Citation
Chemiluminescence of nickel carbonyl and ozone

Material Information

Title:
Chemiluminescence of nickel carbonyl and ozone basis of a novel detection system for monitoring carbon monoxide at sub ppm level
Creator:
Shahgholi, Mona
Publication Date:
Language:
English
Physical Description:
viii, 85 leaves : illustrations ; 29 cm

Subjects

Subjects / Keywords:
Carbon monoxide -- Measurement ( lcsh )
Chemiluminescence ( lcsh )
Carbon monoxide -- Measurement ( fast )
Chemiluminescence ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 83-85).
General Note:
Submitted in partial fulfillment of the requirements for the debree, Master of Science, Department of Chemistry.
Statement of Responsibility:
by Mona Shahgholi.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
22937128 ( OCLC )
ocm22937128
Classification:
LD1190.L46 1989m .S52 ( lcc )

Full Text
CHEMILUMINESCENCE OF NICKEL CARBONYL AND OZONE:
BASIS OF A NOVEL DETECTION SYSTEM FOR MONITORING
CARBON MONOXIDE AT SUB PPM LEVEL
by
Mona Shahgholi
B.S., Oklahoma State University, 1979
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
Master of Science
Department of Chemistry
1989
% J


This thesis for the Master of Science degree by
Mona Shahgholi
has been approved for the
Department of
Chemistry
by
Date JO WftyemLr- ^ 83


Ill
Shahgholi, Mona (M.S., Chemistry)
Chemiluminescence of Nickel Carbonyl and Ozone: Basis of a Novel Detection System
for Monitoring Carbon Monoxide at Sub PPM Level
Thesis directed by Associate Professor John A. Lanning
Monitoring atmospheric carbon monoxide (CO) levels is important for two main
reasons. First, CO is a health hazard at the levels found in polluted urban
environments hence the establishment of National Ambient Air Quality Standards by
the Environmental Protection Agency to protect public health and welfare. Second,
CO is one of the most abundant reactive carbon compounds in the global atmosphere
with a significant role in atmospheric chemistry through reaction with common
intermediates that affect the 03 and HxOy cycles. Although several validated
methodologies exist for monitoring CO at levels found in urban areas, the continuous
monitoring of CO levels in remote atmospheres requires instrumentation with greater
sensitivity and practicality than is currently available. The present study investigated
the feasibility of developing a novel CO monitor based upon the conversion of CO to
nickel carbonyl (Ni(CO)4), and monitoring the chemiluminescent reaction of Ni(CO)4
with ozone. The primary tasks of this project involved the construction of a dynamic
gas dilution and ozone generation system, the evaluation of a catalytic convertor for
the conversion of CO with nickel to form Ni(CO)4 and the assembly of a sensitive
chemiluminescent analyzer Ni(CO)4.
First, the gas dilution system and ozone generator was constructed to provide flows
of sample gas at specific dilutions and pressures to the catalytic convertor as well as
supplying a constant flow of ozone to the chemiluminescent analyzer. Second, nickel
catalysts for the conversion of CO to Ni(CO)4 were prepared with a variety of support
materials by a number of different processes. Third, a chemiluminescent analyzer for
the detection of Ni(CO)4 was constructed and calibrated with standard Ni(CO)4
dilutions. This detector had a detection limit of 0.42 ppm.


IV
The complete monitoring system was assembled and the prepared catalysts were
evaluated for their efficacy in forming Ni(CO)4 from dilutions of CO. The catalysts
demonstrated a wide range of activities, however the most active catalyst produced a
subsequent chemiluminescent signal from less than 1 ppm CO.
These results indicate that, with several modifications, a novel CO monitor can be
developed based upon the formation of Ni(CO)4 and its chemiluminescent reaction with
ozone.
The form and content of this abstract are approved. I recommend
its publication.
Signed


V
ACKNOWLEGDEMENTS
The accomplishment of this thesis would not have been possible without the
grateful assistance of many associates and friends. First, and above, all I thank my
mother, Farah Shahgholi, for her guidance, support and encouragement in fulfilling
this work. Next, I acknowledge my advisors, Drs. John A. Lanning and Larry G.
Anderson for proposing this project. Their advice and assistance will always be
appreciated and remembered. No project can be readily accomplished without the
seemingly endless requests for supplies and chemicals, of course, all of which were
needed "yesterday." For all his help in obtaining these necessities, I am grateful for
Mike Milashs patience and assistance. To Dr. Joe Zirrolli and Lee Miller of the
Health Science Center, I am particularly appreciative of their time and help in
designing and constructing various parts of the monitor.
Finally, to all my friends and colleagues at UCD, "thanks for the memories!"


vi
CONTENTS
Page
LIST OF TABLES.................................................... vii
LIST OF FIGURES .................................................. viii
CHAPTER
I. INTRODUCTION ................................................ 1
II. MATERIALS AND METHODS....................................... 17
Reagents ............................................ 17
Preparation of Nickel Carbonyl ...................... 17
Preparation of Nickel Carbonyl Gas Standards......... 18
Preparation of Heterogeneous Nickel Catalysts ....... 20
Instrumentation..................................... 25
Procedure for Catalyst Evaluation................. 42
III. RESULTS..................................................... 45
Calibration of the Ozone Generator................... 45
Calibration of the Chemiluminescent Analyzer........ 53
Catalyst Evaluations ................................ 53
IV. DISCUSSION.................................................. 68
Catalyst Performance ................................ 73
Chemiluminescence of Ozone and Nickel Carbonyl....... 77
Future Modifications ............................ 81
V. BIBLIOGRAPHY ............................................... 83


vii
LIST OF TABLES
Table Page
1.1. Summary of Carbon Monoxide Monitors ................................. 13
2.1. Physical Characteristics of Zeolites................................. 22
2.2. Physical Properties of Prepared Catalysts............................ 26
2.3. Flow Ranges of Gas Inlet Restrictors................................ 39
3.1. Ozone Generation..................................................... 46
3.2. Ozone Dilution....................................................... 51
3.3. Catalyst Performance: Nickel Adsorbed on
13X Molecular Sieve........................................ 56
3.4. Catalyst Performance: Nickel Adsorbed on
Dowex 50X8-50 Resin ........................................ 57
3.5. Catalyst Performance: Nickel Adsorbed on
Cationotropic Alumina ...................................... 59
3.6. Catalyst Performance: Nickel Adsorbed on
Silica Gel................................................ 60
3.7. Catalyst Performance: Nickel Adsorbed on
Magnesium Oxide............................................. 61
3.8. Catalyst Performance: Nickel Co-precipitated
with Alumina ............................................... 62
3.9. Catalyst Performance: Nickel Adsorbed on
Florisil.................................................... 64
3.10. Catalyst Performance: Nickel Adsorbed on
Amberlite XAD-2 ............................................ 65
3.11. Summary of Catalyst Performance . .....................\ . . 67
4.1. Reported Operating Parameters of Ni(CO)4 Analyzers .................. 72
4.2. Catalysts Prepared and Tested........................................ 74


Vlll
LIST OF FIGURES
Figure Page
1.1. Carbon Monoxide Monitor Schematic ....................................... 3
1.2. NDIR CO Monitor......................................................' 5
1.3. Gas Filter Correlation Spectroscopic Monitor ............................ 8
2.1. Apparatus for Synthesis of Nickel Carbonyl.............................. 19
2.2. Novel Carbon Monoxide Detection System ..................... 27
2.3. Dynamic Gas Dilution System............................................. 28
2.4. Ozone Generator ........................................................ 29
2.5. Flow Characteristic of Restrictor A-l................................... 31
2.6. Flow Characteristic of Restrictor A-2...................-............ 32
2.7. Flow Characteristic of Restrictor B-l................................... 33
2.8. Flow Characteristic of Restrictor B-2........;...................... . 34
2.9. Flow Characteristic of Restrictor C-l.................................. 35
2.10. Flow Characteristic of Restrictor C-2................................... 36
2.11. Flow Characteristic of Restrictor D-l................................... 37
2.12. Flow Characteristic of Restrictor D-2................................... 38
2.13. Catalytic Cell Designs................................................ 41
2.14. Nanoameter Amplifier Schematic.......................................... 43
3.1. Ozone Calibration Run#l................................................. 47
3.2. Ozone Calibration Run#2................................................. 48
3.3. Ozone Calibration Run#4................................................. 49
3.4. Ozone Dilution Curve ................................................... 52
3.5. Nickel Carbonyl Standard Curve ......................................... 54
3.6. Catalyst Performance: Nickel Adsorbed on Dowex 50X8-50 58
3.7. Catalyst Performance: Nickel Co-precipitated with Alumina............ 63
3.8. Catalyst Performance: Nickel Adsorbed on Amberlite XAD-2 .... 66


CHAPTER I
INTRODUCTION
Carbon monoxide is a colorless, odorless gas that has been a subject of interest for
two main reasons; firstly, its toxicity to mammals and secondly, its role in the Oa and
HxOy cycles in the global atmosphere. Global emissions of CO are of both
anthropogenic as well as natural origin. Yearly anthropogenic CO emissions are
estimated as 260 to 1060xl012 grams and the natural sources as 150 to 6700xl012 grams
(Informatics, 1979). The anthropogenic sources are fossil fuel combustion (250 to
lOOOxlO12 grams) and forest fires (10 to 60xl012 grams). The major natural sources
of CO include the oxidation of methane and other naturally occurring hydrocarbons
in the troposphere (110 to 6300x1012 grams), microbial activity in the oceans (20 to
200xl012 grams), and plant emissions (20 to 200xl012 grams). Carbon monoxide is
removed primarily by in situ oxidation to C02 by OH radicals (30x1014 g) and soil
bacterial uptake (4.5xl014 g) (Niki et al 1984).
Most urban areas in the United States, as assessed by the Environmental
Protection Agency (EPA), have excessive CO levels. The EPA has designated CO as
a criteria pollutant, i.e., one possessing documented effects on people, vegetation or
materials at concentrations (0.2 50 ppm) approaching those found in polluted air
(Finlayson-Pitts and Pitts 1986). The EPA has established National Ambient Air
Quality Standards (NAAQS) to protect public health and safety. In the United States
there are two types of NAAQ standards, primary and secondary. The primary carbon
monoxide standards (9 ppm CO per 8 hours average and 35 ppm CO per 1 hour
average) are designed to protect the public health with an adequate margin of safety.
They are set to protect even the most susceptible groups, such as those people with
angina or other cardiovascular diseases, the elderly, fetuses and young infants.


2
Secondary standards have the goal of protecting public welfare. This includes
economic losses to plants and materials, aesthetic effects including visual degradation
and personal discomfort. Although the secondary standards for CO have been revoked
because there appears to be no demonstrated adverse effects from exposure to ambient
CO levels (EPA-450/5-85-007 July 1985), the primary standards are frequently
exceeded in polluted urban environments. Accurate monitoring of CO levels in the
urban atmosphere is thus very important to assess the CO pollution problem and the
effects of mandatory abatement programs.
Equally significant is the monitoring of CO levels in the remote atmosphere.
Carbon monoxide is one of the two most abundant reactive carbon compounds in the
global atmosphere with a significant role in the 03 and HxOy cycles through chemical
interactions with common reactive intermediates (free radicals, e.g. OH, H02, and
CHjOO). Despite its significant role in the global atmosphere the details of COs
atmospheric chemistry will not be discussed here yet we emphasize that COs
anthropogenic perturbation as well as its overall atmospheric impact are of great
interest. Carbon monoxide levels in the remote atmosphere are typically between 0.040
and 0.30 ppm, much lower that the values seen in the urban atmosphere (0.2-50 ppm).
It is apparent that to monitor CO levels in the remote atmosphere requires analytical
techniques with significantly higher sensitivity than what would be sufficient in the
urban atmosphere. A wide variety of analytical techniques for measuring CO levels
in the atmosphere are available yet none are suitable for low concentration
measurements for reasons such as major interferences, cost, lack of precision and
accuracy.
To monitor CO levels requires a sample introduction system, an analyzer system
and a data recording system (EPA-450/5-85-007, July 1985, see Fig 1.1). Our
discussion will focus on the analyzer system. The different analyzers currently in use
vary in their principle of operation as well as their limits of detection and precision.
Following is a brief description of some of the EPA recommended methods capable of


Figure 1.1 Carbon Monoxide Monitor Schematic
ZEnO GAS
SPAN GAS
ANALYZER SYSTEM
CAnBON MONOXIDE ANALYZER
MOISTURE
CONTROL ROTAMETER
PARTICULATE
FILTEn
data nnconoino
AND
DISPLAY SYSTEM
ANALYZER
INDICATOR
-
STRIP CHART
RECOIIOER
IT
DATA
L4ACOU|SITtOP<
SYSTEM
I
I


