 45. doi:10.1007/s. 31. Connor EE, Mwamuka J, Gole A, Murphy CJ, Wyatt MD. Gold nanoparticles are taken up by human cells but do not cause acute cytotoxicity. Small 2005;1(3):325 7. doi:10.1002/smll.200400093. 32. Zhang H, Xia H, Zhao Y. Optically triggered and spatially controllable shape memory polymer gold nanoparticle composite materials. J Mater Chem 2012;22(3):845 849. 33. Song L, Hu W, Zhang H, Wang G, Yang H, Zhu S. In vitro evaluation of chemically cross linked shape memory acrylate methacrylate copolymer networks as ocular implants. J Phys Chem B 2010;114(21):7172 7178. doi:10.1021/jp100876c. 34. Wilson TS, Bearinger JP, Herberg JL, et al. Shape memory polymers based on uniform aliphatic urethane networ ks. J Appl Polym Sci 2007;106(1):540 551. doi:10.1002/app.26593. 35. Haider N, Karlsson S. A rapid ultrasonic extraction technique to identify and quantify additives in poly(ethylene). Analyst 1999;124(5):797 800. doi:10.1039/a809025d. 36. Moraes LGP, Rocha RSF, Menegazzo LM, de Arajo EB, Yukimito K, Moraes JCS. Infrared spectroscopy: a tool for determination of the degree of conversion in dental composites. J Appl Oral Sci 2008;16(2):145 149. doi:10.1590/S1678 77572008000200012. 37. Odian G. Princip les of Polymerization Hoboken: John Wiley & Sons, Inc.; 2004:23 24. 38. Nagahama K, Ueda Y, Ouchi T, Ohya Y. Biodegradable shape memory polymers exhibiting sharp thermal transitions and controlled drug release. Biomacromolecules 2009;10(7):1789 94. doi: 10.1021/bm9002078. 39. Paderni K, Pandini S, Passera S, Pilati F, Toselli M, Messori M. Shape memory polymer networks from sol gel cross linked alkoxysilane caprolactone). J Mater Sci 2012;47(10):4354 4362. doi:10.1007/s10853 012 6289 2 40. Sanders JKM, Hunter BK. Modern NMR spectroscopy: a guide for chemists 2nd ed. Oxford, UK: Oxford University Press; 1993. 41. Keeler J. Understanding NMR Spectroscopy 2nd ed. Chichester, UK: John Wiley & Sons, Inc.; 2010. 42. Jacobsen NE. NMR spectroscopy explained: simplified theory, applications and examples for organic chemistry and structural biology Hoboken, NJ: Wiley Interscience; 2007. 43. Bertmer M, Buda A, Blomenkamp Hfges I, Kelch S, Lendlein A. Biodegradable Shape Memory Polymer N etworks: Characterization with Solid State NMR. Macromolecules 2005;38(9):3793 3799. doi:10.1021/ma0501489.
129 44. Liu C, Chun SB, Mather PT, Zheng L, Haley EH, Coughlin EB. Chemically Cross Linked Polycyclooctene: Synthesis, Characterization, and Shape Mem ory Behavior. Macromolecules 2002;35(27):9868 9874. doi:10.1021/ma021141j. 45. Nielsen LE, Landel RF. Mechanical Properties of Polymers and Composites 2nd ed. New York: Marcel Dekker, Inc.; 1994. 46. Hu M, Yu D, Wei J. Thermal conductivity determinatio n of small polymer samples by differential scanning calorimetry. Polym Test 2007;26(3):333 337. doi:10.1016/j.polymertesting.2006.11.003. 47. Haddadin R, Qian F, Desikan S, Hussain M, Smith RL. Estimation of drug solubility in polymers via differential s canning calorimetry and utilization of the fox equation. Pharm Dev Technol 2009;14(1):18 26. doi:10.1080/10837450802409370. 48. Saywer LC, Grubb DT, Meyers GF. Polymer Microscopy 3rd ed. New York: Springer; 2008. 49. Arima Y, Iwata H. Effect of wettabi lity and surface functional groups on protein adsorption and cell adhesion using well defined mixed self assembled monolayers. Biomaterials 2007;28(20):3074 3082. doi:10.1016/j.biomaterials.2007.03.013. 50. rface free energy of solids. J Achiev Mater Manuf Eng 2007;24(1):137 145. 51. Libera MR, Egerton RF. Advances in the Transmission Electron Microscopy of Polymers. Polym Rev 2010;50(3):321 339. doi:10.1080/15583724.2010.493256. 52. Michler GH. Electron Microscopy of Polymers Leipzig: Springer Verlag; 2008. 53. Lam GKY. Systematic method of formulating liquid phantoms with a given elemental composition and density. Med Phys 1981;8(6):894 896. 