Simple Field Device for Measurement of Dimethyl Sulfide and

Apr 3, 2013 - methanesulfonate, DMS is a major source of cloud con- densation nuclei in the oceanic atmosphere. Consequently, many studies have ...
3 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Simple Field Device for Measurement of Dimethyl Sulfide and Dimethylsulfoniopropionate in Natural Waters, Based on Vapor Generation and Chemiluminescence Detection Takanori Nagahata, Hidetaka Kajiwara, Shin-Ichi Ohira, and Kei Toda* Department of Chemistry, Kumamoto University, 2-39-1 Kurokami, Kumamoto 860-8555, Japan S Supporting Information *

ABSTRACT: A small, simple device was developed for trace analysis of dimethyl sulfide (DMS) and dimethylsulfoniopropionate (DMSP) in natural waters. These compounds are known to be the major sources of cloud condensation nuclei in the oceanic atmosphere and ideally should be measured onsite because of their volatility and instability. First, chemical and physical vapor generations were examined, and simple pressurizing by injection of 30 mL of air using a syringe was adopted. Pressurized headspace air above a 10 mL water sample was introduced to a detection cell as a result of the pressure differential and mixed with ozone to induce chemiluminescence. Although the measurement procedure was simple, the method was very sensitive: sharp peaks appeared within seconds for nanomolar levels of DMS, and the limit of detection was 0.02 nmol L−1 (1 ng L−1). Although interference from methanethiol was significant, this was successfully addressed by adding a small amount of Cd2+ before DMS vapor generation. DMSP was also measured after hydrolysis to DMS, as previously reported. Pond water and seawater samples were analyzed, and DMS was found in both types of sample, whereas DMSP was observed only in seawater. The DMS/DMSP data obtained using the developed method were compared with data obtained by purge/trap and gas chromatography−mass spectrometry, and the data from the two methods agreed, with good correlation (R2 = 0.9956). The developed device is inexpensive, light (5 kg), simple to use, can be applied in the field, and is sensitive enough for fresh- and seawater analysis.

O

DMSP in limnological water are also of interest, but have not been investigated in detail. Onsite analysis is ideal for DMS and DMSP because DMS is volatile and DMSP is unstable and is decomposed by microbes.4 However, it is currently challenging to analyze trace (nanomolar) levels of these compounds in the field. DMS is commonly analyzed using purge and cryotrapping gas chromatography with flame photometric detection (PT−GC− FPD).5−7 DMSP can be measured in the same way after hydrolysis to DMS and acrylate, using by NaOH.8−10 However,

ceanic phytoplankton produce dimethylsulfoniopropionate (DMSP) internally to protect themselves against osmotic pressure from salty water.1 DMSP decomposes naturally to produce dimethyl sulfide (DMS), which is the dominant reduced sulfur compound emitted to the atmosphere from the oceans. 2 Through oxidation to sulfate and methanesulfonate, DMS is a major source of cloud condensation nuclei in the oceanic atmosphere. Consequently, many studies have investigated these compounds in seawater: the global surface DMS database (>50 000 data) is the second largest trace gas database after that for CO2.3 The reported typical concentrations of DMS and DMSP in surface seawater are 0.5−20 and 1−100 nmol L−1 (nM), respectively.4 DMS and © 2013 American Chemical Society

Received: December 31, 2012 Accepted: April 3, 2013 Published: April 3, 2013 4461

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467

Analytical Chemistry

Article

and a vapor transfer line were inserted through the stopper, as shown in Figure 1. Plastic stopcocks (VP455980, Aram, Osaka,

