Anal. Chem.
2010, 82, 7897–7905
Quantification of Carbonate by Gas
Chromatography-Mass Spectrometry†
Dimitrios Tsikas* and Kristine Chobanyan-Jürgens
Institute of Clinical Pharmacology, Hannover Medical School, D-30623 Hannover, Germany
Carbon dioxide and carbonates are widely distributed in
nature, are constituents of inorganic and organic matter, and
are essential in vegetable and animal organisms. CO2 is the
principal greenhouse gas in the atmosphere. In human
blood, CO2/HCO3- is an important buffering system.
Quantification of bicarbonate and carbonate in inorganic
and organic matter and in biological fluids such as blood
or blood plasma by means of the GC-MS technology has
been impossible so far, presumably because of the lack
of suitable derivatization reactions to produce volatile and
thermally stable derivatives. Here, a novel derivatization
reaction is described for carbonate that allows for its
quantification in aqueous alkaline solutions and alkalin-
ized plasma and urine. Carbonate in acetonic solutions
of these matrices (1:4 v/v) and added 13C-labeled car-
bonate for use as the internal standard were heated in
the presence of the derivatization agent pentafluorobenzyl
(PFB) bromide for 60 min and 50 °C. Investigations with
12
CO32-, 13CO32-, (CH3)2CO, and (CD3)2CO in alkaline
solutions and GC-MS and GC-MS/MS analyses under
negative-ion chemical ionization (NICI) or electron ion-
ization (EI) conditions of toluene extracts of the reactants
revealed formation of two minor [i.e., PFB-OCOOH and
OdCO2-(PFB)2] and two major [i.e., CH3COCH2-
C(OH)(OPFB)2 and CH3COCHdC(OPFB)2] carbonate
derivatives. The latter have different retention times (7.9
and 7.5 min, respectively) but virtually identical
EI and NICI mass spectra. It is assumed that
CH3COCH2-C(OH)(OPFB)2 is formed from the reaction
of the carbonate dianion with two molecules of PFB
bromide to form the diPFB ester of carbonic acid, which
further reacts with one molecule of acetone. Subsequent
loss of water finally generates the major derivative
CH3COCHdC(OPFB)2. This derivatization reaction was
utilized to quantify total CO2/HCO3-/CO32- (tCO2) in
human plasma and urine by GC-MS in the NICI mode
by selected ion monitoring of the anions [M - H]- of
CH3COCHdC(OPFB)2 at m/z 461 for the endogenous
species and m/z 462 for the internal standard 13CO32-.
Oral intake of the carboanhydrase inhibitor drug aceta-
zolamide by two healthy volunteers resulted in temporary
†
Part of the special issue “Atmospheric Analysis as Related to Climate
Change”.
* To whom correspondence should be addressed. Address: Institute of Clinical
Pharmacology, Hannover Medical School, Carl-Neuberg-Strasse 1, D-30625
Hannover, Germany. Phone: +49 511 532 3959. Fax: +49 511 532 2750. E-mail:
tsikas.dimitros@mh-hannover.de.
10.1021/ac1007688  2010 American Chemical Society
Published on Web 06/16/2010
increased excretion of tCO2 in the urine. The method is
specific for carbonate, accurate, sensitive and should be
applicable to various matrices including human fluids
and environmental samples.
Entire oxidation of carbon and organic material produces
carbon dioxide (carbonic acid gas, carbonic anhydride). In the
temperate zones of the Earth, the atmosphere contains about 0.04%
(v/v) of CO2. CO2 is constituent of carbonate type of minerals
and products of animal metabolism. CO2 is necessary for the
respiration cycle of plants and animals. CO2 is soluble in water
(88 mL/100 mL H2O; 20 °C, 1 atm). CO2 is absorbed by alkaline
solutions with the formation of carbonates. A very minor part
of CO2 dissolved in water forms H2CO3, which deprotonates
to form HCO3- and CO32-. In many cells and tissues of animals,
notably erythrocytes, renal cortex, gastric mucosa, and muscle,
the enzyme carbonic anhydrase catalyzes the reversible hy-
dratation of CO2 to form HCO3- and H+.1,2 In human blood,
CO2/HCO3- is an important buffering system3 and is mainly
controlled by the lung and the kidney.
CO2 is the main greenhouse gas in the atmosphere. About
30% of the excess released into the atmosphere by human
activities since the industrialization has been absorbed by the
oceans. CO2 has affected and will further affect the human
environment as well as the environmental envelope of oceanic
organisms, notably of coral reef ecosystems.4 The need to
reduce CO2 accumulation in the atmosphere has initiated much
work into developing new technologies for CO2 capture and
utilization.5-8 Interestingly, because of its easy and abundant
availability, CO2 has attracted considerable attention as a
renewable and environmentally friendly source of carbon. CO2
is especially useful as a phosgene substitute and a versatile
synthon for numerous synthetic targets.7
Because of the general interest in CO2, many different
analytical methods have been developed and are being used
to measure CO2, bicarbonate, and carbonate in various biologi-
cal systems. These analytical methods use detectors that utilize
specific physicochemical properties of these species, such as
absorption of light in the infrared region and thermal conduc-
(1) Geers, C.; Gros, G. Physiol. Rev. 2000, 80, 681–715.