54. Marra KG, Szem JW, Kumta PN, DiMilla P a, Weiss LE. In vitro analysis of biodegradable polymer blend/hydroxyapatite composites for bone tissue engineering. J Biomed Mater Res 1999;47(3):324 335. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10487883. 55. Charbonneau C, Ruiz J C, Lequoy P, et al. Chondroit in sulfate and epidermal growth factor immobilization after plasma polymerization: a versatile anti apoptotic coating to promote healing around stent grafts. Macromol Biosci 2012;12(6):812 821. doi:10.1002/mabi.201100447. 56. Tu Q, Li L, Zhang Y, et al. The effect of acetylcholine like biomimetic polymers on neuronal growth. Biomaterials 2011;32(12):3253 3264. doi:10.1016/j.biomaterials.2011.01.044. 57. Breitenbach J, Schrof W, Neumann J. Confocal Raman spectroscopy: analytical approach to solid dispers ions and mapping of drugs. Pharm Res 1999;16(7):1109 1113. doi:10.1023/A:1018956304595. 58. Balss KM, Llanos G, Papandreou G, Maryanoff C a. Quantitative spatial distribution of sirolimus and polymers in drug eluting stents using confocal Raman microscop y. J Biomed Mater Res A 2008;85(1):258 270. doi:10.1002/jbm.a.31535.
130 59. Gianturco C, Anderson H, Wallace S. Mechanical devices for arterial occlusion. Am J Roentgenol 1975;124(3):428 435. 60. Arch Dis Child 1989;64(11):1612 1617. doi:10.1136/adc.64.11.1612. 61. Dowd CF, Halbach W, Higashida RT, Barnwell SL, Hieshima GB. Endovascular coil embolization of unusual posterior inferior cerebellar artery aneurysms. Neurosurgery 1990;27(6):954 961. 62. Higashida RT, Halbach W, Dowd CF, Barnwell SL, Hieshima GB. Interventional neurovascular treatment of a giant intracranial aneurysm using platinum microcoils. Surg Neurol 1991;35(1):64 68. 63. Hilal SK, Solomon RA. Endovascular treatment of aneurysm s with coils. J Neurosurg 1992;76(2):337 339. 64. Viuela F, Duckwiler G, Mawad M. Guglielmi detachable coil embolization of acute intracranial aneurysm: perioperative anatomical and clinical outcome in 403 patients. J Neurosurg 1997;86(3):475 482. doi: 10.3171/JNS/2008/108/4/0832. 65. Akiba Y, Murayama Y, Vinuela F, Lefkowitz MA, Duckwiler GA, Gobin YP. Balloon assisted Guglielmi detachable coiling of wide necked aneurysms: Part I -experimental evaluation. Neurosurgery 1999;45(3):519 527. 66. Lefkowit z MA, Gobin YP, Akiba Y, et al. Balloon assisted Guglielmi detachable coiling of wide necked aneurysms: Part II -clinical results. Neurosurgery 1999;45(3):531 537. 67. Irie K, Negoro M, Hayakawa M, Eayashi J, Kanno T. Stent Assisted Coil Embolization: th e Treatment of Wide necked, Dissecting, and Fusiform Aneurysms. Interv Neuroradiol 2003;9(3):255 61. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3548210&tool=pmcentrez&rend ertype=abstract. 68. Lee Y J, Kim DJ, Suh SH, Lee S K, Kim J, Kim DI. Stent assisted coil embolization of intracranial wide necked aneurysms. Neuroradiology 2005;47(9):680 9. doi:10.1007/s00234 005 1402 8. 69. Kang HS, Han MH, Kwon BJ, et al. Short term outcome of intracranial aneurysms treated with polygly colic acid/lactide copolymer coated coils compared to historical controls treated with bare platinum coils: a single center experience. Am J Neuroradiol 2005;26(8):1921 1928. 70. Roth C, Struffert T, Grunwald IQ, et al. Long term results with Matrix coil s vs. GDC: an angiographic and histopathological comparison. Neuroradiology 2008;50(8):693 699. 71. Smith MJ, Miscitelli J, Santillan A, et al. Bare platinum vs matrix detachable coils for the endovascular treatment of intracranial aneurysms: a multivari ate logistic regression analysis and review of the literature. Neurosurgery 2011;69(3):557 564. 72. Deshaies EM, Adamo MA, Boulos AS. A prospective single center analysis of the safety and efficacy of the hydrocoil embolization system for the treatment o f intracranial aneurysms. J Neurosurg 2007;106(2):226 233.