4
continuous measurement of CO levels.
Non-dispersive Infrared Photometry (NDIR). This method is based upon
absorption of IR radiation over most of the spectrum for a given compound to provide
a quantitative determination of its concentration in a gas mixture. Specifically, the
technique involves determining the difference in IR energy absorption over all the
wavelengths passed by the optical system between the ambient air sample and a
reference path. It is assumed that the difference in IR energy absorbed is proportional
to the concentration of CO in the air sample. The IR absorption band for CO is
centered around 4.6 microns and is the basis for its detection and quantitation by
NDIR. The basic components of a typical IR instrument are a hot filament source of
IR radiation, a chopper, a sample cell and reference cell, and a detector (see Fig 1.2,
APTD-0736, US EPA, July 1971 #58).
The source of radiation is usually a glowing nichrome wire. The wavelength of
interest is selected by a non-dispersive technique such as optical filtering, often
employing the gas to be measured as the filtering medium. The source output may be
split into two parallel beams of nearly equal energy, one directed to the sample cell
and the other into the reference cell. Alternatively two sources may be used, one for
the sample cell and one for the reference cell. The reference cell contains a non-
IR absorbing gas while the sample cell is continuously flushed with the ambient
sample.
Two types of detectors are currently used in NDIR CO analyzers the Luft type
and the flow sensing type. These detectors consist of a two compartment gas cell, both
filled with CO under pressure to absorb radiation typical of CO absorption in the
spectrum. During operation an optical chopper intermittently exposes the' reference
and sample cells to the infrared sources. At the frequency imposed by the chopper,
a constant amount of infrared energy passes through the reference cell to one
compartment of the detector while a varying amount of energy, approximately


INFRARED
SOURCE
RECORDER
REFERENCE
CELL
a dJ Li ca
a s
Figure 1.2 NDIR CO Monitor


6
inversely proportional to the CO concentration in the sample cell, reaches the other
detector cell compartment. These unequal amounts of residual infrared energy reaching
the two compartments of the detector cause a pressure imbalance between the two
compartments. In the Luft type detector the pressure imbalance between the chambers
is converted to a electrical signal that is proportional to the CO concentration.
Water vapor and C02 are the principle interfering substances in determination of
CO by NDIR. Limited interferences from methane and ethane may also be observed.
Dual beam instruments can incorporate an interference cell in line with the sample cell.
The interference cell contains the principal interferents in a concentration sufficient
to block radiation from the overlapping portion of this absorption spectrum.
Additionally the use of narrow bandpass filters can further reduce the effect of
interfering gases. Commercial analyzers have high enough discrimination ratios for
C02, methane and ethane such that none of these gases should contribute more than
0.1 ppm equivalent CO interference even at high ambient concentrations.
Interference by water is not as easily controlled. The concentration of water in the
atmosphere can range from about 1500 ppm (0C, 10% RH) to 50,000 ppm (45C, 90%
RH) (Lodge 1989). The US EPA specifies that the maximum interference from any
substance in the atmosphere should not exceed 1.0 ppm, hence a water discrimination
ratio of 50,000 is required (discrimination ratio is the concentration of interferents
giving a response equivalent to 1.0 ppm CO). Such a discrimination ratio can be
obtained using a flowing reference cell or with a gas filter correlation type analyzer.
In the flowing type reference cell the reference gas is ambient air with CO removed
by a CO convertor. The water effect can also be reduced by removal via drying tubes
or cooling traps.
Non-dispersive IR instruments are not sensitive to flow rate, they require no wet
chemicals, they are reasonably independent of ambient air temperature changes, they
have a wide dynamic range and short response times. Furthermore they can be
operated by nontechnical personnel. NDIR instruments are however very sensitive to


7
vibrations. They have some problems with drift and nonlinerarity and are fairly
expensive. These problems are minimized in newer instruments with good thermostats
and solid state electronics.
Gas Filter Correlation Spectroscopy. This technique is a modification of the
conventional NDIR. It is nondispersive in nature, relying on optical filtering for the
selection of the appropriate bandwidth. A distinguishing feature of GFC is that the
optical filter is independent from the detector and consists of CO that is contained in
a rotating wheel. A reference cell is no longer required. The basic components of a
GFC system are illustrated in Figure 1.3. Radiation from an IR source is chopped and
then passed through an IR transparent gas filter which alternates between CO and
nitrogen due to rotation of the wheel. The gas filter produces a beam which cannot be
further affected by CO while the nitrogen side of the wheel contains frequencies
subject to absorption by CO. After passage through a multiple optical pass flowing
sample cell, the IR radiation is directed to a single detector. In the absence of CO,
the IR energy reaching the detector is equal during both halves of the N2/CO filtering
cycle. If an interfering compound such as water or carbon dioxide is present both the
nitrogen filtered and the CO filtered beams will be reduced in energy by essentially
an equal amount. In the presence of CO a reduction in radiation intensity will occur
during the nitrogen filtered portion of the cycle. The CO filtered portion will not be
affected. This intensity difference between the two halves will be sensed by the
detector and a signal proportional to sample CO concentration will be generated. The
detector can be a simple, wide bandwidth and solid state IR detector. Solid state
detectors do not have the the vibration sensitivity problems associated with Luft type
detectors. Also as there is no reference cell, the sample cell can have multiple pass
optics, significantly enhancing the sensitivity of analysis. A prototype version of this
instrument has a reported minimum sensitivity of 50 ppb (Chaney and McClenny
1977).


Figure 1.3 Gas Filter Correlation Spectroscopic Monitor
Mirror
Sample
Out
Rotating
Gas Filte^N,
Wheel
Sample
In
Pass
Bandpass
Filter
Chopper
1R
Detector
jElectronic
Signal
Processor
oo


9
Gas Chromatography (GO. Gas Chromatography is more suitable for measuring
background levels of CO due to its high sensitivity. A number of gas chromatographs
utilizing different detectors have been used for CO measurements. Among these are
electrochemical detector (Bergman et al 1975), helium detector (Bruner et al 1973),
emission helium plasma detector (Dagnall et al 1973), thermal conductivity detector
(Lodge 1989) and flame ionization detector (FID) (Porter and Volman 1962, Tesarik
and Krejci 1974). The flame ionization detector is by far the most popular. The GC-
FID system consists of a gas sampling valve, a backflush valve, a precolumn, a
molecular sieve column, a catalytic reactor and an FID. Measured volumes of air are
delivered to the FID 4 to 12 times per hour. Total hydrocarbon content (THC) is
determined for these samples. A portion of each air sample is passed through a
column where it is stripped of water, carbon dioxide and non methane hydrocarbons
(NMHC). Methane is then separated from CO by a GC column. Methane which
elutes first passes through the catalytic reduction tube without change. The eluted CO
is however reduced to methane in this catalytic reduction tube before reaching the
FID. Between analyses the stripping column is flushed out. NMHC is determined by
subtracting the methane content from the THC.
Instrumental sensitivity is very good for gas chromatography (detection limit = 0.02
ppm), however, GC instrumentation is relatively complex and requires continuous
maintenance by skilled technicians.
Electrochemical Analyzers. These analyzers fall into three catagories: the
controlled-potential electrochemical analyzer, the galvanic analyzer and the coulometric
analyzer.
In Controlled-Potential Electrochemical Analyzers CO is oxidized in an aqueous
solution at a catalytically active electrode (Bay et al 1972, 1974). The generated
current is linearly proportional to the CO concentration in the 0-100 ppm range (0


10
to 115 mg/m3). Sensitivity is 1.2 mg/m3 (1 ppm) and response time about 10 seconds.
The major interfering substances are acetylene and ethylene (1 part acetylene or
ethylene responds as 11 parts or 0.25 parts CO respectively). Minor interfering
substances include ammonia, hydrogen, hydrogen sulfide, NO, nitrogen dioxide, sulfur
dioxide, natural gas and gasoline vapor.
In galvanic cells designed for continuous CO monitoring (Haagen-Smit 1966; Hersch
1964; Hersch 1966) the air stream containing CO is passed through a heated chamber
(150C) packed with I205 where the following reaction occurs:
I2Os + 5 CO.........> 5 C02 + I2 (150C)
The liberated iodine is absorbed by an electrolyte and transferred to the cathode of the
galvanic cell where it is reduced generating a current that is measured by the
galvanometer. Major interfering substances are water, hydrogen sulfide, mercaptans,
olefins and acetylenes. Water interference can be removed by sampling through a
drying tube. The other interfering substances are removed by sampling through a tube
containing mercuric sulfate on silica gel.
Coulometric analyzers are based on the reaction of CO with I205 as in the galvanic
cells. The current produced from the reduction of generated iodine is measured by
an electrometer. Interfering substances are the same as for the galvanic cells and they
are removed as described above (Dubois et al 1966). The detection limit of 1.2
mg/m3 (1 ppm) is offset by its sensitivity to flow and temperature changes, and a slow
response time.
Mercury Replacement. This method of analysis is based upon the reduction of
mercuric oxide by CO according to the following reaction:


11
CO + HgO----------> C02 + Hg ( 210C)
Mercury vapor is detected photometrically at 253.7 nm. This method is potentially
more sensitive than IR absorption because the oscillator strength of mercury at 253.7
nm is 2000 times that of CO at 4.6 microns. It has a dynamic range of 0.020 10
ppm. Changes of 0.002 ppm are detectable hence it is used for global determination
of CO levels. The primary interfering substances are hydrogen and hydrocarbons
which also reduce mercuric oxide. The degree of intereference has to be assessed
separately for each compound. Also there is thermal decomposition of mercuric oxide
which can increase the background level (Seiler et al 1970).
Fourier Transform Infrared Spectroscopy. Developed in the early 1970s FTIR is
a powerful IR technique which has recently been applied to the measurement of trace
pollutants in ambient air (Hanst et al 1973). Its principal advantage over conventional
IR dispersive systems are sensitivity, speed and improved data processing. The basis
of the instrument is a Michelson interferometer consisting of a fixed mirror, a moving
mirror and a beam splitter. The interferometer generates an interferogram from an IR
light source such as a Nernst glower. The interferogram in an empty cell is
characteristic of the source and beam splitter. Upon addition of an IR absorbing
sample the interferogram will change. The difference between the two interferograms
(with and without an absorbing sample) is characteristic of the IR absorbing species
in the sample. The interferogram reaching the detector is signal intensity as a function
of the path length. This is converted to a signal intensity as a function of wavelength,
that is into an IR spectrum. This mathematical conversion (Fourier Transform) is done
by a computer. FTIR allows for measurement of several species simultaneously. Its
major disadvantage is that the spectrometer and associated optical system are relatively
expensive and require trained operators. Additionally FTIR systems are not easily
portable and therefore not amenable to field studies.


12
Tunable Diode Laser Spectrometry. This technique is based on measuring the
absorbance at specific wavelengths due to absorption of IR radiation by various
pollutants. The light source is a laser of very narrow line width which is tunable over
a smaller (300 cm-1) wavelength range. TDLS is more sensitive than FTIR. Its major
disadvantage is that scanning the entire spectrum quickly is not possible. TDLS
instrumentation is also expensive. The detection limit for CO using TDLS is 0.25 ppb
as calculated for a pathlength of 200 meters (Cassidy and Reid 1982).
Table 1.1 summarizes techniques used to monitor carbon monoxide and lists their
respective limits of detection. The U.S. EPA has designated NDIR as the reference
method for monitoring atmospheric CO levels. Following are the characteristics of the
method:
Range: 0-50 ppm
Detection Limit: 1.0 ppm
Precision: 0.5 ppm
These requirements although satisfactory for polluted atmospheres are not suitable for
remote atmospheres where the CO concentration is in the range Of 0.04 0.3 ppm.
A more sensitive technique is thus required to monitor CO in remote atmospheres.
We thus propose to assess the feasibility of a novel detector for CO based upon
chemiluminescence. Gas phase chemiluminescence has been known for a long time,
however its application as an analytical tool has occurred mainly in the last decade.
The inherent specificity, simplicity and portability of a chemiluminescent analyzer has
made it an attractive alternative to other gas phase analysis techniques should there
exist a specific chemiluminescent reaction for CO. Chemiluminescence, the production
of light energy released in a chemical reaction, is a sensitive and specific technique
used to monitor a variety of pollutants. The air containing the pollutant is reacted
with a species to produce light. The emission intensity is proportional to the
concentration of the pollutant.


