Development of a compound-specific carbon isotope analysis method for atmospheric formaldehyde via NaHSO3 and cysteamine derivatization

Article abstract of DOI:10.1021/ac051616f

A novel method has been developed for the compound-specific carbon isotope analysis of atmospheric formaldehyde using gas chromatography/combustion/isotope ratio mass spectrometry (GC/C/IRMS). The method allows the determination of the δ¹³C value for atmospheric formaldehyde at nanogram levels with higher precision and lower detection limit. In the present work, atmospheric formaldehyde was collected using NaHSO₃-coated Sep-Pak silica gel cartridges, washed out by water, then derivatized by cysteamine of known δ¹³C value, and the δ¹³C value of its derivative (thiazolidine) determined by GC/C/IRMS. Finally, the δ¹³C value of atmospheric formaldehyde could be calculated by a simple mass balance equation between formaldehyde, cysteamine, and thiazolidine. Using three formaldehydes with different δ¹³C values, calibration experiments were carried out over large ranges of formaldehyde concentrations. The carbon isotope analysis method achieved excellent reproducibility and high accuracy. There was no carbon isotopic fractionation throughout the derivatization processes. The differences in the carbon isotopic compositions of thiazolidine between the measured and predicted values were always <0.5‰, within the specifications of the GC/C/IRMS system. The present method was also compared with the previous 2,4-dinitrophenylhydrazine derivatization method, and this method could be performed with lower analytical error and detection limit. Using this method, four 6-h ambient atmospheric formaldehyde samples were consecutively collected from 8 to 9 March 2005. The results showed that the δ¹³C values of atmospheric formaldehyde were different during the daytime and nighttime. This method proved suitable for the routine operation and may provide additional insight on sources and sinks of atmospheric formaldehyde.

Full text of DOI:10.1021/ac051616f

Anal. Chem. 2006, 78, 1206-1211
Development of a Compound-Specific Carbon
Isotope Analysis Method for Atmospheric
Formaldehyde via NaHSO3 and Cysteamine
Derivatization
Y. X. Yu,†,‡ S. Wen,‡ Y. L. Feng,‡ X. H. Bi,‡ X. M. Wang,‡ P. A. Peng,‡ G. Y. Sheng,†,‡ and J. M. Fu*,†,‡
Research Institute of Environmental Pollution and Health, School of Environmental and Chemical Engineering, Shanghai
University, Shanghai 200072, P. R. China, and State Key Laboratory of Organic Geochemistry, Guangdong Key Laboratory
of Environmental Resources Utilization and Protection, Guangzhou Institute of Geochemistry, Chinese Academy of
Sciences, Guangzhou 510640, P. R. China
Research on atmospheric formaldehyde is important not only
because of its adverse effect on human health but also because
of its role in atmospheric chemistry.1,2 Formaldehyde constitutes
the most abundant carbonyl in ambient air and is derived from
direct emissions from industrial processes, vehicle exhaust emis-
sions, and other stationary sources, and also from photochemical
reactions, generally through the oxidation of hydrocarbons in the
presence of nitrogen oxides, OH radicals, and ozone.3-6 Though
many studies have been performed, there are still gaps regarding
its sources and atmospheric reactions.1-6
In the past decade, compound-specific carbon isotopic com-
position measurements have been successfully used to improve
our insight into the budgets and the processes of some atmo-
spheric trace gases, e.g., CO, CO2, CH4, and other trace
compounds.7-9 Gas chromatography/combustion/isotope ratio
mass spectrometry (GC/C/IRMS) has become a powerful analyti-
cal tool to infer the origin and track the fate of organic compounds
in various systems.7-10 This technique permits continuous flow
acquisition of δ13C data for individual components in complex
mixtures at nanogram levels. However, despite the development
of improved methods for compound-specific carbon isotope
analysis of CO, CO2, CH4, and non-methane hydrocarbons, there
are few published studies concerning the stable carbon isotope
analysis of atmospheric formaldehyde.11-13 In one of these
A novel method has been developed for the compound-
specific carbon isotope analysis of atmospheric formal-
dehyde using gas chromatography/combustion/isotope
ratio mass spectrometry (GC/C/IRMS). The method al-
lows the determination of the δ13C value for atmospheric
formaldehyde at nanogram levels with higher precision
and lower detection limit. In the present work, atmo-
spheric formaldehyde was collected using NaHSO3-coated
Sep-Pak silica gel cartridges, washed out by water, then
derivatized by cysteamine of known δ13C value, and the
δ13C value of its derivative (thiazolidine) determined by
GC/C/IRMS. Finally, the δ13C value of atmospheric
formaldehyde could be calculated by a simple mass
balance equation between formaldehyde, cysteamine, and
thiazolidine. Using three formaldehydes with different
δ13C values, calibration experiments were carried out over
large ranges of formaldehyde concentrations. The carbon
isotope analysis method achieved excellent reproducibility
and high accuracy. There was no carbon isotopic frac-
tionation throughout the derivatization processes. The
differences in the carbon isotopic compositions of thia-
zolidine between the measured and predicted values
were always <0.5‰, within the specifications of the
GC/C/IRMS system. The present method was also com-
pared with the previous 2,4-dinitrophenylhydrazine de-
rivatization method, and this method could be performed
with lower analytical error and detection limit. Using this
method, four 6-h ambient atmospheric formaldehyde
samples were consecutively collected from 8 to 9 March
2005. The results showed that the δ13C values of atmo-
spheric formaldehyde were different during the daytime
and nighttime. This method proved suitable for the routine
operation and may provide additional insight on sources
and sinks of atmospheric formaldehyde.
