The near real-time analysis of formaldehyde

Material Information

The near real-time analysis of formaldehyde
Pribil, Michael Jakob
Publication Date:
Physical Description:
78 leaves : ; 28 cm

Thesis/Dissertation Information

Master's ( Master of Science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Chemistry, CU Denver
Degree Disciplines:


Subjects / Keywords:
Formaldehyde -- Analysis ( lcsh )
Air -- Pollution -- Measurement ( lcsh )
Air -- Pollution -- Measurement ( fast )
Formaldehyde -- Analysis ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 77-78).
Department of Chemistry
Statement of Responsibility:
by Michael Jakob Pribil.

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:
45216653 ( OCLC )
LD1190.L46 2000m .P75 ( lcc )

Full Text
Michael Jakob Pribil
B.S., University of Colorado, 1995
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science

Pribil, Michael Jakob (M.S., Chemistry)
The Near Real-Time Analysis of Formaldehyde
Thesis directed by Professor Larry Anderson
The continuous near real-time analysis of formaldehyde was
conducted in a laboratory. A known concentration of gaseous formaldehyde
was generated in a permeation system and collected from the air using a 28-
tum glass coil scrubber and acidified water. The scrubbed aqueous
formaldehyde was combined with a 6 M Nash reagent, at elevated
temperatures, forming the derivative, 3,5 diacetyl-1,4 dihydrolutidine (DDL)
via the Hantzsch reaction. The derivative, DDL, was analyzed using
fluorescent detection. A combination in determining the most efficient
formaldehyde scrubber, optimizing a lower reaction coil temperature, and
increasing the residence time eliminated the need of a de-bubbler. The
instrumental component changes resulted in a signal offset time of
approximately 12 minutes and a mean signal response of 20.6 mV per ppbv
formaldehyde sampling 24 ppbv formaldehyde. The average standard
deviation of zero air was 11 mV and the limit of detection for formaldehyde

Figures ........................................... vii
1. INTRODUCTION......... ........................... 1
Formaldehyde ................................. 1
Sampling Methods...............................5
DNPH Cartridge Methods.........................6
Longpath FT-IR ................................9
Tunable Diode Laser Absorption Spectroscopy.. 11
Differential Optical Absorption Spectroscopy. 13
Fluorometric Analysis........................ 16
2. EXPERIMENTAL ....................................20
Chemicals and Equipment ......................20
Experimental Procedures.......................26
Hanztsch Reaction.............................28

1.1 Oxidation of Methane to Formaldehyde ......................2
1.2 Major Pathways of the removal of Formaldehyde..............5
1.3 DNPH reaction with a Carbonyl................................8
1.4 Formation of 3,3 diacetyl-1,4 dihydrolutidine ............. 18
2.1 Absorption Spectrum of DDL..................................23
2.2 Excitation Filter Absorption Spectrum ......................24
2.3 Emission Filter Absorption Spectrum.........................25
2.4 Aldol Condensation..........................................30
2.5 Michael Reaction............................................31
2.6 Pericyclic Ring Formation ................................. 32
3.1 Permeation System...........................................37
3.2 HCHO Concentrations.........................................38
3.3 Blank 2,4 DNPH Cartridge....................................42
3.4 Sampled 2,4 DNPH Cartridge .................................43
3.5 Scrubber Design ............................................45
3.6 28-Tum Glass Coil Scrubber Design...........................48

3.1 HCHO Concentrations..........................................41
3.2 Scrubber Efficiency......................................... 47
3.3 Nash Reagent Concentration Results...........................55
3.4 Limit of Detection...........................................73

the order of 65 x 106 metric tons for the continental United States (Finlayson-
Pitts and Pitts, 1986). The annual emissions are averaged for the continental
United States, therefore minimizing the effect of natural emissions in urban
environments. Even under ambient conditions methane can become a
dominant formaldehyde precursor. Figure 1.1 represents the oxidation
process of methane to formaldehyde (Finlayson-Pitts and Pitts, 1986).
Figure 1.1
ch4+oh - ch3 + h2o (1)
ch3- + o2+m - ch3 o2 + m (2)
ch3 o2 + no f ch3 o-+no2 (3)
ch3 0- + 02 -> HCHO + H02- (4)
Figure 1.1 Oxidation of Methane to Formaldehyde
The oxidation of biogenic hydrocarbons (ie. isoprene) is similar to the
oxidation process of methane (Figure 1.1) for the production of formaldehyde.
In general, OH reacts with a hydrocarbon forming a radical. The newly
formed radical reacts with 02 to form a peroxy radical. The peroxy radical
can react with NO, figure 1.1 reaction 3, forming a methoxy radical and

the method of photo-oxidation for formaldehyde, its destruction is responsible
for the formation of two hydroperoxyl radicals; the formyl radical reacts with
oxygen in the atmosphere producing hydroperoxyl radicals (H02) (reaction 4
figure 1.2) and the hydrogen atom reacts with oxygen in the atmosphere
producing a second hydroperoxyl radical. The OH radical reaction pathway
basically converts OH into H02i described in figure 1.2 (reactions 3-4). The
production of H02 contributes to the odd hydrogen species in the atmosphere
directly affecting its oxidation capacity. In the presence of sufficient NO
concentrations, H02 will react with NO to form N02 which can react
photochemically to form NO and O. The oxygen atom will then go on to react
with 02to form ozone (Finlayson-Pitts and Pitts, 1986). Environments in
which NO concentrations are low, H02 reacts with itself to form hydrogen
peroxide. It is estimated that up to 60% of odd hydrogen species in the early
morning are produced as a result of radical formation by the photolysis of
formaldehyde as illustrated in figure 1.2 (Fried et al., 1997). Odd hydrogen
species (H, OH, and H02) and odd NOx species (NO and N02) are important
in the understanding of the oxidation ability of the atmosphere and in
establishing more realistic parameters in photochemical models. The two
most important sources of OH formation are the previously described radical
photolysis of HCHO and the removal process of formaldehyde. This removal

laser absorption spectroscopy (TDLAS). Indirect near-real time
measurement of formaldehyde involves the derivatization of formaldehyde
followed by the fluorescent detection of the derivative. Time-averaged
sampling of formaldehyde consists of drawing a known volume of air through
a cartridge containing silica coated with acidified 2,4 dinitrophenylhydrazine
(DNPH) for a set period of time and extracting the hydrazone derivative from
the cartridge. The derivative is then analyzed using high performance liquid
chromatography (HPLC) with uv-vis detection.
DNPH Cartridge Methods
The time-averaged method of sampling formaldehyde is an EPA
approved sampling method (EPA Compendium Method TO-11 A) and has
been modified by the American Society for Testing and Materials (ASTM D-
5197-92, 2000). The principle of both methods for the sampling of
formaldehyde is similar. A known volume of air is sampled through a
cartridge containing acidified 2,4 DNPH coated silica at a known rate and
time period dependent on the concentration of formaldehyde. Typical flow
rates are 1 to 3 L/min. Formaldehyde and other carbonyls react with 2,4
DNPH to form a stable hydrazone derivative which is extracted from the
cartridge with 5 ml acetonitrile (Figure 1.3). The hydrazone derivative is

Figure 1.3
Figure 1.3 DNPH Reaction with a Carbonyl

2390-3000 cm"1 (Tuazon et al., 1980) due to the strong absorptions of H20
and C02 This makes it difficult to observe some of the characteristic
functional group stretching frequencies such as C=0, OH, and C=C. The
majority of air pollutants do however have absorptions in the fingerprint
region. The spectrum acquired is ratioed against a clean background
spectrum. The water absorption is minimized by matching the spectrum to
that of a background water spectrum and also by arithmetic manipulation to
improve the matching of water content.
Sampling using the longpath FT-IR involves drawing ambient air into
the cell at a flow rate of 330 l/sec for a minimum of 4 minutes. This insures
the replacement of the previous sample by 5 volumes of fresh sample.
Experiments were performed to insure that there were no significant losses of
pollutants to the cell walls. Preliminary sampling at 1 cm"1 resolution resulted
in a more exacting cancellation of water lines to identify species such as
HCOOH and HCHO. Improving the resolution to 0.5 cm"1 allows for a more
precise assignment of the Q branches for the species of interest. Infrared
absorptions of formaldehyde for quantitative analysis are the sharp doublets
at 2779 cm"1 and 2781.5 cm"1 (Tuazon et al., 1980) resulting in a detection
limit of 6 ppbv with a pathlength of 1 km.
Improvement in the limit of detection for formaldehyde using FT-IR is

clean and polluted air masses. Time resolution for the measurement of
HCHO is on the order of 3 seconds, basically the residence time of the
sample in the cell. To improve the signal-to-noise ratio, signal averaging for
up to a few minutes may be needed.
The instrument consists of two lasers housed in a Stirling cryocooler
and laser cold head assembly. The lasers are maintained between 65 to 120
K with a stability of 10'3 K. The laser output is modulated at 50 kHz and the
2f signal (two times the modulated frequency) is collected at 100kHz. A
switching mirror directs the radiation of the lasers into the cell. The cell is
either a standard White cell or a modified Hom-Pimentel design cell,
depending on the sensitivity required (a modified Horn-Pimentel is capable of
greater sensitivity). Ambient air is drawn through a PFA-type Teflon tube and
filtered using a 2 pm pore sized Teflon filter. Air exiting the cell is monitored
by a calibrated mass flow meter. Pressure of the sample cell and the
reference cell is monitored and controlled by a baratron pressure gauge,
typically in the range of 10-100 torr. The output from the signal and the
reference detectors are monitored simultaneously. The Hg/Cd/Te detectors
are cooled with liquid nitrogen and their respective signals are corrected for
the power incident on each detector.