131 73. White JB, Cloft HJ, Kallmes DF. But did you use HydroCoil? Perianeurysmal edema and hydrocephalus with bare platinum coils. Am J Neuroradiol 2008;29(2):299 300. 74. White PM, Lewis SC, Ghol kar A, et al. Hydrogel coated coils versus bare platinum coils for the endovascular treatment of intracranial aneurysms (HELPS): a randomised controlled trial. Lancet 2011;377(9778):1655 1662. 75. el D, Ray C, eds. Transcatheter Embolization and Therapy New York; 2010:304. 76. Davis JR, ed. Handbook of Materials for Medical Devices Materials Park: ASM International; 2003. 77. Niinomi M, ed. Metals for biomedical devices Cambridge: Woodhead; 201 0. 78. Kohn DH. Metals in medical applications. Curr Opin Solid State Mater Sci 1998;3(3):309 316. 79. Liu X, Chu PK, Ding C. Surface modification of titanium, titanium alloys, and related materials for biomedical applications. Mater Sci Eng R Reports 2004;47(3 4):49 121. 80. Hedberg Y, Wang X, Hedberg J, Lundin M, Blomberg E, Wallinder IO. Surface protein interactions on different stainless steel grades: effects of protein adsorption, surface changes and metal release. J Mater Sci Mater Med 2013;24(4 ):1015 33. doi:10.1007/s10856 013 4859 8. 81. Geisel D, Gebauer B, Malinowski M, Stockmann M, Denecke T. Comparison of CT and MRI artefacts from coils and vascular plugs used for portal vein embolization. Eur J Radiol 2014;(July 2011):2012 2015. doi:10.1016/j.ejrad.2014.01.004. 82. Khan SN, Rapacchi S, Levi DS, Finn JP. Pediatric cardiovascular interventional devices: effect on CMR images at 1.5 and 3 Tesla. J Cardiovasc Magn Reson 2013;15(1):54. doi:10.1186/1532 429X 15 54. 83. Levine GN, Gome s AS, Arai AE, et al. Safety of magnetic resonance imaging in patients with cardiovascular devices: an American Heart Association scientific statement from the Committee on Diagnostic and Interventional Cardiac Catheterization, Council on Clinical Cardiolo gy, and the Council o. Circulation 2007;116(24):2878 91. doi:10.1161/CIRCULATIONAHA.107.187256. 84. Lambert BJ, Tang FW, Robers WJ. Polymers in Medical Applications Shawbury: Rapra Technology Limited; 2001. 85. Gall K, Yakacki CM, Liu Y, Shandas R, Wil lett N, Anseth KS. Thermomechanics of the shape memory effect in polymers for biomedical applications. J Biomed Mater Res A 2005;73(3):339 48. doi:10.1002/jbm.a.30296. 86. Yakacki CM, Shandas R, Safranski D, Ortega AM, Sassaman K, Gall K. Strong, Tailore d, Biocompatible Shape Memory Polymer Networks. Adv Funct Mater 2008;18(16):2428 2435. doi:10.1002/adfm.200701049.