(1) Carlier, P.; Hannachi, H.; Mouvier, G. Atmos. Environ. 1986, 20, 2079-
2099.
(2) Flores, G. E.; Andraca-Ayala, G.; Ba´ez, A. Atmos. Environ. 2005, 39, 1027-
1034.
(3) Takekawa, H.; Minoura, H.; Yamazaki, S. Atmos. Environ. 2003, 37, 3413-
3424.
(4) Grutter, M.; Flores, E.; Andraca-Ayala, G.; Ba´ez, A. Atmos. Environ. 2005,
39, 1027-1034.
(5) Iannone, R.; Anderson, R. S.; Rudolph, J.; Huang, L.; Ernst, D. Geophys. Res.
Lett. 2003, 30 (13), 1684. Doi: 10.1029/2003GL017221.
(6) Feltham, E. J.; Almond, M. J.; Marston, G.; Ly, V. P.; Wiltshire, K. S.
Spectrochim. Acta, Part A 2000, 56, 2605-2616.
(7) Rudolph, J.; Czuba, E.; Norman, A. L.; Huang, L.; Erns, D. Atmos. Environ.
2002, 36, 1173-1181.
(8) Mook, W. G.; Koopmans, M.; Carter, A. F.; Keeling, C. D. J. Geophys. Res.
1983, 88, 10915-10933.
* Corresponding author: (e-mail) fujm@gig.ac.cn; (fax) +86-20-85290192; (tel)
+86-20-85290196.
(9) Lowe, D. C.; Brenninkmeijer, C. A. M.; Brailsford, G. W.; Lassey, K. R.;
Gomez, A. J. J. Geophys. Res. 1994, 99, 16913-16925.
(10) Zwank, L.; Berg, M.; Schmidt, T. C.; Haderlein, S. B. Anal. Chem. 2003,
75, 5575-5583.
† Shanghai University.
‡ Guangzhou Institute of Geochemistry, Chinese Academy of Sciences.
1206 Analytical Chemistry, Vol. 78, No. 4, February 15, 2006
10.1021/ac051616f CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/24/2005

reports,11 the carbon isotope analyses were performed as fol-
lows: the atmospheric formaldehyde was collected using a
NaHSO3-coated filter, and the product sodium hydroxymethane-
sulfonate (HOCH2SO3Na, denoted here as HMSNa) was analyzed
using a conventional dual inlet isotope ratio mass spectrometry
(IRMS) system. The measured carbon isotopic data were reported
as pertaining to atmospheric formaldehyde. The main drawback
of this method is that the analysis requires the enrichment of
formaldehyde from several hundred cubic meters of air, which
may take one or two days of sampling. Recently, a novel method
determining the compound-specific carbon isotopic composition
for atmospheric formaldehyde at ppb or sub-ppb levels by 2,4-
dinitrophenylhydrazine (DNPH) derivatization was developed, but
its precision is poor.13
It is well known that formaldehyde reacts rapidly with NaHSO3
to form the nonvolatile HMSNa.14 Other studies have shown that
HMSNa decomposes to formaldehyde in strong acid and then
reacts with cysteamine to form thiazolidine,15-20 which is much
more stable and less volatile than formaldehyde. However,
thiazolidine is sufficiently volatile for gas chromatographic (GC)
analysis. In the present work, a novel method is described, which
applies GC/C/IRMS to determine the δ13C value of derivatized
atmospheric formaldehyde at nanogram levels. Calibration experi-
ments, atmospheric sampling, accuracy, and reproducibility of the
method will be discussed in detail, and the stable carbon isotope
effects during the procedure will be evaluated. Also, δ13C data
for atmospheric formaldehyde will be presented to demonstrate
the practical utility of this method.
below. The sampling media commercial Sep-Pak silica gel car-
tridges (Waters, Millipore Corp.) coated with DNPH or NaHSO3.