I0' (A) = \M + (l0(A-,)-l0(A2)) x (A^)/^) 1.2
where is the wavelength at baseline prior to absorption, A2 is the
wavelength of absorption, and A3the wavelength at baseline post absorption.
The differential absorption cross section, a,'(A), is substituted for a(A) in:
c = D/(a(A) L) 1.3
where c is the concentration of the molecule of interest, L is the thickness or
distance from the light source to the detector, and D is the optical density
which is also defined or measured using the differential approach.
D' = log (10. (A)/l (A)) = LI o'(A)(c') 1.4
a, (A) can be determined in a laboratory allowing one to determine the
concentration of a species of interest by measuring its differential optical
c = D'/ (o(A) L) 1.5
c is the concentration of the species being measured, o(A) is replaced with
the laboratory determined value for (Jj'(A), and L is the distance from the
source to the detector.
The instrumental components of a DOAS instrument are dependent on
the species being investigated and the sensitivity needed for its

intensity needed for DOAS would result in high levels of stray light.
The detection limit of formaldehyde using DOAS with a 5-km path
length is 0.4 ppbv using a wavelength interval of 300-360 nm. DOAS for the
detection of formaldehyde are adequate in sensitivity and fully capable of
near-real time analysis. The capability of measuring the molecules
absorption cross section in a laboratory removes the need of calibrating the
instrument in the field. The instrument can be difficult to setup if an unfolded
path is used. The alignment of the instrument of up to several kilometers can
prove to be a difficult task and weather can affect the visibility causing delays
or times when measurements cannot be made (Sigrist, M., 1994).
Fluorometric Analysis
Two techniques for the fluorometric determination of formaldehyde
have been developed and implemented for the field. One is based on the
Hantzsch reaction in which a p-diketone and an amine react with
formaldehyde forming a pericyclic ring (Figure 1.4). The derivative, 3,5
diacetyl-1,4 dihydrolutidine (DDL) is analyzed using a fluorometer (Fan et al.,
1994). The second method is an enzyme-catalyzed reaction in which
formaldehyde is oxidized to formic acid by oxidized nicotinamide adenine
dinucletide (NAD+). The reduced form of nicotinamide adenine dinucletide,

Figure 1.4
Figure 1.4 Formation of 3,3 Diacetyl-1,4 Dihydrolutidine

Chemicals and Equipment
The following chemicals were purchased from Fisher Scientific
(Pittsburgh, PA): ammonium acetate (A639-500) HPLC grade and
hydrochloric acjd (A508-500) Trace Metal. Platinum cured silicone tubing (E-
95802-02) and Norprene tubing (E-06404-14) were purchased from Cole-
Palmer Instrument Company (Vernon Hills, IL). The peristaltic pumps used
for the instrument consisted of a multi-head Gilson Minipuls 2 capable of
adjustable flow rates and the second peristaltic pump was a combination of a
12V DC motor with a drive shaft interfaced to a double port peristaltic housing
(roller and sleeve) also capable of adjustable flow rates. The scrubbing coil,
a 28-tum 3 mm ID glass coil, was blown by Technical Glass, Inc (Aurora,
CO.). Two stainless steel reaction coils in series were used to assist in the
completion of the Hantzsch reaction. The first coil is a 26-tum 1 mm OD
which is submerged in a hot water bath. The second coil is a 10-tum 1 mm
OD at ambient temperature. Dynacal permeation device, formaldehyde (100-
200-2300) 200 ng/ min at 50 C, was purchased from VICI Metronics (Santa
Clara, CA).

increase giving approximately a 10 fold increase in sensitivity. The power
setting of the photo detector for all sampling conducted during this research
was 900 volts. Signal outputs include a digital panel meter with a 3 digit
display on the fluorometer and a recorder output which is connected to a 12
bit data acquisition board in a NEC Power Mate portable computer. The data
was stored on a floppy disk and imported into a spreadsheet program for
Figure 2.1 is the adsorption spectrum of DDL prepared using Nash
reagent versus wavelength in nanometers (nm). DDL is bright yellow
crystalline compound, soluble in water. The absorption spectrum illustrates
an intense absorption band at 254nm and a second strong absorption band
at 412nm. Initial research contributed the 254nm absorption to
decomposition not fluorescence (Kelly et al., 1990), rather the 412nm
absorption leads to fluorescence at 510nm (Belman,1963). Figures 2.2 and
2.3 are absorption spectra of the excitation and emission filters used in the
analysis of formaldehyde. Figure 2.2 is a plot of absorbance versus
wavelength in nanometers of the excitation filter. The peak absorption of
DDL at 254nm is not utilized with this filter, however both the shoulder of the
254nm absorption and the second absorption band of DDL at 412nm overlap
with the transmission of the excitation filter of 275nm to 390nm. The

Figure 2.2
Wavelength (nm)
Figure 2.2 Excitation Filter Absorption Spectrum

emission filter, figure 2.3, is adequate for the measurement of the
fluorescence of DDL. It has strong absorption at wavelengths shorter than
425 nm, at which the excitation filter has an absorption of 3. The filter
combination is complimentary, however once optimization of the components
and reaction temperatures are determined, improvement in the intensity of
excitation wavelength of DDL should be addressed.
A variety of experimental procedures were used in optimizing the
signal response for the near real-time analysis of formaldehyde. Initial
parameters for instrumental analysis were based on previous research and
functioned adequately as a starting point for instrumental setup. The ideal
instrumental setup was determined by conducting smaller experiments on
each component individually and optimizing the signal response for each.
Limitations in some of the equipment directed some of the instrumental
parameters chosen and will be addressed later in this chapter. As discussed
in chapter 1, fluorometric analysis for formaldehyde, multiple components are
combined in the determination of formaldehyde and each major component
will be addressed. Also, some components have a direct effect on other
components and their relationship will be discussed.

in which the formaldehyde is dissolved. Again, minimization of reagent flow
and large signal response is the goal. The effluent flow is the total flow of the
both solutions, scrubber and Nash. The total flow is limited by diameter of
the flow cell on the fluorometer. If too large a total flow is forced through a
small flow cell, back pressure will eventually effect the flow rates of either the
scrubbed formaldehyde or Nash reagent or both.
Hantzsch Reaction
A variety of methods exist for the detection of formaldehyde, however
many of those methods require harsh conditions such as hot sulfuric acid.
The Hantzsch reaction is the cyclization of a p-diketone, an aldehyde and an
amine occurring under relatively mild conditions (Sawicki et al., 1968). The
reagents used in the reaction are common and inexpensive. The pericyclic
ring formed in the reaction is dependent on the p-diketone used. 2,4-
pentanedione was chosen as the p-diketone for multiple reasons which will
be discussed later. The product formed from the reaction of formaldehyde
with 2,4-pentanedione and ammonium acetate is 3,5-diacetyl 1,4-
dihydrolutidine (DDL). The product is formed in a 1:1 ratio with respect to
formaldehyde and can be detected fluorescence (Salthammer, 1993).
2,4-pentanedione was chosen for its high selectivity for formaldehyde