a cryotrapping GC system is large and can only be used for offsite analysis or onboard a research ship. In addition, the analysis is slow (purge time = 15−20 min, plus chromatographic separation), which makes the method unsuitable for onsite and real-time analysis. Recently, a microporous membrane contactor was developed for frequent, onboard analysis of seawater samples by GC-FPD,11 and an atmospheric pressure chemical ionization−mass spectrometer has been equipped with a similar membrane-based device for real-time analysis.12 Kameyama et al. reported a system with a gas/liquidphase equilibrator, which included an air bubbler and proton transfer reaction mass spectrometer.13 Both of the above mass spectrometry (MS) systems are attractive for continuous onboard analysis, but the instruments are heavy (>100 kg) and would be difficult to use in the field. Our group is interested in the roles of DMS and DMSP in limnological water, and a simple device is required for surveys in rural areas. Solid-phase microextraction could be used for aqueous DMS analysis to collect and store DMS in liquid nitrogen.14 Recently, direct analysis of DMSP without conversion to DMS was examined using a high-performance liquid chromatograph equipped with a time-of-flight mass spectrometer. However, even in offsite analysis with this advanced instrumentation, the limits of detection (LOD), 815 and 20 nM,16 were not sufficient for analysis of natural water samples. With 1-pyrenyldiazomethane derivatization, an LOD of 2 nM was obtained for a 100 mL sample.17 To date, a simple instrument with sufficient sensitivity for measuring trace levels of dissolved DMS in the field has not been developed. One strategy for analysis of nanomolar levels of DMS is simple introduction of DMS vapor to a reaction cell without bubbling, followed by detection by ozone-induced chemiluminescence (CL). Gaseous sulfur compounds such as DMS react with ozone to produce CL, and a portable gas analyzer has been developed using CL detection for automated near-real-time gas monitoring.18,19 Very recently, Steinke’s group reported continuous laboratory analysis of DMS production by an algae culture by bubbling of the algae incubation reactor and use of a commercial gas-phase CL meter.20 In the present study, the simple and reproducible introduction of DMS vapor from a water sample was investigated, and trace amounts of aqueous DMS were transferred to a CL cell by “physical shot introduction” (PSI) to detect vapor CL. DMSP was also measured after hydrolysis, as in previous reports.8−10 Even with this simple method, nanomolar levels of DMS and DMSP were detected without serious interference.

Figure 1. The VG−CL system: mAP, miniature airpump; SG, silica gel drying column; FM, airflow meter equipped with a needle valve; GB, frit glass ball gas outlet; SC, stopcock; CV, check valve; PT, plastic tee; PMT, photomultiplier tube; AC, miniature active carbon column; and CL, chemiluminescence.

Japan) were attached to both the air introduction and vapor transfer lines. The generated DMS vapor was introduced to the CL cell via a check valve (CV3030VP, AsOne) and a plastic tee (for 3 mm tubing, VFT306, AsOne). The CL cell was a 100 mL round-bottomed flask with a glass cap connected by a conically tapered ground glass joint provided with gas inlet and outlet lines (3 mm i.d. × 5 mm o.d. glass tubes). The outside of the glass cell and the glass cap, except for the optical window (ϕ 22 mm), were plated with chromium by magnetic radio frequency sputtering to enhance reflectance of CL and covered with black paint to protect the baseline from stray light. The end of the inlet line was 20 mm above the bottom of the flask. A polytetrafluoroethylene tube (0.33 mm i.d. × 1/16 in. o.d.) was inserted into the glass inlet tubing to introduce the DMS vapor. A Teflon insert was fixed in the inlet of the tee with Teflon tubes (AWG 16, 13, 11), and a silicone tube (3 mm i.d. × 6 mm o.d.) was placed concentrically to seal the O3 line. Dry air at 300 mL min−1, produced by a miniature air-pump (CM-1512, Enomoto Micro Pump, Tokyo, Japan), a silica-gel column, and a flow meter equipped with a needle valve (RK200 V, KOFLOC, Kyoto, Japan), was introduced into an O3 generator (1000BT-12, Enaly, Shanghai, China). The produced O3 was introduced through the gap between the glass tube and the Teflon tube, and DMS vapor was mixed with O3 at the end of the tubing in front of a photomultiplier tube (PMT; R3550A ASSY, Hamamatsu Photonics, Hamamatsu, Japan). A high voltage (0.54 kV) was applied to the PMT via a small dc-to-dc converter (Opton-1NC-12, Matsusada Precision, Kusatsu, Japan), and the PMT photocurrent was converted to a voltage using a laboratory-made amplifier circuit comprising an operational amplifier chip (TL082CP, Texas Instruments) and a feedback resistor (10 MΩ). The signal was zero-adjusted,