Cartridge Preparation. Sep-Pak silica gel cartridges were
coated with DNPH according to our previous reports.22,23 The
cartridges were rinsed with 10 mL of ACN and were coated with
7 mL of freshly made acidified DNPH coating solution. When
there was no more solution flowing out of the cartridges, they
were dried with a gentle flow of high-purity nitrogen. Each
cartridge was wrapped in aluminum foil and then wrapped with a
piece of filter paper impregnated with DNPH coating solution to
prevent contamination before use. Last, each cartridge was sealed
in a Teflon bag. All the processes were carried out in a high-
purity nitrogen-filled glovebox (ZKX2, 800 mm × 600 mm × 700
mm, Nanjing University Instrument Plant).23 The prepared car-
tridges were stored at 4 °C until use. Three of each batch were
analyzed using HPLC to test the blank compared to the EPA blank
criteria.24
The NaHSO3-coated Sep-Pak silica gel cartridges were pre-
pared similarly to the DNPH-coated cartridges. The cartridges
were rinsed with 10 mL of ACN, subsequently rinsed with 7 mL
of water, and then coated with 7 mL of freshly made 10% (w/v)
NaHSO3 solution. When there was no more solution flowing out,
they were dried with a gentle flow of high-purity nitrogen. Each
cartridge was wrapped in a piece of filter paper, impregnated with
the NaHSO3 solution to prevent contamination before use, and
then sealed in a Teflon bag. All the processes were carried out in
the high-purity nitrogen-filled glovebox. The NaHSO3-coated
cartridges were stored at 4 °C until use. After washing out and
derivatization as described below, four cartridges were analyzed
using GC/C/IRMS and the results showed no interference.
Preparation of Standard Thiazolidine. Standard thiazolidine
was prepared by reacting equimolar amounts of cysteamine
hydrochloride with formaldehyde in a water solution of pH 8.15-20
After 4 h, the solution was extracted three times with chloroform;
the extract was dried over anhydrous sodium sulfate, filtered, and
then concentrated by a rotary evaporator. The complete procedure
was carried out at room temperature. The purity of the thiazolidine
was verified using GC and GC/MS, respectively, and its δ13C value
was determined by GC/C/IRMS.
EXPERIMENTAL SECTION
Reagents and Materials. Acetonitrile (ACN) purchased from
Merck was HPLC grade. Chloroform was purchased from Shantou
XiLong Chemical Co. Ltd. and distilled three times. Water was
double distilled and filtered by a Milli-Q system. Cysteamine
hydrochloride (97%) purchased from Fluka was recrystallized
twice in ethanol.21 DNPH, purchased from Fluka, was recrystal-
lized twice in HPLC grade ACN and analyzed by HPLC. NaHSO3
was purchased from United Research Institute of Chengdou.
Standard formaldehyde (37% in water solution with 10% methanol)
samples were obtained from three suppliers: Aldrich (F1),
Guangzhou Second Chemical Reagent Factory (F2), and Guang-
zhou Chemical Reagent Factory (F3), respectively.
2,4-Dinitrophenylhydrazine was purchased from ChemService.
A GV-mix standard solution containing C10, C11, and C12 n-alkanes
and methyl decanoate with δ13C values of -28.6, -26.7, -28.6,
and -30.5‰, respectively, was obtained from GV Corp. The
standard thiazolidine was prepared in our laboratory as described
Measurements of δ13C of Standard Formaldehyde. The
method for determining the δ13C value of formaldehyde in a
standard stock solution was as follows:25 aliquots of 37% stock
formaldehyde solution were sealed in glass vials with open screw
caps containing Teflon-lined silica septa. After at least 1 h to reach
equilibrium, ∼1 mL of headspace air containing formaldehyde was
injected into the GC/C/IRMS system for analysis using a
Hamilton airtight locking syringe.
Calibration Experiments for Atmospheric Sampling. Cali-
bration experiments were performed using an airtight system (see
Figure 1), consisting of a 100-L Teflon sample bag (SKC Inc.)
connected to a sampling pump (Thomas). Before each experiment,
the bag was washed at least three times with high-purity (99.99%)
(11) Tanner, R. L.; Zielinska, B.; Uberna, E.; Harshfield, G. J. Geophys. Res. 1996,
101, 28961-28970.
(12) Johnson, B. J.; Dawson, G. A. Environ. Sci. Technol. 1990, 24, 898-902.
(13) Wen, S.; Feng, Y. L.; Yu, Y. X.; Bi, X. H.; Wang, X. M.; Sheng, G. Y.; Fu, J.
M. Environ. Sci. Technol. 2005, 39, 6202-6027.
(14) Andrade, J. B.; Tanner, R. L. Atmos. Environ. 1992, 26A, 819-825.
(15) Goldman, F. H.; Yagoda, H. Ind. Eng. Chem. 1943, 15, 377-378.
(16) Rayner, A. C.; Jephcott, C. M. Anal. Chem. 1961, 33, 627-630.