(2) Supuran, C. T. Curr. Pharm. Des. 2008, 14, 603–614.
(3) Jungas, R. L. Anal. Biochem. 2006, 349, 1–15.
(4) Lough, J. M. J. Environ. Monit. 2008, 10, 21–29.
(5) Aresta, M.; Dibenedetto, A. Dalton Trans. 2007, 36, 2975–2992.
(6) Yu, K. M. K.; Curcic, I.; Gabriel, J.; Tsang, S. C. E. ChemSusChem 2008,
1, 893–899.
(7) Sakakura, T.; Kohno, K. Chem. Commun. 2009, 1312–1330.
(8) Riduan, S. N.; Zhang, Y. Dalton Trans. 2010, 39, 3347–3357.
Analytical Chemistry, Vol. 82, No. 19, October 1, 2010 7897
tivity (by CO2)9 and migration of carbonate in an electric field.10
In medicine, the CO2 partial pressure (pCO2) in blood is
commonly measured potentiometrically, and HCO3- concentra-
tion in plasma is calculated using the Henderson-Hesselbach
equation. In healthy humans, pCO2 and plasma HCO3- con-
centration amount to 36-42 mmHg and 20-27 mM, respec-
tively. In clinical research and practice, gaseous CO2, in
particular the ratio of 13CO2/12CO2, is analyzed by the isotope
ratio mass spectrometry methodology.11
To the best of our knowledge, quantification of bicarbonate
or carbonate in inorganic and organic matter by means of the
GC-MS technology has not been reported so far. The reason for
this is most likely the lack of suitable derivatization reactions to
produce volatile and thermally stable derivatives of HCO3-/CO32-.
However, there are early reports that cyclic carbonate deriva-
tives prepared from vicinal alkyl diols and β,β,β-trichloroeth-
ylcarbonate12 or phosgene13 are susceptible to GC. In previous
work,14 we found that nitrate (NO3-), which is a structural
analog of CO32-, can be converted in aqueous acetone to the
nitric acid pentafluorobenzyl (PFB) ester derivative (PFB-
ONO2) by means of the versatile derivatization agent alpha-
bromo-2,3,4,5,6-pentafluorotoluene (pentafluorobenzyl bromide,
PFB-Br) and quantified in various biological fluids. We assumed
that CO32- would also form with this reagent a thermally stable,
volatile and strongly electron-capturing PFB derivative. In the
present work, a unique derivatization reaction for CO32- with
PFB-Br and acetone in aqueous solution and human plasma
and urine is described. Using commercially available 13C-labeled
CO32- and 2H-labeled acetone, the major PFB/acetone deriva-
tive of CO32- was found to undergo an unusual in-source
rearrangement/fragmentation under negative-ion chemical
ionization (NICI) conditions. The specificity of the derivatization
reaction and the fragmentation allows for the accurate quan-
tification of total CO2/HCO3-/CO32- (tCO2) in biological and
environmental samples by GC-MS using 13CO32- as internal
standard.
EXPERIMENTAL SECTION
Instrumentation. A Hewlett-Packard MS engine 5890A con-
nected directly to a gas chromatograph 5890 series II equipped
with an autosampler Hewlett-Packard model 7673 (Waldbronn,
Germany) was used for GC-MS analyses. GC-MS and GC-MS-
MS analyses were also performed on a triple-stage quadrupole
mass spectrometer ThermoQuest TSQ 7000 (Finnigan MAT, San
Jose, CA) directly interfaced with a Trace 2000 series gas
chromatograph equipped with an autosampler AS 2000 (CE
Instruments, Austin, TX). Fused-silica capillary columns Optima
17 [15 m × 0.25 mm inside diameter, 0.25 µm film thickness] from
Macherey-Nagel (Düren, Germany) were used.
Reagents. Sodium [13C]carbonate (declared isotopic purity
of 99 atom % at 13C) and 2,3,4,5,6-pentafluorobenzyl bromide
(9) Beccaria, A. M.; Poggi, G.; Castello, G. J. Chromatogr. 1987, 395, 641–
647.
(10) Wildman, B. J.; Jackson, P. E.; Jones, W. R.; Alden, P. G. J. Chromatogr.
1991, 546, 459–466.
(11) Benson, S.; Lennard, C.; Maynard, P.; Roux, C. Forensic Sci. Int. 2006,
157, 1–22.
(12) Oehlenschläger, J.; Gercken, G. J. Chromatogr. 1979, 176, 126–128.