■

EXPERIMENTAL SECTION Reagents. Standard solutions of DMS were prepared by diluting commercial pure DMS (Wako Pure Chemical Industries, Osaka, Japan) with Milli-Q water. The solutions were prepared and stored in 50 mL syringes (SS-50ESZ, Terumo, Tokyo, Japan) to prevent formation of headspace air and to prevent loss of DMS. A 50 mM cadmium solution was prepared by dissolving cadmium chloride 2.5 hydrate (Wako Pure Chemical Industries) in Milli-Q water. NaOH was obtained from Nacalai Tesque (Kyoto, Japan). Vapor Generation−CL System. The system comprised a vapor generator (VG), a CL cell, and an ozone generation system. The VG included a 50 mL plastic sample tube plugged with a silicone light stopper (No. 10, AsOne, Tokyo, Japan), and an air introduction tube equipped with a frit glass ball (G1) 4462

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467

Analytical Chemistry

Article

a glass column (4 mm i.d. × 89 mm) packed with 400 mg of Davison silica gel (grade 12, 60/80 mesh, Supelco, Bellefonte, PA) at a flow rate of 200 mL min−1 for 7 min. This silica gel is often used to collect reduced sulfur compounds without a cryo system.21 The silica gel tube was then placed in a thermal desorption system (TurboMatrix 650, PerkinElmer, Waltham, MA) and heated at 150 °C for 3 min, with a helium flow of 50 mL min−1, after purging for 1 min. To focus the analytes, the desorbed DMS was trapped in a small column packed with Tenax TA at −20 °C. The trap was then heated slowly (20 °C s−1) to 150 °C, with a helium flow. In split mode, some of the species produced on heating were transferred to the gas chromatograph at 1.5 mL min−1 (transfer line temperature = 160 °C) and the remainder diverted to waste at 10 mL min−1. The GC−MS system used was a Trace GC Ultra−DSQ (Thermo Fisher Scientific, Waltham, MA) equipped with an Inert Cap 1 capillary column (0.32 mm i.d. × 60 m, GL Science, Tokyo, Japan). The GC oven temperature was initially set at 35 °C for 4 min, increased to 50 °C at 5 °C min−1 and to 200 °C at 40 °C min−1, and then held at 200 °C for 4 min. The signal at m/z 62 was monitored for quantification of DMS, and this peak appeared at a retention time (tR) of 4.6 min. CH3SH and dimethyl disulfide (DMDS) were monitored at m/z 48 (tR = 3.8 min) and 94 (tR = 8.6 min), respectively.

amplified 100 times, and monitored every 100 ms with a notebook computer using an analog-to-digital converter (USB1408FS, Measurement Computing, Norton, MA). The CL cell and PMT were housed in a poly(vinyl chloride) (PVC) pipe (65A, 71 mm i.d. × 76 mm o.d.); both ends were covered with PVC end-caps (65A). The whole system, except for the computer, was operated on 12 V dc generated from 85−260 V ac by an ac adaptor (ADP-45XH, Toshiba, Tokyo, Japan; sold for computer operation) or from a 12 V car battery. The current consumption of the device was 0.13 A (1.5 W) for signal treatment and air-pump operation and 1.64 A (19.7 W) for the ozonator. The waste ozone gas was passed through an activated carbon column. The main part of the VG−CL system weighed 1.9 kg and the ozonator weighed 3.1 kg (total weight = 5.0 kg). Measurement Procedure. Trace level DMS was measured using the VG−CL sysatem as follows. Water samples were collected in 50 mL plastic syringes, and the syringes were plugged while under water, preventing headspace formation. An aliquot of the water sample (10 mL) was transferred to the VG, which was immediately plugged with a silicone stopper. If interference from methanethiol (CH3SH) was a concern, 200 μL of a 50 mM Cd2+ solution was placed in the VG tube in advance. Both stopcocks were closed and the generator tube was gently hand-shaken for 60 s. The stopcock on the air syringe (50 mL, plastic) was opened, and 30 mL of air was introduced from the syringe into the generator through a G1 frit glass ball soaked in sample water. After closing that stopcock, the other stopcock was opened so that the pressurized headspace air was sharply injected (PSI) into the CL cell, where O3 was introduced at a constant flow rate of 300 mL min −1 . A peak corresponding to DMS appeared immediately in the CL signal. This procedure and response signal are shown in the video available as Supporting Information. In the case of DMSP analysis, DMSP was converted to DMS by alkali treatment as follows, and the formed DMS was measured using the VG−CL system. The sample for DMSP analysis was collected in a 500 mL clean plastic bottle. First, dissolved DMS was removed by bubbling the sample (approximately 80 mL) with air (300 mL min−1 for 10 min), which was purified through a small activated carbon column. This water was used to fill a 50 mL sample tube that contained an NaOH pellet (approximately 0.1 g) or 1 mL of 8 M NaOH. After filling, the sample tube was capped to prevent headspace formation. The tube was wrapped with aluminum foil and stored in the dark for 6 h at room temperature. After that, 10 mL of the alkali-treated water was placed in the VG and the formed DMS was analyzed by VG−CL, with Cd2+ treatment. Comparative Analysis by Purge and Trap−Thermal Desorption−Gas Chromatography−Mass Spectrometry (PT−TD−GC−MS). Sample analysis was performed on water samples collected from ponds in a forest (Mt. Tatsuda) near our university and from the shore of the Ariake Sea (Kumamoto prefecture, Japan). The purpose of this study was the comparison of data obtained using two methods, so the samples were packed in the dark in an icebox and brought back to the laboratory. At the same time as the VG−CL analyses, PT−TD−GC−MS analyses were performed for dissolved DMS and for DMS formed from DMSP. An aliquot (10 mL) of the water sample was placed in a 50 mL sample tube and bubbled with purified air. The bubbled air was passed through a CaCl2 column to remove water vapor and subsequently introduced to