(17) Hayashi, T.; Reece, C. A.; Shibamoto, T. J. Assoc. Off. Anal. Chem. 1986,
69, 101-105.
(22) Feng, Y. L.; Wen, S.; Wang, X. M.; Sheng, G. Y.; He, Q. S.; Tang, J. H.; Fu,
J. M. Atmos. Environ. 2004, 38, 103-112.
(23) Feng, Y. L.; Wen, S.; Chen, Y. Z.; Wang, X. M.; Lu¨, H. X.; Bi, X. H.; Sheng,
(18) Yasuhara, A.; Shibamoto, T. J. Assoc. Off. Anal. Chem. 1989, 72, 899-902.
(19) Miyake, T.; Shibamoto, T. J. Agric. Food Chem. 1993, 41, 1968-1970.
(20) Huang, T. C.; Huang, L. Z.; Ho, C. T. J. Agric. Food Chem. 1998, 46, 224-
227.
G. Y.; Fu, J. M. Atmos. Environ. 2005, 39, 1789-1800.
(24) U.S. Environment Protection Agency (US EPA). Compendium method to-
11A. Active Sampling Methodology, 1999.
(25) Wen, S.; Feng, Y. L.; Wang, X. M.; Sheng, G. Y.; Fu, J. M.; Bi, X. H. Rapid
Commun. Mass Spectrom. 2004, 18, 2669-2672.
(21) Taguchi, T.; Kojima, M.; Muro, T. J. Am. Chem. Soc. 1959, 81, 4322-4325.
Analytical Chemistry, Vol. 78, No. 4, February 15, 2006 1207

One field blank was analyzed, and the result showed no interfer-
ence.
At the same time, another four samples were collected using
the DNPH-coated cartridges as described earlier.22,23 The flow rate
was 0.8-1.2 L min-1 through sampling. An ozone denuder (a
copper tube coil, 1.2 m × 0.5 mm i.d. coated with KI inside) was
connected in front of the cartridge to avoid degradation of 2,4-
dinitrophenylhydrazine.27,28 After sampling, the cartridges were
backflushed with ACN under gravity into 2-mL volumetric flasks.
The sample eluates were stored at 4 °C until analysis by HPLC.
One field blank was analyzed, and the result was below the EPA
limit.
Analytical Systems. The concentration of formaldehyde was
determined by HPLC (Hewlett-Packard model 1100) with a
photodiode-array detector and UV light of 360 nm. An SB-C18
reverse column (250 mm × 4.6 mm, 5 µm, Agilent) was used. A
10-µL aliquot was injected into the HPLC system via autosampler
and analyzed as follows: mobile-phase gradient with a flow rate
of 1 mL min-1 from 60 to 70% ACN in water (v/v) for the first 20
min, 70 to 100% ACN for 3 min, 100% ACN isocratic elution for 4
min, 100 to 60% ACN for 1 min, and last 60% ACN 5 min isocratic.
The GC analyses were carried out with an HP 5890 system
(Hewlett-Packard) equipped with an HP-5MS column (30 m ×
0.32 mm × 0.25 µm, Agilent) and flame ionization detector. The
operating conditions were as follows: He carrier flow rate at 1.5
mL min-1, splitless injection, injector and detector temperatures
at 250 and 300 °C, respectively. The oven temperature was
programmed from 45 °C at the rate of 3 °C min-1 to 80 °C.
The GC/MS analyses were performed with an HP 5890 gas
chromatograph combined with an HP 5972 mass selective detector
(Hewlett-Packard) in the full scan mode for 35-350 Da. The other
operating conditions were the same as those used for the GC
analyses above.
The δ13C value of cysteamine hydrochloride was determined
with an elemental analyzer-isotope ratio mass spectrometer
(EA-IRMS, Thermo Finnigan MAT, Bremen Germany, DELTAplus
XL mass spectrometer). The cysteamine hydrochloride samples
were weighed in Sn capsules with negligible backgrounds. The
weighed capsules were dropped into a CuO combustion furnace
in the CE EA1112 C/N/S analyzer through an autosampler and
combusted at 960 °C in an O2 atmosphere. The combustion gases
were swept through a Cu reduction oven operated at 650 °C and
then passed through a GC column where CO2 was separated from
other gases. The CO2 entered the DELTAplus XL MS for measure-
ment of its δ13C value through a Conflo III interface (Finnigan).
At the beginning and end of each batch of samples, carbon black
standards (δ13C ) -36.91‰) were analyzed to evaluate the
reproducibility and accuracy of the instrument.29
Figure 1. Schematic representation of the calibration experiment
for atmospheric sampling (for breakthrough test).