Figure 2.4

Figure 2.4 Aldol Condensation

Figure 2.6
Figure 2.6 Pericyclic Ring Formation

The near real-time analysis of formaldehyde using the wet chemical
method described in the introduction was simulated and modified in a
laboratory environment. Optimization of the instrument for the detection of
formaldehyde was divided into three major components: scrubber, reagent
concentration, and reaction coil temperature. The first component
investigates various scrubber designs available for the removal of
formaldehyde and their ability to remove formaldehyde from air. The criteria
used for the selection of an adequate scrubbing device is also discussed.
The second component addressed is the reagent concentration more
specifically the ammonium acetate concentration of the Nash reagent and its
relationship to signal response. The third component addressed is the
reaction coil temperature. The Hantzsch reaction, as mentioned previously,
eliminates three molecules of water during the formation of the one
fluorescent DDL molecule and is therefore dependent on reaction
temperature. This dependence directly effects the signal response and is
discussed in detail.
Signal response was maximized individually for each of the three

scrubbing solution with a flow rate of 0.8 ml/ min. Figure 3.1 illustrates the
permeation system used during the scrubber efficiency testing. The
formaldehyde source was a permeation tube filled with paraformaldehyde
enclosed in a glass housing submerged in a warm water bath maintained at
40.5 C ( .05 C). A low flow mass transducer regulated the low air flow
through the paraformaldehyde housing at a constant 0.1 L/min continuously -
to ensure adequate air flow through the permeation system at all times. A
high flow mass transducer regulated the diluent air flow which. This diluent
flow was varied to change concentration of formaldehyde generated. The
dilution air is combined with the low flow air in a glass manifold. The
formaldehyde concentrated air (low flow) enters the manifold from a side arm
and the diluting air (high flow) enters from the bottom of the manifold.
Calibration of the high flow transducer with respect to the
concentration of formaldehyde, using 2,4 DNPH cartridges was conducted
with the low air flow rate at constant 0.1 L/min (Figure 3.2). Concentrations of
formaldehyde during all testing ranged from 12 ppbv to 24 ppbv correlating to
high flow rate settings of 118 to 318. Concentrations greater than 24 ppbv
were not possible at the permeation systems temperature, due to the need of
the total air flow rate of the permeation to be greater than the air sampling
rate of the instrument. Lower concentration of formaldehyde is possible by

Figure 3.2
High Air Flow Setting
Figure 3.2 HCHO Concentrations

Table 3.1
High air flow rate HCHO (ppbv) Deviation from the
318 11.5 0.4
318 12.8 0.8
318 11.9 0
318 11.5 0.4
318 12.0 0.1
218 16.6 0.6
218 15.8 0.1
218 15.5 0.5
218 16.0 0
118 24.5 0.2
118 24.4 0.3
118 24.6 0.1
118 25.3 0.6
Table 3.1 HCHO Concentrations

Figure 3.3
Blank Cartridge f a> i i V |

Figure 3.3 Blank 2,4 DNPH Cartridge

determine the concentration of formaldehyde entering the scrubbers. A
typical area count for a blank cartridge was 25-30 and a typical sampled
cartridge area count was 280-350 depending on the volume of air sampled.
The first scrubber tested was a continuous flowing, bubbler type
scrubber consisting of a glass stopper with a straight channeled opening at
the top, narrowing at the tip, reaching through to the bottom of the holding
bottle (Figure 3.5). A second opening on the glass stopper, perpendicular to
the first opening, is connected to the air pump. The bottle into which the
stopper fits is rounded and also contains two openings. The first opening is
located an inch from the bottom and is used to remove the acidified water
containing the scrubbed formaldehyde and combining it with Nash reagent.
The second opening is located at the bottom of the bottle and is used to
introduce clean scrubbing solution. Air is pulled through the straight
channeled opening and pushed out through the side arm. The volume of the
continuous flow bubbler is approximately 5ml. Initial testing resulted in large
bubbles produced during sampling decreasing the air-to-water contact ratio.
The scrubbing efficiency for the standard continuous flowing bubbler is 85.3
(6.5)% (Table 3.2).
The second scrubber resembled the first in many aspects except that
the narrowing opening on the stopper was fitted with a glass frit decreasing

the size of the air bubbles and increasing the air-to-water ratio. Improvement
in the scrubbing efficiency of the bubbler with frit over the standard
continuous bubbler is seen in table 3.2. Improvements of nearly 7% scrubber
efficiency were measured. The standard deviation also improved from 6.5 to
0.5. The choice of scrubber is not solely based on the mean percent
scrubber efficiency, rather a combination of the mean percent scrubber
efficiency, reproducibility based on standard deviation, ruggedness and
simplicity in terms of a maintenance free operation over a long period of time.
Both bubblers incurred a large amount of bubbles which could be drawn into
the reaction coil causing noise in the fluorescence signal.
The third type of scrubber tested was a 28-tum glass coil (Figure 3.6).
The glass coil technique removes formaldehyde from the air by concurrently
pulling ambient air and acidified water through a 3 mm id 28-turn glass coil.
The acidified water coats the walls of the glass coil with a thin film of acidified
water and as the air passes through the glass coil it interacts with the
scrubber solution removing nearly 95 (1.0) % of the formaldehyde. The high
scrubbing efficiency, with minimal bubble formation and the capability to
operate for long periods of time without maintenance, makes this the
scrubber of choice.
The choice of scrubber was simplified from the results of the efficiency

Figure 3.6
air exhaust
Figure 3.6 Scrubber Design

bath consisted of the stainless steel reaction coil placed into a beaker filled
with water placed on a hot plate. A thermometer was attached to the reaction
coil which was submerged in the water bath. Each of the reagents were
tested under similar instrumental conditions. Zero air was sampled for
approximately 5 minutes, followed by the sampling of HCHO at 24
(0.4) ppbv for approximately 20 minutes. Zero air was achieved by pulling
room air through a 2,4 DNPH cartridge. Formaldehyde was generated in the
permeation system as described previously with the temperature of the water
bath of 41.5 C. An air sampling rate of 1 l/min was used during the reagent
concentration testing. The scrubber and reagent flow was 0.8 ml/min. Figure
3.7 is an overlay of the experimental results of each of the four different
reagent concentrations. The 1.3 M ammonium acetate reagent resulted in a
99 (6.7) mV change in signal sampling the 24 (0.4) ppbv formaldehyde
generated air. The remaining three reagents baselines were offset to clarify
the baselines, the rise in signals, and peak signals. Increasing the
concentration of ammonium acetate to 3.0 M improved the signal response to
170 (15.3) mV, a improvement of 71 mV. Nearly doubling the ammonium
acetate concentration to 6.0 M resulted in a 484 (9.8) mV signal response, a
change of 314 mV. The concentration was increased a third time to 7.8 M
resulting in a signal change of 490 (11.4) mV, a 6 mV increase over the 6.0

Figure 3.7
1.30M 2.98M 5.97M-----7.78M
Figure 3.7 Variations in Ammonium Acetate Concentrations

Figure 3.8
2 3 4 5 6 7 8
Ammonium Acetate Concentration (M)
Figure 3.8 Reagent Concentration vs. Signal Change

Table 3.3
Ammonium Molar Response Change Standard
acetate concentration (mV) (mV) deviation
(grams) (M)
10 1.3 99 6.7
23 3.0 170 71 15.3
46 6.0 484 314 9.8
60 7.8 490 6 11.4
Table 3.3 Nash Reagent Concentration Results
scattering contributes to background noise and increasing the ammonium
acetate concentration resulted in a lower background signal possibly due to
greater adsorption of the excitation light scattering by the ammonium acetate
in the Nash reagent during zero air sampling (Kelly et al., 1990). No
decreases in the noise of the zero air signal was evident as the ammonium
acetate was increased. Again, each component of the instrument has an
effect on the signal and makes it difficult to contribute low signal noise to one
single component during the optimization of the instruments components. A
standard deviation of 11.4 mV from a signal response of 490 mV is a
deviation of 2.3 %, which is acceptable. The limit of detection of the
instrument and baseline noise is addressed below.

signal, allowing a clear view of each individual run.
At 24 C, room temperature, an average signal change of 73 mV was
measured with a standard deviation of 7 mV, corresponding to a 3 mV
change per ppbv HCHO. Increasing the temperature to 50 C resulted in a
mean signal response of 347 ( 8) mV, a change of 14.5 mV per ppbv
HCHO. The average and standard deviation of the signal were calculated
based on all of the individual data points collected while the fluorometer was
responding to the signal. The bubble spiked signals and the rise and fall
signal response were not included in the averaging. The change in signal
was determined by subtracting the average baseline signal from the average
peak signal. At 68 C, the signal increased to a mean response change of
489 (12) mV, with a change of 20.5 mV per ppbv HCHO. Increasing the
reaction coil temperature to 82 C increased the mean signal response to 498
(11) mV with a change of 20.9 mV per ppbv HCHO. At a temperature near
92 C, previously reported in the literature, standard deviation of the baseline
increased to 18 mV and the mean change in signal increased to 508 mV
corresponding to a 21.3 mV change per ppbv HCHO and a 0.4 mV increase
in signal per ppbv formaldehyde.
The signal-to-noise ratio for each temperature varied, however the
greatest increase in the standard deviation was noted at the highest reaction