■

RESULTS AND DISCUSSION

Introduction of DMS Vapor from the Water Sample to the CL Cell. For the introduction of DMS vapor into the CL cell, one shot introduction was examined instead of bubbling, which is conventionally used. We call this “physical shot introduction” (PSI), in which 30 mL of air was simply introduced into the plugged VG and then the headspace air was introduced to the CL cell via a stopcock (see Figure 1). Small bubbles that formed at the frit glass ball accelerated vaporization of volatile compounds. The present vapor introduction was rapid and effective compared with the conventional bubbling method for transferring the analyte into the CL cell. Bubbling with air or an inert carrier gas is typically used to transfer a vapor from a liquid sample. We characterized the vaporization of volatile species in the bubbling process using the Henry’s law constant (KH) and the carrier gas flow rate (F), as in eq 1:22 C −Ft = exp C0 KHRTV

(1)

C and C0 are the aqueous DMS concentrations at the vaporization times t and 0, respectively, and R, T, and V are the gas constant, temperature, and water sample volume, respectively. According to eq 1, vaporization of DMS from a 10 mL water sample would theoretically be complete (99% vaporization) in 6.3 min at a 200 mL min−1 flow rate. If the transfer is faster, sharper response peaks are expected. If the transfer does not require a carrier gas cylinder or airpump, the system is simpler than one that does. Even in PSI, KH is an important factor. The KH of DMS is 0.56 M atm−1, which is larger than those of AsH3 (0.0089 M atm−1), H2S (0.10 M atm−1), and CH3SH (0.39 M atm−1), but low enough to transfer DMS molecules to the headspace with bubbling. The KH of DMS is much lower than those of hydrophilic volatile compounds such as NH3 (27 M atm−1), dimethyl sulfoxide (DMSO) (1400 M atm−1), and HCHO (3200 M atm−1). 4463

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467

Analytical Chemistry

Article

An experimental comparison of the proposed vapor generation and bubbling is shown in Figure 2. In the

Figure 2. Comparison of vapor generation and bubbling methods for 50 nM DMS.

bubbling−CL method, all the DMS was introduced into the CL cell, but 6 min was required to complete the transfer from a 10 mL water sample. Measurements could be performed every 10 min, including the baseline, in the bubbling−CL method. The VG−CL peak was higher and much sharper compared with that in the bubbling−CL method. The width at half-height of the peak at 50 nM was only 3 s. This was beneficial for increasing the peak height and reducing the measurement time. As can be seen in Figure 2, the VG−CL measurements could be repeated every few minutes, even for manual operation. We examined chemical shot introduction (CSI)23,24 as well as PSI. The tested CSIs were based on H2 and CO2 bubble formation in the sample water by addition of NaBH4 and NaCO3, respectively, with acid. Compared with CSI, PSI gave a higher peak signal for DMS and the blank signal was negligible. Also, physical introduction is advantageous because it does not require chemical reagents, which may react with compounds in the water sample. Accordingly, PSI was used for further experiments. A detailed comparison of the PSI and CSI results is shown in the Supporting Information. The introduction of air increased the pressure of the generator tube, which is the driving force behind the introduction of DMS vapor to the CL cell. The effects of shaking time and air introduction volume were examined (Figure 3a). A CL signal was obtained for DMS even without shaking, but repeatability was poor because a gas−liquid equilibrium was not established. With shaking for ≥1.0 min, the peak height was constant, with good repeatability. Therefore, 1.0 min was selected as the optimum shaking time. As the volume of injected air increased up to 20 mL, the peak height continued to increase and then was almost constant for air volumes between 30 and 40 mL. The initial peak height increase could be attributed to the increased amount of DMS introduced into the CL cell. Further introduction of air would dilute the O3 in the CL cell and result in no further peak height increases. Furthermore, it was difficult for the generator plug to stay in position when the injected air volume was large. Therefore, 30 mL was selected as the optimum volume for air injection. Optimization of CL Detection. The cell for CL detection was made using a round-bottomed flask (Figure 1), which was better than a 50 mL cylindrical cell because the round shape was better for collecting CL by reflection. The DMS vapor was injected into the cell, and ozone was introduced into the cell continuously at a constant flow rate. Because the cell size would

