high-resolution measurement of volatile organic …high-resolution measurement of volatile organic...

tribute to the formation of photochemicaloxidants and secondary organic aerosols(Donahue and Prinn 1990; Singh et al.1995; Murphy et al. 1998). The ocean sur-face is a substantial natural source of at-mospheric VOCs (5 Tg C yr–1, Guentheret al . 1995) through a variety ofbiogeochemical and photo-induced proc-esses. Light alkenes (e.g., ethene and pro-pene) are produced mainly by photochemi-

High-Resolution Measurement of Volatile OrganicCompounds Dissolved in Seawater Using EquilibratorInlet-Proton Transfer Reaction-Mass Spectrometry(EI-PTR-MS)

Hiroshi Tanimoto1*, Sohiko Kameyama2, Yuko Omori1, Satoshi Inomata1

and Urumu Tsunogai3

1Center for Global Environmental Research, National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305-8506, Japan2Faculty of Environmental Earth Science, Hokkaido University, North 10 West 5, Sapporo, Hokkaido 060-0810, Japan3Graduate School of Environmental Studies, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8601, Japan*E-mail: [email protected]

Abstract. We have developed an equilibrator inlet-proton transfer reaction-mass spectrometry (EI-PTR-MS) system for the high-resolution measurement of the concentrations of multiple volatile organiccompounds (VOCs) dissolved in seawater. The equilibration of six VOCs (dimethyl sulfide (DMS), iso-prene, propene, acetone, acetaldehyde, and methanol) between seawater samples and the carrier gas,and the response time of the system, were evaluated by means of a series of laboratory experiments.Although equilibrium between the seawater sample and the carrier gas in the equilibrator was notachieved for isoprene and propene (likely because of their low water solubility), the other species didreach equilibrium. The EI-PTR-MS system was deployed during a research cruise in the western NorthPacific Ocean. An evaluation of several seawater sampling methods indicated that there was no signifi-cant contamination from the sampling apparatus for the target VOCs. For DMS and isoprene, a com-parison of EI-PTR-MS measurements with measurements obtained with a membrane equilibrator-gaschromatography/mass spectrometry system showed generally good agreement. EI-PTR-MS capturedthe temporal variations of dissolved VOCs, including small-scale variability, which demonstrates thatthe performance of the EI-PTR-MS system was sufficient for simultaneous and continuous measure-ments of multiple VOCs of environmental importance in seawater.

Keywords: Volatile Organic Compounds; PTR-MS; Equilibrator; Western North Pacific Ocean

Introduction

The air-sea exchange of volatile organiccompounds (VOCs) plays an importantrole in the Earth’s biogeochemical cycles,and in the chemistry of the atmosphere. Forexample, nonmethane hydrocarbons(NMHCs), including alkanes, alkenes, andaromatics, and oxygenated VOCs, includ-ing aldehydes, ketones, and alcohols, con-

Western Pacific Air-Sea Interaction Study,Eds. M. Uematsu, Y. Yokouchi, Y. W. Watanabe, S. Takeda, and Y. Yamanaka, pp. 89–115.© by TERRAPUB 2014.

doi:10.5047/w-pass.a02.001

90 H. TANIMOTO et al.

cal reactions of dissolved organic matter(Ratte et al. 1993), and light alkanes (e.g.,ethane and propane) are produced duringthe autolysis of some phytoplanktons(McKay et al. 1996). Isoprene productionis closely related to phytoplankton activ-ity, as indicated by the correlation betweenisoprene concentrations and chlorophyllcontent in surface seawater (Broadgate etal. 1997). Laboratory studies have shownthat isoprene is produced by marinephytoplankton cultures (Moore et al. 1994;Milne et al. 1995; Shaw et al. 2003) andby bacteria (Kuzma et al. 1995). Althoughthe production processes of NMHCs arerelatively well known from previous stud-ies, information about the distribution ofNMHCs in seawater, and their emissionrates to the atmosphere, is limited. Al-though photochemical production has beensuggested as a major source for low-mo-lecular-weight carbonyl compounds in sur-face seawater (Zhou and Mopper 1997),the mechanisms of the formation and de-struction of oxygenated VOCs in seawaterare still largely unknown (Singh et al.2003). As a major natural source of atmos-pheric sulfur, dimethyl sulfide (DMS)plays an important role in climate regula-tion (Charlson et al. 1987). AtmosphericDMS is photo-oxidized to form sulfateaerosols, which may affect the radiativebudget of the atmosphere by serving asprecursors of cloud condensation nuclei.Investigators have proposed thatphytoplankton releasesdimethylsulfoniopropionate (DMSP),which is enzymatically cleaved by micro-organisms to produce DMS (Stefels andVanboekel 1993; Ledyard and Dacey 1994;Liss et al. 1997).

Some of the uncertainty most likelyresults from the limited number of meas-urements of dissolved VOC concentrationsin seawater. A purge-and-trap (P&T)method combined with gas chromato-graphic (GC) detection has been widelyused to measure trace gases dissolved in

seawater. The P&T-GC method permitsmeasurements of dissolved VOC concen-trations with high sensitivity (pmol L–1) inseawater samples collected with Niskinbottles, or by means of shipboard pump-ing systems (e.g., Bonsang et al. 1988;Baker et al. 2000; Kettle et al. 2001). How-ever, this method is not continuous, andeach measurement typically takes minutesto hours, owing to the need for pre-con-centration and pre-separation procedures.Especially for measurements of reactiveVOCs, adsorption during pre-treatment ofsamples sometimes makes obtaining accu-rate and precise measurements by meansof the P&T-GC method difficult. Conse-quently, the development of a highly time-resolved measurement method that doesnot require pre-treatment is stronglyneeded. Such a method would improve ourunderstanding of the global and regionaldistributions of VOCs at the ocean surface,and their production and consumptionprocesses.

Tortell (2005a, b) and Nemcek et al.(2008) developed a membrane inlet-massspectrometry technique for continuous,underway measurement of dissolved DMS,and used the technique to measure DMSconcentrations in marine continental shelfwaters and coastal waters. Marandino etal. (2005, 2007, 2008, 2009) and Saltzmanet al . (2009) made high-frequencyunderway measurements of acetone andDMS in surface seawater using an atmos-pheric pressure chemical ionization massspectrometry system coupled with a con-tinuous-flow membrane equilibrator. Thesensitivity of the system was high enoughto enable the detection of dissolved ac-etone and DMS even in the open ocean.

Proton transfer reaction-massspectrometry (PTR-MS) enables the onlinemeasurement of multiple atmosphericVOCs with relatively high sensitivity (pptvlevels) and a rapid response time (0.1–10s) (Lindinger et al. 1998; de Gouw andWarneke 2007). One of the main weak-

Dissolved VOCs in the Ocean 91

nesses of PTR-MS is that compound iden-tification relies solely on massspectrometry, which means that isobaricspecies cannot be distinguished, particu-larly with a quadrupole mass filter. There-fore, the rigorous comparison of PTR-MSmeasurements to measurements obtainedby a well-established GC technique is stilluseful for the accurate identification andquantification of individual VOCs. PTR-MS has a potential advantage for measure-ments of reactive VOCs, because themethod does not require sample pre-treat-ment (such as dehydration or pre-concen-tration), which sometimes limits the rangeof detectable species. In principle, dis-solved VOCs can be analyzed by PTR-MSby the extraction of VOCs from the liquidphase to the gas phase. Williams et al.(2004) used a P&T PTR-MS method tomeasure several VOCs (e.g., methanol,acetonitrile, acetone, and DMS) in surfacewater.

To measure dissolved VOCs with highsensitivity and time resolution, we devel-oped an equilibrator inlet (EI)-PTR-MSsystem by combining a PTR-MS instru-ment with a bubbling-type equilibrator forequilibration between seawater and air(e.g., Takahashi 1961; Frankignoulle et al.2001). Although a bubbling-type equilibra-tor generally requires a lot of space on re-search vessels, it is less subject to inter-ruption of the water stream than is ashower-type equilibrator (e.g., Keeling etal. 1965; Kelley 1970; Weiss 1981; Butleret al. 1991; Johnson 1999) and less sub-ject to adsorptive loss than is a membrane-type equilibrator (e.g., Saito et al. 1995;Groszko and Moore 1998; Ooki andYokouchi 2008; Loose et al. 2009), de-pending on the membrane material. Herewe report a comprehensive laboratorycharacterization of the EI-PTR-MS systemfor the measurement of six VOCs (DMS,isoprene, propene, acetone, acetaldehyde,and methanol) dissolved in seawater. Wedetermined the dependence of the equili-

brator correction factor on instrumentalparameters, and then investigated the in-strumental response time and the detectionlimits for each VOC. On the basis of fielddeployment of the EI-PTR-MS system ona research cruise, we evaluated seawatersampling methods, compared our resultswith results obtained by a GC method forDMS and isoprene, and obtained the firstVOC concentration data for the westernNorth Pacific Ocean.