132 87. Prell D, Kyriakou Y, Struffert T, Drfler a, Kalender W a. Metal artifact reduction for clipping and coiling in interventional C arm CT. AJNR Am J Neuroradiol 2010;31(4):634 9. doi:10.3174/ajnr.A1883. 88. Meyer JM, Buecker A, Schuermann K, Ruebben A, Gue nther RW. MR evaluation of stent patency: in vitro test of 22 metallic stents and the possibility of determining their patency by MR angiography. Invest Radiol 2000;35(12):739 46. 89. Kwan ESK, Heilman CB, Shucart WA, Klucznik RP. Enlargement of basilar artery aneurysms following balloon occlusion J Neurosurg 1991;75(6):963 968. 90. Ltourneau Guillon L, Soulez G, Beaudoin G, et al. CT and MR Imaging of Nitinol Stents with Radiopaque Distal Markers. J Vasc Interv Radiol 2004;15 (6):615 624. doi:10.1097/01.RVI.00000127898.23424.01. 91. Orth RC, Wallace MJ, Kuo MD. C arm cone beam CT: general principles and technical considerations for use in interventional radiology. J Vasc Interv Radiol 2008;19(6):814 20. doi:10.1016/j.jvir.200 8.02.002. 92. Thomas C, Mller Bierl BM, Rempp H, et al. In vitro assessment of artifacts from commercially available markers for image guided preoperative marking of bone and soft tissue lesions. J Vasc Interv Radiol 2010;21(7):1100 4. doi:10.1016/j.jvi r.2010.04.002. 93. Mahnken AH, Chalabi K, Jalali F, Gnther RW, Buecker A. Magnetic Resonance guided Placement of Aortic Stents Grafts: Feasibility with Real Time Magnetic Resonance Fluoroscopy. J Vasc Interv Radiol 2004;15(2):189 195. doi:10.1097/01.RVI .0000109399.52762.53. 94. Shih M CP, Rogers WJ, Bonatti H, Hagspiel KD. Real time MR guided retrieval of inferior vena cava filters: an in vitro and animal model study. J Vasc Interv Radiol 2011;22(6):843 50. doi:10.1016/j.jvir.2011.01.428. 95. Bartels LW, Smits HF, Bakker CJ, Viergever M a. MR imaging of vascular stents: effects of susceptibility, flow, and radiofrequency eddy currents. J Vasc Interv Radiol 2001;12(3):365 71. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11287516. 96. Todaka Y, McC ormick PG, Tsuchiya K, Umemoto M. Synthesis of Fe Cu Nanoparticles by Mechanochemical Processing Using a Ball Mill. Mater Trans 2002;43(4):667 673. doi:10.2320/matertrans.43.667. 97. Ozdemir I, Ahrens S, Mcklich S, Wielage B. Nanocrystalline Al Al2O3p a nd SiCp composites produced by high energy ball milling. J Mater Process Technol 2008;205(1 3):111 118. doi:10.1016/j.jmatprotec.2007.11.085. 98. Amirkhanlou S, Ketabchi M, Parvin N. Nanocrystalline/nanoparticle ZnO synthesized by high energy ball millin g process. Mater Lett 2012;86:122 124. doi:10.1016/j.matlet.2012.07.041. 99. De Carvalho JF, de Medeiros SN, Morales M a., Dantas a. L, Carrio a. S. Synthesis of magnetite nanoparticles by high energy ball milling. Appl Surf Sci 2013;275:84 87. doi:10. 1016/j.apsusc.2013.01.118.