Figure 3. Optimization of DMS vapor generation (a) and ozoneinduced CL detection (b). Data were obtained for 20 nM DMS.

affect the signal intensity, two flask sizes (100 and 200 mL) were tested with various ozone flow rates. The results are shown in Figure 3b. The smaller cell was better than the bigger cell, and the ozone flow rate affected the peak height. High ozone flow rates were better for effective reaction of DMS, but if the flow rate was too high, the residence time and signal decreased. The residence time is important in gas-phase CL detection.25 We selected an ozone flow rate of 300 mL min−1 and a 100 mL round cell. At this ozone flow rate, the peak was sharp compared with those obtained with lower flow rates. The reaction of DMS with O3 is generally expressed as in eq 2 or 3.25 k

DMS + O3 → product O3

O3

DMS → SO → SO2 * → SO2 + hν

(2)

(λmax = 370 nm) (3)

CL is produced after DMS has reacted twice with O3. The rate constant of the DMS/O3 reaction is not well understood. Thirty-five years ago, Martinez and Herron26 reported a value of k < 1.0 × 10−18 cm3 molecule−1 s−1, and recently two Chinese groups reported k values of 1.0 × 10−19 (Du et al.27) and 2.2 × 10−21 cm3 molecule−1 s−1 (Wang et al.28). The lifetime (τ = 1/(k[O3])) estimated using Du et al.’s value is 150 s at 2500 ppm O3. However, the CL response was completed within several seconds in our experiments. Performance of the Analytical Method. As shown in Figure 2, the VG−CL measurement could be repeated every few minutes, even for manual operation. Measurements were performed for 0, 2, 5, 10, 20, 30, 50, and 100 nM DMS solutions and repeated three times. Response signals for various concentrations up to 20 nM are shown in Figure 4. Sharp and strong signals were obtained even for nanomolar levels of DMS. 4464

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467

Analytical Chemistry

Article

rapidly (kO3 is two orders higher) than DMS (Table 1).31 Isoprene CL is reportedly much smaller than that of DMS in gas analysis.18,32,33 In addition, the concentration of isoprene in water is lower; for example, Matsunaga et al. reported that the concentration of isoprene in surface seawater was on the order of picomolar.34 Several organic sulfur compounds were also tested because some sulfur compounds produce CL.35 DMSO, which is usually present in water samples in relatively high concentrations, did not produce any response, probably because it is difficult to volatilize. Interference from H2S was small, at only 1.5% the intensity of the DMS signal. Sensitivities to 1-propanethiol and DMDS were 1/8 and 1/2 that of DMS, respectively. These compounds are not usually present in higher concentrations than DMS, and their interferences are probably negligible in most unpolluted water samples. However, interference from CH3SH was serious, and its signal was 1.7 times higher than that for DMS. Because CH3SH may exist in natural water samples, this interference should be eliminated. To address this problem, the addition of heavy metals was examined. Heavy metals such as Zn2+ 36,37 and Cd2+ 38,39 are used to remove free dissolved sulfide from water, and thiol compounds may be removed in the same way. Figure 5 shows the effects of Ag+, Ni2+, Zn2+, and Cd2+ on the

Figure 4. Response signals obtained using VG−CL. Responses were obtained for 0, 2, 5, 10, and 20 nM DMS. The series of measurements was repeated three times.

