Borohydride-coupled HPLC-FPD instrumentation and its use in the determination off dimethylsulffonium compounds

Article abstract of DOI:10.1021/ac970167j

Novel HPLC instrumentation has been developed which employs an in-line sodium tetrahydroborate (borohydride) reaction step to generate volatile sulfur species from a variety of sulfonium compounds. Transfer of the resulting volatile sulfur-containing products into the gas phase permits them to be monitored using sulfur-specific flame photometric detection. The system has been evaluated for the determination of a collection of dimethylsulfonium compounds, comprising (dimethylsulfonio)propionate, S-methylmethionine (SMM), (dimethylsulfonio)-2-methylpropionate, dimethylsulfocholine, (dimethylsulfonio)-acetate, (dimethylsulfonio)butanoate, and (dimethylsulfonio)pentanoate. Following their separation by either cation- or anion-exchange HPLC, these compounds react in-line with the tetrahydroborate, generating dimethyl sulfide, which is then swept into a flame photometric detector. The development of chromatographic conditions for the resolution of the seven sulfonium compounds is described. In an example application of the instrumentation, the levels of SMM in parsley and cabbage were found to be 16 and 74 mmol kg^-1, respectively, on a fresh weight basis.

Full text of DOI:10.1021/ac970167j

Anal. Chem. 1997, 69, 2882-2887
Bo ro h yd rid e -Co u p le d HP LC-FP D In s t ru m e n t a t io n
a n d It s Us e in t h e De t e rm in a t io n o f
Dim e t h yls u lfo n iu m Co m p o u n d s
A. G. How a rd* a nd D. W. Rus s e ll
Chemistry Department, University of Southampton, Southampton, Hampshire SO17 1BJ, U.K.
Novel HP LC instrumentation has been developed which
employs an in-line sodium tetrahydroborate (borohydride)
reaction step to generate volatile sulfur species from a
variety of sulfonium compounds. Transfer of the resulting
volatile sulfur-containing products into the gas phase
permits them to be monitored using sulfur-specific flame
photometric detection. The system has been evaluated
for the determination of a collection of dimethylsulfonium
compounds, comprising (dimethylsulfonio)propionate,
S-methylmethionine (SMM), (dimethylsulfonio)-2 -meth-
ylpropionate, dimethylsulfocholine, (dimethylsulfonio)-
acetate, (dimethylsulfonio)butanoate, and (dimethylsul-
fonio)pentanoate. Following their separation by either
cation- or anion-exchange HP LC, these compounds react
in-line with the tetrahydroborate, generating dimethyl
sulfide, which is then swept into a flame photometric
detector. The development of chromatographic condi-
tions for the resolution of the seven sulfonium compounds
is described. In an example application of the instru-
mentation, the levels of SMM in parsley and cabbage were
expected to break down to yield DMS. Such compounds include
S-methylmethionine (found in a wide range of flora including
barley, asparagus, cabbage, onion seedlings, tomatoes,11 and
8
9
10
12
green tea ), (S)-4-(dimethylsulfonio)-2-methoxybutanoate and (R)-
3-(dimethylsulfonio)-2-methoxypropionate,1
macroalgae), gonyauline and gonyol1
3-15
(both found in red
6-18
(found in the dinoflage-
late Gonyaulax polyhedra), and the sulfonium analogue of phos-
phatidylcholine.19 Dimethylsulfocholine (DMSChol) has been
shown to be produced by the hydrolysis of the sulfonium analogue
of phosphatidylcholine from the diatom Nitzschia alba with
cabbage phospholipase D19 (Figure 1).
While HPLC is a highly versatile analytical tool which is well
suited to the separation of ionic sulfonium DMS precursors, the
commonly employed detection systems generally exhibit poor
selectivity and sensitivity. In contrast, GC detectors have been
developed which are capable of high degrees of selectivity and
sensitivity, and much attention has therefore focused on attempts
to adapt gas chromatographic detectors for use in liquid chro-
matography. The obvious problem encountered when interfacing
a liquid chromatograph to a GC detector is the liquid mobile phase
in which the analyte is dissolved. Ideally, the detector should be
able to accept the total column effluent so as to give the best
sensitivity. To date, there have been two principal means by which
an analyte in solution can be introduced to the GC detector, either
by the use of a spraying device (nebulizer) or by evaporation of
-1
found to be 1 6 and 7 4 mmol kg , respectively, on a fresh
weight basis.