Figure 3.10
Figure 3.10 Change in Response versus Reaction Coil Temperature

reduced to 40 minutes, and at 58 C the reaction time was 5 minutes (Nash,
1953). The experiments conducted by Nash used high concentrations of
formaldehyde which also plays a role in the rate of reaction. However, the
rate of reaction is temperature dependent and when modified by increasing
the reaction time and reducing the temperatures, an increase in the signal to
noise ratio can be achieved without decreasing the signal response. Also for
field implementation, a lower reaction temperature minimizes the complexity
of a de-bubbler. Lag time and time resolution at similar temperatures will be
addressed in the next section.
In Lab Analysis of HCHO
Experiments sampling gaseous formaldehyde at a various
concentrations, varying from 12 ppbv to 24 ppbv, were conducted using the
instrumental parameters described previously in a laboratory. Figure 3.11
displays the results of the optimized instrument sampling 24 ppbv
formaldehyde for 60 minutes. Zero air was sampled for 15 minutes with an
adjustment to increase the baseline at 6 minutes. The adjustment ensured a
positive signal in the baseline for the entire run. The mean of the
pre-sampling baseline was 92 (9 mV) after the offset adjustment. The peak
response mean during sampling was 589 (12) mV. The post-sampling

mean of the baseline was 146 ( 8) mV. The standard deviation of the signal
and the mean of the signal were calculated without the spikes caused by the
air bubbles. A signal response change for formaldehyde was calculated to be
21 mV/ppbv. The large spikes in figure 3.11 are the result of air bubbles
passing through the fluorometer, however the signal returns to the response
prior to the air bubble spike. The rise in signal response from 10% to 90%
was 2.2 minutes and the fall from 90% to 10% signal response was 2.4
minutes. The larger standard deviation in the formaldehyde signal response
is attributed to the inconsistent flows of the peristaltic pumps. As discussed
in the experimental section, two peristaltic pumps were used to combine the
reagent components. One was built using the parts of two pumps which
consisted of a 12 V motor and a stacked wheeled housing. The second was
a dual tubing variable flow peristaltic pump. Fluctuation in the motor drive
speed and the frequent crimping of the peristaltic tubing in both pumps varied
the flows of the reagent and the scrubbing solution. Small variations in the
flow rates will change the concentration of formaldehyde resulting in
variations of the sampling signal. The relatively small standard deviation of
both the pre and post-sampling zero air baseline, 9 mV and 8 mV
respectively, suggests that the cause of the variation in the signal response is
related to flow rate variations during the sampling of formaldehyde.

Figure 3.12
Figure 3.12 Sampling of 24 ppbv HCHO at 60 C

Figure 3.13
Figure 3.13 Sampling Various HCHO Concentrations

evident at the higher formaldehyde concentration of 24 (0.4) ppbv as seen in
the figures in the previous section.
A separate experiment was conducted to investigate the rise in the
baseline signal sampling zero air. Figure 3.14 illustrates the rise in the post
sampling baseline with respect to the pre-sampling baseline of zero air. It
also illustrates the rise in signal response with the increase in reaction coil
temperature. A starting temperature of 40 C was used and the temperature
was ramped manually to 62 C over a 55 minute sampling period. The mean
of the pre-sampling baseline was 122 mV and the post-sampling mean was
213 mV. A change of approximately 90 mV from an increase of 22 C.
Fluorometry is temperature dependent and the rise in the zero air signal is
present in some of the sampling data. This is attributable to increases in the
reaction coil temperature. Stable and constant reaction coil temperature is
possibly through the use of an insulated housing wrapped in heat tape.
Commercial items are also available.
Limit of Detection
The limit of detection, as stated previously, can be defined by three
times the signal noise of the baseline and therefore can vary from run to run.
The limits of detections for the runs illustrated in figures 3.9, 3.11, 3.12, and
3.13 were determined and compared to one another. Table 3.4 organizes

the results for each of the four runs conducted at a formaldehyde
concentration of 24 (0.4) ppbv and a reaction coil temperature range of 60 C
to 69 C. The average signal response for the four runs was 20.4 (1.0)
mV/ppbv and the average standard deviation of the baseline or blank signal
was 9.8 (1.7) mV/ppbv. The range of the standard deviation of the baseline
was twice than the average of the signal response per ppbv of formaldehyde
indicating that the limit of detection of the instrument is related more to the
noise in the baseline signal not the variability in the signal response. At a
lower formaldehyde concentration of 16 (0.5) mV/ppbv, the signal response
decreased to an average of 16.9 (1.5) mV/ppbv, corresponding to a limit of
detection of 2 ppbv. Reproducibility at a constant formaldehyde
concentration is evident, with a slight decrease in sensitivity at lower
formaldehyde concentrations. The limit of detection, however is still
comparable to those at higher concentrations. The four runs were conducted
on different days, using separately prepared Nash reagent solutions.

The near real-time analysis of formaldehyde described here provides
the ability to analyze continuously both indoor and, with a few modifications,
outdoor air. The modifications were presented in the results section and
include improvement in the peristaltic pumps implemented in the instruments
design. As displayed in some of the figures, variation in the sampling signal
based on the standard deviation is 2-3 times larger than the zero air signal.
Including a triple peristaltic pump in the instruments design would allow for
the decrease in some of the aqueous solution flow rates. As discussed in the
previous section, decreasing the scrubber flow would improve the rise-fall
times of the instrument along with increasing the signal response decreasing
the reagent flow rate is also possible. The flow rate for the reagent was equal
to the scrubber flow because the two solutions flow rates were controlled by
a single pump. Attempts at changing the flow rates by decreasing the
diameter of the tubing did not work because too small of a diameter tubing
would not function properly in the pump housing. Using too large of a
diameter tubing, after a period of sampling, caused the tubing to be pulled
through the pump, resulting in crimping, which in turn causes a backup in

ASTM D5197-97 Standard Test Method for Determination of
Formaldehyde and Other Carbonyl Compounds in Air (Active
Sampler Methodology), American Society for Testing and
Materials, West Conshocken, P.A. 2000.
Arnts, R.R. and Tejada, S. B. Environ. Sci. Tech. 1989, 23,
Belman, S. Anal. Chim. Acta. 1963, 29, 120-126.
Dong, S. and Dasgupta, P.K. Environ. Sci. Tech. 1987, 21,
EPA Compendium Method TO-11A, Determination of
Formaldehyde in Ambient Air Using Adsorbent Cartridge
followed by HPLC (Active Sampling Methodology). Jan. 1999.
Fan, Q. and Dasgupta, P.K. Anal. Chem. 1994, 66, 551-556.
Finlayson-Pitts, B. and Pitts, J.N. Jr. Atmospheric Chemistry:
Fundamentals and Experimental Techniques, John Wiley &
Sons, Inc., New York, 1986.
Fried, A.; McKeen, S.; Sewell, S.; Harder, J.; Henry, B.; Goldan,
P.; Kuster, W.; Williams, E.; Baumann, K.; Shetter.R.; Cantrell,
C. J. Geophys. Res. 1997, 102,D5, 6283-6296.
Genfa, Z.; Dasgupta, P.K.;Cheng, Y.; Atmos. Environ., 1991,
22, 949-963.
Harder, J.W.; Fried, A.; Sewell, S.; Henry, B. J. Geophys. Res
1997, 102, D5, 6267-6282.
Kelly, T.J. and Fortune, C. R. Intern. J. Environ. Anal. Chem.
1994, 54, 249-263.

Full Text
This thesis for the Master of Science
degree by
Michael Jakob Pribil
has been approved by
Donald C. Zapien
'Zjj, ^ooc>

was 1-2 ppbv.
This abstract accurately represents the content of the candidates research
and thesis. I recommend its publication.
Lsnry Anderson

Reagent Concentration.......................49
Temperature Dependence......................58
In Lab Analysis of HCHO.....................62
Limit of Detection .........................70
4. CONCLUSION.....................................74

3.7 Variations in Ammonium Acetate Concentrations..........51
3.8 Reagent Concentration versus Signal....................53
3.9 Signal Response Dependence on Temperature .............57
3.10 Signal Response versus Reaction Coil Temperature ......60
3.11 Sampling of 24 ppbv HCHO at 69 C ......................63
3.12 Sampling of 24 ppbv HCHO at 60 C ......................66
3.13 Sampling various HCHO Concentrations...................68
3.14 Temperature Ramp.......................................71