133 100. Faraday M. The Bakerian Lecture: Experimental Relations of Gold (and Other Metals) to Light. Philos Trans R Soc London 1857;147(1857):145 181. 101. Frens G. Controlled Nucleation for the Regulation of the Particle Size in Monodisperse Gold Suspensions. Nat Phys Sci 1973;241:20 22. 102. Wilcoxon JP, Williamson RL, Baughman R. Optical properties of gold colloids formed in inverse micelles. J Chem Phys 1993;98(12):9933 9950. doi:10.1063/1.464320. 103. Brust M, Walker M, Bethell D, Schiffrin DJ, Whyman R. Synthesis of thiol derivatised gold nanoparticles in a two phase Liquid Liquid system. J Chem Soc Chem Commun 1994;(7):801 802. 104. Zhou J, Ralsto n J, Sedev R, Beattie D a. Functionalized gold nanoparticles: synthesis, structure and colloid stability. J Colloid Interface Sci 2009;331(2):251 62. doi:10.1016/j.jcis.2008.12.002. 105. Chan WCW, ed. Bio applications of nanoparticles Austin: Landes Bio science; 2007. 106. Philip D. Synthesis and spectroscopic characterization of gold nanoparticles. Spectrochim Acta A Mol Biomol Spectrosc 2008;71(1):80 5. doi:10.1016/j.saa.2007.11.012. 107. Xu X, Rosi NL, Wang Y, Huo F, Mirkin C a. Asymmetric functiona lization of gold nanoparticles with oligonucleotides. J Am Chem Soc 2006;128(29):9286 7. doi:10.1021/ja061980b. 108. Liu Y, Franzen S. Factors determining the efficacy of nuclear delivery of antisense oligonucleotides by gold nanoparticles. Bioconjug Che m 2008;19(5):1009 16. doi:10.1021/bc700421u. 109. Rosi NL, Giljohann D a, Thaxton CS, Lytton Jean AKR, Han MS, Mirkin C a. Oligonucleotide modified gold nanoparticles for intracellular gene regulation. Science 2006;312(5776):1027 30. doi:10.1126/science .1125559. 110. Sperling R a, Parak WJ. Surface modification, functionalization and bioconjugation of colloidal inorganic nanoparticles. Philos Trans A Math Phys Eng Sci 2010;368(1915):1333 83. doi:10.1098/rsta.2009.0273. 111. Love JC, Estroff L a, Krieb el JK, Nuzzo RG, Whitesides GM. Self assembled monolayers of thiolates on metals as a form of nanotechnology. Chem Rev 2005;105(4):1103 69. doi:10.1021/cr0300789. 112. Kang SK, Chah S, Yun CY, Yi J. Aspect ratio controlled synthesis of gold nanorods. Kor ean J Chem Eng 2003;20(6):1145 1148. doi:10.1007/BF02706952. 113. Gole A, Murphy CJ. Seed Mediated Synthesis of Gold Nanorods: Role of the Size and Nature of the Seed. Chem Mater 2004;16(19):3633 3640. doi:10.1021/cm0492336. 114. Garca M a., Bouzas V Carmona N. Influence of stirring in the synthesis of gold nanorods. Mater Chem Phys 2011;127(3):446 450. doi:10.1016/j.matchemphys.2011.02.026. 115. Zijlstra P, Bullen C, Chon JWM, Gu M. High temperature seedless synthesis of gold nanorods. J Phys Chem B 2006;110(39):19315 8. doi:10.1021/jp0635866.
134 116. Smith DK, Korgel B a. The importance of the CTAB surfactant on the colloidal seed mediated synthesis of gold nanorods. Langmuir 2008;24(3):644 9. doi:10.1021/la703625a. 117. Kim F, Song JH, Yang P. P hotochemical synthesis of gold nanorods. J Am Chem Soc 2002;124(48):14316 7. Available at: http://www.ncbi.nlm.nih.gov/pubmed/21043446. 118. Hkkinen H. The gold sulfur interface at the nanoscale. Nat Chem 2012;4(6):443 55. doi:10.1038/nchem.1352. 119. Gronbeck H, Curioni A, Andreoni W. Thiols and Disulfides on the Au (111) Surface: The Headgroup Gold Interaction. J Am Chem Soc 2000;122(16):3839 3842. 120. DRFREDDY. What is the true nature of gold sulfur bonds? Synth Remarks 2013. Available at: http://syntheticremarks.com/?p=3711. Accessed July 17, 2014. 121. Garzon I, Rovira C, Michaelian K, et al. Do thiols merely passivate gold nanoclusters? Phys Rev Lett 2000;85(24):5250 1. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11102233. 