Experimental

EI-PTR-MS instrumentFigure 1 shows a schematic diagram of

the EI-PTR-MS system, including theequipment that supplied artificial and natu-ral seawater samples to the equilibrator.The equilibrator, modeled on the equilibra-tor for fCO2 measurements developed bythe National Institute for EnvironmentalStudies (Murphy et al. 2001), consisted ofbrown (to prevent photolysis) vertical glasstubes (I.D. 15.3 cm) and was equipped atthe bottom with a bubbler head (mesh size20–30 mm) for the carrier gas and aseawater outlet and at the top with a car-rier gas outlet and a seawater inlet. Theheight of the water column in the equili-brator was adjusted to be 40, 60, or 100cm by means of tubes of different lengths(20, 40, and 80 cm) in the middle part ofthe equilibrator. Dissolved VOCs wereextracted into a VOC-free carrier gas(ultrapure N2, >99.99995%, Japan FineProducts Co., Kawasaki, Japan). The car-rier gas and sample seawater stream in theequilibrator flowed in opposite directions,and part of the extracted gas was continu-ously directed to the PTR-MS instrumentat ambient pressure, without pre-treatmentsuch as dehydration and pre-concentration.Perfluoroalkoxy (PFA) tubing (3/8≤ and 1/8≤ O.D.) and fittings were used to deliverthe extracted gas samples to the PTR-MS.Tygon tubing (Saint-Gobain, Courbevoie,France) was used for seawater samples.


The flow of seawater was controlledwith a peristaltic tubing pump (flow rate<2 L min–1, MasterFlex 7554-95, Cole-Parmer Instrument Company, Illinois,USA) or a magnetic drive pump (flow rate>2 L min–1, MD-2DRZ, Iwaki, Tokyo, Ja-pan). The seawater flow rate was continu-ously monitored with a flow meter (LD10-TATAAA-RC and DU-5TGS1, HoribaStec, Kyoto, Japan). Water temperaturewas monitored with a temperature probeand a logger (TidbiT V2 Temp, OnsetComputer Corporation, Massachusetts,USA) at the bottom of the equilibrator.

We used a commercially-availablePTR-MS instrument (PTRMS-FDT-hs,Ionicon Analytik GmbH, Innsbruck, Aus-tria). Briefly, the PTR-MS instrument con-sisted of a discharge ion source to produce

the H3O+ ions; a drift tube, in which theproton transfer reactions between H3O+

and VOCs took place; and a quadrupolemass spectrometer for the detection of rea-gent and product ions. In a hollow cath-ode discharge ion source, H3O+ ions wereproduced from pure water vapor (flow rate7.8 sccm). Sample air was introduced intothe drift tube at approximately 22 sccm;the drift tube pressure was held at 2.1 mbar.The sampling inlet and the drift tube wereheld at 105∞C. In the drift tube, VOCs inthe sample air were ionized by protontransfer reactions as follows:

H3O+ + VOC Æ VOC·H+ + H2O. (1)

A fraction of the reagent ion, H3O+, andthe product ions was extracted through a

Fig. 1. Schematic diagram of the EI-TR-MS system, including equipment that supplies artificialand natural seawater to the equilibrator, used to measure concentrations of VOCs dissolved inseawater.


small orifice into the quadrupole massspectrometer, where they were detected bya secondary electron multiplier operatedin the ion pulse counting mode.

The field strength, E/N, of the drifttube, where E is the electric field strength(V cm–1) and N is the buffer gas numberdensity (molecule cm–3), was set to 108 Td(Td = 10–17 cm2 V molecule–1) to reducethe fragmentation of detected VOCs. Thesource current was 6.0 mA, and the accel-erating voltages of the PTR-MS instrumentfrom the upper stream defined as U4, U5,Udrift, U1, and UNC were 50, 120, 400, 48,and 5.8 V, respectively. The count rate ofH3O+, calculated from the count rate at m/z 21 (H3

18O+) multiplied by 500, was typi-cally 1–2 ¥ 107 cps. To determine watervapor concentrations in the sample air,[H2O]sample, we used a calibration curve forthe plot of [H2O]sample versus the relativeintensity of H3O+·H2O to H3O+ (the ratioof m/z 37 to m/z 19) using air samples withknown water vapor concentrations, as de-scribed in Inomata et al. (2008). The in-tensity of H3O+·H2O was calculated fromthe intensity of the peak at m/z 39(H5O18O+) multiplied by 250, because thesignals at m/z 37 (H3O+·H2O) were intense(~106 cps) enough in humidified air sam-ples to possibly damage the ion detector.The signals at m/z 33, 43, 45, 59, 63, and69 were assigned to methanol, propene,acetaldehyde, acetone, DMS, and isoprene,respectively. These assignments are in ac-cordance with previous measurements, al-though minor contributions from other spe-cies, such as propanal to m/z 59, could notbe ruled out (Lindinger et al . 1998;Williams et al. 2001). The detection sen-sitivities for VOCs under dry conditionswere determined by the dynamic dilutionof gravimetrically prepared standard gaseswith ultrapure N2 gas.

Laboratory experiments with artificialseawater

The EI-PTR-MS system, particularly

the equilibrator, was evaluated in the labo-ratory with artificially-prepared seawatercontaining VOCs. Concentrations of thehighly-water-soluble species (DMS, ac-etone, acetaldehyde, and methanol) in theartificial seawater sample were expectedto be stable, even if portions of these spe-cies were extracted by the carrier gas un-der the equilibrator conditions. We addedNaCl to pure water and controlled its sa-linity at 35‰, and then the salt water wassupplemented with aliquots of authenticreagents for the highly-soluble species todissolve these species. In contrast, concen-trations of species with a low water solu-bility (isoprene and propene) in the artifi-cial seawater were expected to decreaseduring the extraction procedure. In thisstudy, therefore, we used the apparatusdescribed by Ooki and Yokouchi (2008) todissolve these species in salt water (Fig.1). Artificial seawater for the low-solubil-ity VOCs was continuously produced witha silicone hollow fiber membrane module(NAGASEP, Nagayanagi Co., Tokyo, Ja-pan; 6000 silicone tubes; length 20 cm;O.D. 0.25 mm; I.D. 0.17 mm). Samplegases, prepared by dynamic dilution of thegravimetric gas standards, were suppliedto the module at 5 L min–1. The samplegases passing through the module werebubbled through water in a bucket (watervolume 15 L). Thus, VOCs in the samplegases permeated the silicone of the mod-ule and dissolved in the circulation water.

The degree of equilibration between thegas phase and the liquid phase depends onthe contact time between the carrier gasand the water in the equilibrator and onthe solubility of the target gases (Johnson1999). We defined the degree of equilibra-tion (D) as p/p¢, where p was the observedpartial pressure in the carrier gas streamand p¢ was the expected Henry’s Law par-tial pressure (D < 100% for disequilibriumconditions). Furthermore, as for low-solu-bility species, the concentrations in artifi-cial seawater decrease in the equilibrator


because the carrier gas removes a substan-tial fraction of the dissolved gases fromthe liquid phase. This “stripping effect (S)”would be 1 for high-solubility species, but<1 for low-solubility species such as iso-prene and propene. Consequently, we ob-tained an equilibrator correction factor(Feq), which was the product of both D andS: Feq = D ¥ S, through a series of labora-tory experiments. We conducted threetypes of experiments in the laboratory toinvestigate Feq in various setups of theequilibrator: (1) We varied the carrier gasflow rate in the range of 75–1000 sccm todetermine how the variation affected theequilibrium of the target species betweenthe carrier gas and the artificial seawater.(2) For species in disequilibrium at any ofthe carrier gas flow rates tested, we exam-ined the dependence of Feq on the waterflow rate and the height of the water col-umn. The flow rate of artificial seawaterwas varied from 0.8 to 6.5 L min–1, andthe tested heights of the water column were40, 60, and 100 cm. (3) The dependenceof Feq on the water temperature, an impor-tant determinant of gas solubility, was alsoexamined in the range from 17 to 34∞C.

Field observation during a research cruiseField observations were made during

the SPEEDS/SOLAS (Subarctic PacificExperiment for Ecosystem DynamicsStudy/Surface Ocean Lower AtmosphereStudy) 2008 cruise of the R/V Hakuho-Maru in the western subarctic North Pa-cific Ocean in July–August 2008. We col-lected seawater samples from the sea sur-face by means of three different continu-ous pumping methods (Fig. 1) to evaluatethe possible contamination of VOCs fromthe sampling apparatus.

The first method involved the collec-tion of surface seawater with the built-inpumping system of the R/V Hakuho-Maru(hereafter referred to as BI sampling). Sur-face seawater was pumped from an inletat the bottom of the ship (5 m below sea

level) with a residence time of approxi-mately 1 min. The use of BI sampling al-lowed us to collect a large volume ofseawater continuously, both underway andduring heave-to periods. However, thesampling line, some of which consisted ofmetals (stainless steel), remained con-nected and in use for long periods and wasa potential source of interference (e.g., ar-tifacts, contamination, adsorption/desorption) for some VOCs and their pre-cursors, such as organic matter in theseawater samples.