13
Table 1.1 Summary of Carbon Monoxide Monitors
Techniaue1 Detection Limit Reference
GFCS 50 ppb Informatics, Inc. (1979)
NDIR 1 ppm Informatics, Inc. (1979)
FTIR not reported
GC-TCD 500 ppm . Lodge (1988)
GC-FID 0.02 ppm Informatics, Inc. (1979)
TDLS 250 ppt Cassidy & Reid (1982)
Electrochemical
Sensors 1 ppm Informatics, Inc. (1979)
Mercury Replacement 20 ppb Informatics, Inc. (1979)
Colorimetric Analysis > 1 ppm Informatics, Inc. (1979)
1. GFCS = Gas Filter Correlation Spectroscopy
NDIR = Non-dispersive Infrared Photometry
FTIR = Fourier Transform Infrared Spectroscopy
GC-TCD = Gas Chromatography Thermal Conductivity Detector
GC-FID = Gas Chromatography Flame Ionization Detector
TDLS = Tunable Diode Laser Spectrometry


14
However for emission to be detected by a photomultiplier tube (PMT), its
wavelength must lie in the range of 200 800 nm. Hence the emitting molecule ,M*,
must possess an excitation energy between 37 140 kcal per mole. This implies that
the reaction leading to production of M must have such exothermicity. The energy
of this exothermic reaction must get channeled into an electronically excited molecular
state with an allowed transition to the ground state. Additionally this reaction must
be fast such that there is significant product formation.
Chemiluminescent reactions are very susceptible to quenching, hence instrument
calibration and testing for interferents are essential. Quenching notwithstanding,
sensitivity is one of the best attributes of chemiluminescent analyzers. These analyzers
are also very selective. For another molecule to interfere it must chemiluminesce with
the analyte and emit in the same spectral region. Chemiluminescent analyzers are
simple, compact, easy to operate and adaptable for field analysis. They are for the
most part continuous mode instruments giving real-time data, hence their application
in air pollution field studies.
The defining equation for chemiluminescence is
Signal Fsampie nx %iemilum Collect ^detect
where Fgample is the flow of the analyte, nx is the mixing ratio of the unknown in the
sample, ^hemiium 1S t^e emission efficiency and ^ollectSdetect is the product of cell
design and photodetection technique. Chemiluminescence is intrinsically sensitive to
analyte flow. The analyte enters through an intake system and proceeds to the reactor
where it is mixed with an excess of reagent. Light emission is detected by a PMT and
the reaction products are pumped away by means of a pumping system. A trap after
the reactor can be inserted to remove excess reagents that might damage the pump or
be poisonous.
Although no direct chemiluminescent reactions are known for CO, carbon monoxide


15
is known to form metal carbonyls with a variety of metals. These carbonyl metal
compounds react rapidly in the gaseous phase and at low pressures with oxygen,
hydrogen or nitrogen atoms to give electronically excited products with characteristic
spectra (Becker and Schurgers 1971). Morris and Niki (Morris and Niki 1970) observed
light emission in the range of 500 650 nm when CO was mixed with ozonized oxygen
in a flow reactor. This emission with a broad maxima at 5650, 5900 and 6200 A0 was
attributed to FeO and NiO. It was concluded that carbonyl impurities, Fe(CO)5 and
Ni(CO)4, produced from the storage of CO in steel tanks, reacting with ozone were
responsible for the emission. A peculiar feature of the reaction is that it requires the
addition of excess 100% carbon monoxide. This addition enhances the observed
chemiluminescence.
The chemiluminescent reaction of nickel carbonyl and ozone was used in
developing a portable detector to monitor Ni(CO)4 levels in nickel refineries (Stedman
et al 1979). Indeed this detector was reported to be highly sensitive with a detection
limit of 2 ppb.
Based upon the preceeding discussion, the intention of this study is to evaluate the
feasibility of utilizing a nickel carbonyl detector in a novel carbon monoxide monitor
for determining CO levels in remote atmospheres. This monitor, to be functional, must
be capable of measuring CO at less than 40 ppb for real time continuous CO
monitoring.
Such a monitoring system would consist of a gas inlet system (both a dynamic gas
dilution system and ozone generator), a catalytic convertor for reacting carbon
monoxide with nickel to form nickel carbonyl and a chemiluminescent analyzer for
nickel carbonyl. Therefore this study is divided into three major tasks: construction
of the gas inlet system; the design, preparation and evaluation of the catalytic
convertor; and assembly of the chemiluminescent analyzer and reproduction of the
nickel carbonyl ozone chemiluminescent reactor.
The most crucial task is the conversion of carbon monoxide to nickel carbonyl via


16
a suitable catalyst (task #2). However both the first and third tasks must be completed
prior to testing of the conversion process. The heterogeneous reaction between gaseous
carbon monoxide and metallic nickel is indeed the principle of the Mond process of
nickel refining (Mond et al 1890). In this process impure nickel is reacted with pure
CO at 50 C and the carbonyl produced is fractionated several times prior to pyrolysis
at 200C. This leads to the deposition of metallic nickel with a purity of 99.90 -
99.99% (Nicholls 1973). The Mond reaction occurs under relatively mild conditions,
at least from a thermodynamic point of view. However no information is currently
available regarding the kinetics of the reaction. Therefore, a major effort in this study
is to investigate whether carbon monoxide to nickel carbonyl conversion can occur at
such rates to allow the monitoring of carbon monoxide levels via a chemiluminescent
nickel carbonyl detector.


CHAPTER II
MATERIALS AND METHODS
Reagents
All chemicals were analytical reagent grade or equivalent. Distilled, deionized
water was used in all procedures. Specifically the following reagent chemicals were
used: nickel nitrate hexahydrate, Type 13X molecular sieve and Florisil (Fisher
Scientific, Pittsburgh, PA), Type 4A molecular sieve, magnesium oxide and ammonium
carbonate monohydrate (J.T. Baker, Phillipsburgh, NJ), aluminum nitrate, Amberlite
XAD-2 (Mallinckrodt, St. Louis, MO), Dowex-50W (Sigma Chemical Co., St. Louis,
MO), cationotropic aluminum oxide (M. Woelm.Eshwege, Ger.), silica gel (Matheson,
Coleman and Bell, East Rutherford, NJ), Raney nickel (Alpha Products, Danvers, MA).
Oxygen and air were U.S.P. grade (General Air Service, Denver, CO). Carbon
monoxide was C.P. grade, 99.3%, (Air Products, Allentown, PA) and 200 ppm carbon
monoxide in air (Alphagaz).
Preparation of Nickel Carbonvl
Nickel carbonyl is a poisonous, colorless, volatile liquid with a boiling point of
43C and a melting point of -19.3C. This versatile chemical reagent is commercially
available (K&K Labs, Plainview, NY; Alpha, Danvers, MA) but only in relatively large
amounts of one pound units. Thus to use nickel carbonyl on the microgram scale, it
was prepared twice, once to verify the Mond catalytic process (see INTRODUCTION)
and observe its physical properties, and the second time to prepare standard stock
dilutions for optimizing reaction and detection parameters. In both instances the
catalyst (nickel impregnated 4A type molecular sieve) was first reduced' in a stream of
hydrogen at 400C for 2 hours, cooled to room temperature and then reacted with a


18
flow of carbon monoxide. The output of the catalyst cell was passed through two
cold traps placed in series and then through a trap of ammoniacal copper chloride
solution to destroy any escaping nickel carbonyl. The entire reaction was performed
in a well ventilated exhaust hood due to the toxicity of nickel carbonyl. The
ammoniacal copper chloride solution was prepared by adding ammonium hydroxide
(cone.) to 2M copper chloride until all Cu(OH)2 had been complexed as Cu(NH3)4 as
indicated by a deep bluish color. The first two traps were cooled in Dewars of dry
ice-acetone (Figure 2.1 A). After 4 hours the carbon monoxide flow (250 mL/min) was
stopped and white nickel carbonyl crystals were observed on the walls of the dry ice
acetone traps. The crystals which liquified rapidly upon removal from the cold trap
were lost through evaporation during sample collection.
The experimental setup was modified for subsequent preparations of nickel
carbonyl to enhance the yield and facilitate trapping the product (Figure 2.IB). In this
procedure the 50X cell, containing nickel impregnated 4A type molecular sieve, was
used instead of the 15X cell to increase the reactive surface area and available nickel
for the reaction. The outflow of this cell was connected to a series of two conical
collecting septum vials (4 mL and 1 mL) via a section of 0.32 mm i.d. fused silica
tubing. Traps were chilled in an ice-salt bath (-15C) and again a small ammoniacal
copper chloride trap was employed to destroy any escaping nickel carbonyl. After 2
hours approximately 40 microliters of nickel carbonyl were collected.
Preparation of Nickel Carbonvl Gas Standards
Approximately 9.7 mg of nickel carbonyl were transferred to a 15-liter evacuated
plastic gas tank via a fused silica tube (0.32 mm X 10 cm) through a septum inlet and
diluted with carbon monoxide to a final pressure of 8 psi (approximately 67 ppm).
The nickel carbonyl was stored in purified carbon monoxide because of a rapid
loss of Ni(CO)4 in air or nitrogen diluent. Nickel carbonyl decomposition in air is


19
2.1-A
2.1-B
Flowmeter
trap
Hydrogen
Figure 2.1 Apparatus for Synthesis of Nickel Carbonyl


20
strongly dependent on the concentration of carbon monoxide. In the absence of CO,
the lifetime in air at 296 K and at atmospheric pressure is reported to be 60 + 5
seconds (Stedman and Hikade 1980).
Preparation of Heterogeneous Nickel Catalysts
The preparation of metallic catalysts is a highly mystical art with the objective
being to obtain the metal in a highly dispersed state to maximize the available reactive
surface area. In these studies a second desirable feature is high gas throughput over
the catalytic bed for air sampling. The general practice in preparing metallic catalysts
is to incorporate onto an inert support an ionic form of the metal, usually from
aqueous solution, by processes such as adsorption, impregnation, ion exchange, co-
precipitation or deposition (see specific methods below). Catalysts are then dried,
calcined and reduced with hydrogen. The calcination procedure is generally a high
temperature reaction to convert nickel hydroxide or nickel nitrate to nickel oxide
which can then be reduced to metallic nickel. Ionic metallic compounds are chosen
such that relatively low calcination temperatures are sufficient for the reaction as high
temperatures can alter both the nature of the support material as well as increase the
average size of the metal particle formed on the support (thus decreasing the effective
metal surface area). For example high temperature calcination of silica and alumina
supports results in the formation of a thin sheet of mixed oxides over absorbed
metallic complexes which then diminishes the efficiency of the following reduction
step. Typical supports employed are silica, alumina, silica-alumina mixtures, zeolites
and carbon; although in principle virtually any solid can be used as a catalyst support
for the dispersed metal. The general nature of the supports used in this work is
briefly discussed below.
Silica. High surface area silica, e.g. silica gel, is amorphous and formed by the
polymerization of silicic acid in which silanol groups (SiOH) are condensed to siloxane


21
groups (Si-O-Si) producing a silica hydrosol (Anderson 1975). These sol particles are
approximately spherical and consist mainly of a non-ordered arrangement of Si04
tetrahedra (Anderson 1975). The diameter of the primary particles in silica gel is
widely variable and dependent on the manner of preparation. However the gel has
a highly porous structure and is an ideal support material due to its high surface area
and inert nature.
Alumina. The structural chemistry of alumina is highly dependent on the method
of preparation. Two types of alumina were used in these studies: a cationotropic
alumina that was commercially obtained and another made by precipitation using
aluminum nitrate. Although precipitation conditions such as temperature, time and pH
as well as dehydration temperature greatly affect the final nature of the alumina
support, none of these conditions were extensively studied. Alumina, as silica, is
widely used as a support due to its high surface area and inertness.
Magnesia. Much like alumina the physical nature of magnesia depends greatly on
dehydration temperature, but it is also relatively inert and was chosen as a candidate
support material for this quality.
Zeolites. Zeolites are cage alumino-silicate compounds composed of ordered
arrangements of Si04 and A104 tetrahedra linked by corner sharing. The synthetic
zeolites (molecular sieves) can be thought of as interconnecting cavities formed by an
aluminosilicate skeleton. Zeolites vary in their cage size, Si/Al ratio and cation type
(cations provide electrical balance, neutralizing the negative A104 groups). The
characteristics of the two types of zeolites used in these studies, A and X, are listed
in Table 2.1. All zeolites are oxides and much of their surface activity depends on
the extent of water adsorption and subsequent hydration reactions to form hydroxyl


22
TABLE 2.1 Physical Characteristics of Zeolites
Type Si/Al Ratio Approximate free dimensions of main features Approximate number of guest molecules per cavity
Main Cavity diameter
X,Y -1.25 (X) >2.4 (Y) ~1.2nm -0.8-1.Onm 32 H20 17-19 Ar, N2, 02 5.6 cyclopentane 5.4 benzene
Main Cavity diameter
A 1 1.1-1.2 nm (a-cage) 0.4-0.5 nm (window) 29 H20 19-20 NH3
14-16 Ar, N2, 02
15 H,S
12 CHgOH
10 so2
9 C02