(13) Koppenhoefer, B.; Lin, B. J. Chromatogr. 1989, 481, 17–26.
(14) Tsikas, D. Anal. Chem. 2000, 72, 4064–4072.
7898 Analytical Chemistry, Vol. 82, No. 19, October 1, 2010
(PFB-Br) were obtained from Aldrich (Steinheim, Germany).
Acetone, hexadeutero-acetone (declared isotopic purity of 99
atom % at 2H), sodium carbonate, sodium bicarbonate, and
sodium hydroxide were bought from Merck (Darmstadt,
Germany). For quantitative measurements stock solutions
(each 100 mM) of unlabeled and 13C-labeled carbonate in
distilled water were freshly prepared.
Safety Considerations. PFB-Br is corrossive and an eye
irritant. Inhalation and contact with skin and eyes should be
avoided. All work should be performed in a well-ventilated fume
hood.
Derivatization Procedure. Acetone, aqueous solutions, and
biological fluids were stored in closed vials in an ice-bath. In
quantitative analyses, derivatization with PFB-Br was performed by
mixing a carbonate containing matrix (100 µL) with acetone (400
µL) and PFB-Br (10 µL) and by incubating the mixture at 50 °C for
60 min. For quantitative measurements, biological fluids (100 µL)
were combined with a 100-µL aliquot of a solution of the internal
standard Na213CO3 in distilled water to reach final added concen-
trations of 5, 20, or 50 mM. PFB-Br derivatization was performed
by adding acetone (800 µL) and PFB-Br (20 µL) and by incubating
the mixture at 50 °C for 60 min. After derivatization samples were
cooled to room temperature, acetone was evaporated under a
nitrogen stream and reaction products were extracted from the
remaining aqueous phase by vortex-mixing with toluene (1 mL)
for 1 min without preceding pH correction. One 800-µL aliquot of
the organic phase was transferred into a clean 1.8-mL autosampler
glass vial.
GC-MS and GC-MS-MS Conditions. In both instruments,
toluene aliquots (1 µL) were injected in the splitless mode using
the following temperature program: the column temperature was
held at 70 °C for 1 min, then increased to 280 and 300 °C at a rate
of 30 °C/min and 10 °C/min, respectively. The oven temperature
was held at 300 °C for 1 min.
In GC-MS analyses on the MS engine 5890A, helium (50 kPa)
and methane (200 Pa) were used as the carrier and the reagent
gas for NICI, respectively. Electron energy and electron current
were set to 230 eV and 300 µA, respectively, for NICI. Electron
energy was 70 eV in the EI mode. Constant temperatures of 180,
280, and 200 °C were kept at the ion source, interface, and injector,
respectively. Quantification of carbonate was performed in the
NICI mode by selected ion monitoring (SIM) of the ions with
m/z 461 for carbonate and m/z 462 for the internal standard
Na213CO3 with a dwell time of 50 ms for each ion. In quantitative
analyses the electron multiplier voltage was set to values
between 1600 and 2000 V.
GC-MS and GC-MS-MS spectra were also generated on the
TSQ 7000 instrument. Interface, injector, and ion-source were kept
at 280, 280, and 180 °C, respectively. Electron energy and electron
current was set to 200 eV and 300 µA, respectively. Helium (50
kPa), methane (530 Pa), and argon (0.13 Pa collision pressure)
were used as carrier, reagent, and collision gases, respectively.
Collision energy was set to 15 eV.
Statistical Analysis. If not otherwise specified, quantitative
analyses were performed in triplicate. Values are presented as
mean ± standard deviation.
Table 1. Ions in the Mass Spectra of the Major GC Peaka of [12C]carbonate and [13C]carbonate after Derivatization
with Pentafluorobenzyl Bromide in CH3COCH3 or CD3COCD3
12
carbonate/acetone CO32-/CH3COCH3
negative-ion chemical ionization
[M-H/D]- 461 (31)
[M-H/D-H2O]- 443 (2)
[M-MeCO]- 419 (5)
[M-H/D-*CH(OH)3]- 397 (100)
[M-H/D-*CH(OH)3-H/DF]- 377 (10)
281 (5)
217 (5)
electron ionization
[M-H/D]•+ 461 (2)
[PFB]•+ 181 (32)
[MeCO]•+ 43 (100)
a
Derivative 2; MS engine 5890A. Me, CH3 or CD3; *C indicates 12C or 13C; PFB, pentafluorobenzyl.