The second pumping method involvedan external pumping system (hereafter re-ferred to as EX sampling); this method wasemployed only in heave-to periods. Thescrew pump (WS-20, Nishigaki Pump,Gifu, Japan) in the EX sampling line wasmade partly of metals, but the water flowrate was high (>15 L min–1). The line forthe sample stream was made from newfluororesin tubing and Tygon tubing.

The third method was an underwaysampling method involving a towed tor-pedo-shaped fish to keep the sampling in-let below the sea surface (hereafter referredto as TF sampling). Seawater samples werecollected at 0.5–1.5 m below sea level. Thesampling system, which included tubingand a pump (AstiPure PFD2 bellows pumpand AstiPure AMC pulsation damper,Saint-Gobain), was made of Teflon. TFsampling was available only duringunderway periods. Although the pumpingrate of surface water was 7 L min–1, theseawater flow rate introduced to the equi-librator was limited to approximately 1 Lmin–1 when the TF sample was shared byseveral groups. None of the seawater sam-ples collected by any of the three methodswere filtered, because filtering might haveinterrupted the seawater stream and dis-rupted planktonic cells.

We used an equilibrator with a 60-cmcolumn height (water volume 10 L) withcarrier gas and water flow rates of 75 sccmand 1 L min–1, respectively, during the


cruise. PTR-MS measurements were con-ducted in a multiple ion detection (MID)mode at 5-s integration for each mass percycle, to obtain mass signals at 1-min in-tervals. We also carried out the analysis ina scan mode (m/z 21–100 with 0.5- or 1-sdata integration per cycle) several timesduring the observation period to determinewhether species other than those monitoredin the MID mode could be detected.

In the field measurements, we com-pared the EI-PTR-MS results for DMS andisoprene to those obtained by membraneequilibrator-gas chromatography/massspectrometry (ME-GC/MS). The ME-GC/MS system was originally developed forthe automatic measurement of trace ma-rine halocarbons, typical concentrations ofwhich range from 0.1 to hundreds ofpicomoles per liter. The GC/MS systemwas coupled with a silicone membraneequilibrator, as described in detail in Ookiand Yokouchi (2008). Surface seawatersamples for ME-GC/MS were obtained bymeans of BI and EX sampling. Seawatersamples were continuously passed througha silicone membrane equilibrator at 15 Lmin–1; the equilibrator consisted of six sili-cone tubes (length 10 m; O.D. 2.0 mm; I.D.1.5 mm, material Q7-4780, 80SH, FujiSystem Co., Tokyo, Japan) housed in apolyvinyl chloride pipe (I.D. 16 mm). Dis-solved VOCs (including isoprene) perme-ated the silicone membrane into the gasphase within the silicone tubing. Ultrapureair was continuously supplied to the equi-librator inlet at 25 sccm. The collectedVOC samples in the gas phase were dehy-drated with a Nafion dryer (Perma PureLLC, New Jersey, USA) and transferredto a pre-concentration/capillary GC/MSsystem. The VOC sample was collected ina trap containing Carboxene 1000 andCarbopak B cooled to –50∞C in a smallfreezer for 30 min. Concentrated VOCs inthe first trap were thermally (200∞C)desorbed and transferred to a second trap,which contained Tenax TA and Carboxene

1000 cooled to –50∞C. Then the secondtrap was heated to 200∞C, and the desorbedcomponents were transferred to a capillarycolumn (Porabond Q, 0.32 mm ¥ 50 m)for GC/MS analysis. A gravimetrically-prepared standard gas (isoprene 970 pptv;Taiyo Nippon Sanso Co., Tokyo, Japan)was analyzed for quantification throughthe pre-concentration/capillary GC/MSsystem in the same manner used for theextracted gas sample measurements.

We confirmed that DMS and isoprenereached equilibrium between the liquidphase and the gas phase in the membraneequilibrator. Other polar organics, such asacetone and methanol, were not detectablewith this ME-GC/MS system because theNafion dryer was used. The procedure forchecking the DMS and isoprene equilib-rium was essentially same as the procedureused for the laboratory EI-PTR-MS experi-ments. Ambient air was introduced into themembrane module at a flow rate of 10 Lmin–1 to dissolve DMS and isoprene in thesample water. The concentrations of DMSand isoprene in the ambient air and in theair extracted through the membrane equi-librator were compared by alternate meas-urements at 1-h intervals. Agreement of theratio of the concentrations in the ambientair and the extracted air at the closest meas-urement times was calculated to be 0.99,indicating that complete equilibrium be-tween the liquid and gas phases for DMSand isoprene was achieved with the mem-brane equilibrator, and that the efficiencyof the pre-concentration system did notdepend on the humidity in the sample gas.Analytical precision (2s), based on re-peated analyses of the standard gas, was2%. Isoprene, DMS, and severalhalocarbons were measured with this sys-tem every 70 min during the cruise.

Quantification of dissolved VOC concen-trations

To calculate the VOC concentrations insample water, we applied correction fac-


tors to the concentrations in the sample gasextracted from the equilibrator on the ba-sis of thermodynamic principles. Giventhat VOCs were in equilibrium between thegas and liquid phases, the concentrationsof the VOCs in the liquid phase (ca) shouldhave been equal to the product of the Hen-ry’s law constant (kH) and the partial pres-sure of VOCs in the gas phase (pg):

ca = kH ¥ pg. (2)

Henry’s law constants depend on tempera-ture (T):

kH = kH0 ¥ exp[–DsolnH/R ¥ (1/T – 1/T0)],(3)

where kH0 is the Henry’s law constant un-der standard conditions (T0 = 298.15 K),DsolnH is the enthalpy of the solution, andR is the gas constant. The temperature de-pendence is given by:

–dlnkH/d(1/T) = –DsolnH/R. (4)

In this study, we used data for Henry’s lawconstants reported in previous studies in

seawater (salinity ª 35‰) for DMS, ac-etone and acetaldehyde and in freshwaterfor isoprene, propene, and methanol. Weestimated the Henry’s law constants andtheir temperature dependence by averag-ing all the data (Table 1). The constant kH0,and its temperature dependence, were cal-culated from the fitting curve based on Eq.(3). We used the calculated Henry’s lawconstants to calculate the concentration ofeach VOC dissolved in seawater.

Results and Discussion

Humidity dependence of detection sensi-tivity

In the EI-PTR-MS system, the samplegases extracted through the equilibratorcontained a lot of water vapor. Therefore,we measured the humidity dependence ofthe detection sensitivity for VOCs by vary-ing the humidity of gas mixtures as de-scribed in detail by Inomata et al. (2008).The humidity was varied by mixing 100%humidified pure N2 and dry pure N2 as adiluent gas by means of a humidity gen-erator/controller (SRG-1R-10, SHINYEI,Kobe, Japan). The water vapor concentra-

m/z (amu) kHO (M/atm) -dlnkH/d(1/T) (K) Water type References

DMS 63 0.42 3593 Seawater Cline and Bates (1983)Przyjazny et al. (1983)Dacey et al. (1984)De Bruyn et al. (1995)Wong and Wang (1997)

Isoprene 69 0.013 9950 Freshwater Mackay and Shiu (1981)Karl et al. (2003)

Propene 43 0.00074 3400 Freshwater Wilhelm et al. (1977)

Acetone 59 27.0 ± 1.8 5090 ± 285 Seawater Zhou and Mopper (1990)

Benkelberg et al. (1995)

Acetaldehyde 45 12.3 ± 1.2 2476 ± 743 Seawater Zhou and Mopper (1990)

Benkelberg et al. (1995)

Methanol 33 220 5200 Freshwater Snider and Dawson (1985)

Table 1. Henry’s law constant and its temperature dependence for six VOCs.


tions, [H2O]sample, in VOC standard gasesmixed with the moist air generated withthe humidity controller were in the rangeof 0–19 mmol mol–1 for methanol and 0–36 mmol mol–1 for other VOCs.

The detection sensitivity for the sixVOCs depended markedly on the humid-ity, which indicates that the VOCs reactedwith not only H3O+ ions but also clusterions in the drift tube. For example, thedetection sensitivity for isoprene showeda significant dependence on humidity (Fig.2). Because the number of cluster ions in-creased with increasing humidity in the airsamples, a simple function was used tocorrelate the humidity and the detectionsensitivity. For all VOCs, the humiditydependence of the detection sensitivitieswas better fitted by a quadratic functionthan a linear one, so we used the follow-ing quadratic functions to correct the de-tected signals:

RS = 0.000253 ¥ {[H2O]sample}2 + 1,

R2 = 0.98 for DMS, (5.a)

RS = 0.000235 ¥ {[H2O]sample}2 + 1,

R2 = 0.98 for isoprene, (5.b)

RS = 0.000157 ¥ {[H2O]sample}2 + 1,

R2 = 0.97 for propene, (5.c)

RS = 0.000287 ¥ {[H2O]sample}2 + 1,

R2 = 0.98 for acetone, (5.d)

RS = 0.000274 ¥ {[H2O]sample}2 + 1,

R2 = 0.98 for acetaldehyde, (5.e)

RS = 0.000218 ¥ {[H2O]sample}2 + 1,

R2 = 0.86 for methanol, (5.f)

[H2O]sample = 0–36 mmol mol–1;0–19 mmol mol–1 for methanol,

where RS is the detection sensitivity rela-tive to that under dry conditions. The mea-sured detection sensitivity under dry con-ditions for DMS, isoprene, propene, ac-etone, acetaldehyde, and methanol were6.0, 5.1, 4.8, 9.5, 13.8, and 10.6 normal-ized counts per second (ncps) ppbv–1, re-spectively; the count rates of the productions are expressed as ion count rates nor-malized to an H3O+ intensity of 106 cps.