Formaldehyde plays an important role in photochemical processes in
the atmosphere. Concentrations of 10 parts per billion by volume (ppbv) to
100 ppbv have been measured in urban air (Fan et al., 1994), 0.1 ppbv to 5
ppbv in continental rural air (Harder et al., 1997) and approximately 0.2 ppbv
in remote regions of the ocean (Lazrus et al., 1988). The presence of
formaldehyde in urban, rural and remote air suggests that different sources of
formaldehyde exist. Sources of formaldehyde can be categorized as
anthropogenic or biogenic. Anthropogenic emissions include both primary
and secondary origins. The most prevalent primary source is the combustion
of fossil fuels, which includes automobile exhaust, industrial emissions and
energy production. Secondary sources of formaldehyde related to
anthropogenic sources result from the photo-oxidation of emitted
hydrocarbons. Biogenic sources release reactive non-methane hydrocarbons
such as isoprene and monoterpenes into the troposphere which undergo
photo-oxidation similar to the anthropogenic secondary sources. Total
annual emissions of biogenic precursors of formaldehyde are substantial, on

nitrogen dioxide. The methoxy radical can react with 02, figure 1.1 reaction
4, to form formaldehyde.
Formaldehyde also posses a health hazard both directly and indirectly.
Directly, it is a suspected carcinogen and irritant with OSHA exposure limits
currently at 750 ppbv and an action level of 375 ppbv over an 8 hour period.
The byproducts from the decomposition of formaldehyde such as, formyl
radicals and hydroperoxyl radicals, are important in the oxidative ability of the
atmosphere and also could possibly posses a health hazards of their own.
The relative importance of the production versus the destruction of
formaldehyde in ambient air is what governs its concentration. Two major
pathways for the removal of formaldehyde include photo-decomposition and
reaction with the hydroxyl radical. Figure 1.2 illustrates the major pathways of
formaldehyde removal from the troposphere. Product formation from the
photolysis of formaldehyde is dependent on the wavelength of the absorbed
radiation causing the decomposition. Shorter wavelengths, 290-320 nm,
primarily form H atom and the formyl radical. At longer wavelengths, 330-370
nm, hydrogen gas and carbon monoxide are the principle products. In
remote air masses (those free from anthropogenic inputs) reaction 2 in figure
1.2 is the primary source for carbon monoxide. The radical reaction with OH,
generates the formyl radical and carbon monoxide. If reaction 1 figure 1.2 is

Figure 1.2
HCHO + hA (290-320nm)
HCO- + 02
H+ 02
H + HCO- 0)
- h2 + co (2)
- HCO- + H20 (3)
* ho2- + CO (4)
ho2- (5)
Figure 1.2 Major Pathways of the Removal of Formaldehyde.
process for formaldehyde and its influence in the global mixing ratio of ozone
and OH (Vairaramurthy et al., 1992) defines the importance for the near-real
time measurement of formaldehyde in order to fully understand the oxidation
capacity of the troposphere.
Sampling Methods
Currently a large variety of sampling methods for formaldehyde exist,
including real time analysis and time-averaged analysis. Real time direct
measurements of the concentration of formaldehyde have been conducted
using the following techniques: Fourier transform infrared spectroscopy
(FTIR), differential optical absorption spectroscopy (DOAS), tunable diode

analyzed by reverse phase HPLC equipped with a uv-vis detector operating
at 360 nm. Interferences by other compounds with the measurement of
formaldehyde have been investigated. The most prevalent interfering
compound with the measurement of formaldehyde is ozone (Amts et
al.,1989). However, using an ozone scrubber prior to the cartridge has
proven to eliminate the interference of ozone for the measurement of
formaldehyde. This method of sampling can also be used for the
measurement of many other carbonyls. The detection limit using this method
of sampling is less than 1 ppbv (Kleindienst et al., 1988).
A second method of measuring formaldehyde using 2,4 DNPH uses a
pyrex coil gas-liquid scrubber in series with a high performance liquid
chromatograph (HPLC) equipped with a ultraviolet-visible detector (Lee et al.,
1993). Carbonyls that have solubilities in water similar to that of
formaldehyde are removed from ambient air by concurrently pulling
atmospheric air and acidified 2,4 DNPH through a 10-tum round coil. A 6 ml
sample is collected, corresponding to a 20 minute sampling period. The
sample is allowed to sit overnight for the derivatization reaction to occur,
typically 16-24 hours. Standards were analyzed using of 0.5 ml injection
volumes, however to increase sensitivity 1.23 ml volume samples were
injected into HPLC for the field sample analysis. Interference testing of other

major atmospheric species concluded minimal effects for the analysis of
formaldehyde. Ozone, which has a low solubility in water, had essentially no
effect on formaldehyde even at ozone concentrations of 400ppb. Detection
limits using this DNPH technique for formaldehyde is on the order of 0.02
ppb. Improvements in the sample collection can be made by automation of
the scrubbing method, though the time resolution cannot be improved better
than the HPLC run times, which for this research was 30 minutes.
Longpath FT-IR
Ambient concentrations of formaldehyde have also been determined
using long-path FT-IR spectroscopy (Sigrist, 1994). The instrument consists
of an eight mirror multiple reflection cell, with four 30 cm diameter mirrors at
one end of the cell and four rectangular mirrors (3 measure 6.4 cm x 32 cm,
with the fourth slightly smaller) at the other end. The two sets of mirrors are
22.5 m apart and through a sequence of reflections, the pathlength of the cell
measures 1080 m (~1km). A total of 47 reflections through the cell yielded a
transmitted intensity 0.39 times the incident intensity. To achieve this
transmitted intensity all of the mirrors were gold coated and optically polished
to maximize reflectivity. At longer pathlengths (1 km or greater) the infrared
region of detection is limited to the 760-1300 cm'1, 2000-2230 cm'1, and

achieved by improving the signal to noise ratio and by increasing the size of
the absorption of the species of interest. Lowering the temperature of the
detector (HgCdTe) to liquid nitrogen temperatures reduces the noise level by
100 times to that of a room temperature operated detector. Shortening the
pathlength to 100 m allows for interpretation of the 1600 cm'1, 1360 cm'1 and
674 cm'1 bands which are characteristic absorption bands for N02, S02, and
benzene respectively. The limit of detection for formaldehyde with a 100 m
pathlength, 0.5 cm'1 resolution, and a noise level at 1C4 is 3.0 ppbv (Sigrist,
Tunable Diode Laser Absorption Spectroscopy
Tunable diode laser absorption spectroscopy (TDLAS) has been
incorporated in trace gas measurements producing the sensitivity and fast
response required (Sigrist, M., 1994). For this analysis, typically the middle
infrared region, 2 to 15 pm, is used for its characteristic spectral region for
atmospheric trace gases. A large variety of atmospheric species have strong
absorption bands in this region. Measuring formaldehyde absorption at 2781
cm'1 allows for a detection limit of 0.25 ppb with a pathlength of 150m (Sigrist,
M., 1994). The high spectral resolution of the laser source allows for the
measurement of single ro-vibrational lines of the species of interest in both

Differential Optical Absorption Spectroscopy
Differential optical absorption spectroscopy (DOAS) can be used in
determining the concentration of atmospheric species of interest by
measuring the termed differential absorption. Differential absorption is a
portion of the total absorption of the molecule of interest rapidly varying with
wavelength. The absorption cross section of a given molecule can be
described by:
0|(A) = OiO(A) + a,'(A) 1.1
where O;0(A) varies slowly with respect to wavelength and ^'(A) varies
rapidly with respect to wavelength. Slow variation can be described by a
general slope, Raleigh and Mie scattering are assumed to vary slowly. Rapid
variation is an absorption line produced by the species of interest. The
definition of the terms rapid and slow variation are dependent on the
observed wavelength interval and width of the absorption band. It is not with
respect to time rather change with respect to wavelength. The true intensity
of the light source at the detector in the absence of any absorption by
atmospheric molecules is not important when using the differential absorption
approach. The intensity of the light source can be interpolated from the light
intensity at either side of a narrow absorption line of the species of interest

determination. An overview of the major components is outlined for the
determination of formaldehyde.
A variety of light sources can be used in the design of a DOAS system.
The primary concern for selecting a light source is that the intensity varies
slowly with a wavelength. Some common light sources include: thermal
(incandescent or arc lamps), lasers, and natural light such as the sun or
moon. The design of the path of the instrument can also vary. A reflector
returning the light to the detector would double the path length over choosing
to place the light source at one end of the path and the detector at the other
end. The path can also be folded using White cell or Hom-Pimentel cell. For
measurements at reduced pressures a closed cell would be required.
Detectors that are used for DOAS include photo multipliers (PMT), non-
dispersive semiconductors, and photodiode array. Photodiode array solid
state detectors are typically used for medium to strong absorption bands and
when weak absorption lines are to be measured optomechanical scanning
device with a PMT is used. Minimization of the stray light generated from the
spectrometer is essential in the selection of the type of spectrometer to be
used. The Czerny-Tumer spectrometer is a common spectrometer used for
DOAS due to its inherently low level of stray light. A Flat-field spectrometer is
also used, however with the flat-fields holographic grating the high