RESULTS AND DISCUSSION
Identity of Reaction Products. GC-MS analysis of toluene
extracts from the derivatization with PFB-Br of aqueous saturated
solutions of Na212CO3 and Na213CO3 in CH3COCH3 or
CD3COCD3 yielded several GC peaks. The mass spectra of four
peaks showed corresponding ions with a difference of 1 amu
in the EI and in the NICI mode due to the mass difference of
the 13C and 12C isotopes. The retention times (tR) of the two
major GC peaks were 7.9 min (derivative 1) and 7.5 min
(derivative 2). The EI and NICI mass spectra of the larger GC
peak (derivative 2) of the PFB derivatives of Na212CO3 and
Na213CO3 produced in CH3COCH3 or CD3COCD3 are shown
in Supporting Information, Figure 1. The most intense ions
present in the EI and NICI mass spectra of derivative 2 of
Na212CO3 and Na213CO3 obtained from derivatization reactions
performed in CH3COCH3 or CD3COCD3 are summarized in
Table 1. The most intense ions in the NICI mass spectra of the
PFB derivative of Na213CO3 of derivative 1 were m/z 462
(intensity, 100%) and m/z 317 (90%) in CH3COCH3 and m/z
465 (intensity, 80%) and m/z 317 (100%) in CD3COCD3.
The largest ions in the EI and NICI mass spectra of derivatized
Na213CO3 in CH3COCH3 had m/z values of 461 and 462,
respectively (Supporting Information, Figure 1). The largest ions
in the EI and NICI mass spectra of derivatized Na213CO3 in
CD3COCD3 had m/z values of 464 and 465, respectively (Table
1). These findings suggest that derivative 2 contains one 13C atom
from Na213CO3 and three D atoms from CD3COCD3. The
intense ions at m/z 43 ([CH3CO]+•) and m/z 46 ([CD3CO]+•)
in the EI mass spectra suggest that these derivatives contain
in their molecules one acetyl group from CH3COCH3 and
CD3COCD3, respectively. The intense mass fragment at m/z
181 was present in the EI mass spectra of the carbonate
derivatives analyzed and corresponds to the PFB radical cation
([C6F5CH2]+•). To achieve molecular ions on the order of m/z
461 to m/z 465 (Table 1), the PFB derivatives of Na212CO3/
Na213CO3 must contain two PFB moieties (362 Da).
The data of Table 1 suggest that the carbonate dianion reacts
with two PFB-Br molecules in a nucleophilic substitution, most
likely through a SN2 mechanism, to produce the diPFB ester
of carbonate. Because of the considerable nucleophilicity of
acetone (pKa 20) and the alkaline conditions of the derivatiza-
tion reaction (pH about 12.5), it can be assumed that acetone,
which is present at a very high molar excess over both
13
CO32-/CH3COCH3 12
CO32-/CD3COCD3 13
CO32-/CD3COCD3
462 (35) 464 (23) 465 (17)
444 (2) 446 (1) 447 (1)
420 (7) 419 (3) 420 (4)
397 (100) 400 (100) 400 (100)
377 (10), 378 (11) 380 (9) 381 (10)
282 (5) not detected 285 (8)
217 (6) 221 (5) 221 (4)
462 (3) 464 (5) 465 (7)
181 (28) 181 (80) 181 (85)
43 (100) 46 (100) 46 (100)
carbonate and PFB-Br, reacts with the diPFB ester of
carbonate by the methylene carbanion of acetone to produce
two derivatives which have virtually identical NICI and EI
mass spectra (Figure 1). Because of the clearly different GC
behavior, derivative 1 is more polar and less volatile (tR ) 7.9
min) than derivative 2 (tR ) 7.5 min). Possible structures could
be CD3COCD2-C(OH)(OPFB)2 (MW 483) for derivative 1 and
CD3COCDdC(OPFB)2 (MW 464) for derivative 2 from
Na212CO3 in CD3COCD3. The appearance of the ion at m/z 461
in the EI and NICI mass spectra of Na212CO3 suggests that
ionization of CH3COCHdC(OPFB)2 produces the radical cation
[CH3COC)C(OPFB)2]+• (i.e., [M-H]+•) and the anion
[CH3COCdC(OPFB)2]- (i.e., [M-H]-) by loss of one H• or
H+, respectively (Table 1). That derivative 1 and derivative 2 have
virtually identical NICI and EI mass spectra suggests that
ionization of CD3COCD2-C(OH)(OPFB)2 is accompanied by loss
of one water molecule (HOD; see Figure 1).
Formation of the diPFB/acetone derivatives of carbonate raises
the possibility of formation of monoPFB/acetone derivatives, as
well as formation of PFB derivatives without involvement of
acetone. Indeed, using Na212CO3 and Na213CO3 we observed very
minor GC peaks with relative retention times of 0.788 (deriva-
tive 3) and 0.912 (derivative 4) with respect to the derivative
2. These derivatives are likely to be PFBO-12COOH (MW 242)
and PFBO-13COOH (MW 243) for derivative 3 and
(PFBO)2-12CO (MW 422) and (PFBO)2-13CO (MW 423) for
derivative 4 (Supporting Information, Figure 2).
Taken altogether, under alkaline conditions in aqueous acetone,
carbonate reacts with PFB-Br to form at least four derivatives. The
most abundant derivatives are CH3COCHdC(OPFB)2 and its
putative precursor CH3COCH2-C(OH)(OPFB)2, and they have
virtually identical MS properties under NICI and EI conditions.