Equilibrator correction factorFigure 3 shows the dependence of the

equilibrator correction factor (Feq) on the

Fig. 2. Dependence of RS for isoprene on water vapor concentrations in the samples, whereRS is the detection sensitivity relative to that under dry conditions. Vertical and horizontal barsrepresent 2s.


carrier gas flow rate for target VOCs. Theartificial seawater flow rate and the watercolumn height were set to 1 L min–1 and60 cm, respectively. Artificial seawaterwas circulated without carrier gas flowingfor long enough (at least a half day) so thatequilibrium between the VOCs in the equi-librator’s headspace and those dissolved inthe seawater was achieved. Then we de-termined Feq from the ratio of the signalsin the equilibrium experiments to the sig-nals in the headspace gas measurement. Feq

for acetone, acetaldehyde, methanol, andDMS reached 100% at flow rates rangingfrom 75 to 1000 sccm (Fig. 3(a)). The car-rier gas did not remove a substantial frac-tion of these three gases dissolved in theartificial seawater with this equilibratorsetup. The solubility of these VOCs, whichreached equilibrium at all the carrier gasflow rates, was higher than that of DMS(Henry’s law constant, 0.42 M atm–1 at25∞C). These results suggest that specieswith a water solubility higher than that of

Fig. 3. Dependence of equilibrator correction factor of (a) acetone, acetaldehyde, methanol,DMS, (b) isoprene, and (c) propene on carrier gas flow rate. Vertical bars represent 2s. Thecolored lines indicate the theoretically determined degree of equilibration with different car-rier gas flow rates if dissolved gases were extracted at a constant ratio of equilibration regard-less of the carrier gas flow rate.


DMS were extractable to the gas phase inthe complete equilibrium with the liquidphase from seawater using the equilibra-tor at the seawater flow rate and columnheight used for the experiments.

In contrast, Feq for isoprene and pro-pene decreased markedly with increasingcarrier gas flow rate (Figs. 3(b) and (c)).Concentrations of isoprene and propenedissolved in artificial seawater decreasedin the equilibrator, because the carrier gasremoved a substantial fraction of the dis-solved gases from the liquid phase. Assum-ing that isoprene in the carrier gas reachedequilibrium with isoprene in the liquidphase, the concentrations of isoprene dis-solved in seawater theoretically decreasedby approximately 20% relative to the con-centrations in the incoming seawater at acarrier gas flow rate of 75 sccm, a seawaterflow rate of 1 L min–1, and a water volumeof 10 L (60 cm column height). Therefore,both the disequilibration between the gasand liquid phases and the decrease of theconcentrations in the equilibrator shouldhave contributed to Feq for isoprene andpropene. The lines shown in Figs. 3(b) and(c) indicate the theoretically-derived de-pendence of the degree of equilibration (D)on the carrier gas flow rate, if the dissolvedgases were extracted at a constant ratio ofequilibration regardless of the carrier gasflow rate. Assuming complete equilibriumbetween carrier and dissolved gases, Feqshould have coincided with the 100% lines,but the highest data for isoprene and pro-pene did not reach the 100% lines. Thisresult suggests that extracted gases werenot in complete equilibrium with the dis-solved gases for either isoprene or propene.If the bubble size was constant regardlessof the carrier gas flow rate, then the ratioof equilibration between the gas and liq-uid phases ought to have been stable, andtherefore the variations in Feq should havedepended only on the decrease of the con-centrations of isoprene and propene in theliquid phase. However, we found that de-

crease of Feq was greater than that expectedfrom the “stripping effect (S)” alone, whichimplies that the high carrier gas flow rateinduced the production of bigger bubbles.Hence, the reduction of the gross contactarea between the carrier gas and theseawater contributed to the decrease in theexchange rate. We found that Feq includ-ing S for isoprene and propene were ap-proximately 67% and 28%, respectively,at the lowest carrier gas flow rate (75sccm). We confirmed good consistency forFeq between the start and the end of meas-urements during a series of laboratory ex-periment with the same carrier gas flowrate (75 sccm), which suggests that theconcentrations of VOCs in circulated arti-ficial seawater, which decreased in theequilibrator, recovered to the initial con-centrations before the seawater re-enteredthe equilibrator after being mixed in thelarge buffer (15 L) in the bucket and afterre-addition of isoprene and propene by thegenerator of VOC-containing artificialseawater.

Figures 4(a) and (b) show the depend-ence of Feq of isoprene on the artificialseawater flow rate and the height of thewater column. Feq significantly increasedwith increasing artificial seawater flowrate and water column height at all the car-rier gas flow rates. However, Feq did notreach 100%; that is, the equilibrium con-dition, taking S in the equilibrator into con-sideration. Therefore, we had to correct themeasurement data using Feq to determinethe concentrations of the dissolved VOCswith a low water solubility (e.g., isopreneand propene), which were not in equilib-rium with the dissolved gases in the equi-librator.

Figure 4(c) shows the dependence ofFeq of isoprene on the water temperature.In this experiment, the carrier gas flowrate, seawater flow rate, and water columnheight were set at 75 sccm, 1 L min–1, and60 cm, respectively. Feq showed no signifi-cant dependence on water temperature in


Fig. 4. Dependence of equilibrator correction factor for isoprene on (a) artificial seawaterflow rate at 60-cm water column height, (b) water column height at a seawater flow rate of 1 Lmin–1, and (c) water temperature. The dependence on carrier gas flow rate is also shown by thecolored lines in views (a) and (b). Vertical bars represent 2s.


Fig.

5.

R

esp

on

se o

f th

e EI

-PTR

-MS

syst

em t

o a

bru

pt

chan

ges

in t

he c

on

cen

trat

ion

s o

f ta

rget

VO

Cs

in th

e ar

tifi

cial

sea

wat

er s

trea

m u

nd

er v

ario

us

exp

erim

enta

l co

nd

itio

ns:

wit

h a

PFA

-Tef

lon

tub

e at

lo

w w

ater

tem

per

atu

re (

~15∞C

, b

lue)

an

d h

igh

wat

er t

emp

erat

ure

(~2

5∞C

, re

d)

wit

hw

ater

dro

ple

ts o

n t

he in

ner

wal

l of

the

head

spac

e, w

ith

a st

ain

less

ste

el t

ub

e (g

reen

) at

~1

5∞C

wit

h w

ater

dro

ple

ts,

and

wit

h a

PFA

-Tef

lon

tu

be

at ~

15∞C

wit

hou

t w

ater

dro

ple

ts (

bla

ck).

The

arti

fici

al s

eaw

ater

was

sw

itch

ed b

etw

een

tw

o b

atch

es o

f dif

fere

nt

con

cen

trat

ion

s at

the

tim

e o

f0

min

. Th

e fi

nal

co

nce

ntr

atio

ns

wer

e ca

lcu

late

d b

y ex

trap

ola

tin

g ch

ange

s in

VO

C c

on

cen

tra-

tio

ns.

Dat

a ar

e n

orm

aliz

ed t

o r

atio

s o

f co

nce

ntr

atio

n c

han

ges

fro

m i

nit

ial

con

cen

trat

ion

(0

%)

to f

inal

co

nce

ntr

atio

n (

10

0%

). D

ata

colle

cted

at

1-s

inte

grat

ion

are

plo

tted

at

7-s

inte

rval

s. F

or

mo

re d

etai

ls,

see

Kam

eyam

a et

al.

(20

10

)


the range of 17–34∞C. The changes in solu-bility of isoprene due to differences in thewater temperature (calculated as 0.005–0.033 M atm–1 with the Henry’s law con-stant for isoprene) did not contribute to Feqin the range of 17–34∞C. We suspect thatthe higher (lower) D led to the lower(higher) concentration of VOCs in theequilibrator, and that these two effects can-celled each other out. Feq depended onlyon the equilibrator conditions (carrier gasflow rate, seawater flow rate, and heightof water column); therefore, Feq is alsoapplicable to field observations through acareful characterization of its dependenceon the equilibrator conditions in the labo-ratory.