23
groups {S-02_+ H20 = S-(OH)2"} and associated stereochemical changes. Hydrated
surfaces of zeolites exhibit ion exchange properties and it is this property which is
exploited in charging these supports with ionic nickel.
Resins. Ion exchange resins can also be used as supports although their main
limitations are thermal and chemical instabilities. The two resins used were thermally
stable to 150C and the strongly acidic cation exchange resin (a polystyrene sulfonic
acid) adsorbed ionic nickel very effectively.
The support materials investigated were Kieselguhr (silica gel), Zeolite (cage
aluminosilicates), alumina, magnesium oxide, Florisil (synthetic magnesium containing
silica gel) glass wool, and ionic and non-ionic polyvinylbenzene resins. Specific
catalyst preparation procedures are now described.
Impregnation method. Nickel nitrate hexahydrate (58 grams) dissolved in 80 mL
of water is added to 50 grams of the desired inert support (coarse silica gel or type 4A
molecular sieve). This mixture is ground in a mortar, for approximately one hour,
until it is relatively homogeneous and flows with the consistency of heavy lubricating
oil (approximately one hour). Homogeneity was not achieved with 6-12 mesh silica
gel support as these particles were very coarse and abrasive. However 4A molecular
sieve does form a thick solution when ground. This mixture is added to a solution of
34 grams of ammonium carbonate monohydrate in 200 mL of water, stirred, filtered
in vacuo, washed twice with 100 mL of water and dried at 100C. It is then calcined
at 500 C for 3 hours, transferred to an appropriate catalyst cell and reduced in situ
at 400C for 1.5-2 hours.
Adsorption method. A selected amount of support (coarse silica gel, type 13X
molecular sieve, Florisil, cationotropic alumina or magnesium oxide) is suspended in


24
a solution of nickel nitrate hexahydrate as above and the mixture remains suspended
at room temperature for five days to allow maximal adsorption of nickel ions on the
support. The mixture is then filtered in vacuo, washed with two 100 mL portions of
water and dried at 100C. Calcination now follows to convert nitrate ions (probably
by evolution of nitric acid and other oxides of nitrogen) and the catalyst is reduced
as above. However the calcination and reduction steps were omitted in the preparation
of catalysts involving resinous supports as they are thermally unstable above 150C.
The cationic exchange resin used was of the strongly acidic type and required extensive
cleaning before use. It was washed with several portions of hot IN NaOH, IN HC1,
deionized water, and then boiled in methanol.
In a modified adsorption method, Raney nickel (50% nickel, 50% alumina) was
adsorbed/suspended on glass wool. Glass wool was first coarsely packed in the catalyst
cell and then the selected amount of Raney nickel added. The cell was vibrated and
sonicated until the nickel dust appeared to evenly coat the glass wool support. The use
of metallic nickel is advantageous as it obviates the calcination and reduction steps
required with ionic nickel.
A third adsorption method involved a magnesia support prepared from a paste of
magnesium oxide and water. The paste was transferred to a large 50 mL syringe and
long strips of material were extruded, air dried and then oven dried (1000C, 48
hours). The magnesia pellets formed from these broken strips were wetted with a
nickel nitrate solution, filtered, dried and calcined and reduced as above.
Co-orecipitation method. A solution of sodium hydroxide (100 grams in 500 mL
of water) was added with vigorous stirring over a 4 hour period to a solution of
aluminum nitrate (227 grams of A1(N03)3-9H20 in 1.5 L of water). Another solution
was prepared by dissolving 51 grams of nickel nitrate hexahydrate in 300 mL of water
containing 23 mL of nitric acid (cone.). This nickel solution was then added slowly
with vigorous stirring to the alkaline aluminum nitrate. The resultant gelatinous


25
mixture was filtered, resuspended in 1 L of water, stirred for 15 minutes and
refiltered. These wash steps were repeated six times. The final product was oven
dried (210C, 16 hours) and sized to the desired mesh.
The physical properties and characteristics of the supports prior to and after
charging with nickel are shown in Table 2.2.
Instrumentation
Novel Carbon Monoxide Detection System. A novel carbon monoxide detection
system was designed and constructed based upon the chemiluminescent reaction of
nickel carbonyl and ozone. This system is composed of the following components: a
custom made dynamic gas dilution system, carbon monoxide to nickel carbonyl
convertor and the chemiluminescent reactor/detector as shown in Figure 2.2.
Dynamic Gas Dilution System. The gas dilution system, constructed by the
author, performs two main functions. First, it contains an ozone generator and
provides a calibrated flow of ozone to the chemiluminescent reactor/detector. Second,
it produces and delivers a calibrated flow of carbon monoxide from stock tanks. This
unit as shown in Figure 2.3 contains an ozone generator, a glass dilution chamber, a
glass reaction chamber and four gas inlet ports each with their own pressure gauge,
pressure regulator, set of flow restrictors and valves. The glass reaction chamber (2.1
X 2.8 X 11.3 cm, volume 66.4 cc) receives the outflow from Port A where it can react
with ozone prior to entering the dilution chamber. The glass dilution chamber (d =
5.715 cm, h = 6.8 cm, volume 174 cc) receives outflows from all ports and the reaction
chamber and provides dynamic dilution of input gases. The ozone generator, extracted
from a Beckman model 952 NOx analyzer, consists of a mercury vapor UV lamp
housed in a Teflon cylinder (Figure 2.4). Ozone can be directed to either the reaction
chamber or to the chemiluminescent reactor/detector. Calibration of this generator is


26
Table 2.2 Physical Properties of Prepared Catalysts
Catalvst Phvsical ProDerties
Nickel adsorbed on silica gel coarse mesh size 8-18, abrasive granules black with bluish sheen
Nickel impregnated 4A molecular sieve approximate mesh size 30 200 gray with light green tint
Nickel adsorbed on 13X molecular sieve mesh size 8-12 tan, gray
Nickel co-precipitated on alumina mesh size 30 200 green
Nickel adsorbed on magnesium oxide mesh size 8 20, long pellets, soft gray,tan
Nickel adsorbed on cationotropic alumina approximate mesh size 200, fine powder light blue
Nickel adsorbed on Florisil approximate mesh size 200, fine particles gray
Nickel adsorbed on XAD-2 mesh size 20 50 (wet) light yellow to tan
Nickel adsorbed on acidic cation exchange resin mesh size 20 50 golden yellow particles
Raney nickel on glass wool very fine powder, 50% Al-Ni on glass wool gray


Figure 2.2 Novel Carbon Monoxide Detection system
Variac
to


Direct
Inlet A Outlet
Direct
Outlet
Inlet 6
0
1
TV
I t 2 $
r
RC
0
1
-Q
06
t t
if 2?
Outlet
. >
it
MC


Inlet .0
t '
lj 2%
Inlet D
o
f-
n
I* 21
o Pressure Gauge Restrictor
o Pressure Regulator RC Reaction Chamber
A Open/Close Valve to Restrictor OG Ozone Generator
V Toggle Switch for flow measurements at the front panel MC Mixing Chamber
Figure 2.3 Dynamic Gas Dilution System


Figure 2.4 Ozone Generator
Lamp Base 0-Ring
Lamp Assembly End-Gap Air-Inlet Fitting
N>
VD


30
described below. Port A, designated for an auxiliary gas, is currently used for delivery
of pure CO to the catalyst convertor cell. The outflow of A can also be directed to
the reaction chamber where it reacts with ozone prior to entering the dilution chamber.
Port B connects to an oxygen tank and feeds to the ozone generator. Port C connects
to a compressed air tank for dilution of gases in the dilution chamber. Port D is used
for delivery of carbon monoxide (typically 200 ppm) to the catalyst convertor.
Associated with each inlet port are two gas flow restrictors having unique flow
characteristics as determined by the pressure set at the inlet port regulators. Each
restrictor was calibrated over an inlet pressure range and their flow characteristics are
shown in Figures 2.5 2.12. All restrictors exhibited a linear relationship between the
inlet pressure set at their respective regulators and the delivered gas flow in their mid
range settings. All experiments were performed at restrictor settings in this linear
range. These restrictors are summarized in Table 2.3. The flow restrictors were made
by compression of 1/8 or 1/16 inch stainless steel tubing, except for A-l which is a
fused silica tube (0.25 mm X 1 m). Selected flows were achieved through repeated
flow measurements and compression until the target flow was obtained.
Ozone analyzer. Ozone concentrations were determined with a Dasibi
Environmental Corporation Model 1003-AH Ozone Analyzer. This analyzer operates
on the principle of absorption of ultraviolet light by ozone. The intensity of UV light
diminishes when it passes through a sample of air containing ozone. Light intensity
is a function of sample pathlength, ozone concentration and the wavelength of the
light. Using the Beer Lambert law, the instrument measures the ozone concentration
in the air sample. The ozone analyzer is attached to the dynamic gas dilution system.
Ozone and air are fed individually into the analyzers sampling column where mixing
occurs. The analyzer was allowed to warm up for one hour with no ozone flow,
followed by a second hour with relatively high concentrations of ozone flowing through
it. The measurements were recorded every 30 minutes after the ozone concentration


31
Figure 2.5 Flow Characteristic of Restrictor A-l


32
Figure 2.6 Flow Characteristic of Restrictor A-2


33
Figure 2.7 Flow Characteristic of Restrictor B-l


34
Figure 2.8 Flow Characteristic of Restrictor B-2


35
Figure 2.9 Flow Characteristic of Restrictor C-l


36
Figure 2.10 Flow Characteristic of Restrictor C-2


37
Figure 2.11 Flow Characteristic of Restrictor D-l


38
Figure 2.12 Flow Characteristic of Restrictor D-2


39
Table 2.3 Flow Ranges of Gas Inlet Restrictors
Pressure Flow1 Pressure Flow1
(psig) mL/min (psig) mL/min
Restrictor Restrictor
A-l 6 11 A-2 6 52
11 29 11 125
16 51 16 206
21 76 21 296
26 103 26 390
B-l 10 0.8 B-2 10 6
15 2.6 15 15
20 3.7 20 26
25 4.9 25 52
30 6.2 30 65
C-l 5 215 C-2 5 1.09xl03
10 411 10 1.87xl03
15 618 15 2.48xl03
20 820 20 3.09x10s
25 1.03xl03 25 3.59xl03
30 1.25xl03 30 4.13xl03
D-l 5 1.0 D-2 5 11
10 2.3 10 26
15 4.0 15 42
20 5.9 20 62
25 8.2 25 85
30 10.9 30 108
1. All flows were measured at ambient atmospheric pressure (approximately 0.8
atm).


40
was changed by changing the pressure on the ozone port of the dynamic gas dilution
system. The results for the calibration of the ozone generator are described in Chapter
III.
Catalyst convertor cell. A crucial aspect of this carbon monoxide detector system
is the conversion of CO to nickel carbonyl. This reaction occurs in the catalytic
convertor consisting of a catalytic cell filled with a selected catalyst. The cell was
optionally heated by means of a Briskheat Flexible Electric Heating Tape (120 v, 156
W, 1/2" w X 2 1) and temperature was regulated by Variac control. This cell was
also used for the hydrogen reduction step in the preparation of catalysts.
Five catalyst cells differing in geometric design were investigated. These cells are
shown in Figure 2.13. For simplicity the cells are labeled 15X, 25X, 50X, 75X and
XX where the number refers to the approximate volume (cc) of the cell. Cell XX was
constructed from a coiled glass gas chromatographic column (1/4 inch X 3 ft).
However, the pressure drop through this cell was too high after packing with the
catalyst to be of any practical use in the convertor system.
Chemiluminescent detector. Nickel carbonyl formed in the catalytic convertor is
combined with an additional flow of carbon monoxide and reacted with ozone in the
chemiluminescent reactor (see Figure 2.2). This unit is based upon a modified
Beckman NOx Analyzer, model 952 (Beckman Instruments, Fullerton, CA) consisting
of a reaction chamber, photomultiplier detector and signal amplifier. The reaction
chamber and the photomultiplier are housed face-to-face in a thermoelectrically cooled
block. A red cut-on filter passing wavelengths greater than 600 nm is fitted over the
reaction chamber. Reactants flow into the chamber through two 1/16 inch orifices
located toward the center of the disk shaped cell. These inlets are located in close
proximity to each other. Across from the inlets is the outlet orifice connected to an
exhaust forepump. The pressure inside the reaction cell is measured with a capacitance


41
i.d. 1/8"
6"
8"
i.d. 1/4"
Catalytic Convertor Cell 15X
Approximate Volume: 15 mL
i.d. 1/8"
c
c
*----4.5"------
-------------- 5" ------------
Catalytic Convertor Cell 50X
4.5"----
DOT 3E 1800 Whitney
4EK0 78/ 75cc.
i.d. 1/2"
--------------- 5.03---------
Catalytic Convertor Cell 75X
Figure 2.13 Catalytic Cell Designs


42
manometer (10 volts full scale equals 1000 torr).
The chemiluminescent reaction is measured with a ten stage Venetian blind type
dynode photomultiplier operated at 1.0 kV. The resultant noise at this setting is 0.10 -
0.12 nA with typically signal currents ranging in the nanoampere range. This small
output current was initially measured with a picoameter (electrometer 610, Keithley
Instruments). To improve signal detection a new amplification system was designed
to replace this electrometer. This amplifier, as shown in Figure 2.14, was constructed
by installing two preamplifiers in series at the anode of the PMT. The first amplifier
sets the gain, converts current to voltage and amplifies the signal. The second
amplifier controls the zero offset and serves as an electronic buffer between the first
preamplifier and the rest of the circuitry. The output of this system is in the 1-2
volt range for an input signal from 0 1000 nA and can then be sent directly to a
computer, a strip chart recorder or, in this case, a digital voltmeter. All electronic
components used for signal amplification and handling are powered by a + 15 volt
power supply.
Procedure for Catalyst Evaluation
The experimental procedure for testing a catalyst consisted of three stages: first,
preparation of the catalyst convertor; second, warm-up of the ozone generator and
chemiluminescent reactor; and third, measurement of the conversion of carbon
monoxide to nickel carbonyl by the chemiluminescent reaction of ozone and nickel
carbonyl.
During the first stage, the catalyst cell is filled with the desired catalyst and heated
to 400C under a flow of hydrogen for two hours. The outflow of the catalyst cell
is disconnected from the chemiluminescent analyzer during the heating period to
prevent flow of water vapor into the analyzer. The catalyst cell is subsequently cooled
to 40C and reconnected to the chemiluminescent analyzer.