122. Mc Neil SE, ed. Characterization of nanoparticles intended for drug delivery 1st ed. New York: Springer Science; 2011. 123. Carotenuto G, Nicolais L. Size controlled synthesis of thiol derivatized gold clusters. J Mater Chem 2003;13(5):1038 1041. 124. Hir amatsu H, Osterloh FE. A Simple Large Scale Synthesis of Nearly Monodisperse Gold and Silver Nanoparticles with Adjustable Sizes and with Exchangeable Surfactants. Chem Mater 2004;16(13):801 802. 125. Yang YW, Fan LJ. High Resolution XPS Study of Decanet hiol on Au ( 111 ): Single Sulfur Gold Bonding Interaction. Langmuir 2002;18(4):1157 1164. 126. Haiss W, Thanh NTK, Aveyard J, Fernig DG. Determination of size and concentration of gold nanoparticles from UV vis spectra. Anal Chem 2007;79(11):4215 21. doi:10.1021/ac0702084. 127. Spectroscopy. J Phys Chem C 2009;113(11):4277 4285. doi:10.1021/jp8082425. 128. DeLong R, Schaeffer A. Functionalized gold nanoparticles for the bindi ng, stabilization, and delivery of therapeutic DNA, RNA, and other biological macromolecules. Nanotechnol Sci Appl 2010;3:53 63. doi:10.2147/NSA.S8984. 129. Wang Z, Ma L. Gold nanoparticle probes. Coord Chem Rev 2009;253(11 12):1607 1618. doi:10.1016/j. ccr.2009.01.005. 130. Mallidi S, Larson T, Aaron J, Sokolov K, Emelianov S. Molecular specific optoacoustic imaging with plasmonic nanoparticles. Opt Express 2007;15(11):6583 8. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19546967. 131. Gibson JD, Khanal BP, Zubarev ER. Paclitaxel functionalized gold nanoparticles. J Am Chem Soc 2007;129(37):11653 61. doi:10.1021/ja075181k.
135 132. Bhattacharya R, Patra CR, Verma R, Kumar S, Greipp PR, Mukherjee P. Gold Nanoparticles Inhibit the Proliferation of Mult iple Myeloma Cells. Adv Mater 2007;19(5):711 716. doi:10.1002/adma.200602098. 133. Shukla R, Bansal V, Chaudhary M, Basu A, Bhonde RR, Sastry M. Biocompatibility of gold nanoparticles and their endocytotic fate inside the cellular compartment: a microsco pic overview. Langmuir 2005;21(23):10644 54. doi:10.1021/la0513712. 134. Pan Y, Neuss S, Leifert A, et al. Size dependent cytotoxicity of gold nanoparticles. Small 2007;3(11):1941 9. doi:10.1002/smll.200700378. 135. Hauck TS, Ghazani A a, Chan WCW. Ass essing the effect of surface chemistry on gold nanorod uptake, toxicity, and gene expression in mammalian cells. Small 2008;4(1):153 9. doi:10.1002/smll.200700217. 136. Yu K, Westbrook KK, Kao PH, Leng J, Qi HJ. Design considerations for shape memory pol ymer composites with magnetic particles. J Compos Mater 2013;47(1):51 63. doi:10.1177/0021998312447647. 137. Park J A, Kim H K, Kim J H, et al. Gold nanoparticles functionalized by gadolinium DTPA conjugate of cysteine as a multimodal bioimaging agent. B ioorg Med Chem Lett 2010;20(7):2287 91. doi:10.1016/j.bmcl.2010.02.002. 138. Chanda N, Kattumuri V, Shukla R, et al. Bombesin functionalized gold nanoparticles show in vitro and in vivo cancer receptor specificity. Proc Natl Acad Sci U S A 2010;107(19): 8760 8765. doi:10.1073/pnas.1002143107. 139. Alric C, Serduc R, Mandon C, et al. Gold nanoparticles designed for combining dual modality imaging and radiotherapy. Gold Bull 2008;41(2):90 97. 140. Sun C, Lee JSH, Zhang M. Magnetic nanoparticles in MR ima ging and drug delivery. Adv Drug Deliv Rev 2008;60(11):1252 65. doi:10.1016/j.addr.2008.03.018. 141. Moriggi L, Cannizzo C, Dumas E, Mayer CR, Ulianov A, Helm L. Gold nanoparticles functionalized with gadolinium chelates as high relaxivity MRI contrast agents. J Am Chem Soc 2009;131(31):10828 9. doi:10.1021/ja904094t. 142. Sun H, Yuan Q, Zhang B, Ai K, Pengguo Z, Lu L. GdIII functionalized gold nanorods for multimodal imaging applications. Nanoscale 2011;3(5):1990 1996.