N2. After that, the standard formaldehyde stock solution was
injected into the bag, heated to aid evaporation, and then kept
for 30 min to reach equilibrium. Then the gas (containing
formaldehyde) was drawn out and through the sampling cartridge
(NaHSO3-coated, DNPH-coated, or both) using the sampling
pump. To test the recovery efficiency during sampling for
formaldehyde, two kinds of experiments were performed: (1) for
breakthrough tests, the formaldehyde was collected by drawing
the gases first through a NaHSO3-coated cartridge and then a
DNPH-coated cartridge in series; and (2) for the adsorption test
of the whole system, only a DNPH-coated cartridge was used to
measure the formaldehyde concentration. By comparison of the
measured and theoretical formaldehyde concentrations, we could
determine whether the adsorption of the whole system was major
or minor. In the calibration experiments for atmospheric sampling,
just a NaHSO3-coated cartridge was used. The flow rate was 1 or
2 L min-1 measured by a rotameter, which was calibrated with a
digital flow meter (DryCal DC Lite, Bios Corp.) before use. After
sampling, the DNPH-coated cartridge was eluted with ACN into
2-mL volumetric flask, and the eluate was analyzed using HPLC.
The NaHSO3-coated cartridge was treated as described below.
Sample Preparation for GC/C/IRMS. After sampling, the
NaHSO3-coated cartridge was eluted with 10 mL of water into a
25-mL beaker and the pH of the solution was adjusted to 1-2;
the solution was then placed in a 60 °C water bath for 20 min to
decompose the HMSNa and evaporate the SO2, removed, and 1
mL of 19 µg µL-1 cysteamine hydrochloride solution with known
δ13C value added. The final pH of the solution was adjusted to
8-9 with sodium hydroxide solution. After 4 h, the solution was
extracted three times with 10 mL of chloroform, and the combined
extract was dried over anhydrous sodium sulfate. Then the extract
was filtered, concentrated with a rotary evaporator at 20 °C,
transferred into a 2-mL vial, and blown down to ∼100 µL using a
gentle flow of high-purity N2. Finally, 1 µl of the concentrated
sample was injected into the GC/C/IRMS for analysis.
Air Sampling. The samples were collected on the roof of an
office building 15 m above ground at the Guangzhou Institute of
Geochemistry, Chinese Academy of Sciences located in the Tianhe
District of Guangzhou, China. The Guangyuan expressway and a
railway are located ∼100 m away. All samples were collected at a
height of 1 m above the floor.
The δ13C values of formaldehyde were determined using an
HP 6890 GC (Agilent) equipped with an HP-PLOT Q column (30
m × 0.32 mm × 20 µm, Hewlett-Packard) coupled to a combustion
furnace and DELTAplus XL isotope ratio mass spectrometer (GC/
C/IRMS). The CuO/Ni/Pt combustion furnace and Cu reduction
Four 6-h samples were collected consecutively from 7:00 a.m.
on 8 March 2005 to 7:00 a.m. on 9 March 2005. NaHSO3-coated
cartridges were used as sampling media, and air was drawn
through at a flow rate of 2 L min-1 by the sampling pump. After
sampling, the cartridges were treated as described above, and
the δ13C values of the thiazolidine determined by GC/C/IRMS.
(26) Yu, Y. X.; Wen, S.; Feng, Y. L.; Bi, X. H.; Wang, X. M.; Sheng, G. Y.; Fu, J.
M. Rapid Commun. Mass Spectrom. 2005, 19, 2469-2472.
(27) Christensen, C. S.; Skov, H.; Lohse, C. Atmos. Environ. 2000, 34, 287-
296.
(28) Mu¨ller, K.; Pelzing, M.; Gnauk, T.; Kappe, A.; Teichmann, U.; Spindler, G.;
Haferkorn, S.; Jahn, Y.; Herrmann, H. Chemosphere 2002, 49, 1247-1256.
(29) Jia, G. D.; Peng, P. A. Mar. Chem. 2003, 82, 47-54.
1208 Analytical Chemistry, Vol. 78, No. 4, February 15, 2006

for three batches of formaldehyde (with different δ13C values) was
evaluated. Their δ13C values and those of the corresponding
thiazolidine derivatives are given in Table 2. The analytical errors
(standard deviation) obtained for three or five GC/C/IRMS
analyses of formaldehyde from the same supplier ranged from
0.13 to 0.40‰, with an average of 0.28 ( 0.14‰, and for
thiazolidine, the analytical errors were from 0.15 to 0.26‰, with
an average of 0.20 ( 0.03‰. The accuracy of these results was
well within the technical specifications of the GC/C/IRMS system.
The reproducibility was also excellent compared with that of the
previous studies.7,13,25,30,31
Storage Stability of Cartridge Samples. Different storage
conditions were discussed. Samples of F3 were stored for 30 days
with one batch at 4 °C and another batch at room temperature
(3-26 °C) (Table 2). These samples were derivatized to thiazo-
lidine and analyzed by GC/C/IRMS. The δ13C value differences
among these two batches of samples are negligible (all within
instrument specifications). The results also show that the form-
aldehyde-NaHSO3 adducts (HMSNa) were stable and the car-
tridges could be stored over long periods.