The release of dimethyl sulfide (DMS) into the atmosphere
from aquatic and terrestrial sources is now recognized as being
1
an important natural component of the global sulfur cycle which
the solvent from the column effluent. While these processes have
had some success with NPD and ECD detectors,20 there is a
leads to both increased acidification of precipitation2 and the
possible control of climate through the generation of cloud
condensation nuclei.3 It has been estimated that the oceanic
release of DMS may account for 90% of the sulfur flux from the
oceans.4 Highly specific HPLC instrumentation having been
previously developed for the measurement of the compound which
is believed to be the most commonly occurring natural precursor
of DMS, â-(dimethylsulfonio)propionate (DMSP),5 it became
necessary to extend the procedures to include a wider range of
potential DMS sources.
problem with the FPD, namely, the quenching potential of the
cointroduced solvent.21 Julin et al.22 nebulized up to 5 mL/ min
of column effluent and directed it into a cool hydrogen flame and
(
8) Dufour, J. P. J. Am. Soc. Brew. Chem. 1 9 8 6 , 44, 1-5.
(9) Challenger, F.; Haywood, B. J. Chem. Ind. 1 9 5 4 , 729.
(
(
10) McRorie, R. A.; Sutherland, G. L.; Lewis, M. S.; Barton, A. D.; Glazener, M.
R.; Shive, W. J. Am. Chem. Soc. 1 9 5 4 , 76, 115-118.
11) Hattula, T.; Granroth, B. J. Sci. Food Agric. 1 9 7 4 , 25, 1517-1521.
-7
(12) Kiribuchi, T.; Yamanshi, T. Agric. Biol. Chem. 1 9 6 3 , 27, 56-59.
(
(
13) Sciuto, S.; Piatelli, M.; Chillimi, R. Phytochemistry 1 9 8 2 , 21, 227-228.
14) Patti, A.; Morrone, R.; Chillimi, R.; Piatelli, M.; Sciuto, S. J. Nat. Prod. 1 9 9 2 ,
In recent years, a number of naturally occurring dimethylsul-
fonium compounds have been identified which might also be
5
5, 53-57.
(
15) Patti, A.; Morrone, R.; Chillimi, R.; Piatelli, M.; Sciuto, S. J. Nat. Prod. 1 9 9 3 ,
(
1) Lovelock, J. E.; Maggs, R. J.; Rasmussen, R. A. Nature 1 9 7 2 , 237, 452-
53.
2) Charlson, R. J.; Rodhe, H. Nature 1 9 8 2 , 295, 683-685.
56, 432-435.
4
(16) Nakamura, H. J. Chem. Soc. , Perkin Trans. 1 9 9 0 , 3219.
(17) Nakamura, H.; Ontoshi, M.; Sampei, O.; Akashi, Y.; Murai, A. Tetrahedron
Lett. 1 9 9 2 , 33, 2821.
(
(
3) Charlson, R. J.; Lovelock, J. E.; Andreae, M. O.; Warren, S. G. Nature 1 9 8 7 ,
3
26, 655-661.
(18) Nakamura, H.; Fujimaki, K.; Sampei, O.; Murai, A. Tetrahedron Lett. 1 9 9 3 ,
34, 8481.
(
(
(
4) Andreae, M. O. Mar. Chem. 1 9 9 0 , 30, 1-29.
5) Howard, A. G.; Russell, D. W. Anal. Chem. 1 9 9 5 , 67, 1293-1295.
6) Howard, A. G.; Russell, D. W. In Biological and Environmental Chemistry of
DMSP and Related Sulfonium Compounds; Kiene, R. P., Visscher, P. T.,
Keller, M. D., Kirst, G. O., Eds.; Plenum: New York, 1996; p 65.
7) Russell, D. W.; Howard, A. G. In ref 6, p 155.
(19) Anderson, R.; Kates, M.; Volcani, B. E. Nature 1 9 7 6 , 263, 51-53.
(20) Kientz, C. E.; Brinkman, U. A. T. Trends Anal. Chem. 1 9 9 3 , 12, 363-373.
(21) Chang, H.-C. K.; Taylor, L. T. Anal. Chem. 1 9 9 1 , 63, 486-490.