Figure 2.14 Nanoameter Amplifier Schematic
10 Turn Pot
100M.Q 501CQ


44
The second stage allows for warm-up and equilibration of the instruments. The
ozone generator is allowed to warm up for two hours with oxygen flowing through
it at 65 mL/min as set by restrictor B-2 on the dynamic gas dilution system. The
vacuum pump is turned on to evacuate the analyzer. The voltmeter dedicated to the
capacitance manometer, the PMT power supply, and the electrometer are switched on
at the same time.
The third stage is the actual catalyst test. Ports D and C respectively deliver
purified 200 ppm CO and air to the mixing chamber of the dynamic gas dilution
system and the output of the mixing chamber is connected to the catalytic convertor.
Dilutions of 200 ppm CO are delivered over the catalyst in the catalytic convertor by
changing the pressures on ports C and D. Additional purified 100% CO may be
delivered to the input side of the chemiluminescent analyzer as desired through Port
A (See Figure 2.2). The ozone flow is always maintained at 65 mL/min by setting
restrictor B-2 at 30 psig and the analyzer cell pressure is controlled by valve V-2 (See
Figure 2.2). The chemiluminescent response is measured by reading the electrometer
after a ten minute equilibration period each time the experimental parameters are
changed.


CHAPTER III
RESULTS
The primary tasks of this project were outlined in Chapter I and include the
construction of a dynamic gas dilution system, the evaluation of a catalytic convertor
for the reaction of carbon monoxide with nickel to form nickel carbonyl and the
assembly of a sensitive chemiluminescent analyzer for nickel carbonyl. Of these tasks
the design, construction and determination of gas flows of the dynamic gas dilution
system have been described in Chapter II. The remaining tasks include calibration of
the ozone generator (a component of the gas dilution system), calibration of the
chemiluminescent analyzer and evaluation of the catalysts.
Calibration of the Ozone Generator
Air (U.S.P. grade) was delivered to the mixing chamber of the analyzer through
port C of the dynamic gas dilution system with both restrictors, C-l and C-2 (Figure
2.3), fully open an at inlet pressure of 30 psig. This resulted in a total flow of 5.4xl03
mL/min. Oxygen was delivered to the ozone generator through port B at various inlet
pressures to regulate its flow. Five consecutive samples were taken every thirty
minutes after changing the pressure setting on port B. Background reading was taken
when the oxygen was delivered without being ozonized and the value subtracted from
all other readings.
The ozone generator was calibrated with a series of four consecutive runs over a
period of two days. The data from these calibration runs are summarized in Table 3.1
and plotted in Figures 3.1-3.3. The first run (Figure 3.1) was carried out after the
ozone generator had been allowed to warm up for one hour without ozone flowing


46
Table 3.1 Ozone Generation
Run #1* *
Run #2
Run #3*
Run #4*
Flow1 Flow1 Ozone Ozone Concentration
C-l + C-2 B-l Generator (ppm)
5.4xl03 2.6 off 0
5.4x10s 0.8 on 1.1x10s
5.4x10s 2.6 on 0.7x103
5.4x10s 3.7 on 0.9xl03
5.4x10s 5.0 on l.lxlO3
5.4x10s 6.2 on 1.3xl03
5.4x103 2.6 on l.lxlO3
5.4xl03 0.8 on 2.0xl03
5.4xl03 5.0 on 1.2xl03
5.4x103 3.7 on l.lxlO3
5.4x103 5.0 on 1.4xl03
5.4x103 6.2 on 1.4xl03
B-2
5.4x103 6.0 on 1.8xl03
5.4xl03 14.9 on 2.0xl03
5.4x103 26.2 on 2.3x10s
5.4x103 38.1 on 2.6x103
1. Flow rates for restrictors are listed in mL/min.
C-l + C-2: Air
B-l: Oxygen
B-2: Oxygen
*Run #1: Morning 8/17/89
Run #2: Afternoon 8/17/89
Run #3: Morning 8/18/89 (after overnight operation)
Run #4: Afternoon 8/18/89


47
Figure 3.1
Ozone Calibration Run#l


48
Figure 3.2 Ozone Calibration Run#2


49
Figure 3.3 Ozone Calibration Run#4


50
through it, followed by one hour of high ozone concentration passing through the
system to condition it. The second run was carried out the same day after the ozone
generator had been in operation for four hours (Figure 3.2). The third and fourth runs
were performed the following day with the generator and monitor both in operation
overnight (Figure 3.3).
Additionally a final calibration run was performed on the dynamic gas dilution
system as well as the the ozone generator by making several dilutions of 03 with air
in the systems mixing chamber. These data are tabulated in Table 3.2 and plotted in
Figure 3.4.
The results of the ozone generator calibration runs indicate that there is a time
dependency in ozone concentration with the concentration running higher after the
generator had been in operation for a certain length of time, as shown by the
difference between the morning and afternoon values. Also there is a delay in
establishing equilibrium concentrations upon modification of the regulator setting
because the volume of the generator is large (approximately 200 mL) and the flows
through the restrictor are low, ranging from 0.8 to 6.2 mL/min with B-l and 6.0 to
65 mL/min with B-2 (Table 3.1). Checking the dynamic gas dilution system by
maintaining the ozone flow constant and varying the air flow indicated that the
generator had a reasonably steady output and that deviations from linearity reflect the
non-linear pressure versus flow behavior of these restrictors and the time lag in
establishing equilibrium concentrations in the ozone generator, as shown in Figures 3.1
through 3.4.
In our studies the most favorable ozone flow was established (see Calibration of
Chemiluminescent Analyzer, p. 53) and used throughout the investigation of various
catalysts, thereby minimizing any time dependency variability. Additionally a two
hour warm-up period was allowed each time the generator was used in order to
achieve a more consistent and stable ozone production.


Table 3.2 Ozone Dilution
Flow1 Flow1 Ozone Dilution Final Ozone
C-l + C-2 B-l Concentration1 2 Factor Concentration
5.4xl03 2.6 0.518 2.1xl03 l.lxlO3
6.3xl03 2.6 0.419 2.4x103 l.OxlO3
7.1xl03 2.6 0.359 2.7xl03 1.0x10s
1. Flow rates for restrictors are listed as mL/min
C-l + C-2: Air
B-l: Oxygen
2. Mean values of five measurements in ppm.
3. Listed in ppm.


B-l Restrictor Pressure Cpsig)
52
25 r
20;
15;
10;
5 -
0L
900
1 I L i
1000 1100
Ozone Concentration (ppm)
___i
1200
Figure 3.4 Ozone Dilution Curve


53
Calibration of the Chemiluminescent Analyzer
The chemiluminescent analyzer was calibrated with standard gas dilution of nickel
carbonyl. Approximately 9.7 mg of synthesized Ni(CO)4 was transferred to a 15-
liter evacuated plastic reservoir. This was then pressurized with purified carbon
monoxide to a pressure of 8 psig to yield a final concentration of approximately 60-
67 ppm nickel carbonyl. This standard gas dilution was then delivered to the
chemiluminescent analyzer through port D of the gas dilution system while added CO,
ozone and air were delivered through ports A, B and C respectively. The optimum
ozone flow was achieved with a 30 psig inlet pressure on restrictor B-2 corresponding
to a concentration of approximately 3.2xl03 ppm ozone in the analyzer. This
optimized ozone flow was held constant through the remainder of the calibration runs.
Purified carbon monoxide flow was then varied to produce the maximal signal
enhancement. This flow was achieved with an inlet pressure of 16 psig at restrictor
A-2 corresponding to a CO flow of 210 mL/min. The concentration of nickel
carbonyl was varied by either altering the flow of gaseous Ni(CO)4 through port D
or changing the air flow through port C. The chemiluminescent signal intensity was
linear with the nickel carbonyl concentration over the 0.6 2.6 ppm tested range. The
detection limit, which is determined by the PMT noise, is approximately 0.42 ppm as
shown in Figure 3.5.
Catalyst Evaluations
Cell Design. As described in Chapter II and shown in Figure 2.13, five catalytic
cells of varying geometry and construction were used in the preliminary tests of
catalysts. The intention was to construct a cell that would provide a long pathlength
to increase sample residence time and a large volume to enhance the surface area
available for the heterogeneous gas metal reaction. Of the five cells tested, four


Signal Intensity (nA)
54
Ni (CO)4 Concentration (ppm)
Figure 3.5 Nickel Carbonyl Standard Curve


55
were made of glass (15X, 25X, 50X, XX) and one was made of metal (75X). The
XX cell, a 1/8 inch glass chromatography column, was impractical because of its long
length and the large resistance to gas flow when packed with catalyst. The remaining
cells all work satisfactorily with no obvious distinction among them. The cell that was
chosen for all the catalyst evaluations was the 75X cell, a metal container with a
volume of approximately 75 mL. The cell was chosen simply on the basis of its ease
of handling (non-breakable and convenient to load) and heating properties.
Catalyst Performance. The catalysts prepared as described in Chapter II were
tested for their ability to facilitate the conversion of carbon monoxide to nickel
carbonyl. The 75X cell was used for all these evaluations and the cell was packed
fully and evenly. All but three of the catalysts (as listed below) were initially reduced
at 400C under hydrogen for approximately two hours then cooled to 40C and tested
by passing over them purified carbon monoxide at 200 ppm. The three catalysts that
were not reduced were Dowex 50X8-50 ion exchange resin, the Amberlite XAD-2
non-ionic polymeric adsorbent and the cationotropic alumina. The resins were not
reduced due to their thermal labilities. The alumina was used as a check of the active
catalytic state of nickel. Although the ion exchange resin had been charged with Ni2+
and not reduced, this catalyst was very efficient in converting CO to Ni(CO)4. We
therefore used the alumina support to investigate if another catalyst with cationic
affinity would behave in a similar manner.
Described below are the evaluations of these catalysts. Also prior to each
experiment both 100% and 200 ppm carbon monoxide were tested for carbonyl
contaminants by flowing them directly into the chemiluminescent analyzer and
determining their reactivities to ozone.
The performance of nickel adsorbed on 13X molecular sieve is shown in Table 3.3.
The main problem with this catalyst is its large mesh size (8-12) resulting in a low


56
residency time of the carbon monoxide sample in the catalytic convertor. Increasing
the pressure in the reactor by partial closure of the exit valve (V-2) resulted in an
enhanced signal. This is explained by realizing that this partial closure of V-2 results
into a slower flow rate in the catalytic convertor and a concomitant increase in both
the pressure and residence time in the catalytic convertor.
Table 3.3 Catalyst Performance: Nickel Adsorbed on 13X Molecular Sieve
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 3.8 40 0.12
0 65 0 26 3.8 40 0.12
210 65 0 26 40 0.12
0 65 0 62 4; 40 0.17
0 65 215 62 40 40 0.17
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample
* Pressure in cell increased by closing valve V-2.
As shown in Table 3.4, the catalyst prepared by adsorption of nickel on the cation
exchange resin is active in producing nickel carbonyl only when heated to 150C.
Various concentrations of CO were delivered over this catalyst and the resulting signals
plotted as a function of their respective CO concentrations are shown in Figure 3.6.
All measurements were made with additional 100% CO flow into the analyzer cell as
this was found to slightly enhance the signal (from 0.15 to 0.18 nA). However this


57
catalyst became deactivated after 90 minutes. The used catalyst was later leached with
2N HC1 and found to still contain Ni2+.
Table 3.4 Catalyst Performance: Nickel Adsorbed on Dowex 50X8-50 Resin
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 4 40 0.12
210 65 0 26 4 40 0.12
0 65 0 62 20 40 0.13
0 65 0 85 22 40 0.14
0 65 0 42 21 40 0.13
0 65 215 42 23 40 0.12
0 65 215 0 22 40 0.11
210 65 0 62 23 150 0.17
210 65 410 62 30 150 0.14
210 65 620 62 35 150 0.14
0 65 820 42 36 150 0.15
210 65 820 42 36 150 0.18
210 65 620 42 36 150 0.19
210 65 410 42 36 150 0.22
210 65 215 42 36 150 0.26
210 65 0 42 36 150 0.35
0 65 0 42 36 150 0.41
0 65 0 62 36 150 0.42
0 65 0 85 36 150 0.42
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample


CO Concentration (ppm)
-*------------
\
100
Figure 3.6 Catalyst Performance: Nickel Adsorbed on Dowex 50X8-50