Under the derivatization conditions used (i.e., basic pH, 50 °C,
60 min) acetone not only serves as a solvent but also is a reactant
and completes the reaction between carbonate and PFB-Br.
Because derivative 2, the acetale CH3COCHdC(OPFB)2, is the
major reaction product, we focused on this derivative.
In-Source Rearrangement/Fragmentation and Collision-
Induced Dissociation in NICI Mode. The most intense ions in
the NICI mass spectra of the derivatives of Na212CO3 and
Na213CO3 in unlabeled acetone had the same m/z value of 397,
suggesting that this ion does not contain the 12C/13C atom from
Analytical Chemistry, Vol. 82, No. 19, October 1, 2010 7899
Figure 1. Proposed mechanism for the reaction of Na212CO3 with PFB-Br and CD3COCD3 in aqueous alkaline solution to form two derivatives,
that is, derivative 1 and derivative 2. Derivative 2 is likely to be the acetale CD3COCDdC(OPFB)2, which is produced from derivative 1, that is,
CD3COCD2-C(OH)(OPFB)2, by loss of water (DOH). Derivatives 1 and 2 have different retention times in chromatography but produce virtually
identical mass spectra in NICI and EI conditions. Because of its hydroxyl group, CD3COCD2-C(OH)(OPFB)2 is less volatile emerges later from
the GC column than CD3COCDdC(OPFB)2.

Na212CO3/Na213CO3 (Table 1 and Supporting Information,
Figure 2A and B). Similarly, the most intense ions in the NICI
mass spectra of the derivatives of Na212CO3 and Na213CO3 formed
in CD3COCD3 had the same m/z value of 400, suggesting that
this ion does not contain the 12C/13C atom from Na212CO3/
Na213CO3, but it contains the CD3 group from CD3COCD3. The
loss of the carbonate-C atom from [CH3COCdC(OPFB)2]-
suggests that this ion rearranges within the ion-source under
NICI conditions, whereby a species with the molecular mass
of 64 Da leaves the anion. The corresponding neutral loss from
[CH3COCd13C(OPFB)2]- is 13CH(OH)3 (65 Da).
To investigate the origin of the common ion at m/z 397 in the
mass spectra of derivative 2 of Na212CO3 and Na213CO3 (Table 1
and Supporting Information, Figure 2A and B), the precursor
anions at m/z 461 and m/z 462 were subjected to collision-induced
dissociation (CID). The tandem mass spectra shown in Figure 2
indicate numerous product ions. Some product ions have same
m/z values, whereas other product ions differ each by 1 amu
because of 13C. Interestingly, no product ions at m/z 397 were
observed from CID of m/z 461 (Na212CO3) and m/z 462
(Na213CO3). This finding strongly suggests that m/z 397 is
formed in the ion-source during NICI.
In Figure 3, a mechanism for this unusual rearrangement for
[CD3COCDdC(OPFB)2]- is proposed. In accordance with this
7900 Analytical Chemistry, Vol. 82, No. 19, October 1, 2010
mechanism, a 1,3-rearrangement, fragmentation starts with the
intramolecular nucleophilic attack of the mesomeric alcoholate
anion (from CD3COCD3) on the C-atom (from carbonate). After
a sequence of several steps, this rearrangement is completed
by the neutral loss of CH(OH)3 forming the substituted
cyclobutadiene anion (Figure 3). Because of the highly conju-
gated system and the abundant strongly electron-capturing F
atoms, the anion [CD3C4(C6F5)2]- should be very stable and
strongly electron-capturing. The driving forces for this rear-
rangement and the subsequent fragmentation could be (1) the
inability of [M-H]- to directly lose CO2, (2) the possibility of
the formation of the neutral loss CH(OH)3, and (3) the
formation of a highly conjugated, stabilized anion. Indeed, the
anion [CD3C4(C6F5)2]- (m/z 400) undergoes only minor
fragmentation to m/z 381 because of loss of a fluoride anion
(Table 1). Subjection of the ions at m/z 397 ([CH3C4(C6F5)2]-)
of derivative 2 from Na212CO3 and Na213CO3 (Table 1) to CID
(15 eV collision energy) resulted in virtually identical product ion
mass spectra with three product ions at m/z 377, m/z 357, and
m/z 337 because of the loss of 1, 2, and 3 HF molecules,
respectively, with the 3 H most likely stemming from the original
methyl group of [CH3C4(C6F5)2]- (Supporting Information,
Figure 3A and B).
Figure 2. GC-MS/MS spectra produced by collision-induced dissociation of the parent anions (P-) of the derivative 2 at m/z 461 of
CH3COCHd12C(OPFB)2 from Na212CO3 (A) and m/z 462 of CH3COCHd13C(OPFB)2 from Na213CO3 (B). Common product ions are highlighted
in bold. The instrument TSQ 7000 was used. Collision energy was 15 eV each.