(22) Julin, B. G.; Vandenborn, H. W.; Kirkland, J. J. J. Chromatogr. 1 9 7 5 , 112,
443-453.
(
2882 Analytical Chemistry, Vol. 69, No. 15, August 1, 1997
S0003-2700(97)00167-4 CCC: $14.00 © 1997 American Chemical Society

2
5.89 g, 416.7 mmol). Subsequently, 100 mL of calcium sulfate-
dried toluene was added, and HCl gas was bubbled through the
mix. After 2 h, a brown oil was produced which, upon cooling
overnight, became a yellow-white solid. The crude product was
recrystallized from a 50/ 50 ethanol/ diethyl ether mixture to give
brilliant white crystals. Yield: 5.58 g, 30.21 mmol, 8.54%. 1H NMR
(
(
D
2
O): δ 1.4 (m, 3H), 3.0 (s, 6H), 3.15 (m, 1H), 3.55 (m, 2H). IR
cm-1, Nujol): ν 1720 (CdO stretch), 1180 (C-O stretch). CHN
microanalysis. calcd: C, 39.02; H, 7.10. Found: C, 38.29; H, 6.34.
Dimethylsulfocholine (DMSChol). Bromoethanol (Aldrich,
1
5 mL; 26.445 g, 211.6 mmol) was mixed with a small excess of
DMS (Janssen, 16 mL; 13.536 g, 217.9 mmol) and refluxed at 55
C for 2 days. Subsequently, the unreacted starting material was
°
evaporated off under reduced pressure, leaving a viscous liquid
Fig u re 1 . Chemical structures of potential dimethyl sulfide precur-
sors employed in this study.
that was frozen at -20 °C. Yield: 30.593 g, 163.51 mmol, 77.3%.
1
13
H NMR (D
2
O): δ 3.1 (s, 6H), 3.65 (t, 2H), 4.15 (t, 2H). C NMR
(
D
2
O): δ 28.25, 49.15, 58.78. IR (cm-1): ν 3300 (O-H stretch),
then monitored the emission at 383 nm, but only a relatively poor
detection limit of 6.25 mmol of sulfur could be achieved. The
quenching effect arising from cointroduced solvent can be reduced
by the use of a dual-flame photometric detector (DFPD) system.
Bernard et al.23 achieved reduced quenching with their elaborate
HPLC/ FPD interface which involved vaporizing a methanolic
column effluent at 750 °C to fuel a methanol-oxygen flame.
Subsequently hydrogen was added, generating hydrogen sulfide
in a reducing combustion step. Finally, this mixture was burned
in the presence of oxygen, and it was the emission from this
second flame that was recorded.
1420 (O-H bend), 1050 (C-O stretch).
(Dimethylsulfonio)acetate (DMSAcet). Bromoacetic acid
(Aldrich; 19.80 g, 142.5 mmol) was mixed with DMS (Janssen,
13 mL; 10.998 g, 177.0 mmol) and heated to 55 °C. The reaction
was essentially complete within 2 min, giving a white solid. The
product was recrystallized twice from ethanol, the last crystals
being assisted out of solution by the addition of a little diethyl
1
ether. Yield: 17.48 g, 86.9 mmol, 61.0%. H NMR (D O): δ 3.0
2
(s). IR (cm- , Nujol): ν 1720 (CdO stretch), 1180 (C-O stretch).
CHN microanalysis. Calcd: C, 23.89; H, 4.51. Found: C, 23.70;
H, 4.16.
1
The development of packed capillary HPLC columns has
permitted the sulfur chemiluminescence detector to be directly
coupled to an HPLC column eluted at flow rates up to 10 µL/
(
Dimethylsulfonio)butanoate (DMSBut). Bromobutyric
acid (Lancaster; 5.04 g, 30.2 mmol) was refluxed at 70 °C for 7 h
with DMS (Janssen, 3 mL; 2.538 g, 40.8 mmol). No solid formed,
but evaporation under reduced pressure gave an oily brown liquid
which was shown to be DMSBut. Yield: 5.231 g, 22.8 mmol,
2
3
min.
Detection limits of ∼3 pg of S/ s were quoted using
methanol/ water systems.
In the above examples, the aim was to produce a general
purpose detector that would detect a range of sulfur compounds
without preference, with sensitivity being of secondary importance.