NADH, is fluorescent and is used for the determination of formaldehyde in air
(Lazrus et al., 1988). These two fluorometric methods used for the
determination of formaldehyde incorporate similar scrubbing techniques for
the removal of formaldehyde from the air. The most popular are the
membrane-based diffusion scrubbers (Dasgupta, 1990) and the 28-turn glass
coil which will be discussed in more detail in the experimental section. In
both methods formaldehyde reacts with a reagent to form a fluorescent
compound. The chemical derivatization of formaldehyde, DDL from the
Hantzsch reaction, coupled to a fluorescent detector can be utilized for the
near-real time determination of formaldehyde (Kelly et al., 1994). Gaseous
formaldehyde is removed from the ambient air by concurrently pulling air at 1
to 1.5 L/min and acidified water (pH 2) at 0.8 mL/min through a 28-turn glass
coil (Figure 3.6). Nash reagent (ammonium acetate, glacial acetic acid and
2,4 pentanedione) is introduced at a rate of 0.4 ml/min through a side
opening and mixes with dissolved formaldehyde in the scrubber solution at
the bottom of the 28-tum glass coil. The mixture is pulled through a reaction
coil maintained at 95 C. The elevated temperature increases the rate of
reaction of formaldehyde and Nash reagent in forming the DDL derivative.
The effluent passes through a fluorometer where excitation at 412 nm or 254
nm is achieved by bandpass filters on the excitation side. High-pass filters

are typically used on the emission side. The fluorometric response is either
recorded by use of a chart recorder or more commonly by a computer. A
more detailed description of the technique is addressed in the experimental
section. The enzymatic approach for the determination of formaldehyde
removes formaldehyde from the air by scrubbing it with acidified water,
typically adjusted to a pH of 2 with hydrochloric acid. The aqueous
formaldehyde is combined with a buffer solution at a pH of 8 and
introduced into a five-turn mixing coil. Formaldehyde dehydrogenase (FDH)
and nicotinamide dinucleotide (NAD+) are added to the reaction and the
mixture is introduced to a second five-turn mixing coil, followed by a reaction
time delay coil (13.2 minute delay) which allows for the formation of the
coenzyme (NADH):
H2C(OH)2 + FDH HCOO' + H+ + H2FDH (1)
H2FDH + NAD+ NADH + FDH + H+ (2)
NADH is fluorescent and is measured using a fiuorometer. It has an
excitation wavelength of 340 nm and emission wavelength of 460 nm.

Nash reagent consisting of ammonium acetate, glacial acetic acid and
2,4-pentanedione was prepared using Milli-Pore water. Prior to using the
Nash reagent, it was allowed to sit for a minimum of 24 hours in order for the
keto-enol tautomerism of the beta-diketone to reach equilibrium. The Nash
reagent was stored in a refrigerator prior to and between sampling. The
signal response during the determination of formaldehyde decreased after 4-
5 days when the reagent remained on the laboratory bench at room
temperature and exposed to light. Time was allotted prior to sampling for the
reagent to reach room temperature. The Hantzsch reaction is a temperature
dependent reaction.
A Fluoro-Tec Filter Fluorometer (St. John Associates, Inc.) was used
to measure the fluorescence of the DDL, the formaldehyde derivative formed
from the Hantzsch reaction. The excitation source is a 4 watt low pressure
mercury vapor lamp. The flow cell assembly consists of a 75 pL quartz
cylindrical flow cell with a coupled 90 degree optical path that minimizes stray
light and also acts as a shutter when in the closed position. The detector is a
cage-dynode, side window 931 B photo multiplier with an analog current-to-
voltage conversion followed by a low-pass active filter. Power to the photo
detector is temperature compensated with switchable settings of voltages
ranging from 450 volts to 900 volts in 150 volt intervals with each 150 volts

Figure 2.1
Figure 2.1 Absorption Spectrum of DDL

Figure 2.3
Wavelength (nm)
Figure 2.3 Emission Filter Absorption Spectrum

An air sampling rate of 1 L/min was used for the efficiency testing of
the three different formaldehyde scrubbers. Ideally, the larger amount of
formaldehyde in solution the larger amount of DDL formed and the larger the
signal in the detector, the limiting factor is the ability of the scrubber and
scrubber solution to remove all of the formaldehyde from the air. As the air
sampling rate is increased the amount of formaldehyde needed to be
removed from the air increases. The idea is to maximize the air sampling
rate and minimized the scrubber solution rate. Three scrubbers of various
design were tested to determine the most efficient and practical scrubber for
the removal of formaldehyde from air. The results are discussed in detail in
the following section.
Flow rates for each of the solvents are different and resulted in the use
of two peristaltic pumps. The flow of the scrubber solution is based on the air
flow and scrubbing efficiency. The theory is similar to that of the scrubber,
the larger amount of formaldehyde scrubbed by the scrubbing solution the
larger the signal. As the scrubber solution flow rate is increased the volume
in which formaldehyde is dissolved also increases, decreasing the overall
concentration of formaldehyde in solution. The goal was to minimize the
scrubber solution flow rate and keep the scrubbing efficiency of formaldehyde
high. The Nash reagent flow also plays an important role in the total volume

over other carbonyls and methanol. Selectivity studies were conducted
resulting with formaldehyde being 33,000 times more likely to react with 2,4-
pentanedione in the presence of ammonium acetate than acetaldehyde,
11,700 times greater than acrolein, and 87,000 times greater methanol (Dong
et al., 1987). The great selectivity for formaldehyde and the mild conditions
were the reasons for choosing 2,4-pentanedione as the p-diketone for use in
the analysis of formaldehyde.
The cyclization of a p-diketone with an aldehyde and an amine is
proposed to occur in three steps (Dong et al., 1987). The first step is an aldol
condensation, where a carbonyl is converted partially to enolate anion and an
a carbon carbon bond is formed through a nucleophilic addition (Figure 2.4).
The aldol condensation reaction is followed by a Michael addition, which
involve the nucleophilic addition of carbanions to a,p unsaturated ketones.
The conjugate additions occurs with a second enolate anion acting as weak
base towards the p-diketones (Figure 2.5). The last step of the Hantzsch
reaction is the closure of the pericyclic ring. Ammonia acts as a nucleophile
towards two carbonyls forming a pericyclic ring (Figure 2.6). The formation of
the pericyclic ring results in the production of a three water molecules. The
formation of water is relatively slow at room temperature. Therefore, heat is
added to decrease the reaction time. Estimations derived from experimental

Figure 2.5
car ban ion intermediate
Figure 2.5 Michael Reaction

data predicts that the reaction time to reach 99% formation of DDL at 20 C is
5 hours, at 37 C the reaction time is reduced to 40 minutes, and at 58 C the
reaction time for 99% completion is 5 minutes (Nash, 1953). Current
literature for analysis of formaldehyde using the wet chemical fluorescent
method maintains the reaction coil temperature at 95 C (Dong et al., 1987).
The reagent composition effects the sensitivity measurement. The
components of Nash reagent are ammonium acetate, 2,4-pentanedione, and
glacial acetic acid. Reagent concentration versus signal studies were
conducted to determine the optimal reagent composition. Past literature
determined the concentration of ammonium acetate to have the largest affect
on the signal (Kelly et al., 1994). Typically 500ml of Nash reagent at a time
was prepared for sampling, in which 230 g of ammonium acetate is used. An
experiment measuring the signal of a constant concentration of formaldehyde
and varying the concentration of ammonium acetate from 10 g to 60 g in 100
ml was conducted to determine the ideal concentration of a ammonium
acetate and is described in detail in the experimental results section.

components using existing detection procedures initially. Flow rates,
scrubber design and reagent concentration were optimized based on
individual component results. The detection limit of an instrument can be
defined as the concentration giving a signal which is three times the noise
level of the baseline or blank. A detection limit on the order of 1ppbv
formaldehyde is needed for outdoor ambient sampling where concentration of
formaldehyde ranges from less than 0.2 ppbv during the night and up to 13
ppbv during the day (Kelly, et al., 1994). Kelly et al achieved a detection limit
of 0.2 ppbv with their continuous formaldehyde analyzer. The detection limit
for indoor air sampling is not as stringent since indoor levels of formaldehyde
range from 10 to 100 ppbv. Time resolution is also an important parameter
for the outdoor ambient sampling of formaldehyde and for this instrument it is
dependent on both the flow rate of the scrubber solution and the volume of
the scrubber. The limit of detection, time resolution, and response time are
addressed later in this section.
Three formaldehyde scrubbers of different designs were tested to
determine the best scrubber for the near real-time analysis of formaldehyde.
The three scrubbers were tested using similar experimental parameters.
Acidified water (0.1 N sulfuric acid in Millipore water) was used as the