136 APPENDIX A Protocol for Size Controlled Synthesis of H ydrophobic Thiol Derivatized Gold Clusters Using O leylamine Modified scale synthesis of nearly monodisperse gold and silver nanoparticles with adjustable sizes and Chemistry of Materials, 2004, Vol 13, Issue 13 1. Clean all glassware to come in contact with nanoparticles (directly or indirectly) in aqua regia (3 parts HCl, 1 part HNO 3 turns vivid yellow, then gold, and then orange). Rinse with double distilled water. Oven dry. 2. Required supplies a. Direct contact: i. 20 ml glass vial ii. Glass Pasteur pipette (2x) for weighting out HAuCl 4 iii. Condenser iv. 500 ml round bottom boiling flasks (2x) v. Stir bars (2x) vi. 500 ml beakers (2x) vii. 1 L beaker viii. 5 ml glass serological pipette (2x) ix. 10 ml glass serological pipette (3x) b. Indirect (regular cleaning or disposable new) i. 50 ml conical polypropylene tubes (12x) ii. 250 ml graduated cylinder iii. 100 ml graduated cylinder iv. Methanol squirt bottle (keep separate for use as a reagent) c. Other
137 i. Ice bucket (to chill methanol) ii. Water line tubing for condenser 3. Reagen ts needed: a. Oleylamine: ~29 ml b. Toluene: ~125 ml c. Methanol: ~1 L 4. Bring 20 g (~24.6 ml) of oleylamine and 120 ml toluene to a boil in a 500 ml round bottom flask at 135C and 200 rpm stirring. 5. Prepare a mixture of 300 mg HAuCl 4 3 g (3.6 ml) oleylamine, and 3 ml toluene. Add this immediately to the round bottom flask and allow to react for 2 hours. If desired, take 1ml samples from flask and acquire UV Vis spectrum (350 to 750 nm) in real time to track reaction progress. Chill m ethanol on ice prior to next step. 6. Decant reaction volume into a 1L beaker and add cold methanol to bring volume to 500 ml 7. Dispense all 500 ml into ten 50 ml conical polypropylene tubes. Cap and centri fuge at > 1000 rcf for 15 min (4C). 8. Decant supernatant and resuspend each pellet in 15 20 ml cold methanol (using squirt bottle). Repeat centrifugation twice. 9. Dry in vacuo (no heat) overnight. 10. In order to perform a ligand exchange with dodecanethiol (DDT, MW 202.4 g/mol, density of 0.845 g/ml), first bring 50 ml of toluene and 5 ml of DDT to a boil in a round bottom flask at 135C and 200 rpm stirring.
138 11. Resuspend the vacuum dried AuNPs in toluene (add 15 ml to each tube, mix with a glass serological pipette then add to a chilled beaker repeat two more times). Add the resuspended particles to the round bottom flask and react for 1 hour. 12. If desired, filter the reaction volume through a ~1 m pore size syringe filter to remove micron scale contaminants and a ggregates. Separate the hydrophobic thiol coated gold nanoparticles from the solution by centrifugation at > 1000 RCF for 15 min at 4C (remember to weigh the centrifuge tubes first!). Note: you can divide the reaction volume up into 50 ml conical unequall y in order to produce pellets of varying masses. Decant and wash with 15 ml toluene per pellet. Repeat centrifugation and wash twice. 13. Dry in vacuo (no heat) overnight. 14. Measure the UV Vis spectrum of the sample from 350 to 750 nm using the standard monomer mix as reference. Apply the Mie or Mie Gans theory to estimate nanoparticle size and concentration. Return the content of the cuvette to the remaining sample volume. 15. Cast and polymerize using short UV exposures (to minimize heating the sample). Expect the resulting polymer to have a red purple hue.