For this work, however, both low detection limits and good
selectivity toward potential DMS precursors were required. The
HPLC/ FPD apparatus reported here satisfies these criteria by
employing a chemical derivatization step to transfer the sulfur from
the liquid phase to a gas phase, thereby circumventing the
problems encountered with the direct introduction of liquids into
a flame photometric detector.
5.6%. 1H NMR (D
.4 (2H, t). IR (cm-1): ν 1730 (CdO stretch), 1200 (C-O stretch).
CHN microanalysis. Calcd: C, 31.45; H, 5.72). Found: C, 30.27;
H, 5.19.
7
3
2
O): δ 2.15 (2H, t), 2.6 (2H, m), 2.9 (6H, s),
(
Dimethylsulfonio)pentanoate (DMSP ent). Bromovaleric
acid (Aldrich; 5.01 g, 27.7 mmol) was refluxed for 5 h at 55 °C
with DMS (Janssen, 3 mL; 2.538 g, 40.8 mmol). Upon cooling, a
white solid formed which was recrystallized from ethanol. Yield:
.682 g, 6.916 mmol, 24.9%. 1H NMR (D
t, 2H); 2.9 (s, 6H), 3.4 (t, 2H). IR (cm-1, Nujol): ν 1720 (CdO
stretch), 1200 (C-O stretch). CHN microanalysis. Calcd: C,
1
2
O): δ 1.82 (m, 4H), 2.5
(
EXPERIMENTAL SECTION
3
5.19; H, 6.50. Found: C, 34.81; H, 5.64.
(
i) P reparation of Compounds. Dimethylsulfoniopropi-
Gonyol and Gonyauline. Samples of synthetic gonyauline
onate (DMSP ) was prepared by the addition of DMS to acrylic
acid using the method described by Larher et al.24
and natural gonyol were kindly provided by Prof. Nakamura of
the University of Hokkaido, Sapporo, Japan.
S-Methylmethionine (SMM) was obtained by the methylation
of methionine with iodomethane in a formic acid/ acetic acid
Chromatographic Eluents. Eluents covering the pH ranges
3
.13-3.66 and 5.31-7.04 were obtained by adjusting the pH of
KH
PO (0.05 mol dm ) with hydrochloric acid or sodium
2 4
mixture according to the method described by Toennies and
Kolb.25
-
3
hydroxide solution, respectively. pH 4.51 and 4.77 were obtained
(
Dimethylsulfonio)-2 -methylpropionate (DMS-2 -MP ) was
6 4 2
H (COO)
) (0.05 mol dm-3) with
by adjusting the pH of KH(C
sodium hydroxide solution.
prepared using a modification of the method employed for the
preparation of DMSP. Methacrylic acid (Lancaster, 30 mL; 30.45
g, 353.7 mmol) was stirred for 3 days with 30 mL of DMS (Janssen;
Samples. Cabbage and parsley were purchased from a local
grocers. Pine needles (Abies grandis) were collected from
Ashurst, West Sussex, U.K.
(
23) Bernard, J.; Nicodemo, T.; Barthakur, N. N.; Blais, J. S. Analyst 1 9 9 4 , 119,
475-1481.
1
(
ii) Instrumentation. The instrumentation employed in this
(
(
24) Larher, F.; Hamelin, J.; Stewart, G. R. Phytochemistry 1 9 7 7 , 16, 2019-2020.
25) Teonnies, G.; Kolb, J. J. J. Am. Chem. Soc. 1 9 4 5 , 67, 849-851.
work consisted of three distinct modules (Figure 2).
Analytical Chemistry, Vol. 69, No. 15, August 1, 1997 2883

Fig u re 3 . Isoresponse profile showing the variation of instrument
sensitivity to â-(dimethylsulfonio)propionate with borohydride con-
centration and acid strength (% hydrochloric acid (v/v)). [Points denote
experimental conditions.]
Fig u re 2 . Schematic diagram of the HPLC-borohydride-FPD instru-
ment.
The first of these, the HPLC system, consisted of a duPont
high-pressure pump delivering eluent through a Rheodyne 7125
loop injection valve to an ion-exchange column at a typical flow
rate of 1 mL min-1; 200 µL injection volumes were employed.