Response timeTo estimate the response time of the EI-

PTR-MS system for the target VOCs, weevaluated the change in concentrationswhen two batches of artificial seawaterwith greatly different VOC concentrationswere switched at the inlet of the equilibra-tor. We prepared a VOC-containing batch,as described earlier, and a VOC-free batch,consisting of salted Milli-Q water with noadded VOCs. By switching the intake tubebetween the two batches, we induced aninstantaneous change in VOC concentra-tion. The carrier gas flow rate, seawaterflow rate, and water column height wereset at 75 sccm, 1 L min–1, and 60 cm, re-spectively (i.e., the same settings used foroceanic observation).

Figure 5 shows the changes in the con-centrations measured by EI-PTR-MS cor-responding to the abrupt changes of theartificial seawater streams. We found nosignificant difference in response time dueto water temperature (15 and 25∞C), or tub-ing materials (PFA and stainless steel tub-ing), with our equilibrator setup. The re-sponse time (defined as the e-folding time)of the entire measurement system clearlydepended on the VOC species (Table 2).The response for DMS was quite rapid,

DM

SIs

op

ren

ePr

ope

neA

ceto

neA

ceta

ldeh

yde

Met

hano

l

Det

ectio

n lim

ita

50 p

mo

l L-

13

pm

ol

L-1

7 p

mo

l L-

14

nmo

l L-

13

nmo

l L-

136

nm

ol

L-1

Res

po

nse

time

(min

)b

With

wat

er d

rop

lets

in

head

spac

e1

12-1

315

6-7

3-4

18-1

9

With

out

wat

er d

ropl

ets

in h

eads

pac

e1

12N

Ac2

22

Tab

le 2

.

Det

ecti

on

l im

it a

nd

res

po

nse

tim

e o

f th

e EI

-PTR

-MS.

a Det

ecti

on

lim

it i

s es

tim

ated

as

the

low

est

sign

al w

hich

is

stat

isti

cal ly

dif

fere

nt

fro

m t

he b

ackg

rou

nd

sig

nal

at

95

% c

on

fid

ence

lev

el.

Co

nce

ntr

atio

n i

s ca

lcu

late

das

sum

ing

the

wat

er t

emp

erat

ure

to

be

25∞C

, an

d t

he w

ater

vap

or

con

cen

trat

ion

to

be

satu

rate

d v

apo

r co

nce

ntr

atio

n a

t 2

5∞C

(2

8.7

mm

ol m

ol–

1).

bR

esp

on

se t

ime

is d

efin

ed a

s e -

fold

ing

tim

e o

f ch

ange

of

det

ecte

d c

on

cen

trat

ion

.c N

A,

no

t an

alyz

ed.


with a response time of ~1 min. Acetalde-hyde also had a short response time (3–4min). The residence times of the carrier gasin the headspace, and of seawater in theequilibrator, were calculated to be 8 and10 min, respectively; that is, the instrumen-tal response time for DMS and acetalde-hyde, and acetone (6–7 min) was alsomuch shorter than the residence times inthe equilibrator. The shorter-than-expectedresponse time most likely resulted from animmediate achievement of equilibriumbetween seawater directed from the inletand gases residing in the headspace. Al-though methanol was one of the speciesthat reached equilibrium, the response timewas relatively slow (18–19 min). A longerresponse time tended to be associated withspecies with a high solubility among theequilibration species, which suggests theinfluence of the uptake of these solublespecies into water droplets present on theinner wall of the headspace. The sprayfrom bubbles breaking at the surface of thewater column in the equilibrator is respon-sible for the production of water dropletsin the headspace. We minimized the influ-ence of water droplets on the response timeby floating a polyethylene sheet on thesurface of the water column to minimizespray. Notches were added to thepolyethylene sheet to prevent water andcarrier gas from stagnating at the surface.We confirmed that the polyethylene sheetdid not adsorb the target VOCs. Underthese conditions, the response time ofmethanol was improved greatly to 2 min,and was shorter than that of any other equi-libration species in the absence of waterdroplets. These results suggest that the re-sponse times for acetone, acetaldehyde,and methanol were delayed by the pres-ence of water droplets and that the influ-ence of the droplets was larger for themore-soluble species.

In contrast, the response times of iso-prene and propene were relatively long(12–13 and 15 min, respectively). Isoprene

and propene did not reach equilibrium inour equilibrator. The response time of iso-prene was not influenced by the presenceof water droplets. Although we did notcarry out the experiments for propene, wehypothesized that propene would not beaffected by water droplets, because itssolubility is lower than that of isoprene.Thus the achievement of equilibrium, andsolubility, were likely the dominant fac-tors in controlling the response time of theVOCs.

We expected the response time duringthe oceanic observations to be similar tothat determined in the laboratory with re-gard to the influence of water droplets,because the spray shield was not used dur-ing the oceanic observations. Minimizingthe presence of “aged” water droplets onthe headspace wall is important for achiev-ing a shorter response time. For the next-generation instrument, we will cover thespray as we did in the laboratory, or rede-sign and rebuild the headspace to main-tain the flow of fresh seawater. Heating thetube between the equilibrator and PTR-MSis also needed to minimize condensation.

Background signals and detection limitsTo determine the background signals

for VOCs in the system, we analyzed N2-purged pure water samples. Backgroundsignals were measured in the MID modeat 1-s integration for each mass per cycle.The pure water (Milli-Q water) stored inthe equilibrator was purged with pure N2gas at 1.0 L min–1 for 6 h, and then theVOCs signals were analyzed at a carriergas flow of 75 sccm. Theoretically, the dis-solved isoprene and propene were almostcompletely extracted into the gas phase bymeans of this degassing treatment. In con-trast, the highly-soluble species (DMS,acetaldehyde, acetone, and methanol) re-mained in the stored water after the purg-ing treatment for a long time, prohibitingus from accurately estimating the back-ground signals for those VOCs. For DMS,


Fig. 6. PTR mass spectra of extracted gas from seawater samples with BI sampling (red bars)and with EX sampling (blue bars), along with difference spectra (BI minus EX, green bars). PTRmass spectra of extracted gas from N2-purged pure water (solid bars) and of pure N2 gas (openbars) are also shown. The ion signals at m/z 21, 29, 30, 31, 32, 34, 36, 37, 38, 39, 55, and 57affected by primary ions are masked.


acetaldehyde, acetone, and methanol,therefore, the signals obtained when moni-toring the pure N2 gas that did not passthrough the equilibrator were assumed tobe the lower-limit background signals.Background signals were measured withN2-purged pure water and pure N2 gas 3and 4 times, respectively, during the cruise.These measurements provided the oppor-tunity to account for the temporal variabil-ity of background signals during the cruiseon the order of weeks. All the oceanic datawere corrected by subtracting the back-ground signals.

Detection limits were defined as thelowest signals that were statistically dif-ferent from the background signals at the95% confidence level. Concentrations ofdetection limits for each species were cal-culated assuming the water temperature tobe 25∞C and the water vapor concentrationto be the saturated vapor concentration at25∞C (Table 2). Only acetaldehyde wasundetectable in the open ocean by compar-ing the detection limit (3 nmol L–1) to thevalue reported in a previous study (1.38 ±0.08 nmol L–1, Zhou and Mopper 1997).The detection limits of other VOCs werelow enough to permit detection of the dis-solved VOCs in the open ocean. Detectionlimits depend on the intensity of the back-ground signals and the integration time ofmass detection. By reducing the influenceof the background signals and extendingthe integration time, we expect to be ableto reduce the detection limits sufficiently,in the future, to enable measurements ofdissolved acetaldehyde in the open ocean.

PTR mass spectra and evaluation of sam-pling methods

Figure 6 shows examples of PTR massspectra of extracted gas from samples ob-tained by BI sampling (with the vessel’sbuilt-in pumping system) and EX sampling(with the external pumping system em-ployed in the heave-to period) at an obser-vational station (42∞N, 160∞E) during the

cruise. Both samples exhibited significantsignals for VOCs at m/z 33, 43, 45, 47, 49,59, 63, 69, 71, and 85, which were assignedto methanol, propene, acetaldehyde, etha-nol, methanetiol, acetone, DMS, isoprene,total C5 alkenes (C5H10·H+), and total C6alkenes (C6H12·H+), respectively. The sig-nals for methantiol (m/z 49) were higherin the BI samples than in the EX samples,which suggests that some of the methantioldetected in the BI samples resulted fromcontamination in the BI sampling line.Selection of the sampling method and sam-pling line will enable us to measure dis-solved methantiol continuously. The dif-ference between the DMS signal intensitiesof the BI and EX mass spectra was only 5ncps, which was smaller than the statisti-cal fluctuations (approximately 30 ncps)resulting from the intensity of the DMSsignals. Therefore, this difference was notdue to contamination during BI sampling.