139 APPENDIX B Protocol for Size Controlled Synthesis of Hydrophobic Thiol Derivatized Gold C lusters using Ethylene G lycol and P olyvinylpyrrolidone Modified version of the protocol developed by G. Carotenuto controlled synthesis of thiol J. Mater. Chem., 2003, 13, 1038 1041 1. Prepare large volume stock of ethylene glycol/poly(N vinyl prrolidone) (EG/PVP) for use over multiple reactions. a. Dissolve 200 g PVP in 500 ml EG (requires heat and several hours to dissolve completely, but stable at room temp. after that) i. Note: PVP density = 1.2 g/ml final concentration = 200g/(166+ 500 ml) = 0.3 g/ml (30%) 2. Clean all glassware to come in contact with nanoparticles (dire ctly or indirectly) in aqua regia (3 parts HCl, 1 part HNO 3 turns vivid yellow, then gold, and then orange). Rinse with double distilled water. Oven dry. a. One 2 ml vial b. Two 250 ml round bottom (RB) boiling flasks c. One condenser and tubes for water line d. One 1 L beaker e. Twelve 50 ml conical polypropylene centrifuge tubes f. Two glass or Teflon spatulas g. Two Teflon stir bars 3. Chill ~400 ml of acetone on ice (for step 5)
140 4. Mix the following under stirring at 400 rpm in a 250 ml round bottomed boiling flask (with conde nser and paraffin oil filled boiling tray) until a ruby red color is achieved (5 10 min). Note that ethylene glycol (EG) acts as both the solvent and reducing agent. PVP acts as the polymeric capping agent. Nanoparticle size can be regulated by varying the reaction temperature, or the HAuCl 4 :PVP ratio. Increasing the reaction temperature or decreasing the HAuCl 4 :PVP ratio yields smaller particles. a. 100 ml of PVP/EG solution b. Solution of 200 mg HAuCl 4 in 1 ml EG. Note: HAuCl 4 corrodes steel use glass or Tefl on spatulas Note: Reaction solution should turn ruby red within 5 min, but continue for 10 min. 5. Cast the reaction mixture into the chilled acetone and sonicate for a 5 minutes. This ends the reaction and separates the nanoparticles. The Au PVP nanocomposit e is achieved after flocculation. Collect the samples by centrifugation at 1000 RCF for 15 min in 50 ml conical tubes. 6. Wash the product 3x with acetone (~25 ml per pellet each time) and dry at room temperature under a vacuum (~ 15 in. Hg) for > 5 hours. Th e result will be a thick syrupy red pellet. 7. Dissolve the Au PVP system into a dilute ethanol solution of dodecanethiol (DDT) (100 ml of 50 mM) under stirring in a round bottomed flask for 2 hours. a. DDT has a MW of 202.4 g/mol and a density of 0.845 g/ml 1 .2 ml DDT in 100 ml total yields 50 mM b. Get 50 ml ethanol + 1.2 ml DDT stirring in 250 ml RB flask. The remaining 50 ml of ethanol is needed to resuspend the AU PVP pellet. c. Use a glass spatula to break up the Au PVP pellet from each centrifuge tube in the presence of a small volume of ethanol. Once all pellets are in solution, pour into the RB flask. The reaction mixture will turn purple, then red, and then clear
141 with very fi ne dark purple particles (or dark purple/black if concentration is very high). The figure below illustrates this ligand exchange reaction: 8. Separate the hydrophobic thiol coated gold nanoparticles from the PVP ethanol solution by centrifugation at 1000 RCF for 10 min (remember to weigh the centrifuge tubes first!). 9. Wash the solid 3x with ethanol to remove impurities and dry under vacuum (~ 15 in. Hg for 3 hrs). The product should be a golden solid. Weigh the centrifuge tubes again and determine the mass of collected AuNPs. 10. Disperse AuNP pellet in desired volume of monomer mixture (2 ml required for a standard coupon) by sonication (20% power fo r 2 min on ice). The centrifuge tube(s) will have to be cut at mid height to allow the sonicator probe to reach the sample. Exercise caution when sonicating volatile liquids. Use aluminum foil to cover the opening and introduce a hole just wide enough to a ccommodate the probe. This, along with the ice, will reduce evaporative losses. Filter through a ~1 m pore size syringe filter to remove micron scale contaminants and aggregates. 11. Measure the UV Vis spectrum of the sample from 350 to 750 nm using the stand ard monomer mix as reference. Apply the Mie or Mie Gans theory to estimate nanoparticle size and concentration. Use the paper by Haiss et al. concentration of gold nanoparticles from UV Anal. Chem., 2007, 79, 421 5 4221. as a reference. Return the content of the cuvette to the remaining sample volume. 12. Cast and polymerize using short UV exposures (to minimize heating the sample). Expect the resulting polymer to have a red purple hue. F