The calibration curve was straight, with R2 = 0.9991, in the concentration range from 0 to 100 nM. The dynamic range was up to 1000 nM (R2 = 0.9975). The present method was very sensitive, even although it is a simple analytical procedure. Also, see the abstract artwork for more detailed responses to DMS at low concentrations, namely, 2 and 5 nM. The LOD obtained from three times the signal-to-noise ratio was 0.02 nM (1 ng L−1). The absolute LOD corresponded to 0.2 pmol (10 pg). In addition, water vapor was not introduced into the CL cell continuously in this system so that the CL cell could be used without condensation. One problem in DMS analysis is maintaining the concentration of the standard solution. In the present study, this problem was solved by using a syringe for preparation of the standard solution. The signal for a standard solution prepared in a volumetric flask decreased dramatically with standing time, and the DMS concentration decreased linearly at a rate of 0.7% min−1 (Figure S1 in the Supporting Information). In contrast, the response for a standard solution prepared in a syringe was constant over 100 min, with a relative standard deviation of 0.8%. Similar trends were observed with a water sample, and a syringe was used as the container in dissolved DMS sample analysis. Interference from Other Compounds and Elimination of Interference. This method was based on vaporization and subsequent detection of gas-phase CL, and other dissolved volatile compounds may interfere with the DMS analysis. Some organic compounds such as isoprene react with ozone to produce CL.18,29 Phytoplankton typically produce isoprene at a rate of 0.1−1 pmol 106 cell−1 d−1.30 However, the normalized CL intensity for aqueous isoprene measured using VG−CL was only 1/60 that of DMS, even though isoprene is more volatile (KH is 43 times smaller) than DMS and reacts with O3 more

Figure 5. Removal of CH3SH interference by addition of heavy metals. The examination was performed for 20 nM DMS and CH3SH solutions.

responses to CH3SH and DMS. All these metal ions decreased the intensity of the response for CH3SH, and large decreases were observed with higher metal ion concentrations. The Ag+ ion removed CH3SH very effectively, but it affected the DMS signal. The effect of Ag+ on the DMS signal was small in the

Table 1. Interferences from Various Compounds

a

compd

concn (nM)

response (mV)

relative response

DMS MMa MM with Cd2+ DMDS 1-propanethiol isoprene H2S DMSO

20 20 20 100 1000 2000 10 000 10 000 000

268 445 10 624 1676 420 321 nd

1.00 1.67 0.04 0.467 0.125 0.0157 0.0024

KH (M atm−1) 0.56 0.39 0.92 0.25 0.013 0.10 1400

kO3b (cm3 molecule−1 s−1) 1.0 × 10−19

1.4 × 10−17

MM: methanethiol. bkO3: second-order rate constant for reaction with O3. 4465

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467

Analytical Chemistry

Article

Figure 6. Comparison of analytical data for pond water and seawater samples, obtained using the present method and PT−TD−GC−MS. An example of the raw data obtained using VG−CL and GC−MS is shown on the right.

presence of 3.5% NaCl, which corresponds to the salt concentration in seawater, but the signal was still ≈10% lower. With 1 mM Cd2+, the CH3SH signal was not observed, whereas the DMS response intensity was not affected by addition of Cd2+. The Cd2+ procedure, involving addition of 200 μL of 50 mM cadmium nitrate to a 10 mL water sample, was simple. This countermeasure was only useful with the PSI method. If CO2 generation was used in the CSI, the water sample became acidic and the heavy metal did not remove CH3SH. The effect of salinity was also examined. The signal for 20 nM DMS was constant within a 2.1% relative standard deviation for addition of 0, 10, 20, 30, or 40 mM NaCl, and even 1 M NaCl did not affect the signal. This suggests that the method is suitable for seawater as well as freshwater. Analysis of DMS and DMSP in Water Samples. Seawater and freshwater samples were analyzed using VG− CL and PT−TD−GC−MS. In addition to DMS, DMSP contained in the water samples was analyzed after the transformation. The dissolved free DMSP was decomposed to DMS with a lifetime τDMSP = 5.5 min, and quantitative transformation could be performed in 15 min in our experiments, but a long reaction time (6 h) was needed for total DMSP, including particulate DMSP, as previously reported.8−10 DMS produced by alkaline treatment corresponded to the sum of dissolved and particulate DMSP. In the sample analyses, DMS was detected in all samples of pond water and seawater. However, DMSP was detected only in the seawater samples. This suggests that DMSP is characteristic of oceanic biology. The DMS concentrations in pond water (2.2− 11.1 nM) were lower than those in seawater (5.14−13.8 nM). The DMSP concentrations in seawater (6.5−37.1 nM) were higher than the DMS concentrations in the same samples. Data from analysis of the same samples using VG−CL are plotted against the PT−TD−GC−MS data in Figure 6. Good correlation was obtained between the data:

In the PT−TD−GC−MS analysis, the LODs for DMS and CH3SH were 0.17 and 0.23 nM under the experimental conditions. CH3SH was detected in most of the samples, but only at subnanomolar levels. The DMDS concentration was negligible, as illustrated in the example in the right-hand panel of Figure 6. These experiments were performed using asobtained water samples. Obvious differences were not observed between the data obtained using the two methods, one of which was based on pressure difference and the other on bubbling. Gentle filtering is sometimes used to avoid artificial production of DMS during bubbling, and the same pretreatment can be adopted in this method for plankton bloom waters. Note that strong filtering may produce DMS by destroying plankton, and the filtering process may lead to loss of dissolved DMS by vaporization. Bell et al. showed that 16% of DMS was lost by filtering.3 The pretreatment should be selected in accordance with the sample characteristics. The data in Figure 6 include both DMS and DMSP values, and both sets of data are on the same straight line. This suggests that the effect of artificial DMS formation during bubbling and in the present method were the same, or negligible, in our experiments.

■

VG−CL data (nM) = (0.9517 ± 0.0124) PT−TD−GC−MS (nM) + (0.4138 ± 0.1715) R2 = 0.9956,

n = 28

CONCLUSIONS

Although VG−CL is a very simple analytical method, it is highly sensitive and reliable for onsite analysis of natural waters. With this system on standby, each analysis can be performed within a few minutes. Nanomolar levels of DMS can be detected from a 10 mL sample. It should be noted that prevention of headspace formation in stored samples and standard solutions is necessary for reliable DMS measurements, or the DMS concentration will decrease to half its original value within 1 h in the presence of headspace air. By converting DMSP to DMS, this system can also be used for DMSP analysis. We are currently investigating application of this device to DMS and DMSP in lake water samples in isolated areas in the field. DMSP is produced even in freshwater under specific conditions. Sample analyses (n = 800) have been performed in winter and summer on the shores of Lake Baikal, where the motorized vehicles in close proximity in summer are boats, because of the absence of roads. The results from these analyses will be reported in the near future.

(4) 4466

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467

Analytical Chemistry

■

Article

(30) Bonsang, B.; Gros, V.; Peeken, I.; Yassaa, N.; Bluhm, K.; Zoellner, E.; Sarda-Esteve, R.; Williams, J. Environ. Chem. 2010, 7, 554−563. (31) Atkinson, R.; Winer, A. M.; Pitts, J. N. Atmos. Environ. 1982, 16, 1017−1020. (32) Ohira, S.; Li, J.; Lonneman, W. A.; Dasgupta, P. K.; Toda, K. Anal. Chem. 2007, 79, 2641−2649. (33) Toda, K.; Hirota, K.; Tokunaga, W.; Suda, D.; Gushiken, Y.; Ohira, S. Bunseki Kagaku 2011, 60, 489−498. (34) Matsunaga, S.; Mochida, M.; Saito, T.; Kawamura, K. Atmos. Environ. 2002, 36, 6051−6057. (35) Glinski, R. J.; Sedarski, J. A.; Dixon, D. A. J. Am. Chem. Soc. 1982, 104, 1126−1128. (36) Johnson, C. M.; Nishita, H. Anal. Chem. 1952, 24, 736−742. (37) Pawlak, Z.; Pawlak, A. S. Talanta 1999, 48, 347−353. (38) Jacobs, M. B.; Braverman, M. M.; Hochheiser, S. Anal. Chem. 1957, 29, 1349−1351. (39) Bamesberger, W. L.; Adams, D. F. Environ. Sci. Technol. 1969, 3, 258−261.