Chromatographic separations were carried out using 25 cm × 4.6
mm id columns containing anion- (5 µm particle size Techsphere
SAX) and cation-exchange (5 µm particle size Techsphere SCX)
packings. The columns were supplied by HPLC Technology Co.
Ltd., Macclesfield, U.K.
In the derivatization step, the column eluent was mixed with
sodium borohydride solution (0.75 mL/ min, typically ∼2 M) and
hydrochloric acid (0.75 mL/ min, typically ∼0.1 M). The reaction
mixture was passed through a PTFE coil (2.10 m, 0.74 mm i.d.)
held in a duPont column oven at 85 °C. A gas flow of nitrogen
Fig u re 4 . Isoresponse profile showing the variation of instrument
sensitivity to S-methylmethionine with borohydride concentration and
acid strength (% hydrochloric acid (v/v)). [Points denote experimental
conditions.]
(
typically 40 mL/ min) then carried the mixture on to a simple
gas-liquid separator consisting of a glass T tube leading the liquid
products to waste.
(
(
i) ∼10 g of material was blotted dry and weighed;
ii) the sample was homogenized to a fine powder by grinding
The gas stream was dried by passing through two, 2 mL drying
chambers. The first was empty and was employed to trap residual
spray; the second was filled with anhydrous magnesium perchlo-
rate (granular ACS reagent grade, Aldrich). In view of the
inevitable band-broadening resulting from the introduction of these
chambers into the gas stream, attempts were made to replace this
drying system with a Nafion dryer held in molecular sieve. This
approach, however, failed due to the inability of the Nafion system
to remove the volume of water from the gas stream, resulting in
excessive signal noise and occasional extinction of the flame due
to water droplets. In this apparatus, a small amount of apparent
chromatographic efficiency had to be sacrificed to minimize the
problems associated with water transport to the detector.
The dried gas stream then mixed with hydrogen and com-
busted in the flame photometric detector. This detector, which
was constructed in-house, was based on an EMI 6256B photo-
multiplier tube viewing an air/ hydrogen flame through a BG12
in liquid nitrogen;
iii) without allowing the sample to thaw, the sample was
extracted into 10 mL of 2% (v/ v) HCl;
(
(
(
(
iv) the mixture was centrifuged for 10 min (2800g);
v) the supernatant liquid was filtered (Whatman GF/ C); and
vi) immediately prior to the chromatographic analysis, the
aqueous extract was buffered to the column eluent pH and diluted
to give a response in the calibration range.
RESULTS AND DISCUSSION
(i) The Borohydride Link Step. Reaction Conditions.
Reactant concentrations were optimized using DMSP and SMM,
as these are currently believed to be of greatest environmental
significance and because they differ significantly in their suscep-
tibility to base hydrolysis. The responses of the system to DMSP
and SMM solutions (1.2 × 10 and 2.0 × 10 mol dm-3
-
5
-5
,
-
(
Oriel) glass filter having a transmission maximum at 400 nm.
respectively) were recorded under a range of BH
concentrations (Figures 3 and 4).
4
and HCl
Hydrogen and air flow rates were typically 140 and 150 mL/ min,
respectively, in the final configuration, but due to sensitivity
changes resulting from hydrogen generated by the borohydride
reaction, these flow rates were optimized to obtain maximum
sensitivity for each set of new acid/ borohydride conditions. The
photomultiplier current signal was converted to a voltage and then
amplified. The resulting output was monitored using both a chart
recorder (Bryans 28000) and a chromatographic computing
integrator (Axxiom 717).