Table 1. Sampling Efficiency and Breakthrough of
Formaldehyde in Calibration Experiment for
Atmospheric Sampling
concentrations of
formaldehyde
formaldehyde/
total
rate flow
cartridgea (µg)
(AM ( SD)b
volume (L)
(L min-1
)
(µg m-3
)
100
1
2
2
2
20.28-312.00c
20.28-312.00c
15.6
0.07 ( 0.02 (n ) 6)
0.05 ( 0.01 (n ) 6)
0.06 ( 0.03 (n ) 3)
0.05 ( 0.02 (n ) 3)
0.05 ( 0.03 (n ) 3)
100
800
1200
lab blank
7.8
a Amount of formaldehyde in one cartridge. b The arithmetic means
and standard deviations. c Concentrations of formaldehyde in calibra-
tion experiment for atmospheric sampling at six different concentra-
tions of 20.28, 40.56, 62.40, 99.84, 199.68, and 312.00 µg m-3
respectively. n, number of samples analyzed by HPLC.
,
oven were operated at 940 and 650 °C, respectively.13 Split injection
(∼10:1) was used, and the GC oven temperature was kept at 180
°C. Standard CO2 (δ13C ) -26.65‰) was used as the external
reference gas. Standard CH4 (δ13C ) -36.30‰) was used to
evaluate the reproducibility and accuracy of the IRMS.13
The δ13C values of thiazolidine were determined using the
same GC/C/IRMS system as above. The GC was equipped with
an HP-5MS column and operated as above. Over 60 ng of
thiazolidine was needed for every injection to obtain data of
acceptable accuracy and precision. The reproducibility and ac-
curacy of the GC/C/IRMS system were checked daily with two
isotopic standards: a GV-mix standard solution and a standard
thiazolidine solution. Triplicate standard analyses yielded excellent
accuracy and reproducibility. All 13C/12C ratios in this work are
expressed in conventional δ notation, as per mil (‰) relative to
the Vienna PDB standard.
Isotope Effects of the Method. Theoretically, the substrates
and products should exhibit δ13C compositions that reflect the
relative contributions of carbon from each component and their
respective δ13C values. In this method, the following reactions of
occur:
HCHO + NaHSO3 ) HOCH2SO3Na
HOCH2SO3Na + HCl ) HCHO + NaCl + H2O + SO2v
(2)
(1)
RESULTS AND DISCUSSION
During the course of the reactions, if formaldehyde reacts
quantitatively with NaHSO3, and HMSNa discomposes completely,
there should be no carbon isotopic fractionation; and if formal-
dehyde also quantitatively reacts with cysteamine in accordance
with the stoichiometric mass balance equation, there should also
be no carbon isotopic fractionation during that reaction. Thus,
the stable carbon isotopic compositions of formaldehyde, HMSNa,
cysteamine, and thiazolidine should comply with the following
equations:
δ13C Analysis of Standard Thiazolidine. The δ13C value of
standard thiazolidine was determined from 30 measurements by
GC/C/IRMS to be -33.00 ( 0.20‰. The thiazolidine was stored
at 4 °C as a standard stock solution and was used with the
GV-mix standard solution as laboratory standards to check
reproducibility and accuracy.
Efficiency of Formaldehyde Recovery in the Calibration
Experiment. The results of breakthrough tests indicated that no
formaldehyde appeared in the DNPH-coated cartridge (Table 1).
In the adsorption test of the whole system, using different
formaldehyde concentrations (similar to Table 1), the recovery
efficiencies were all over 90%,34 which means the loss within the
system was minor. Therefore, formaldehyde was efficiently col-
lected in the calibration experiment.
δ13CHCHO ) δ13CHMSNa
(4)
1/3δ13CHCHO + 2/3δ13Ccysteamine ) δ13Cthiazolidine (5)
δ13C Analysis of Formaldehyde and its Derivative (Thi-
azoline). The reproducibility of the carbon isotope composition
where 3 and 2 are the mole C numbers of thiazolidine and
cysteamine, respectively. According to eq 5, the stable carbon
isotopic composition of the original formaldehyde could be
calculated from the known δ13C values of cysteamine and thiazo-
lidine determined by EA-IRMS and GC/C/IRMS, respectively.
According to Rieley’s discussion of kinetic isotope effects,32
the primary isotope effect, whereby a bond containing the atom
under consideration is changed in the rate-determining step, is
the most significant.33 If no carbon bond changed in the rate-
determining step, and indeed if no carbon-containing bond is
(30) Saito, T.; Tsunogai, U.; Kawamura, K.; Nakatsuka, T.; Yoshida, N. J. Geophys.