Significant signals at m/z 47 and 71 inN2-purged pure water were found. Thesesignals were comparable to those detectedin the BI and EX samples. In contrast, sig-nals at m/z 47 and 71 in pure N2 gas weresmaller than those in the BI and EX sam-ples, and in N2-purged pure water. Theseresults suggest that the m/z 47 and 71 weredue to contaminants in the equilibrator.Again, because of the total C6 alkenes, wecould not explicitly identify the species atm/z 85. The signals at m/z 47, 49, 71, and85 are not discussed further in this paper.A significant signal at m/z 43 (propene)was found in N2-purged pure water, al-though this signal was smaller than the sig-nal for the BI and EX samples. This signalprobably resulted from fragmentation ofother species (e.g., propanol and aceticacid) because propene was expected to bealmost completely extracted during thedegassing treatment. The signal at m/z 45(acetaldehyde) in the N2-purged pure wa-ter and the pure N2 gas was significant,making up a large fraction of the signalsdetected in the BI and EX samples. This


result indicates that there was a large blankassociated with contamination in a part ofthe measurement l ine used for bothseawater and N2 gas measurements (e.g.,the inlet of the PTR-MS). Evaluating thebackground signals of both the equilibra-tor and the line is important for accuratemeasurements.

Figures 7(a) and (b) show the temporalvariations of dissolved VOC concentra-tions during measurements of samples ob-tained by EX and TF sampling (underway

sampling with a towed fish), comparedwith samples obtained by BI sampling. Themeasurements were made alternately tominimize the influence of natural varia-tions. The use of the EX and TF samplingmethods likely minimized contaminationduring the sampling procedure because thesampling lines were newly prepared for thecruise, and most of the lines contained nometals. Although signals during alternatemeasurements were associated with natu-ral variations, we found no significant dif-

Fig. 7. Temporal variation of dissolved VOC concentrations during alternate measurementsmade by means of (a) BI sampling and EX sampling and (b) BI sampling and TF sampling. Datacollected at 5-s integration are plotted at 1-min intervals. Averages were calculated from thedata collected during the last 10 min of each sampling period. Data for acetaldehyde are notshown, because large portions of the data were below the detection limit.


ferences before or after the change in sam-pling methods in the alternate measure-ments. This result indicates that the targetVOCs were not influenced by contamina-tion from the BI sampling line of the R/VHakuho-Maru, and hereafter, field obser-vation data are given without mention ofthe sampling methods.

Field observation and comparison of EI-PTR-MS to ME-GC/MS

Figure 8 shows examples of the tem-poral variations of the concentrations ofsix VOCs dissolved in surface seawatermeasured by EI-PTR-MS during the

cruise. The temporal variations of the VOCconcentrations were captured well bymeans of EI-PTR-MS. The temporal vari-ations for DMS and isoprene during thecruise were large, likely reflecting greatspatial variations in DMS and isoprenedistributions associated with highly-vari-able biological activities in the westernNorth Pacific Ocean. The phases in theconcentration variations were well corre-lated to those of chlorophyll-a concentra-tion variations (e.g., 12–13 August), whichsuggest that phytoplankton biomass is amajor controlling factor for DMS and iso-prene concentrations in the ocean surface.

Fig. 7. (continued).


Fig.

8.

T

rack

of

the

SPEE

DS/

SOLA

S 2

00

8 c

ruis

e an

d e

xam

ple

s o

f te

mp

ora

l var

iati

on

s o

f se

asu

rfac

e w

ater

tem

per

atu

re (

do

tted

gra

y lin

e),

chlo

rop

hyll-

a co

nce

ntr

atio

ns

(bla

ck l

ine)

, an

dV

OC

s d

isso

lved

in s

urf

ace

seaw

ater

(b

lack

do

ts).

VO

C d

ata

fro

m p

art

of

the

cru

ise

(bo

ld li

ne)

are

plo

tted

. M

E-G

C/M

S d

ata

are

also

sho

wn

fo

r D

MS

and

iso

pre

ne

(op

en c

ircl

es).

Dat

a co

l-le

cted

at

5-s

in

tegr

atio

n a

re p

lott

ed a

t 1

-min

in

terv

als

for

EI-P

TR-M

S. N

ote

tha

t th

e d

ata

for

acet

ald

ehyd

e b

elo

w t

he d

etec

tio

n li

mit

are

plo

tted

as

the

con

cen

trat

ion

of

a ha

lf o

f th

e d

etec

-ti

on

lim

it.


Fig.

9.

S

catt

er p

lots

of

EI-P

TR-M

S d

ata

vers

us

ME-

GC

/MS

dat

a fo

r D

MS

and

iso

pre

ne.

The

solid

lin

e is

a r

egre

ssio

n l

ine;

the

das

hed

lin

e in

dic

ates

1:1

co

rres

po

nd

ence

. Th

e er

ror

ran

ges

for

the

slo

pe

and

inte

rcep

t o

f th

e re

gres

sio

n li

ne

are

def

ined

at

the

95

% c

on

fid

ence

leve

l.


The variation patterns of propene and ac-etone were well correlated. Although manyof the data for acetaldehyde were belowthe detection limit, episodes of high acetal-dehyde concentrations (e.g., 10–11 Au-gust) were detected in synchrony with epi-sodes of high propene and acetone concen-trations. The similar production sources ofthese species (photolysis of organic mat-ter has been proposed; Ratte et al. 1993;Zhou and Mopper 1997) probably contrib-uted to the simultaneous occurrence. EI-PTR-MS detected small-scale variationsfor DMS, isoprene, propene, and acetone(e.g., 12–13 August), likely resulting fromrapid production or consumption processesdue to diverse biophysicochemical param-eters. In contrast, variations in the metha-nol concentrations were generally broad;therefore, we expect that its production andconsumption processes were not inducedby regional variability of biological activ-ity. In general, although there were someoffsets in EI-PTR-MS data, agreement inthe variability for DMS and isoprene be-tween EI-PTR-MS and MI-GC/MS wasreasonably good. ME-GC/MS sometimesdid not capture the high variability (e.g.,early morning on 13 August) because ofthe method’s limited measurement fre-quency (i.e., 70-min intervals at maxi-mum). This result indicates that EI-PTR-MS is a powerful tool for monitoringhighly-variable features of VOCs in a widedynamic region of the open ocean.

DMS and isoprene concentrations de-termined by EI-PTR-MS were comparedto those obtained by ME-GC/MS (Fig. 9).The slopes and intercepts of the regressionlines are based on the reduced-major-axis(RMA) regression method because bothdata sets are measured variables and, thus,both are subject to error (Ayers 2001). TheEI-PTR-MS concentrations compared herewere averaged over 30 min to match thetime resolution of ME-GC/MS.

For DMS, both data sets were tightlycorrelated. The slope of the overall regres-

sion line for DMS was 0.90 ± 0.02, andthe intercept was negligible (–0.03 ± 0.30nmol L–1, R2 = 0.99). EI-PTR-MS tendedto underestimate DMS concentrationscompared to ME-GC/MS at high concen-trations (>10 nmol L–1) were detected in arelatively warm water region (>16∞C),where the humidity of extracted gasesreached 35 mmol mol–1. In the high-hu-midity range, the uncertainty associatedwith the fitting curve for humidity correc-tion was large (±4% at 35 mmol mol–1 hu-midity). This uncertainty likely contributedto the difference between the EI-PTR-MSand the ME-GC/MS data. Other possibleexplanations include uncertainties in thecorrection factors for background signals(±1% at 10 nmol L–1) for EI-PTR-MS. Asfor ME-GC/MS, the background signalsdue to artifacts may have been underesti-mated at higher concentrations.

For isoprene, the two data sets wereconsistent with each other at the 95% con-fidence interval for the slope of 0.98 ±0.09. We found a small but significant in-tercept of 7.0 ± 6.1 pmol L–1 at the 95%confidence interval. Although isoprene isoften the dominant species at m/z 69among various VOCs (de Gouw andWarneke 2007), the detection of severalother VOC species such as methyl butanal,ethyl vinyl alcohol, and furan at m/z 69 inatmospheric observations has also beenreported (Karl et al. 2001; de Gouw et al.2003; Christian et al. 2004). We calculatedthe ratio between m/z 69 and 70 to narrowdown the potential contributors to the m/z69 signals. If C5H9

+ contributed to m/z 69,the ratio of m/z 70 to m/z 69 (M70/M69)can be calculated to be approximately 5%,owing to the contribution of 13C; other-wise, M70/M69 would tend to be lower,indicating the contribution from other spe-cies including oxygen or sulfur (e.g., furanor C3S). The M70/M69 ratio observed dur-ing the oceanic observations was approxi-mately 6%, which indicates that the offsetof m/z 69 detected during the oceanic ob-


servation probably resulted from the C5H9+

ion. Efforts to identify the species inter-fering with the m/z 69 signals in seawaterwill be made in future work; for example,by using a GC interfaced with the PTR-MS system.