A distinct difference was observed in the behavior of the two
compounds. At a borohydride concentration of 1%, for example,
more highly acidic conditions were required to obtain a maximum
signal from SMM than were necessary for DMSP. This is believed
to reflect the relative ease by which DMSP undergoes simple base
hydrolysis to dimethyl sulfide and acrylic acid under the alkaline
conditions generated by the sodium borohydride. Both plots point
to optimum responses being obtained at the highest borohydride
concentrations. Working with such high concentrations (∼3.0 mol
(
iii) Sample Extraction P rocedure. Four subsamples were
-
3
taken from each sample, and each subsample was extracted using
the following procedure:
dm ), however, was impracticable due to the excessive signal
noise which resulted from the large amount of hydrogen that was
2884 Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

Ta b le 1 . Re la t ive Re s p o n s e Fa c t o rs fo r Bo ro h yd rid e a n d Ac id Flo w In je c t io n S ys t e m
3
-
2.0 mol dm-3 BH4 , 1.0% HCl
-
1.0 mol dm-3 BH4 , 2.0% HCl
-
0.5 mol dm-3 BH4 , 1.0% HCl
-
compd
DMSP
3.0 mol dm- BH4 , 1.0% HCl
1.0
1.0
1.0
1.0
DMS-2-MP
DMSBut
SMM
DMSAcet
DMSPent
DMSChol
0.91 ( 0.10
0.37 ( 0.01
0.24 ( 0.03
1.19 ( 0.10
0.68 ( 0.05
0.45 ( 0.03
0.84 ( 0.06
0.35 ( 0.02
0.13 ( 0.01
1.18 ( 0.07
0.64 ( 0.06
0.44 ( 0.04
0.90 ( 0.06
0.34 ( 0.03
0.56 ( 0.06
4.41 ( 0.87
0.29 ( 0.05
0.78 ( 0.08
0.67 ( 0.08
0.21 ( 0.03
0.25 ( 0.03
1.90 ( 0.18
0.31 ( 0.05
0.57 ( 0.07
2
produced. The conditions selected to give reasonable response
k k ) R/ [S ] ) k
1
2
liq
S
from both DMSP and SMM were a borohydride concentration of
2
(
.0 mol dm- and a hydrochloric acid concentration of 1.0% (v/ v)
3
0.115 mol dm-3).
S
The constant k reflects the sensitivity of the apparatus to the
dimethylsulfonium compound under test; a high value implies a
high yield of DMS and a high response. If DMSP is taken as the
reference compound, then the sensitivity of a test compound
Nature of the Sulfur-Volatile. The nature of the sulfur
volatile generated in the system was investigated by gas chro-
matographic analysis using flame photometric detection. A PTFE
1
GC column (350 cm × /
8
in.) containing Chromosorb 101 was
(kS(test)) can be related to the sensitivity to DMSP, kS(DMSP):
employed at 150 °C with a carrier gas flow rate of 30 mL/ min. All
seven dimethylsulfonium compounds generated a single recorded
sulfur compound, having a retention time identical to that of
dimethyl sulfide.
Ktest ) kS(test)/ kS(DMSP)
where Ktest is the relative response factor for the test compound.
All studied sulfonium compounds gave relative response factors
that were within an order of magnitude of that of DMSP (Table
P erformance. Calibration. The characteristic nonlinear
response of the FPD dictates that a log/ log plot must be used to
give a linear calibration curve; the gradient of this plot is usually
below the theoretical value of 2 (typical values ∼1.8). During
routine operation, the detection limit (∼3 σ) of the system is ∼2.0
1
).
Some relative response factors varied as the borohydride and
-
6
-3
acid concentrations were altered. It was apparent that, under
conditions where the ratio of acid to borohydride was higher, the
response from SMM was favored, as were the responses from
DMSChol and DMSAcet.
×
10 mol of DMSP dm . With a 200 mL injection loop, this
corresponds to an absolute mass of 12.8 ng of sulfur.
Relative Response Factors. The sensitivity of the detection
system to the various sulfonium compounds was determined by
feeding solutions of each compound into the detection system at
a point corresponding to the end of the HPLC column. In this
way, a steady state response could be achieved without being
subject to the effects of peak width which would have occurred if
sensitivity had been assessed by injection onto the chromatog-
raphy column.
In assessing the sensitivity of the system to the various
compounds, it was assumed that the detector response (R) to a
compound was proportional to the square of the gas phase sulfur
concentration, [Sgas], and that the gas phase sulfur concentration
was linearly related to the concentration of sulfonium compound
in solution, [Sliq], prior to the borohydride reaction step. Sensitivi-
ties were then normalized to that of DMSP.
Gonyol (1.12 × 10 mol dm-3) and gonyauline (2.57 × 10
-
5
-5
-
3
mol dm ) produced measurable detector responses, but, unfor-
tunately, insufficient material was available for the investigation
of response factors.
(iii) Chromatography of Dimethylsulfonium Compounds.
Cation-Exchange HP LC. Cation-exchange chromatography was
carried out using a Technosphere SCX 5 mm cation-exchange
column (250 mm × 4.6 mm i.d.) eluted with aqueous eluents at
1
mL/ min. The retention times of the six compounds were
recorded over a range of eluent pH values (Figure 5).