Figure 3.1
TOTAL FLOW > 2.25 L/min
Figure 3.1 Permeation System

increasing the temperature of the permeation system. The markers in figure
3.2 represent the formaldehyde concentration with a data regression line.
Table 3.1 shows the reproducibility of not only the permeation system, but
also the HPLC, the 2,4 DNPH cartridges and air sampling system. Five
separate tests at a high air flow rate setting of 318 were conducted and
resulted with a formaldehyde concentration of 12 ppbv (0.5). At a high air
flow rate settings of 218 and 118, four individual tests at each air flow rate
were conducted. At a high air flow rate of 218 the mean formaldehyde
concentration was 16 ppbv (0.5). Decreasing the high air flow rate to 118
increased the formaldehyde concentration to 24 ppbv (0.4). The
reproducibility of the concentration of formaldehyde at the different high air
flow rates was very good (0.4 to 0.5) when one takes into account the
variability of each component involved in the tests.
DNPH cartridges were used to determine both the concentration of
formaldehyde being pulled through the scrubber and the concentration of the
unscrubbed formaldehyde in the post-scrubbed air of the scrubber. Two
DNPH cartridges were used for each scrubber efficiency test. A union T
fitting was placed into the teflon tubing extending from the dilution manifold to
measure the concentration of formaldehyde in the air entering the scrubber.
A second cartridge was placed on the exhaust end of the scrubber and was

used to determine the concentration of formaldehyde not removed by the
scrubber (Figure 3.1). Efficiency runs consisted of sampling approximately 1
L/min of air over a four hour period. The total flow of the permeation system
was kept at 0.25 L/min greater than the total air sampling flow rate by placing
an air flow rate meter at the exhaust of the glass manifold. Atypical
formaldehyde concentrations were measured when the air sampling flow rate
was equal or greater than the total air flow rate of the permeation system.
Total volumes of air sampled by each cartridge was greater than 250 L. The
percent efficiencies were calculated by dividing the amount of formaldehyde
on the post-scrubber cartridge by the amount of formaldehyde in the air being
scrubbed, multiplying by 100 and subtracting that number from 100. Blank
background concentrations of formaldehyde are present on the 2,4 DNPH
cartridges, however background subtraction was not done for the samples in
the efficiency study. The lack of background subtraction ensured that
scrubbing efficiencies greater than 100% were not calculated. The lack of
background subtraction for formaldehyde from the 2,4 DNPH cartridges
underestimates the scrubbing efficiency of all of the scrubbers tested.
Figure 3.3 is a HPLC chromatogram illustrating the presence of formaldehyde
on a blank 2,4 DNPH cartridge. Figure 3.4 is a HPLC chromatogram of a
sampled 2,4 DNPH cartridge. The sampled cartridges were used to

Figure 3.4
Figure 3.4 Sampled 2,4 DNPH Cartridge

Figure 3.5
ii }
Figure 3.5 Scrubber Design

Table 3.2
Scrubber Type Mean Percent Standard Deviation
28-turn glass coil 94.9 1.0
bubbler w/frit 92.1 0.5
bubbler wo/frit 85.3 6.5
Table 3.2 Scrubber Efficiency
testing. The first 2 scrubbers involved continuous monitoring for excessive
bubble formation, when present led to the decrease in flow of the HCHO
scrubbed solution in the bottle section of the scrubber. The flow rate of the
scrubber solution was constant however, the flow rate of the HCHO scrubbed
solution fluctuated with relationship to the amount of bubbles formed. An
increase in bubbles resulted in the decrease in HCHO scrubbed solution flow
rate. Interference in the removal of formaldehyde using the 28-tum glass coil
was not evident in the laboratory. It has been suggested through particle
testing that losses of submicron particles of nearly 50% was evident with a
glass coil scrubber design using similar air flow rates described previously
(Genfa et. al.j 1991). As addressed by Kelly (Kelly et, al.j1994), the testing
emphasized polymer and silver aerosols under dry conditions. The method
of scrubbing implemented for the removal of formaldehyde is wet, therefore

minimizing electrostatic losses of aerosols or particles to the sides of the
glass coil. Field comparison studies also conducted by Kelly during separate
testing showed no inherent interference of aerosol or particle losses. It
should also be noted that HCHO is predominately gaseous in the atmosphere
and that outdoor sampling effect of particle deposition for formaldehyde is
inconsequential (Kelly et al., 1994).
Reagent Concentration
Experiments were conducted to determine an ideal ammonium acetate
concentration in the Nash reagent. Previous research (Kelly et al., 1990)
determined that the concentration of the (3-diketone and the glacial acetic
acid used in the preparation of the Nash reagent had little effect on the
sensitivity of the instrument, rather the sensitivity had a greater dependence
on the concentration of ammonium acetate in the Nash reagent. Four
individual Nash reagents consisting in ammonium acetate concentration of
1.3 M (10 g), 3.0 M (23 g), 6.0 M (46 g), and 7.8 M (60 g) were prepared in
100ml volumetric flasks as described in the experimental section. Similar
amounts of the other two components were used in each reagent, 0.3 ml of
2.4 pentanedione and 3 ml of glacial acetic acid. A reaction coil bath
temperature of 68 C was used for each of the reagents. The reaction coil

M ammonium acetate Nash reagent. An offset allowing the overlapping of
the 6.0 M reagent with the 7.8 M was applied to illustrate the minimal
illustrate the minimal improvement for the increase in ammonium acetate
concentration. The results of the reagent concentration was plotted on an x-y
graph, plotting peak response with respect to the ammonium acetate
concentration of the Nash reagent. 6.0 M is the optimum concentration of
ammonium acetate for the Nash reagent (Figure 3.8). The largest
improvement is noted by the increase in ammonium acetate concentration
with the largest slope, the increase from 3.0 M to 6.0 M. The small slope of
the line between 6.0 M and 7.8 M is near zero and signifies a minimal
improvement for the increase in the ammonium acetate concentration. Table
3.3 organizes the results of the ammonium acetate concentration
experiments, including the signal response, change in signal response and
standard deviation of the baseline for the various Nash reagents. A
concentration of 1.3 M ammonium acetate resulted in a response of 100 mV.
Slightly more than doubling the concentration of ammonium acetate to 3 M
resulted in an increase in the signal response of 71 mV. The largest increase
in the signal response was the ammonium acetate concentration increase
from 3 M to 6 M which resulted in an increase of 314 mV. The most
concentrated ammonium acetate reagent, 7.8 M had a signal response of

490 mV, a change of 6 mV from the 6 M reagent. Vigorous shaking and
allowing the reagent to stand for the 24 hours needed for the keto-enol
tautomerization to equilibrate was necessary for the reagent to completely
dissolve. Solubility problems with ammonium acetate at such high
concentrations and the cost of ammonium acetate makes the 6.0 M solution
more practical for the analysis of formaldehyde.
Previous research on the concentration of ammonium acetate resulted
in a lower background signal in fluorescence in the absence of formaldehyde
at higher concentrations of ammonium acetate (Kelly et al., 1990). The
standard deviations of the baseline for the four concentrations of ammonium
acetate were random (Table 3.3). The four solutions were prepared from the
same lot of chemicals and at the same time. The only difference was the day
they were tested. The 1.3 M and the 3.0 M reagents were tested on a the
same day and the 6.0 M and the 7.8 M reagents were tested two days later.
The variability in the standard deviation of the different concentrations of
ammonium acetate is inherent to the instrument. More specifically the
components outlined in this research. Small variations in flow caused by tube
crimping or peristaltic pump anomalies are propagated through to the signal
response, resulting in a larger standard deviation of both the baseline signal
and peak response signal. Previous research noted that the excitation light