It is worth mentioning that unusual fragmentation has been indicate that the gas-phase anion chemistry in the laboratory may
observed for R-nitro-pentafluorotoluene, although under CID be very complex and may lead to structurally unusual anions like
conditions in GC-MS/MS.15 These and previous observations16-18 those detected in remote galactic regions.19-24 Interestingly, in-

(15) Tsikas, D.; Schwedhelm, E.; Frölich, J. C. J. Chromatogr. A 2005, 1067, (17) Ingemann, S.; Nibbering, N. M. M.; Sullivan, S. A.; DeyPuy, C. H. J. Am.
337–345. Chem. Soc. 1982, 104, 6520–6527.
(16) Dawson, J. H. J.; Nibbering, N. M. M. Int. J. Mass Spectrom. Ion Phys. 1980, (18) Büker, H. H.; Nibbering, N. M. M.; Espinosa, D.; Mongin, F.; Schlosser,
33, 3–19. M. Tetrahedron Lett. 1997, 38, 8519–8522.

Analytical Chemistry, Vol. 82, No. 19, October 1, 2010 7901

Figure 3. Proposed mechanism for the rearrangement and fragmentation of the anion of CD3COCDdC(OPFB)2 at m/z 464 under NICI conditions
leading to the anions m/z 400 and m/z 381 and to the neutral loss of the alcohol CH(OH)3 (MW 64).
source fragmentation and decay are quite common phenomena
in mass spectrometry and can be utilized, for instance, for protein
identification by MALDI mass spectrometry.25
Analysis of Carbonate in Aqueous Solution. Standard
curves of Na2CO3 (range 0-20 mM) were prepared in aqueous
solutions of Na213CO3 (5 mM, 20 mM or 50 mM). Expectedly,
plotting of the peak area ratio (PAR) of m/z 461-462 versus
the concentration of Na2CO3 did not result in a straight line
but yielded a hyperbolic curve when Na213CO3 was used at 5
mM. This observation is mainly attributed to the 13C-isotope
(natural relative abundance, 1.107%) of the anion
[CH3COCdC(OPFB)2]- (m/z 461), which originates from
Na2CO3. The ion m/z 461 considerably contributes to the anion
[CH3COCd13C(OPFB)2]- (m/z 462), which originates from the
internal standard Na213CO3. Considering eighteen 13C atoms,
three 17O atoms (natural relative abundance, 0.037%), and seven
(19) Langer, W. D.; Velusamy, T.; Kuiper, T. B.; Peng, R.; McCarthy, M. C.;
Travers, M. J.; Kovacs, A.; Gottlieb, C. A.; Thaddeus, P. Astrophys. J. 1997,
480, L63–L66.
(20) Xu, L.; Huang, Y.; Giese, R. W. J. Mass Spectrom. 1998, 33, 615–620.
(21) Blansky, J. S.; Bowie, J. H. J. Mass Spectrom. Rev. 1999, 18, 131–151.
(22) Dua, S.; Blansky, S. J.; Bowie, J. H. Rapid Commun. Mass Spectrom. 2000,
14, 118–121.
(23) Sierra, M. A.; Gomez-Gallego, M.; Mancheno, M. J.; Martinez-Alvarez, R.;
Raminez-Lopez, P.; Kayali, N.; Gonzalez, A. J. Mass Spectrom. 2003, 38,
151–156.
(24) Julian, R. R.; May, J. A.; Stoltz, B. M.; Beauchamp, J. L. J. Am. Chem. Soc.
2003, 125, 4478–4486.
(25) Debois, D.; Bertrand, V.; Quinton, L.; De Pauw-Gillet, M. C.; De Pauw, E.
Anal. Chem. 2010, 82, 4036–4045.
7902 Analytical Chemistry, Vol. 82, No. 19, October 1, 2010
2
H atoms (natural relative abundance, 0.015%), the abundance
ratio of m/z 461 to m/z 462, that is, PM+1/PM, was calculated
according to Biemann26 as 0.204. The concentration of Na2CO3
([Na2CO3]) in aqueous solutions and biological fluids was
calculated by means of the formula (F1) which considers the
contribution of the 13C-isotope.
[Na2CO3] ) ([Na213CO3] × PAR)/(1 - 0.204 × PAR) (F1)
whereas [Na213CO3] is the concentration of the internal standard
Na213CO3 in the sample.
Linear regression analysis of the calculated Na2CO3 concen-
tration (y) and the concentration (x) of Na2CO3 added to an
alkaline aqueous solution of Na213CO3 (5 mM) in the range of
0-20 mM resulted in straight lines for both GC peaks with
the regression equations y ) 2.89 + 1.06x (R ) 0.99686) for
the major GC peak and y ) 2.91 + 1.08x (R ) 0.99428) for the
minor GC peak. The slope values of both regression equations
indicate mean accuracy values of 106% and 108%, respectively.