Res. 2002, 107 (D4). Doi: 10.1029/2000JD000127.
(31) Abrajano, T. A.; Murphy, D. E., Jr.; Comet, P. A.; Brooks, J. Org. Geochem.
1994, 21, 611-617.
(32) Rieley, G. Analyst 1994, 119, 915-919.
(33) Melander, L.; Saunders, W. H. Reaction Rates of Isotopic Molecules; John
Wiley: NewYork 1980.
(34) Yu Y. X. Ph.D. Thesis. The Graduate School of the Chinese Academy of
Sciences, 2005; pp 41-42.
Analytical Chemistry, Vol. 78, No. 4, February 15, 2006 1209

Table 2. Stable Carbon Isotopic Compositions of Formaldehyde and Derivatives Measured and Calculated in the
Calibration Experiment for Atmospheric Sampling
δ13C(%)a
formaldehyde
measured
measured
calculated
calculated
absolute
valuesh
suppliers
F1i
concn (µg m-3
)
formaldehydeb,c
thiazolidineb-e
formaldehydef
thiazolidineg
312.0
-45.08 ( 0.40
(n ) 5)
-32.99 ( 0.22
-44.66 ( 0.67
-33.14
0.15
202.8
101.4
31.2
-33.39 ( 0.18
-33.01 ( 0.18
-32.88 ( 0.20
-32.76 ( 0.20
-32.30 ( 0.17
-45.86 ( 0.56
-44.71 ( 0.53
-44.42 ( 0.45
-44.56 ( 0.56
-42.57 ( 0.38
0.25
0.13
0.26
0.38
0.15
7.8
F2i
312.0
-42.11 ( 0.13
(n ) 3)
-32.15
202.8
101.4
31.2
-32.32 ( 0.26
-32.40 ( 0.18
-32.39 ( 0.17
-32.30 ( 0.18
-31.44 ( 0.20
-42.65 ( 0.78
-42.88 ( 0.54
-42.85 ( 0.51
-42.59 ( 0.54
-40.01 ( 0.61
0.17
0.25
0.24
0.15
0.23
7.8
F3j
312.0
-39.30 ( 0.31
(n ) 5)
-31.21
-31.21
202.8
101.4
312.0
-31.25 ( 0.18
-31.37 ( 0.24
-31.27 ( 0.26
-39.42 ( 0.53
-39.78 ( 0.73
-39.49 ( 0.78
0.04
0.16
0.06
F3k
-39.30 ( 0.31
(n ) 5)
202.8
101.4
-31.16 ( 0.24
-31.36 ( 0.17
-39.15 ( 0.71
-39.76 ( 0.52
0.05
0.15
a Stable carbon isotopic compositions reported in per mil relative to VPDB. b Determined by GC/C/IRMS. c The arithmetic means and standard
deviations. d Derivatized with cysteamine hydrochloride of δ13C ) -27.16 ( 0.02‰, from 10 analysis determined by EA-IRMS. e Three times repeated
in GC/C/IRMS analysis. f δ13C values of calculated δ13C values based on mass balance relationship eq 5 and standard deviations (S) calculated
according to eq7. g δ13C values of calculated thiazolidine based on mass balance relationship eq 5. h Absolute values of the difference between the
δ13C values of measured and calculated thiazolidine. i The samples were eluted and derivatized immediately after sampling. j The samples were
stored in 4 °C for 30 days and then eluted and derivatized. k The samples were stored in room temperature (3-26 °C) for 30 days and then eluted
and derivatized. n, number of analysis for each sample.
involved in the reaction, then there is not likely a primary isotope
Table 3. Comparison between This and the DNPH
Method
effect on δ13C. In this work, during the derivatization procedure,
formaldehyde reacts with cysteamine as in reaction 3. It is clear
DNPH methoda
this method
that only formaldehyde contributes a carbon atom whose bond is
altered in the reaction, while the reaction positions of cysteamine
are the nitrogen and sulfur atoms. So any carbon kinetic isotope
effect is mostly related to the formaldehyde. In our experiments,
although the formaldehyde has a carbon bond altered, the
cysteamine was always used in extreme excess and the formal-
dehyde reacted quantitatively. Thus, no carbon isotope fraction-
ation should be introduced. This prediction was confirmed by the
results shown in Table 2: the differences between the measured
and predicted (calculated according to eq 5) δ13C values of
thiazolidine were in the range of 0.04-0.38‰, within instrument
specifications. These results imply that this method is promising
for measuring the δ13C of atmospheric formaldehyde.
Sb,c (δ13C, ‰)
1.74-2.19
>200
0.15-0.57
injection (ng)
>60
sampling duration (h)
20-24
2
6
2
flow rate (L min-1
)
concentration (µg m-3
)
14.07-66.97
4.52-10.16
a The data were cited from ref 13. b The standard deviations were
calculated with mass balance eq 6 or eq 7. c δ13C reported relative to
VPDB.