Figure 3.9
T=24 C---- T=50 C----- T=68 C---- T=82 C---- T=92 C
Figure 3.9 Signal Response Dependence on Temperature

temperature. The standard deviation in the response signal at 92 C (145mV)
was caused by the production of bubbles at higher reaction coil temperatures
as seen by the sharp peaks in figure 3.9. At 24 C, room temperature, an
average signal change of 73 mV was measured with a standard deviation of
7 mV, corresponding to a 3 mV change per ppbv HCHO. Increasing the
temperature to 50 C resulted in a mean signal response of 347 ( 8) mV, a
change of 14.5 mV per ppbv HCHO. At 68 C, the signal increased to a mean
response change of 489 (12) mV, with a change of 20.5 mV per ppbv
HCHO. Increasing the reaction coil temperature to 82 C increased the mean
signal response to 498 (11) mV with a change of 20.9 mV per ppbv HCHO.
At a temperature near 92 C, previously reported in the literature, standard
deviation of the baseline increased to 18 mV and the mean change in signal
increased to 508 mV corresponding to a 21.3 mV change per ppbv HCHO
and a 0.4 mV increase in signal per ppbv formaldehyde.
Figure 3.10 is an x-y plot of the change in response (mV) versus the
temperature of the reaction coil. A significant increase in the change in signal
per ppbv HCHO is noted by the large slope of the line between 24 C and 50
C. Increasing the temperature to 50 C increased the signal response to 14.5
mV per ppbv HCHO. Greater increase in the response change per ppbv
HCHO is evident from 50 C to 68 C. The reaction coil temperature of 68 C

resulted in a signal response change of 20.5 mV. Minimal increase in the
response change, 20.9 mV and 21.3 mV respectively, was measured at the
higher temperatures resulting in a small slope in the line from 68 C to 92 C.
A temperature in the range of 60 C was chosen as the reaction coil
temperature based on the small standard deviation and large response per
ppbv HCHO. 82 C also resulted in a small standard deviation and large
signal response, however the increase in the temperature by 14 C for a 0.4
mV per ppbv HCHO change does not warrant the increase in the reaction coil
temperature. The instrument is ultimately to be placed in the field and the
lower reaction coil temperature will result in less maintenance for field
sampling. At the higher reaction coil temperatures a debubbler would be
necessary to remove the air bubbles generated.
The reaction coil temperature is also dependent on the resident time of
the components in the system. As mentioned previously, during this testing
two double coiled stainless steel reaction coils were used. The first was
immersed into the hot water bath and the second was placed between the
bath and the fluorometer. Experiments investigating the time requirements
for 99% completion for the Hantzsch reaction with 8 pg formaldehyde
reacting with a similar Nash reagent (in component concentrations) was
determined at 20 C to be approximately 5 hours, at 37 C the time was

Figure 3.11
Figure 3.11 Sampling of 24 ppbv HCHO at 69 C

A second experiment sampling 24 (0.4) ppbv HCHO was conducted
under similar conditions as the previously mentioned run (Figure 3.12). The
reaction coil bath temperature was 66 C and the sampling time was also 60
minutes. Zero air was sampled for 12 minutes. The mean signal during the
zero air sampling was 104 (8) mV. Gaseous formaldehyde with a
concentration of 24(0.4) ppbv was sampled for 13 minutes with a mean in
signal of 640 (15) mV. A signal response of 22 mV per ppbv formaldehyde
was calculated. The mean of the post-sampling zero air was 115 (9) mV.
An increase of 11 mV over the pre-sampling baseline. The rise time from
10% to 90% signal response was 2.0 minutes, a small increase in response
time over the previous experiment. The fall time from 90% to 10% was 2.3
minutes. Relatively similar to the previous experiment, where the rise time
was 2.2 minutes and the fall time was 2.4 minutes. The rise and fall times
are slightly longer than both the times reported by Kelly which approximately
80 seconds (Kelly et al., 1994) and the enzymatic approach to sampling
formaldehyde. The rise and fall times are dependent on the volume of
acidified water in the scrubber and the resident time of the aqueous flow in
the system. Decreasing the volume of solution could improve the rise and fall
times and could be achieved by decreasing the scrubber flow rate.
Decreasing the scrubber flow would also increase the signal response by

decreasing the total volume of the aqueous flow. Also, decreasing the
resident time of the aqueous flow would minimize the longitudal diffusion of
the scrubbed formaldehyde, decreasing the rise and fall times and increasing
the time resolution. Lag time is another component which is effected by the
flow rates and resident time of the aqueous flow. The lag time is larger than
other methods due to the decrease in reaction coil temperature. A current
lag time of approximately 12 minutes is evident. Lag time, time resolution
and rise and fall times will be addressed further in the conclusion section.
Another experiment sampling various known concentrations of HCHO was
conducted during a 90 minute instrument run, with a reaction coil temperature
of 68 C (Figure 3.13). The run began by the sampling of zero air until a
stable baseline was achieved. At time equals 0 minutes, 12 (0.5) ppbv
HCHO was sampled. A change of 204mV was measured with a delay of 12
minutes from sampling zero air to 90% peak signal. The signal response
time delay from 10% peak signal to 90% peak signal was 2.3 minutes. The
concentration of HCHO was increased to 16 (0.5) ppbv by decreasing the
diluent air flow. A change of 287mV from zero air was measured with a
signal response time delay of 2.4 minutes. The concentration of HCHO was
increased a second time to 24 (0.4) ppbv and 544mV increase from
baseline was measured, corresponding to a signal change of 22.6 mV ppbv

formaldehyde. The signal response time delay was 2.6 minutes. The
concentration of HCHO was decreased to 16 (0.5) ppbv and the change in
mV from baseline was 292, a 5mV difference in the initial measurement of 16
(0.5) ppbv HCHO. Zero air was sampled again at the end of the run. An
increase in the pre-sampling and post-sampling baseline and signal response
in sampling 16 (0.5) ppbv HCHO is evident. The increase in the baseline
signal from pre-sampling to post-sampling was 44 mV and the increase in
signal response for sampling 16 (0.5) ppbv HCHO was 52 mV. The
similarity in the signal increase in the pre and post-sampling baseline and the
sampling of 16 (0.5) ppbv HCHO. During initial sampling the water bath
hadnt reached 68 C and the rise in the baseline signal is attributed to the
stabilization of the reaction coil bath temperature. The difference in the
signal response for the sampling of 16 (0.5) ppbv HCHO was calculated by
subtracting the average of the peak signal response from the average of the
pre-sampling and post-sampling baseline signal. The change in mV per ppbv
HCHO for each sampling period was 16 mV/ppbv HCHO, 17 mV/ppbv
HCHO, 20 mV/ppbv HCHO, and 18mV/ppbv HCHO respectively.
Corresponding to a limit of detection of 2-3 ppbv at these higher
formaldehyde concentrations. Reproducible responses are noted for
sampling at 16 (0.5) ppbv during this run and day-to-day reproducibility is

Figure 3.14
Figure 3.14 Temperature Ramp

Table 3.4
Figure HCHO concentration (ppbv) Response mV/ppbv Standard Deviation Baseline (mV) LOD (ppbv)
3.9 24 20.4 11.7 2
3.11 24 21.9 7.6 1
3.12 24 19.8 10.5 2
3.13 24 19.6 9.5 1
Table 3.4 Limit of Detection

response. All of the modifications are related and would need to be
addressed in a systematic approach.
The continuous near real-time analysis of formaldehyde was
accomplished in the laboratory. Identification of a rugged and efficient
scrubber, determination of the ideal reagent concentration and a lower
reaction coil temperature eliminating need of a debubbler was achieved. The
derivatization of formaldehyde using the Hantzsch reaction provides a
portable, basic, and practical approach for the continuous monitoring of
formaldehyde. The instruments reproducibility for the detection of elevated
formaldehyde was good. Five separate runs conducted on different days and
different reagent resulted in a mean signal change of 20.6 (1.1) mV/ppbv. A
signal change on the order of 20 mV/ppbv corresponds to a detection limit of
1-2 ppbv depending on the signal noise of the baseline. Similar results were
also demonstrated at the previously mentioned formaldehyde concentrations
of 12 and 16 ppbv. A detection limit on the order of 1 ppbv may be attainable
with the modifications suggested.

Kelly, T.J. and Bames, R.H. Development of Real-Time
Monitors for Gaseous Formaldehyde, Final Report to U.S.
Environmental Protection Agency, Battelle, Columbus, Ohio,
Report No. EPA/600/3-90/088, November, 1990.
Kleindienst, T.E.; Shepson, P.B.; Nero, C.M.; Amts, R.R.;
Tejada, S.B.; Macky, G.I.; Mayne, L.K.; Schiff, H.I.; Lind, J.A.;
Kok, G.L.; Lazrus, A.L. Atmospheric Environment, 1988, Vol.
22, No. 9, 1931-1939.
Lazrus, A.L.; Fong, K.L.; Lind, J.A. Anal. Chem. 1988, 60,1074-
Lee, N. and Zhou, X. Environ. Sci. Tech. 1993, 27, 749-756.
Nash, T. Biochem. 1953, 55, 416-421.
Salthammer, T. J. Photochem. Photobiol. A: Chem. 1993,74,
Sawicki, E. and Cames, R.A. Mikrochim. Acta. 1968,1, MS-
Sigrist, M. W. Air Monitoring by Spectroscopic Techniques,
John Wiley & Sons, Inc., New York, Vol. 127, 1994.
Tuazon, E.C.; Winer, A.M.; Pitts, J.N. Jr. Adv. Envir. Sci. Tech.
1980, 10, 259-275.
Vairaramurthy, A.; Roberts, J.M.; Newman, L. Atmos. Environ.
1992, 26A, 1965-1993.