Six of the seven available compounds were zwitterionic.
DMSChol cannot be recovered from the column under the elution
conditions investigated here. Interpretation of the observed
chromatographic behavior would be assisted by a knowledge of
For a compound under test, therefore,
2
a
the pK values of the sulfonium compounds. The authors are,
R ∝ [Sgas
]
however, not aware of these values being currently available in
the literature.
2
or
R ) k [Sgas]
1
A pH of 5.66 was chosen to provide a satisfactory separation
of the compounds (Figure 6). A secondary effect of altering the
pH of the column eluent was that, with low pH values, the
sensitivity of the detection system toward DMSP and DMS-2-MP
was reduced. This is believed to be due the acidity of the eluent
reducing the initial solution conditions at the point of entry of the
sodium borohydride solution, reducing the base hydrolysis of
these two compounds. If it had been necessary to do so, the
problem could have been readily overcome by adjustment of the
conditions employed in the postcolumn reaction step.
If the concentration of sulfur in the gas phase, [Sgas], is linearly
related to the concentration of the sulfonium compound in
solution, [Sliq], i.e., [Sgas] ) k
2
[Sliq], then
2
R ) k k [S ]
1
2
liq
giving
Analytical Chemistry, Vol. 69, No. 15, August 1, 1997 2885

Fig u re 5 . Effect of phosphate eluent pH on the apparent retention
times of sulfonium compounds separated by cation-exchange chro-
matography. [Apparent retention times due to postcolumn holdup in
the detection system.]
Fig u re 7 . Effect of phosphate eluent pH on the apparent retention
times of sulfonium compounds separated by anion-exchange chro-
matography. [Apparent retention times due to postcolumn holdup in
the detection system.]
Fig u re 6 . HPLC/FPD chromatogram of sulfonium compounds
separated by cation-exchange chromatography. Eluent pH ) 5.66.
Fig u re 8 . HPLC/FPD chromatograms of sulfonium compounds
separated by anion-exchange chromatography, showing the effect
of changing eluent concentration (pH ) 7.22). (a) ) DMSAcet, (b)
DMSP, (c) DMS-2-MP, (d) DMSBut, (e) DMSPent, (f) DMSChol, (g)
SMM.
Anion-Exchange HP LC. The pH dependency of the reten-
tion times of all seven prepared DMS precursors was assessed
using a Technosphere SAX column (5 mm particle size, 250 mm
(
iv) SMM in Terrestrial P lants. While the presence of
×
4.6 mm i.d.) eluted with a range of aqueous pH-adjusted,
DMSP in the oceanic environment is well established, the origins
of DMS which has been detected in a number of terrestrial
phosphate-based eluents (Figure 7). Two distinct trends were
observed in the retention times of the separated compounds, with
the retention times of SMM and DMSChol being more severely
influenced by pH changes above pH 6.5 than the other com-
pounds.
ecosystems, including forests, grasslands, and freshwater
wetlands,1
,25-31
has yet to be identified. One possible source of
this terrestrial release of DMS is the breakdown of SMM, a
compound which has been shown to be present in a number of
plants, including asparagus,8,9,32 tomatoes,33 and tea.11 The HPLC/
FPD apparatus not only provides the means of surveying a range
of samples for potential DMS precursors such as SMM but also
can be employed in the quantitative evaluation of these sources.
Cabbage, fresh parsley, and pine needles (Abies grandis) were
extracted into dilute hydrochloric acid, filtered, centrifuged,
In general, the anion-exchange separation of the compounds
was inferior to the cation-exchange separations, with many of the
compounds being grouped together and unable to be separated
by altering the eluent pH. At the selected pH of 7.22 (Figure 8),
as at all pH values, SMM and DMSChol were unresolved. The
concentration of the eluent (pH 7.22) was reduced in order to
investigate whether the resolution could be improved (Figure 8).
DMSChol and SMM were only resolved over a narrow range
of eluent concentrations, the best separation being achieved with
(
(
(
26) Otte, M. L.; Morris, J. T. Aquat. Bot. 1 9 9 4 , 48, 239-259.
27) Berresheim, H. Atmos. Environ. 1 9 9 3 , 27A, 211-221.