The detection limits for both methods are also given in Table
3. For analytical reproducibility, a minimum concentration of the
formaldehyde derivative was needed for each analysis. The
minimum amounts were 60 and 200 ng, respectively, for this and
the DNPH methods. In addition, the sampling times were 6 and
20-24 h, respectively, where the former was at lower formalde-
hyde concentrations. Therefore, the detection limit of this method
is much lower. Thus, it is obvious that this method has a lower
detection limit and less analytical error compared to the DNPH
method.
Measurements of Atmospheric Formaldehyde. Atmo-
spheric formaldehyde samples were collected as described above.
The target thiazolidine was chromatographically well resolved
(Figure 2). The δ13C values and concentrations of formaldehyde
are summarized in Table 4. The formaldehyde was quantitatively
collected by the NaHSO3-coated cartridge at the prevalent
atmospheric concentrations. The data show that the δ13C values
Comparison with the DNPH Method.13 For both methods,
the analytical error of the calculated data for underivatized
formaldehyde (usually expressed as the standard deviation, S)
should follow eqs 6 and 7, according to the propagation of random
error:
S2formaldehyde ) 49S2DNPH-derivative + 36S2
(6)
(7)
DNPH
S2formaldehyde ) 9S2thiazolidine + 4S2
cysteamine
where 49, 36, 9, and 4 are the squared numbers of carbon atoms
in the respective compounds. According to eqs 6 and 7, the
analytical errors (S) of formaldehyde δ13C values are lowest for
the present method (0.15-0.57 versus 1.74-2.19‰, Table 3).
1210 Analytical Chemistry, Vol. 78, No. 4, February 15, 2006

Figure 2. Typical GC/C/IRMS chromatograms for thiazolidine. (a) Standard thiazolidine; (b) sample collected on top of an office building in
Guangzhou in 2005.
Table 4. Concentrations and δ13C of Atmospheric
Formaldehyde Measured in Guangzhou, (8-9 March
2005)
to trap the formaldehyde, and the adduct is derivatized with
cysteamine of known isotopic composition. The δ13C of the
resultant thiazolidine derivative is measured, and the formalde-
hyde δ13C values are then calculated according to the mass
balance equation among formaldehyde, cysteamine, and thiazo-
lidine. Based on a calibration experiment, excellent reproducibility
and accuracy for δ13C are achieved without carbon isotopic
fractionation. Also, a comparison between this and the previous
DNPH method demonstrates the improvement of this method.
The δ13C values of atmospheric formaldehyde were determined
with this method. The application of this method for more ambient
δ13C data can provide additional information on the sources and
sinks of atmospheric formaldehyde.
δ13C (‰)a
measured
thiazolidineb,d
(AM ( SD)c
calculated
formaldehyde
(AM( S)e
sampling
time
formaldehyde
(µg m-3
)
7:00-13:00
13:00-19:00
19:00-1:00
1:00-7:00
10.16
9.45
4.52
5.80
-31.67 ( 0.05 (n)3) -40.69 ( 0.15
-31.30 ( 0.16 (n)3) -39.58 ( 0.48
-31.16 ( 0.16 (n)3) -39.14 ( 0.48
-32.10 ( 0.19 (n)3) -41.96 ( 0.57
a δ13C reported relative to VPDB. b Determined by GC/C/IRMS.
c The arithmetic means and standard deviations for three analysis.
d Derivatized with cysteamine hydrochloride of δ13C ) -27.16 ( 0.02‰,
from 10 analyses determined by EA-IRMS. e The arithmetic means of
calculated δ13C values based on mass balance relationship eq 5 and
standard deviations (S) calculated according to eq 7. n, replicate
measurements by GC/C/IRMS.
ACKNOWLEDGMENT
This research was financially supported by the Natural
Science Foundation of China (40590392), the 973 foundation
(2002CB410803), Foundation of CAS (GIGCX-03-030311106 and
KZCX3-SW-429031112). We sincerely thank Dr. J. Z. Liu and Dr.
W. L. Jia for their help in EA-IRMS and GC/C/IRMS analysis
and Dr. T. S. Xiang in GC/MS analysis.
of atmospheric formaldehyde were different during the daytime
and nighttime, which may due to the complex reactions of this
compound or its different sources.
CONCLUSIONS
Received for review September 9, 2005. Accepted
November 30, 2005.
A novel method for determining the compound-specific carbon
isotopic compositions of atmospheric formaldehyde was developed
and tested. It uses NaHSO3-coated Sep-Pak silica gel cartridges
AC051616F
Analytical Chemistry, Vol. 78, No. 4, February 15, 2006 1211