59
As stated earlier cationotropic alumina was used as a support to study the possible
active state of nickel in these catalysts. The catalyst was prepared by soaking the
support in a nickel solution followed by washing with water, drying and calcination.
The calcination step is thought to remove nitrates as nitric acid and NOx gases leaving
nickel on the support in the form of NiO and NixAlyOz. However this catalyst was
not reduced before testing and the results listed in Table 3.5 show that it was not
active at 40C. Activity was observed at higher temperatures but the response time
was very slow due to the fine particulate nature of the support. Addition of purified
100% CO did enhance the signal (0.28 to 0.38 nA) but this effect appears to result
from CO flowing back from the reactor cell to the convertor, opposite to the sample
flow and caused by partial closure of V-2. Despite the manner in which this catalyst
was prepared it is not possible to know with certainty the state of nickel on the
support except that it is essentially non-reduced. The performance then of this catalyst
is particularly interesting as the literature regarding gaseous nickel carbonyl preparation
refer to nickel as being in the metallic state.
Table 3.5 Catalyst Performance: Nickel Adsorbed on Cationotropic Alumina
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 4 40 0.12
0 65 0 85 36 40 0.14
210 65 0 85 36 40 0.14
0 65 215 42 36 40 0.14
0 65 0 42 36 150 0.28
210 65 0 42 36 150 0.38
0 65 0 42 36 150 0.48
0 65 0 62 36 150 7
0 65 0 0 36 150 7
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample


60
Table 3.6 shows the catalytic activity of nickel adsorbed on silica gel. These data
indicate that this catalyst is quite inactive for our purposes. Addition of purified 100%
CO seems to increase the signal intensity to a slight degree. However examination of
the collected data reveals that signal enhancement occurred again only upon increasing
the reactor cell pressure by partial closure of V-2. This effect was observed both with
and without delivery of additional purified 100% CO to the chemiluminescent analyzer
indicating that closure of this valve also results in backflow of sample CO into the
catalytic convertor cell with an increase in both the pressure and CO residence time
over the catalyst. This indicates that the catalyst is slightly active but only at much
longer residence times and higher pressures. Again this observation is reasonable
because the silica gel used was very coarse (mesh 6-12) and consequently the sample
pressure and residence time becomes too low for efficient conversion of CO to
Ni(CO)4.
Table 3.6 Catalyst Performance: Nickel Adsorbed on Silica Gel
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 4 40 0.12
0 65 0 42 4 40 0.12
0 65 0 42 20 40 0.12
0 65 0 85 36 40 0.14
210 65 0 85 36 40 0.15
0 65 0 85 42 40 0.17
210 65 0 85 42 40 0.17
210 65 0 0 42 40 0.17
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample


61
Table 3.7 illustrates the catalytic behavior of nickel adsorbed on magnesium oxide
which was tested after calcination and hydrogen reduction. The catalyst does not
respond to 200 ppm CO when the reactor cell pressure is 4.2 torr, however a signal is
observed as the reactor pressure is increased by partial closure of V-2. Again this
suggests that the CO sample residence time and pressure in the catalytic convertor are
too low for efficient CO to Ni(CO)4 conversion. Peculiar to this catalyst is the reduced
signal intensity upon addition of purified 100% CO to the reactor cell.
Table 3.7 Catalyst Performance: Nickel Adsorbed on Magnesium Oxide
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 4 40 0.12
. 0 65 0 42 4 40 0.12
0 65 0 42 36 40 0.22
52 65 0 42 36 40 0.19
210 65 0 42 36 40 0.19
0 65 0 42 36 40 0.19
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample
The evaluation of the nickel co-precipitated with alumina catalyst was performed
with the catalytic convertor cell only partially filled due to the unavailability of
sufficient support to prepare the catalyst. This catalyst has a wide mesh size, ranging
from 30 to 200. It was subjected to both calcination and hydrogen reduction. Larger


62
signals were obtained at higher reactor cell pressures probably due to lengthened
sample residence time in the convertor cell (see Table 3.8). These higher reactor cell
pressures were required to generate a signal possibly because much less catalyst was
used. Addition of purified 100% CO did not enhance the signal intensity, rather at
one point it slightly reduced it. It is possible that backflow of 100% CO into the
catalytic convertor might have exhausted all the available catalytic sites and deactivated
the catalyst.
A series of dilutions of the 200 ppm CO sample were delivered over the catalyst
and those results are plotted in Figure 3.7. As shown the response to CO is linear
even though the activity (that is the signal response) is quite low. The response might
have been better with a completely filled catalytic cell, and this catalyst should be re-
evaluated.
Table 3.8 Catalyst Performance: Nickel Co-precipitated with Alumina
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 4 40 0.12
0 65 0 42 4 40 0.14
0 65 0 42 36 40 0.15
210 65 0 42 36 40 0.15
0 65 0 85 60 40 0.20
210 65 0 85 60 40 0.18
0 65 0 85 60 40 0.28
0 65 215 85 60 40 0.20
0 65 410 85 60 40 0.18
0 65 620 85 60 40 0.16
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample


63
=c
c
+->
iH
0)
c
0J
c
o
c
cn
*I
LO
Figure 3.7 Catalyst Performance: Nickel Co-precipitated with Alumina


64
Nickel adsorbed on Florisil does not demonstrate much catalytic activity with 200
ppm CO at 40C even upon increasing the reactor cell pressure to 36 torr. However
addition of purified 100% CO at the same pressure (36 torr) did slightly enhance the
signal. This indicates that the nickel on Florisil catalyst can respond to 100% CO but
not lower concentrations.
Table 3.9 Catalyst Performance: Nickel Adsorbed on Florisil
Flow1 Flow1 Flow1 Flow1 Pressure Temperature Signal
A-2 B-2 C-l D-2 Cell(torr) Cell (C) (nA)
0 65 0 0 4 40 0.12
0 65 0 42 4 40 0.12
0 65 0 42 36 40 0.12
210 65 0 42 36 40 0.17
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample
The catalysts Raney nickel on glass wool and nickel impregnated 4A type molecular
sieve were also non-responsive (data not shown) to 200 ppm CO at a flow rate of 42
ml/min under either low (4 torr) or higher pressures (36 torr). Addition of 100% CO
did not produce any increase in signal intensity, except under conditions where the
pure CO leaked into the convertor cell.
The data for the catalytic activity of Amberlite XAD-2 nonionic macroreticular
polymer charged with nickel are listed in Table 3.10. Although these evaluations were
performed with the cell only 67% full, the reactivity was quite high even at 40 C.
With no sample CO flow over the catalyst a signal 1.7 times the noise level was


65
observed simply due to the drawing of ambient air into the cell by the suction of the
exhaust vacuum pump. Upon delivery of 200 ppm CO the signal maximized at 14 torr
pressure in the chemiluminescent analyzer and was constant. Addition of purified
100% CO did not enhance the signal and was not used in subsequent dilution
experiments. These results are plotted in Figure 3.8 and show a minimal signal of 0.17
nA for 0.90 ppm CO.
Table 3.10 Catalyst Performance: Nickel Adsorbed on Amberlite XAD-2
Flow1 Flow1 Flow1 Flow
A-2 B-2 C-l D-2
0 65 0 0
0 65 0 9
0 65 0 9
0 65 0 9
210 65 0 9
0 65 0 9
0 65 0 11
0 65 215 11
0 65 410 11
0 65 620 11
0 65 820 11
0 65 l.OxlO3 11
0 65 1.2xl03 11
0 65 1.4xl03 11
0 65 1.7xl03 11
0 65 2.3x10s 11
0 65 0 11
0 65 0 26
0 65 0 42
0 65 0 11
0 65 l.lxlO3 11
0 65 1.0x10s 11
0 65 620 11
0 65 410 11
Pressure Temperature Signal
Cell(torr) Cell (C) (nA)
4 40 0.12
4 40 0.25
11 40 0.36
14 40 0.40
14 40 0.35
14 40 0.44
14 40 0.46
14 40 0.37
14 40 0.32
14 40 0.29
14 40 0.25
14 40 0.24
14 40 0.21
14 40 0.20
14 40 0.18
14 40 0.17
14 40 0.45
14 40 0.47
14 40 0.47
14 40 0.50
14 40 0.22
14 40 0.24
14 40 0.30
14 40 0.34
1. Flow rates at restrictors are listed as mL/min.
A-2: Added CO (100%)
B-2: Ozone
C-2: Air
D-2: CO Sample


66
+J
* rH
0)
c
0)
+->
c
D
c.
CD
* rH
CO
CO Concentration (ppm)
Figure 3.8 Catalyst Performance: Nickel Adsorbed on Amberlite XAD-2


67
In summary Table 3.11 lists the minimal chemiluminescent signal detected with its
corresponding carbon monoxide concentration. These data should be reviewed with
caution as important factors such as residence time, cell pressure and temperature were
not precisely controlled nor examined. However it is noteworthy that in this survey
evaluation of catalyst materials and preparations, several catalysts responded quite
strongly. Indeed, one preparation, the nonionic XAD-2 produced a chemiluminescent
signal with less than 1 ppm of carbon monoxide in air.
Table 3.11 Summary of Catalyst Performance
Catalyst
(nickel on)
Response1
( nA / ppm CO)
Nonionic Amberlite XAD-2
Dowex 50X8-50 Resin
Alumina
Florisil
Cationotropic Alumina
13X Molecular Sieve
Raney Nickel (on glass wool)
4A Molecular Sieve
Magnesium Oxide Pellets
Silica Gel
0.17 / 0.9
0.18 / 7.4
0.16 / 22
0.17 / 27
0.28 / 79
0.17 / 97
NR
NR
NR
NR
1. NR = no response at any CO level
Although many questions still need to be answered the primary tasks set forth in
this project have been addressed. A dynamic gas dilution and ozone generation system
have been constructed, a variety of catalysts have been prepared by several methods
with different support materials, and a sensitive chemiluminescent detector for nickel
carbonyl has been assembled. Further, the detection of sub parts-per-million CO levels
has been accomplished with this novel type of CO monitor.


CHAPTER IV
DISCUSSION
The main objective of this investigation was to assess the feasibility of a more
sensitive technique for measuring carbon monoxide below 1.0 ppm. Two independent
discoveries, the Mond process for the purification of nickel (Mond et al 1890) which
resulted in the production of the first metal carbonyl, namely nickel carbonyl, and the
chemiluminescent reaction between Ni(CO)4 and ozone (Morris and Niki 1970)
suggested that it might be possible to detect CO at very low levels. The
chemiluminescent reaction between nickel carbonyl and ozone became the basis for the
Ni(CO)4 detector developed by Stedman et al (Stedman et al 1979). This detector
offers very high sensitivity with a reported nickel carbonyl detection limit of 2 ppb.
The reaction for the formation of nickel carbonyl has long been known and industrially
applied to the production of pure nickel. However there are still no documented
studies on the kinetics of this reaction. Our interest was to determine if the reaction
would occur at sufficient rates to allow measuring CO in "real time". Although the
emphasis of the project is in the CO to Ni(CO)4 conversion via suitable catalyst, the
subsequent step, that of nickel carbonyl detection had to be developed in order to
assess the success of that conversion process. Hence the project started with the
custom fabrication of an Ni(CO)4 analyzer. The nickel carbonyl analyzer developed
by Stedman and the results of our analyzer will be discussed in the first part of this
chapter. The second and main phase of the project was investigating the conversion
of carbon monoxide to nickel carbonyl. Initially the production of Ni(CO)4 was
attempted by a slightly modified Mond process and found to be successful. In the
Mond process for purifying nickel, carbon monoxide is passed Over finely divided
metallic nickel that had been previously reduced at 400C under a flow of hydrogen


69
and then cooled to room temperature (Mond et al 1890). We however used a
laboratory prepared nickel catalyst to perform this same conversion. The next task was
to prepare a variety of catalysts to determine if this heterogeneous conversion was fast
enough to be useful in our detection scheme. The catalyst preparation phase was
therefore undertaken. With the underlying assumption that large surface area translates
into more conversion, the intention was to charge nickel onto a stable, solid support
in a highly dispersed state. The results of these catalysts activities are discussed in
the second part of this chapter. Unexpectedly another question arose during these
studies, namely what is the exact form or forms of nickel that are catalytically active.
Lastly we came to realize that a better understanding of the actual nickel carbonyl and
ozone reaction was necessary. Prior research of this reaction (Stedman et al 1979,
Morris and Niki 1970) had shown that the nickel carbonyl response is enhanced upon
addition of carbon monoxide. We did not consistently observe this effect. This aspect
will be reviewed along with what is currently thought about the chemiluminescent
reaction of ozone and nickel carbonyl.
Stedman et al (Stedman et al 1979) developed a detector for the measurement of
Ni(CO)4 in air based upon the chemiluminescent reaction of nickel carbonyl with
ozone. Their instrument, capable of selectively detecting Ni(CO)4 in air over the range
of 2 100 ppb using a CO modulation technique, was' built by modifying a
chemiluminescent analyzer (Thermo Electron Corp., model 8A). The modifications
included the following:
1. The red filter between the PMT and the reaction chamber was replaced by a
2 mm Pyrex window.
2. The sample flow capillary was replaced with a larger bore capillary (0.025 inch
i.d.) to increase sample flow.
3. The ozonizer was connected directly to a 110 V AC line to eliminate the 0.05
Hz pulsing circuit.