Comparison with previous measurementsTable 3 shows statistics for the concen-

trations of VOCs dissolved in surfaceseawater during the cruise. The mean DMSconcentration for all samples was 4.6 ± 3.9nmol L–1, which agrees well with summer-time levels of 4.3 ± 3.2 nmol L–1 in thesame region obtained by means of a gen-eral GC method (Aranami and Tsunogai2004). We transformed all the observeddata into grid data once, taking an averageat 0.1∞ intervals, and then we obtained sta-tistics for the grid data. As for acetalde-hyde, many of the data were below thedetection limit (3 nmol L–1) during thecruise. The isoprene statistics were com-parable to the concentration range of <12–94 pmol L–1 (Matsunaga et al. 2002) pre-viously reported for the western NorthPacific Ocean. The concentrations of ac-etone measured in this study were similarto a previously-reported concentration(12.1 ± 3.0 nmol L–1) for the eastern NorthPacific Ocean (Marandino et al. 2005).Data for propene and methanol in surfaceseawater are sparse; we found no previous

measurements of these species in the west-ern North Pacific Ocean. In the other re-gions, concentrations in the surfaceseawater of 20–148 pmol L–1 for propene(Plass-Dülmer et al. 1995) and 118.4 ±48.2 nmol L–1 for methanol have been re-ported (Williams et al. 2004). The propeneconcentrations we measured in this studytended to be higher than the values re-ported in a previous study (Plass-Dülmeret al. 1995). Gist and Lewis (2006) re-ported a drastic increase in the propeneconcentration in the summer, an increasecorresponding to increases in the intensityof insolation and phytoplankton biomass.This result suggests that the fact that thecruise occurred during the summer season(July–August), and that the biological pro-ductivity in the research area was high,contributed to the high propene concentra-tion. Our data for acetone and methanolwere slightly higher than but consistentwith the values reported in previous stud-ies within the range of variation. The highbiological productivity in the subarcticNorth Pacific Ocean may also have con-tributed to the higher concentrations foracetone and methanol that we observed.

Conclusions

We developed an EI-PTR-MS systemfor continuous measurements of the con-

*S.D., standard deviation.**BDL, below detection limit.

Average S.D.* Minimum Median Maximum

DMS (pmol L-1) 4.6 3.9 1.1 3.4 30.4

Isoprene (pmol L-1) 70.6 17.3 36.9 71.6 118.7

Propene (pmol L-1) 174.3 34.9 84.9 169.3 286.4

Acetone (nmol L-1) 19.0 4.4 4.4 18.9 41.3

Acetaldehyde (nmol L-1) BDL** BDL BDL BDL 5.9

Methanol (nmol L-1) 158.9 33.1 77.9 162.9 325.0

Table 3. Statistics for concentrations of VOCs dissolved in surface seawater measured duringthe SPEEDS/SOLAS 2008 cruise.


centrations of multiple VOCs (DMS, iso-prene, propene, acetone, acetaldehyde, andmethanol) dissolved in seawater(Kameyama et al. 2009, 2010). Importantcorrection factors including the humiditydependence of the detection sensitivity,Feq, and the subtraction of background sig-nals were investigated so that the dissolvedVOCs could be accurately quantified bymeans of the EI-PTR-MS system with ahigh time-resolution. However, for low-solubility species such as isoprene and pro-pene, Feq may vary according to changesin the bubble size and/or stripping effectrelated to the existence of surfactants insurface seawater. In order to evaluate theuncertainties, inter-comparison with theGC/MS method for each species should beaddressed under a variety of conditions(such as coastal and open ocean waters) infuture field studies. The inter-comparisonof EI-PTR-MS to GC-PTR-MS measure-ments is a most important subject for fu-ture work because it can evaluate not onlyFeq but also in-situ background signals in-cluding interferences, having the samemass number of the target species. Thisinvestigation should assess all target VOCsother than DMS and isoprene and undervarious conditions to test whether the EI-PTR-MS system is universally applicable.In particular, for the low-solubility species,S should be minimized by using a greaterrate of water flow for seawater samples.

Although the measurement of acetalde-hyde concentrations will require a longerintegration time than that used in thisstudy, the detection limits of the EI-PTR-MS system for DMS and isoprene, pro-pene, acetone, and methanol were lowenough for detection in the open ocean.Reduction of contamination in the sam-pling line and the EI-PTR-MS system willenable the detection of other VOCs (e.g.,ethanol and methantiol) by means of theEI-PTR-MS. Thus, the EI-PTR-MS systemhas the potential to enhance the spatial andtemporal coverage of dissolved VOC con-centrations in the open ocean surface, andwill be useful for an investigation into theproduction and consumption of VOCs tochanges in biological activities; for exam-ple, as seen in iron enrichment experimentsand natural phytoplanktonic bloom. There-fore, we conclude that it will be a usefultool for improving our understanding ofthe global budget and production processesof VOCs.

Acknowledgements

We express our sincere thanks to Dr. Y. Nojiri forfruitful comments on the development of the equili-brator. Our thanks go to Drs. Y. Yokouchi, A. Ooki,S. Takeda, H. Obata, A. Tsuda, M. Uematsu, and thecaptain and crew of the R/V Hakuho-Maru for theirsupport during the cruise. This research is a contri-bution to the Surface Ocean Lower AtmosphereStudy (SOLAS) Core Project of the InternationalGeosphere-Biosphere Programme (IGBP).

References

Aranami K, Tsunogai S (2004) Seasonal and regional comparison of oceanic and atmosphericdimethylsulfide in the northern North Pacific: Dilution effects on its concentration during winter. J.Geophys. Res. 109: D12303, doi:1029/2003JD004288.

Ayers GP (2001) Comment on regression analysis of air quality data. Atmos. Environ. 35: 2423–2425.Baker AR, Turner SM, Broadgate WJ, Thompson A, McFiggans GB, Vesperii O, Nightengale PD, Liss PS,

Jickells TD (2000) Distribution and sea-air fluxes of biogenic trace gases in the eastern AtlanticOcean. Global Biogeochem. Cycles 14: 871–886.

Benkelberg HJ, Hamm S, Warneck P (1995) Henry’s law coefficients for aqueous solutions of acetone,acetaldehyde and acetonitrile, equilibrium constants for the addition-compounds of acetone andacetaldehyde with bisulfite. J. Atmos. Chem. 20: 17–34.

Bonsang B, Kanakidou M, Lambert G, Monfray P (1988) The marine source of C2–C6 aliphatic hydro-


carbons. J. Atmos. Chem. 6: 3–20.Broadgate WJ, Liss PS, Penkett SA (1997) Seasonal emissions of isoprene and other reactive hydrocar-

bon gases from the ocean. Geophys. Res. Lett. 24: 2675–2678.Butler JH, Elkins JW, Thompson TM, Hall BD, Swanson TH, Koropalov V (1991) Oceanic consumption

of CH3CCl3: implications for tropospheric OH. J. Geophys. Res. 96: 22347–22355.Charlson RJ, Lovelock JE, Andreae MO, Warren SG (1987) Oceanic phytoplankton, atmospheric sulfur,

cloud albedo and climate. Nature 326: 655–661.Christian TJ, Kleiss B, Yokelson RJ, Holzinger R, Crutzen PJ, Hao WM, Shirai T, Blake DR (2004) Com-

prehensive laboratory measurements of biomass-burning emissions: 2. First intercomparison of open-path FTIR, PTR-MS, and GC-MS/FID/ECD. J. Geophys. Res. 109: doi:10.1029/2003JD003874.

Cline JD, Bates TS (1983) Dimethylsulfide in the equatorial Pacific Ocean—A natural source of sulfur tothe atmosphere. Geophys. Res. Lett. 10: 949–952.

Dacey JWH, Wakeham SG, Howes BL (1984) Henry law constants for dimethylsulfide in fresh-waterand seawater. Geophys. Res. Lett. 11: 991–994.

De Bruyn WJ, Swartz E, Hu JH, Shorter JA, Davidovits P, Worsnop DR, Zahniser MS, Kolb CE (1995)Henry’s law solubilities and Setchenow coefficients for biogenic reduced sulfur species obtainedfrom gas-liquid uptake measurements. J. Geophys. Res. 100: 7245–7251.

de Gouw J, Warneke C (2007) Measurements of volatile organic compounds in the Earth’s atmosphereusing proton-transfer-reaction mass spectrometry. Mass Spectrom. Rev. 26: 223–257, doi:10.1002/mas.20119.

de Gouw J, Warneke C, Karl T, Eerdekens G, van der Veen C, Fall R (2003) Sensitivity and specificity ofatmospheric trace gas detection by proton-transfer-reaction mass spectrometry. Int. J. Mass Spectrom.223: 365–382.

Donahue NM, Prinn RG (1990) Nonmethane hydrocarbon chemistry in the remote marine boundary-layer. J. Geophys. Res. 95: 18387–18411.

Frankignoulle M, Borges A, Biondo R (2001) A new design of equilibrator to monitor carbon dioxide inhighly dynamic and turbid environments. Water Res. 35: 1344–1347.

Gist N, Lewis AC (2006) Seasonal variations of dissolved alkenes in coastal waters. Mar. Chem. 100: 1–10.

Groszko W, Moore RM (1998) A semipermeable membrane equilibrator for halomethanes in seawater.Chemosphere 36: 3083–3092.