28) Goldan, P. D.; Kuster, W. C.; Albritton, D. L.; Fehesenfeld, F. C. J. Atmos.
Chem. 1 9 8 7 , 5, 439-467.
an eluent concentration of 0.0025 mol dm-3
.
At these lower
(29) Cooper, W. J.; Cooper, D. J.; Saltzman, E. S.; de Mello, W. Z.; Savoie, D. L.;
Zika, R. G.; Prospero, J. M. Atmos. Environ. 1 9 8 7 , 21, 1491-1495.
concentrations, however, the resolution of DMSP and DMS-2-MP
worsens. As was found when investigating the pH of the eluent,
the behavior of SMM and DMSChol seems to differ from those
of the other five compounds, in that their retention times were
much more susceptible to changes in the buffer strength.
(
30) Lamb, B.; Westberg, H.; Allwine, G.; Bamesberger, L.; Guenther, A. J. Atmos.
Chem. 1 9 9 2 , 5, 469-491.
(
(
(
31) Hines, M. E.; Morrison, M. C. J. Geophys. Res. 1 9 9 2 , 97D, 16703-16707.
32) Berresheim, H.; Vulcan, V. D. Atmos. Environ. 1 9 9 2 , 26A, 2031-2036.
33) Challenger, F. Aspects of the Organic Chemistry of Sulphur; Butterworths
Scientific: London, 1959.
2886 Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

-
1
a
nature and the presence of these compounds and, as importantly,
the quantities involved. The instrument described in this paper
goes some way to providing a reliable, selective, and sensitive
instrumental means of both identifying and quantifying such
compounds. The selectivity inherent in the postcolumn reaction
allows naturally occurring sulfonium compounds to be measured
in relatively impure sample extracts, obviating the need for lengthy
sample “cleanup” protocols.
T
a
b
l
e
2
.
S
M
M
C
o
n
t
e
n
t
(
m
m
o
l
k
g
)
o
f
T
e
r
r
e
s
t
r
i
a
l
F
l
o
r
a
SMM content
SMM content
-
1
-1
(
mmol kg FW)
(mmol kg DW)
cabbage
parsley
74.32 ( 7.00
16.11 ( 0.78
701.16 ( 66.05
130.27 ( 6.32
a
Data are represented as mean ( standard deviation, n ) 4. FW,
fresh weight; DW, dry weight.
The chemical derivatization concept developed in this paper
could be extended to a wide range of additional compound groups
and reagent systems. Further enhancement of the instrument
sensitivity could potentially be obtained by replacing the conven-
tional flame photometric detector by one of the now commercially
available sulfur chemiluminescence or pulsed flame photometric
detectors.
buffered to the column eluent pH, and then analyzed using the
HPLC-borohydride-FPD instrumentation. No measurable DMS
precursor could be found in the pine needle extracts, but SMM
was found to be present in both the cabbage and parsley (Table
2
).
The HPLC/ FPD instrumentation described in this paper is not
restricted to applications in the environmental field. SMM, for
CONCLUSIONS
The measurement of sulfonium compounds in the environment
example, is important in a wide range of biochemical processes.
DMS is an important flavor component of beer,34 the source of
which has been identified as the SMM35-37 which is present in
provides important information on the contribution made by these
compounds to the global sulfur cycle. The analysis of these
compounds has, in the past, been particularly difficult, which has
resulted in little information being available on both the chemical
green malt.38
(
(
(
34) Wong, F. F.; Carson, J. F. J. Agric. Food Chem. 1 9 6 6 , 14, 247-249.
35) Wainwright, T. Brew. Dig. 1 9 7 2 , 47, 78-84.
36) Hysert, D. W.; Morrison, N. M.; Weaver, R. L. J. Am. Soc. Brew. Chem.
Received for review February 10, 1997. Accepted May 1,
X
1997.
1
9 7 9 , 37, 169-174.
AC970167J
(
(
37) Dickenson, C. J. J. Inst. Brew. 1 9 8 3 , 89, 41-46.
38) It should be noted that, in the brewing industry, DMSP means DMS
X
precursors and not (dimethylsulfonio)propionate.
Abstract published in Advance ACS Abstracts, July 1, 1997.
Analytical Chemistry, Vol. 69, No. 15, August 1, 1997 2887