70
4. Purified CP grade carbon monoxide was introduced into the chamber along with
ozone and the sample.
The sample flow rate was typically 2.2 L/min but operation at lower flow rates with
a concomitant decreased instrument sensitivity was possible. Carbon monoxide flow
rate averaged 100 mL/min (alternating 200 mL/min for one minute, then no flow for
the next minute) although linearly increasing this rate up to 500 mL/min produced an
enhanced linear response to nickel carbonyl. At all CO flow rates the response to
Ni(CO)4 was linear over the range 0-100 ppb. The air flow rate to the ozonizer was
200 mL/min. Although atmospheric humidity did quench the chemiluminesecent
reaction by a factor of 4, the addition of a permeation tube drier to the inlet system
reduced this water effect to less than 5%.
The purpose of carbon monoxide modulation was to remove any interferences by
nitrous oxide (NO/Ni(CO)4 rejection ratio > 25,000). Without CO modulation the
instrument was only 3 times more sensitive to Ni(CO)4 than NO. The rejection ratio
for Fe(CO)5 was about 6. This can be improved by using a green sensitive PMT and
interference filter at the expense of a reduced signal strength of about two orders of
magnitude. Overall the minimum detection limit for Ni(CO)4 in dry air was 2 ppb
using an average flow of 100 mL/min carbon monoxide.
As mentioned earlier in Chapter II the chemiluminescent detector used in the
present studies is a modified Beckman NOx analyzer, model 952. The major
modifications are:
1. The PMT operating voltage was reduced from 2.0 to 1.0 kV which reduced the
background noise level to 0.12 0.14 nA.
2. The signal was measured mostly with an electrometer although a new
amplification system with two outputs (digital and strip chart recorder) was
constructed for the later stages of the project.


71
3. The ozone generator was installed into a newly constructed dynamic gas dilution
system.
This detector was capable of detecting Ni(CO)4 at 0.42 ppm.
A number of factors could be contributing to the difference in sensitivities
between the nickel carbonyl detector described by Stedman (Stedman et al 1979) and
our instrument. These factors include flow rates, operating cell pressures, analyzer cell
volume, optical filtering and PMT cooling. Listed in Table 4.1 are the reported
operating conditions for nickel carbonyl detection and, as stated, above the factors that
appear most critical for optimum sensitivity are cell volume and total flow rates which
determine cell residence time.
During the final stages of this study we became aware of an instrumental design
problem.' This problem involves the method by which gas flows were controlled in our
system. The dynamic gas dilution system was designed to allow the control of flows
as well as make dilutions of gases. However when two gases are diluted to a desired
degree, the flow of the the diluted gas is the sum Of the flows of the individual
components. Another flow meter and valve are required on the exit side of the
dilution chamber to control the flow of the diluted gas and vent any excess. Our
instrument calibration with nickel carbonyl was performed in such a way that each
time a dilution was made the total flow through the reactor was increased. This would
effectively reduce the residence time of nickel carbonyl in the chemiluminescent
analyzer. To partially counteract this effect, the cell pressure was held relatively
constant, however the enhanced flow still affects residence time as well as quenching.
Complete information about the analyzers cell volumes is not available. Our
reactor cell has a relatively small volume of 4 mL and this would also decrease the
sample residence time before the PMT detector.
Of the other parameters, the operating reactor cell pressure did not change
significantly among the detectors. However the operating PMT parameters of voltage,
PMT temperature and signal to noise ratio did differ from the present work, and could


72
Table 4.1 Reported Operating Parameters of Ni(CO)4 Analyzers
Reference: Stedman 1976
Instrument: Thermo-Electron Model 12A
Optical Filtering: Kodak #74 Wratten Filter (A = 5390 A0)
Reactor Pressure: 5 torr
Reactor Volume: NR
Sample Flow: 50 mL/min
Added CO Flow: 50 mL/min
Ozone Flow: NR
PMT Voltage: 1400 V
Signal Response: 102 nA equivalent to 100 ppb Ni(CO)4
Reference: Stedman 1979
Instrument: Lab Flow System
Optical Filtering: Red cutoff filter passing A> 600 nm
Reactor Pressure: 1.5 torr
Reactor Volume: 150 mL
Sample Flow: 150 mL/min
Added CO Flow: 100 mL/min
Ozone Flow: 30 mL/min
PMT Temperature: 0C
Signal Response: 68 nA equivalent to 10 ppb Ni(CO)4
Reference: Stedman 1979
Instrument: Thermo-Electron Model 8A
Optical Filtering: none; CO modulation to remove NO interference
Reactor Pressure: NR
Reactor Volume: NR
Sample Flow: 2.2 L/min
Added CO Flow: 200 mL/min
Ozone Flow: NR
PMT Parameters: NR
Signal Response: no current reported, 2 ppb Ni(CO)4 detection limit
Reference: Present Study
Instrument: Beckman Model 952 NOx Analyzer
Optical Filtering: Red cut-on filter passing A> 600 nm
Reactor Pressure: 6.8 8.7 torr
Reactor Volume: 4 mL
Sample Flow: 11.0 mL/min
Added CO Flow: 206.2 mL/min
Ozone Flow: 65.3 mL/min
PMT Voltage: 1000 V
PMT Noise: 0.12 nA
Signal Response: 0.24 nA equivalent to 420 ppb Ni(CO)4
Note: NR = not reported


73
account for some of the discrepancies. Cooling the PMT to zero degree, increasing the
PMT operating voltage and maximizing the signal to noise ratio would improve the
overall instrument sensitivity. An investigation of these PMT parameters was not
performed in the present studies. The PMT voltage was set to 1.0 kV from 2.05 kV,
which is the lower end of its operating voltage. Noise at this voltage averaged 0.12
nA without any cooling of either the PMT or analyzer block.
Another factor involves the preparation of standard gas stocks of nickel carbonyl.
Although nickel carbonyl is a gas that liquifies at 43 C, it has a high vapor pressure
(261 mm Hg at 15C) which makes handling and precise transfers very difficult.
Therefore we are reasonably cautious of estimating the precise concentration of our
nickel carbonyl standard and without a second, independent analysis of Ni(CO)4
content we view these dilutions as approximate values.
Catalyst Performance
Table 4.2 lists the catalysts which were evaluated in this study and their method
of treatment prior to testing. Most of these catalysts were prepared following standard
procedures as reported in the literature, that is they were first charged with nickel,
calcined at 400 600 C for 4 hours and then reduced at 400 C under a flow of
hydrogen for approximately 2 hours prior to testing. Several of the catalysts were
treated slightly different as there were indications that calcination could deactivate the
nickel to varying degrees dependent on the type of support to which it was bound,
or because the support was thermally labile at the calcination temperature.
With this in mind, two catalysts were prepared but not calcined, the Dowex 50X8-
50 ion exchange resin and the Amberlite XAD-2. All catalysts except the Dowex
resin, Amberlite XAD-2, and the cationotropic alumina were reduced with hydrogen
at 400C and then cooled to 40 C prior to testing.


Table 4.2 Catalysts Prepared and Tested
Catalyst Mesh Treatment Response1
(nickel on) Calcination Reduction
13X Molecular Sieve 8-12 Yes Yes Active
Amberlite XAD-2 20 50 No No Reactive
Dowex 50X8-50 Resin 20 50 No No Reactive
Cationotropic Alumina 200 Yes No Active
Magnesium Oxide Pellets 8-20 Yes Yes Active
Alumina 30 200 Yes Yes Active
Florisil 200 Yes Yes Non-reactive
Raney Nickel Fine No Yes Non-reactive
4A Molecular Sieve 30 200 Yes Yes Non-reactive
Silica Gel 8-18 Yes Yes Non-reactive
1. Reactive = response to <200 ppm CO
Active = response to 200 ppm CO
Non-reactive = response only to 100% CO


75
As shown in Table 4.2 a number of the catalysts did successfully convert carbon
monoxide to nickel carbonyl. These catalysts can be broadly catagorized into three
groups, a reactive group that responds to CO below 200 ppm, an active group that
reponds to CO at 200 ppm and an inactive group that responds only to 100% CO. No
conclusive correlation has been determined between catalytic activity and the type of
catalyst treatment (calcination/reduction) as it appears that these treatments could
produce different effects with different support materials.
In addition to the type of treatment of the catalysts, the results indicate that the
surface area (mesh size) of the catalyst is a predominant factor in the conversion
process. Too large of an area, as with the cationotropic alumina (mesh 200), produced
a large signal but the response time was very slow. Such a long response time would
make the catalyst unsuitable for continuous air measurements. Catalysts with slightly
less surface area (both resins, Dowex 50X8-50 and Amberlite XAD-2, mesh 20 50)
were very reactive and responded rapidly to less than 200 ppm CO. Catalysts with less
surface area, such as the coarse supports including 13X molecular sieve and silica gel,
were much less efficient in converting CO to Ni(CO)4. The observed response is a
combined effect of low surface area and short residence time of the sample CO in the
convertor cell.
A third contributing factor which is also of peculiar interest is the positive
response obtained with the Ni2+ charged on the strongly acidic cation exchange resin,
the cationotropic alumina and the nonionic macroreticular Amberlite XAD-2. Of these
catalysts, none were reduced with hydrogen at 400C and yet all three were very
reactive in converting CO to Ni(CO)4. Literature reports on the preparation of nickel
carbonyl suggest that it is elemental nickel which reacts with carbon monoxide to form
nickel carbonyl. However the form of nickel charged on these two supports is believed
to be predominantly other than Ni.
The resins as well as the cationotropic alumina are liquid chromatographic supports
designed for usage in solutions. As such they either trap cations by adsorption to their


76
negatively charged surface or by size exclusion. Nickel is most likely still in the +2
oxidation state on the cationic resin supports after charging and drying. The state of
nickel on the nonionic XAD-2 is not clear, yet it appears to be readily available for
reacting with CO.
A reasonable assumption is that Ni2+ is reduced to Ni by carbon monoxide prior
to the formation of nickel carbonyl. It is possible that the loss of catalytic activity
observed with the cation exchange resin results from the formation of a net negative
surface charge as Ni2+ is reduced and carbonylated. This negative surface charge
would bind the remaining nickel ions more tightly and consequently less available for
reaction.
In summary two factors appear to be crucial in the conversion of CO to Ni(CO)4:
the surface area available for the catalytic conversion and the state of nickel on the
support. The surface area effect is logical because larger surface area (smaller mesh
size) translates into a greater number of catalytic sites for the conversion and a longer
CO residence time in the catalyst cell. This was observed with the resins and the
cationotropic alumina. Yet Florisil and the 4A molecuar sieve impregnated with nickel
did not react significantly inspite of their fine particulate nature. Thus other factors
are important in the conversion process. One such factor is the chemical state of
nickel on the support material. All three catalysts that were not reduced (the resins
and cationotropic alumina) produced a positive conversion response. This behavior was
unexpected in light of literature reports that elemental nickel is involved in the
conversion reaction.
Other factors that must be contributing to the activity of these catalysts is the type
of support material and the method of treatment in preparation. Some of the catalysts
that were calcined and reduced responded positively (eg. 13X molecular sieve and
co-precipitated alumina) whereas others did not. Of course it is premature to limit
discussion of catalytic activity to the above factors without further investigation.
However despite the ambiguities regarding the exact parameters controlling catalytic


77
activity, the data indicate that the kinetics of the CO to Ni(CO)4 conversion reaction
are favorable and that a more rigorous investigation of the above factors would allow
one to design a workable catalyst for this detection system.
Chemiluminescence of Ozone and Nickel Carbonvl
Although the reaction between ozone and nickel carbonyl is the basis of the
proposed CO monitoring system, the exact mechanism for this chemiluminescent
reaction is not fully understood. In a study on the reactions of carbonyl metal
compounds with oxygen atoms (Becker and Schurgers 1971) MO* emissions were
observed in all reactions of the type
M(CO)n + 0 (1)
except for
Mo(CO)6 + O (2)
*
These MO emissions were reported to result from direct formation of this species
from oxygen atoms and metal atoms, with the metal atoms being formed from carbonyl
metal compounds by stepwise degradation of the CO ligand and the formation of C02
(Becker and Schurgers 1971).
M(CO)n + nO-------> M(g) + nC02 (3)
However the reaction of ozone with nickel carbonyl is reported to result in the
formation of NiO as shown in equation 4 (Groth et al 1974).