ASSOCIATED CONTENT

S Supporting Information *

A video showing the measurement procedure and a comparison of DMS transfer methods for analysis of DMS. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +81-96-342-3389. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Yoch, D. C. Appl. Environ. Microbiol. 2002, 68, 5804−5815. (2) Lovelock, J. E.; Maggs, R. J.; Rasmussen, R. A. Nature 1972, 237, 452−453. (3) Bell, T. G.; Malin, G.; Lee, G. A.; Stefels, J.; Archer, S.; Steinke, M.; Matrai, P. Biogeochemistry 2012, 110, 147−161. (4) Turner, S. M.; Malin, G.; Liss, P. S.; Harbour, D. S.; Holligan, P. M. Limnol. Oceanogr. 1988, 33, 364−375. (5) Leck, C.; Bagander, L. E. Anal. Chem. 1988, 60, 1680−1683. (6) Andress, M. O.; Barnard, W. R. Anal. Chem. 1983, 55, 608−612. (7) Yang, G.-P.; Zhuang, G.-C.; Zhang, H.-H.; Dong, Y.; Yang, J. Mar. Chem. 2012, 138−139, 21−31. (8) Dacey, J. W. H.; Blough, N. V. Geophys. Res. Lett. 1987, 14, 1246−1249. (9) Simó, R.; Grimalt, J. O.; Albaigés, J. Anal. Chem. 1996, 68, 1493− 1498. (10) Li, M.; Yuan, D. X.; Li, Q. L.; Jin, X. Y. Chin. Chem. Lett. 2007, 18, 99−102. (11) Park, K.; Lee, K. Limnol. Oceanogr. Methods 2008, 6, 548−577. (12) Saltzman, E. S.; De Bruyn, W. J.; Lawler, M. J.; Marandino, C. A.; McCormick, C. A. Ocean Sci. 2009, 5, 534−546. (13) Kameyama, S.; Tanimoto, H.; Inomata, S.; Tsunogai, U.; Ooki, A.; Yokouchi, Y.; Takeda, S.; Obata, H.; Uematsu, M. Anal. Chem. 2009, 81, 9021−9026. (14) Sakamoto, A.; Niki, T.; Watanabe, Y. W. Anal. Chem. 2006, 78, 4593−4597. (15) Li, C.; Hill, R. W.; Jones, A. D. J. Chromatogr. B 2010, 878, 1809−1816. (16) Spielmeyer, A.; Georg, P. J. Chromatogr. B 2010, 878, 3238− 3242. (17) Spielmeyer, A.; Gebser, B.; Pohnert, G. Mar. Chem. 2011, 124, 48−56. (18) Azad, M. A. K.; Ohira, S.; Toda, K. Anal. Chem. 2006, 78, 6252− 6259. (19) Toda, K.; Obata, T.; Obolkin, V. A.; Potemkin, V. L.; Hirota, K.; Takeuchi, M.; Arita, S.; Khodzher, T. V.; Grachev, M. A. Atmos. Environ. 2010, 44, 2427−2433. (20) Green, B. C.; Suggett, D. J.; Hills, A.; Steinke, M. Biogeochemistry 2012, 110, 163−172. (21) Devai, I.; Delaune, R. D. Org. Geochem. 1996, 24, 941−944. (22) Toda, K.; Kuwahara, H.; Ohira, S. Environ. Sci. Technol. 2011, 45, 5622−5628. (23) Idowu, A. D.; Dasgupta, P. K.; Genfa, Z.; Toda, K.; Garbarino, J. R. Anal. Chem. 2006, 78, 7088−7097. (24) Hashem, M. A.; Jodai, T.; Ohira, S.; Wakuda, K.; Toda, K. Anal. Sci. 2011, 27, 733−738. (25) Toda, K.; Dasgupta, P. K. Chem. Eng. Commun. 2008, 195, 82− 97. (26) Martinez, R. I.; Herron, J. T. Int. J. Chem. Kinet. 1978, 10, 433− 452. (27) Du, L.; Xu, Y.; Ge, M.; Jia, L.; Yao, L.; Wang, W. Chem. Phys. Lett. 2007, 436, 36−40. (28) Wang, H.; Zhang, Y.; Mu, Y. J. Environ. Sci. 2007, 19, 641−643. (29) Kummer, W. A.; Pitts, J. N.; Steer, R. P. Environ. Sci. Technol. 1971, 5, 1045−1047. 4467

dx.doi.org/10.1021/ac303803w | Anal. Chem. 2013, 85, 4461−4467