The y-axis intercepts of 2.89 and 2.91 indicate presence of
Na2CO3 at a concentration of about 2.9 mM in the alkaline
solution used to prepare dilutions of Na2CO3.
Imprecision (RSD) in these analyses ranged between 0.2% and
5.6%. Instrumental precision was determined by 6-fold injection
of 1-µL aliquots of the toluene extract from the derivatization of a
(26) Biemann, K. Mass Spectrometry, Organic Chemical Application; McGraw-
Hill Book Company, Inc.: New York, 1962.
5-mM solution of Na213CO3. On the assumption of complete
derivatization of carbonate and quantitative extraction of the
PFB derivative, the amount injected corresponds to 500 pmol
of Na213CO3. This amount was detected with an imprecision
(RSD) of 0.7% and a signal-to-noise (S/N) ratio of 2529:1 (RSD,
6.9%) for the major GC peak. A 6-fold injection of 1-µL aliquots
of a 1:1000-dilution (v/v) of the toluene extract mentioned
above resulted in an S/N ratio of 4.2 ± 0.67: 1 (RSD, 16%) for
the major GC peak (derivative 2). Thus, the amount of 500 fmol
of Na213CO3 is considered the lower limit of detection of the
method.
The shortcomings associated with the hyperbolic curves can
be overcome by using the internal standard Na213CO3 at concen-
trations higher than the highest expectable concentration of
Na2CO3 in the sample. Thus, using Na213CO3 at an added final
concentration of 50 mM linear regression analysis between the
PAR measured (y) and added Na212CO3 (x) resulted in straight
lines for both GC peaks: y ) 0.08 + 0.018x, R ) 0.99962 for
the major GC peak (derivative 2), and y ) 0.09 + 0.017x, R )
0.99962 for the minor GC peak (derivative 1). In such cases
Na2CO3 concentrations can be calculated by using Formula F2
which refuses the contribution of the 13C-isotope of Na212CO3
to the internal standard Na213CO3. Linear regression analysis
between the calculated Na2CO3 concentration (y) and the
concentration (x) of Na2CO3 added (0, 10, 20, 30, 40 mM) to
an alkaline aqueous solution of Na213CO3 (50 mM) resulted in
straight lines for both GC peaks with the regression equations
y ) 3.28 + 1.07x (R ) 0.99919) for the major peak and y ) Figure 4. Representative partial chromatograms from the GC-MS
4.07 + 1.02x (R ) 0.99986) for the minor peak. The slope values analysis of carbonate (SIM of m/z 461 and m/z 462) in the NICI mode
of both regression equations indicate mean accuracy values of in urine of a healthy volunteer before (A) and after (B) addition of 20
mM of Na2CO3 to the urine sample. The concentration of the internal
107% and 102%, respectively.
standard Na213CO3 was 50 mM with respect to the urine sample. The
PFB derivatives eluted at 7.33 (major GC peak, derivative 2) and 7.85
min (minor GC peak, derivative 1).
[Na2CO3] ) [Na213CO3] × PAR (F2)
up to 24 h after drug administration. Urine specimens were
Analysis of Carbonate in Human Urine. For quantitative stored at 4 °C in closed vials until measurement of carbonate,
analyses in urine samples the minor GC peak was considered creatinine and pH. Figure 5 shows that administration of the
because the first peak of the internal standard was not baseline- diuretics led to comparable pH increases in both volunteers. By
separated from an unknown compound with an m/z value of 462 contrast, the pharmacokinetics of carbonate excretion was quite
(Figure 4). The method was validated using a urine sample of a different in the volunteers. Creatinine-corrected tCO2 excretion
pH value of 6.5 and a Na213CO3 final concentration of 50 mM. reached a sharp maximum value of about 40 mmol/mmol
For Na2CO3 added to urine (0, 10, 20, 30, 40 mM) in duplicate, creatinine in volunteer A, whereas in volunteer B the maximum
recovery (91.6-106%) and imprecision (0.2-18%) were within was broader ranging between 23 and 25 mmol/mmol creati-
generally acceptable ranges. Linear regression analysis be- nine. Seven hours after drug administration the tCO2 excretion
tween measured (y) and added (x) Na2CO3 concentration rate decreased only slowly in the remaining 17 h in volunteer
resulted in a straight line with the regression equation y ) 9.87 B, but decreased more quickly in volunteer A. These differ-
+ 0.959x, R ) 0.99994. For NaHCO3 added to urine (0, 10, 20, ences are likely to be the result of the different pharmaceutical
30, 40 mM) in duplicate, recovery (93.2% to 103%) and preparations of the same drug, that is, regular release (Glau-
imprecision (3.3-5.2%) were also satisfactory. Linear regression pax) versus retarded release (Diamox).