78
Ni(CO)4 + 03--------> NiO + products (4)
Furthermore the NiO* emission (460 750 nm) was observed with pure Ni(CO)4 and
03 only in the presence of carbon monoxide (Morris and Niki 1970).
NiO + CO--------> Ni(gj + C02
Ni(g) + 03------> NiO* + 02
Without added CO a black deposit forms on the walls of the reaction vessel which was
reported to be indicative of a nonluminous reaction. NiO emission could be
regenerated upon addition of pure CO downstream. This implies that a nickel
compound had travelled some distance before it reacted with carbon monoxide to
produce NiO chemiluminescence. The most likely species would be NiO which could
easily be reduced by CO (equation 5). The subsequent reaction of ozone with the
reduced Ni^gj is exothermic by 76 kcal (equation 6) and the spectroscopically observed
*
species, NiO has an excitation energy of about 57 kcal/mol. The exothermicity of
equation 5 would provide the needed energy for the observed emission.
NiO*------> NiO + hv (7)

In their studies of nickel carbonyl ozone chemiluminescence, Stedman et al (Stedman
et al 1979) observed a signal enhancement of approximately 100 times upon addition
of pure carbon monoxide. This is in agreement with the aforementioned proposed
mechanism that CO must be involved in the formation of emission precursors (equation
5). Addition of hydrogen or sulfur dioxide instead of carbon monoxide did not
produce this enhancement effect.
In summary then, a combined mechanism which is consistent with both studies can
be outlined as follows:
(5)
(6)


79
03 + Ni(CO)4 > NiO + products (4)
NiO + CO > Ni + C02 (5)
Nl(g) + 3 > NiO* + 02 (6)
NiO* > NiO + hi/ (7)
NiO* + M > NiO + M (8)
During the course of our investigations we tested the effect of added pure CO on
the emission from nickel carbonyl ozone chemiluminescent reaction by measuring the
signal in the absence as well as the presence of added carbon monoxide. Also
to ensure that NiO was not being deposited on the walls of the reactor in a non-
luminous reaction, the analyzer was disassembled and inspected for black deposits on
the reactors walls after several tests. In all cases no deposits were observed and no
significant enhancement of signal intensity was ever observed upon the addition of
pure carbon monoxide. In fact, in certain instances, excess CO had a simple diluent
effect and reduced the chemiluminescent response. However when the calibration of
the chemiluminescent analyzer was performed with standard Ni(CO)4 plus added 100%
carbon monoxide, a slight enhancement of the signal occurred and the flow of added
CO was optimized at 210 mL/min. This flow was then used in subsequent evaluations
of the catalysts. In certain tests, the chemiluminescent signal appeared to be increased
by this added CO, but in others the signal was unaffected (MgO, coprecipitated
alumina, XAD-2). Of the tests in which the signal was enhanced, the signal increase
was attributed to CO backflushing into the catalytic convertor in all (Florisil,
cationotropic alumina) but one test (Dowex resin).
As our instrument is not presently as sensitive as the previously reported detector,
it is possible that added CO does enhance the nickel carbonyl ozone signal, but that
it is effectively observable at lower Ni(CO)4 concentrations. Certainly we have not


80
experienced any enhancement approaching the 100 fold increase reported to be caused
by added CO at the concentrations of nickel carbonyl tested.
The proposed mechanism suggesting that CO is reducing NiO to gaseous elemental
i|i
nickel (equation 5), followed by the reaction of that nickel with ozone to form NiO
(equation 6) is reasonable. Also acceptable is the proposal that Ni(CO)4 reacts with
ozone to form gaseous nickel (equation 3) with subsequent reaction with ozone to form
NiO*, i.e.
Ni(CO)4 + 4 Oa------> 4 C02 + 4 Oz + Ni(g) (9)
Ni(g) + Oa-----> NiO* + 02 (10)
Both mechanisms are plausible and further investigation is required to elucidate the
exact nature of the nickel carbonyl ozone reaction. However, in applying this
reaction to monitor carbon monoxide levels, it would obviously be advantageous if
equations 9 and 10 were favorable. This would then obviate the need for added CO
to enhance the reaction and prevent complications in measuring extremely low levels
of atmospheric carbon monoxide in a monitor employing 100% CO. Having to use
CO for detecting the chemiluminescent signal would necessitate at least an additional
trap that could destroy the used carbon monoxide.
Regardless of the exact mechanism by which the nickel carbonyl ozone
chemiluminescent reaction occurs, this study has demonstrated that it can be utilized
in a novel monitor for carbon monoxide. The conversion of carbon monoxide to
nickel carbonyl occurs under relatively mild conditions, at least from a thermodynamic
viewpoint (Brief et al 1967), and the present studies also indicate that the kinetics of
the reaction are favorable. Although there are a number of areas for improvement,
in all three systems of this monitor: the gas dilution system, the catalytic convertor and


81
the chemiluminescent detector, the overall assessment of the feasibility of a carbon
monoxide monitor based on the two reactions
Ni + CO--------> Ni(CO)4
and
Ni(CO)4 + 03-------> light
looks promising and warrants further study.
Future Modifications
Preparation of Nickel Carbonvl Gas Standard. Although nickel carbonyl is a gas
that liquifies at 43 C, the handling and transfer of liquid nickel carbonyl is difficult
due to its high vapor pressure (261 mmHg at 15C). These difficulties can introduce
an estimated + 10% error in the actual concentration of a nickel carbonyl gas standard
during its preparation. Therefore, nickel carbonyl gas standards should be analyzed
by an alternate method, other than gravimetric, to determine its precise concentration.
One such method would use IR spectroscopy and monitor the stretching frequencies
of nickel carbonyl in carbon monoxide at 2130 and 2043 cm"1 (Braterman 1975).
Dynamic Gas Dilution System. The initial design of the dynamic gas dilution
system was intended to allow the control of gas flows as well as to make dilutions of
gases. However, as the system is presently constructed, the dilution of a stock gas with
a diluent gas results in a flow that is the sum of the flows of both gases, the stock and
diluent gases. Therefore each dilution of the stock gas results in a new flow and it
is important to control the gas flow into the chemiluminescent analyzer for precise
measurements of the nickel carbonyl and ozone reaction. Thus another flowmeter and
(ID
(12)


82
relief valve should be installed on the exit side of the dilution chamber to control the
flow of the diluted gas by venting the excess flow.
Catalytic Convertor. To optimize the reaction of nickel carbonyl with ozone, it is
desirable to vary the pressure within the chemiluminescent reactor cell without altering
the residence time of carbon monoxide in the catalytic convertor. However during the
evaluation of catalysts, each time the reactor cell pressure was increased by partial
closure of valve V-2, the contents of the reactor backflushed into the catalytic
convertor cell. To prevent this, a one way flow device should be installed on the exit
side of the catalytic convertor such that gas residence times in the reactor and catalytic
convertor can be controlled independently.
Chemiluminescent Detector. The sensitivity of the chemiluminescent detector
could be increased by optimizing the operating paramters of the PMT. The PMT
housing should be cooled to reduce noise and the operating voltage increased to achieve
the maximum signal to noise ratio.
These refinements in technique, instrumental design and operating conditions will
improve this method of studying the conversion of carbon monoxide to nickel carbonyl
and the subsequent detection of nickel carbonyl.


BIBLIOGRAPHY
Anderson, J. R.: Structure of Metallic Catalysts. Academic Press Inc. Ltd., London,
1975. pp 41, 74-76.
Bay, H. W., Blurton, K. E., Lieb, H. C. and Oswin, H. G.: Electrochemical
measurements of carbon monoxide. Am. Lab. 4: 57-58, 60-61, 1972.
Bay, H. W., Blurton, K. F., Sedlak, J. M. and Valentine, A. M.: Electrochemical
technique for the measurement of carbon monoxide. Anal. Chem. 46: 1837-1839,
1974.
Becker, K. H. and Schurgers, M.: Atomic reactions with carbonyl metal compounds.
Angew. Chem. Internat. Edit. 10: 934-935, 1971.
Bergman, I., Cleman, J.E. and Evans, D.: A simple gas chromatograph with an
electrochemical detector for the measurement of hydrogen and carbon monoxide
in the parts per million range, applied to exhaled air. Chromatogaphia, 8: 581 -
583, 1975.
Braterman, P.S.: Metal Carbonyl Spectra. Academic Press Inc., New York, NY, 1975,
pp 189-190.
Brief, R. S., Venable, F. S. and Ajemion, R. S.: Am. Ind. Hyg. Assoc. J. 28: 21, 1967.
Bruner, F., Ciccioli, P, and Rastelli, R.: Determination of carbon monoxide in air in
the parts per billion range by means of a helium detector. J. Chromatogr. 77:
125-129, 1973.
Cassidy, D.T. and Reid, J.: Atmospheric pressure monitoring of trace gases using
tunable diode lasers. Appl. Opt. 21: 1185, 1982
Chaney, L. W. and McClenny, A. S.: Unique ambient carbon monoxide monitor based
on gas filter correlation: Performance and application. Environ. Sci. Tecnol. JJ.:
1186-1190, 1977.
Ciapetta, F. G. and Plank, C. J.: In: Emmett, P. H. (ed) Catalysis. Reinhold, NY,
NY, 1954, pp. 315.
Covert, L. W., Connor, R. and Adkins, H.: J. Amer. Chem. Soc. 54: 1651, 1932.
Dagnall, R. M., Johnson, D. J. and West, T. S.: A method for the determination of
carbon monoxide, carbon dioxide, nitrous oxide and sulfur dioxide in air by gas
chromatography using an emissive helium plasma detector. Spectrosc. Lett. 6:
87-95, 1973.
Dubois, L., Zdrojewski, A. and Monkman, J. L.: The analysis of carbon monoxide in
urban air at the ppm level, and the normal carbon monoxide value. J. Air Pollut.
Control Assoc. 16: 135-139, 1966.


84
EPA Doc.: Field operations guide for automatic air monitoring equipment. PEDCO-
Environmental Specialists. APTD-0736, U. S. Enivronmental Protection Agency,
Research Triangle Park, NC, July 1971.
EPA Doc.: Regulatory impact analysis of the National Ambient Air Quality Standards
for carbon monoxide. EPA-450/5-85-007, July 1985.
Finlayson-Pitts, B. J. and Pitts, J. N.: Atmospheric Chemistry: Fundamentals.and
Experimental Techniques. John Wiley & Sons, NY,NY, 1986.
Goth, W., Schurath, U. and Weber, K.: z. Phys. Chem. (Frankfurt am Main). 93: 159,
1974
Haagen-Smit, A. J.: Carbon monoxide levels in city driving. Arch. Environ. Health
12: 548-551, 1966.
Hanst, P. L., Lefohn, A. S. and Gay Jr., B. W.: Detection of atmospheric pollution at
parts-per-billion levels by ir spectroscopy. Appl. Spectrosc. 27: 188-198, 1973.
Hersch, P.: Galvanic analysis. Adv. Anal. Chem. Instrum. 3: 183-249, 1964.
Hersch, P.: Method and Apparatus for Measuring the Carbon Monoxide Content of
a Gas Stream. U.S. Patent No. 3,258,411, U.S. Patent Office, Washington, D.C.,
June 28, 1966.
Informatics, Inc.: Air Quality Criteria for CO. US EPA Report 600/8-79-0922,
October, 1979.
Lodge, J. P. Jr. (ed): Methods of Air Sampling and Analysis. 3rd Edition. Lewis
Publishers, Iric., 1989.
Mond, L., Langer, C. and Quincke, F.: J. Chem. Soc. 57: 749, 1890.
Morris, Jr., E. D. and Niki, H.: Chemiluminescent reactions of iron and nickel
carbonyl with ozone. J. Amer. Chem. Soc. 92: 5741-5742, 1970.
Nicholls, D.: Comprehensive Inorganic Chemistry Vol. 3. Pergamon Press, New York,
NY, 1973, pp 1109.
Niki, H., Duce, R. and Dickinson, R.: Global Tropospheric Chemistry: A Plan For
Action. National Academy Press, Washington, D.C., 1984.
Porter, K. and Volman, D.H.: Flame ionization of carbon monoxide for gas
chromatographic analysis. Anal Chem. 34: 748-749, 1962.
Salvi, G., Fiumari, A. and Riganti, V.: J. Catalysis 10: 307, 1968.
Seiler, W. and Junge, C.: Carbon monoxide in the atmosphere. J. Geophys. Res. 75:
2217-2226, 1970.
Stedman, D. H. and Tammaro, D. A.: Chemiluminescent measurement of parts-per-
billion levels of nickel carbonyl in air. Anal. Lett. 9: 81, 1976.
Stedman, D. H., Branch, D. K. and Pearson Jr., R.: Chemiluminescence detector for
the measurement of nickel carbonyl in air. Anal. Chem. 51: 2340, 1979.


85
Steelman, D. H. and Hikade, D. A.: Nickel carbonyl: Decomposition in air and related
kinetic studies. Science 208: 1029-1030, 1980.
Stedman, D. and Fraser.: In J. G. Burr (ed.), Chemi- and Bioluminescence. Marcel
Dekkar, Inc., N.Y., N.Y., 1985.
Tesarik, K. and Krejci, M.: Chromatographic determination of carbon monoxide below
the 1 ppm level. J. Chromatogr. 91: 539-544, 1974.
Zelinskii, N. and Komorewsky, W.: Ber. 57: 667, 1924.