analysis between measured (y) and added (x) NaHCO3 con- Analysis of Carbonate in Human Plasma. Unlike in urine,
centration resulted in a straight line with the regression in plasma samples both GC peaks (derivative 2 and derivative 1)
equation y ) 7.74 + 0.976x, R ) 0.99909. can be considered for carbonate quantification because the
The applicability of the method for urinary tCO2 was demon- unknown compound eluting in front of the derivative 2 did not
strated in a self-experiment on the diuretics acetazolamide, a interfere in urine samples with the internal standard (Figure 6).
carbonic anhydrase drug inhibitor.3 One author (volunteer A) The method was validated using Na213CO3 at a final concentration
received orally one and a half Glaupax 250-mg tablets (6.3 mg of 50 mM. Comparison of Na2CO3 concentrations obtained from
per kg bodyweight). The other author (volunteer B) received
(27) Tsikas, D.; Wolf, A.; Mitschke, A.; Gutzki, F. M.; Will, W.; Bader, M.
orally a 500-mg Diamox retard capsule (5 mg per kg body- J. Chromatogr. B [Online early access]. DOI: 10.1016/j.jchromb.2010.04.025.
weight). Urine samples were collected immediately before and Published Online April 24, 2010.

Analytical Chemistry, Vol. 82, No. 19, October 1, 2010 7903

Figure 5. Urine pH and creatinine-corrected carbonate excretion
before and after oral intake of the diuretics acetazolamide by the
authors. Volunteer A received one and a half 250-mg tablet Glaupax
(OmniVision, Puchheim, Germany), resulting in a dose of 6.3 mg/kg
bodyweight. Volunteer B received a 500-mg Diamox retard capsule
(Goldshield Pharmaceuticals Ltd., U.K.), resulting in a dose of 5 mg/
kg bodyweight. The time point of administration was few minutes after
collection of the first urine sample and is indicated by an arrow. The
concentration of the internal standard Na213CO3 was 50 mM in all
samples. The PFB derivatives eluted at 7.3 (major) and 7.8 min (minor
peak). The minor GC peak was used for quantification. Creatinine in
10-µL aliquots of urine samples was measured by GC-MS using
trideuteromethyl-creatinine as the internal standard (10 mM).27

both GC peaks yielded statistically insignificantly different

results (P ) 0.44, paired t test). The results were combined
for statistical analysis. For Na2CO3 added to pooled plasma (0,
12, 24, 36, 48 mM) in duplicate, recovery (99.4% to 112%), and
imprecision (0.1% to 7.3%) were within generally acceptable
ranges. Linear regression analysis between measured (y) and
added (x) Na2CO3 concentration resulted in a straight line with
the regression equation y ) 22.7 + 1.15x, R ) 0.99793, Figure 6. Representative partial chromatograms from the GC-MS
suggesting a mean basal Na2CO3 (actually tCO2) concentration analysis of carbonate (SIM of m/z 461 and m/z 462) in the NICI mode
in plasma of a healthy volunteer before (A, C) and after (B, D) addition
of 22.7 mM and a mean recovery of 115% for added Na2CO3. of 20 mM of Na2CO3 or NaHCO3, respectively The concentration of
the internal standard Na213CO3 was 50 mM with respect to the plasma
sample. The PFB derivatives eluted at 7.35 and 7.87 min. The
CONCLUSIONS
concentration of endogenous carbonate, that is, of the CO2/HCO3-/
The reaction of carbonate with PFB-Br and acetone in aqueous
CO32- system in this freshly obtained plasma sample was measured
phase yields two thermally stable and strongly electron-capturing to be 28 mM.
PFB derivatives. The major PFB derivative of carbonate was
identified as the acetale CH3COCHdC(OPFB)2. Under NICI out to the complex gas phase anion chemistry. This derivati-
conditions the anion of CH3COCHdC(OPFB)2 undergoes zation reaction offers the unique opportunity to derivatize and
unusual in-source rearrangement and fragmentation pointing quantify carbonate in environmental samples and in biological
7904 Analytical Chemistry, Vol. 82, No. 19, October 1, 2010
fluids by GC-MS with high accuracy, precision and sensitivity. SUPPORTING INFORMATION AVAILABLE
This method should be useful for routine quantitative analysis Electron ionization (EI) and negative-ion chemical ionization
of the CO2/HCO3-/CO32- system in human environmental and (NICI) mass spectra, and product ion mass spectra. This material
biological samples. is available free of charge via the Internet at http://pubs.acs.org.

ACKNOWLEDGMENT
The expert laboratory assistance of A. Mitschke is gratefully Received for review March 25, 2010. Accepted June 1,
acknowledged. The authors thank F.M. Gutzki for performing GC- 2010.
MS/MS analyses. AC1007688

Analytical Chemistry, Vol. 82, No. 19, October 1, 2010 7905