Guenther A, Hewitt CN, Erickson D, Fall R, Geron C, Graedel T, Harley P, Klinger L, Lerdau M, McKayWA, Pierce T, Scholes B, Steinbrecher R, Tallamraju R, Taylor J, Zimmerman P (1995) A globalmodel of natural volatile organic compound emissions. J. Geophys. Res. 100: 8873–8892.

Inomata S, Tanimoto H, Kameyama S, Tsunogai U, Irie H, Kanaya Y, Wang Z (2008) Technical note:Determination of formaldehyde mixing ratios in air with PTR-MS: Laboratory experiments and fieldmeasurements. Atmos. Chem. Phys. 8: 273–284.

Johnson JE (1999) Evaluation of a seawater equilibrator for shipboard analysis of dissolved oceanictrace gases. Anal. Chim. Acta 395: 119–132.

Kameyama S, Tanimoto H, Inomata S, Tsunogai U, Ooki A, Yokouchi Y, Takeda S, Obata H, Uematsu M(2009) Equilibrator inlet-proton transfer reaction-mass spectrometry (EI-PTR-MS) for sensitive, high-resolution measurement of dimethyl sulfide dissolved in seawater. Anal. Chem. 81: 9021–9026,doi:10.1021/ac901630h.

Kameyama S, Tanimoto H, Inomata S, Tsunogai U, Ooki A, Takeda S, Obata H, Tsuda A, Uematsu M(2010) High-resolution measurement of multiple volatile organic compounds dissolved in seawaterusing equilibrator inlet-proton transfer reaction-mass spectrometry (EI-PTR-MS). Mar. Chem. 122:59–73.

Karl T, Fall R, Crutzen PJ, Jordan A, Lindinger W (2001) High concentrations of reactive biogenic VOCsat a high altitude site in late autumn. Geophys. Res. Lett. 28: 507–510.

Karl T, Yeretzian C, Jordan A, Lindinger W (2003) Dynamic measurements of partition coefficientsusing proton-transfer-reaction mass spectrometry (PTR-MS). Int. J. Mass Spectrom. 223: 383–395.

Keeling CD, Rakestraw NW, Waterman LS (1965) Carbon dioxide in surface waters of the Pacific Ocean.1. Measurement of the distribution. J. Geophys. Res. 70: 6087–6097.

Kelley JJ (1970) Carbon dioxide in the surface waters of the North Atlantic Ocean and the Barents andKara Seas. Limnol. Oceanogr. 15: 80–97.

Kettle AJ, Rhee TS, von Hobe M, Poulton A, Aiken J, Andreae MO (2001) Assessing the flux of differentvolatile sulfur gases from the ocean to the atmosphere. J. Geophys. Res. 106: 12193–12209.

Kuzma J, Nemecek-Marshall M, Pollock WH, Fall R (1995) Bacteria produce the volatile hydrocarbon


isoprene. Curr. Microbiol. 30: 97–103.Ledyard KM, Dacey JWH (1994) Dimethylsulfide production from dimethylsulfoniopropionate by a

marine bacterium. Mar. Ecol. Proc. Ser. 110: 95–103.Lindinger W, Hansel A, Jordan A (1998) On-line monitoring of volatile organic compounds at pptv

levels by means of proton-transfer-reaction mass spectrometry (PTR-MS) medical applications, foodcontrol and environmental research. Int. J. Mass Spectrom. 173: 191–241.

Liss PS, Hatton AD, Malin G, Nightingale PD, Turner SM (1997) Marine sulphur emissions. Philos.Trans. R. Soc. B. 352: 159–168.

Loose B, Stute M, Alexander P, Smethie WM (2009) Design and deployment of a portable membraneequilibrator for sampling aqueous dissolved gases. Water Resour. Res. 45: doi:10.1029/2008WR006969.

Mackay D, Shiu WY (1981) A critical review of Henry’s law constants for chemicals of environmentalinterest. J. Phys. Chem. Ref. Data 10: 1175–1199.

Marandino CA, De Bruyn WJ, Miller SD, Prather MJ, Saltzman ES (2005) Oceanic uptake and theglobal atmospheric acetone budget. Geophys. Res. Lett. 32: L15806, doi:10.1029/2005GL023285.

Marandino CA, De Bruyn WJ, Miller SD, Saltzman ES (2007) Eddy correlation measurements of the air/sea flux of dimethylsulfide over the North Pacific Ocean. J. Geophys. Res. 112: D03301, doi:10.1029/2006JD007293.

Marandino CA, De Bruyn WJ, Miller SD, Saltzman ES (2008) DMS air/sea flux and gas transfer coeffi-cients from the North Atlantic summertime coccolithophore bloom. Geophys. Res. Lett. 35: L23812,doi:10.1029/2008GL036370.

Marandino CA, De Bruyn WJ, Miller SD, Saltzman ES (2009) Open ocean DMS air/sea fluxes over theeastern South Pacific Ocean. Atmos. Chem. Phys. 9: 345–356.

Matsunaga S, Mochida M, Saito T, Kawamura K (2002) In situ measurement of isoprene in the marineair and surface seawater from the western North Pacific. Atmos. Environ. 36: 6051–6057.

McKay WA, Turner MF, Jones BMR, Halliwell CM (1996) Emissions of hydrocarbons from marinephytoplankton—Some results from controlled laboratory experiments. Atmos. Environ. 30: 2583–2593.

Milne PJ, Riemer DD, Zika RG, Brand LE (1995) Measurement of vertical distribution of isoprene insurface seawater, its chemical fate, and its emission from several phytoplankton monocultures.Mar. Chem. 48: 237–244.

Moore RM, Oram DE, Penkett SA (1994) Production of isoprene by marine phytoplankton cultures.Geophys. Res. Lett. 21: 2507–2510.

Murphy DM, Thomson DS, Mahoney TMJ (1998) In situ measurements of organics, meteoritic mate-rial, mercury, and other elements in aerosols at 5 to 19 kilometers. Science 282: 1664–1669.

Murphy PP, Nojiri Y, Fujinuma Y, Wong CS, Zeng J, Kimoto T, Kimoto H (2001) Measurements ofsurface seawater fCO2 from volunteer commercial ships: Techniques and experiences from Skaugran.J. Atmos. Ocean. Technol. 18: 1719–1734.

Nemcek N, Ianson D, Tortell PD (2008) A high-resolution survey of DMS, CO2, and O2/Ar distributionsin productive coastal waters. Global Biogeochem. Cycles 22: GB2009, doi:10.1029/2006GB002879.

Ooki A, Yokouchi Y (2008) Development of a silicone membrane tube equilibrator for measuring par-tial pressures of volatile organic compounds in natural water. Environ. Sci. Technol. 42: 5706–5711, doi:10.1021/es800912j.

Plass-Dülmer C, Koppmann R, Ratte M, Rudolph J (1995) Light nonmethane hydrocarbons in seawater.Global Biogeochem. Cycles 9, 79–100.

Przyjazny A, Janicki W, Chrzanowski W, Staszewski R (1983) Headspace gas-chromatographic deter-mination of distribution coefficients of selected organosulfur compounds and their dependenceon some parameters. J. Chromatogr. 280: 249–260.

Ratte M, Plass-Dülmer C, Koppmann R, Rudolph J, Denga J (1993) Production mechanism of C2–C4hydrocarbons in seawater: Field measurements and experiments. Global Biogeochem. Cycles 7:369–378.

Saito H, Tamura N, Kitano H, Mito A, Takahashi C, Suzuki A, Kayanne H (1995) A compact seawaterpCO2 measurement system with membrane equilibrator and nondispersive infrared gas analyzer.Deep-Sea Res. I 42: 2025–2033.

Saltzman ES, De Bruyn WJ, Lawler MJ, Marandino CA, McCormick CA (2009) A chemical ionizationmass spectrometer for continuous underway shipboard analysis of dimethylsulfide in near-surfaceseawater. Ocean Sci. 5(4): 537–546.

Shaw SL, Chisholm SW, Prinn RG (2003) Isoprene production by Prochlorococcus, a marine


cyanobacterium, and other phytoplankton. Mar. Chem. 80: 227–245.Singh HB, Kanakidou M, Crutzen PJ, Jacob DJ (1995) High concentrations and photochemical fate of

oxygenated hydrocarbons in the global troposphere. Nature 378: 50–54.Singh HB, Tabazadeh A, Evans MJ, Field BD, Jacob DJ, Sachse G, Crawford JH, Shetter R, Brune WH

(2003) Oxygenated volatile organic chemicals in the oceans: Inferences and implications based onatmospheric observations and air-sea exchange models. Geophys. Res. Lett. 30: doi:10.1029/2003GL017933.

Snider JR, Dawson GA (1985) Tropospheric light alcohols, carbonyls, and acetonitrile: Concentrationsin the southwestern United States and Henry’s law data. J. Geophys. Res. 90: 3797–3805.

Stefels J, Vanboekel WHM (1993) Production of DMS from dissolved DMSP in axenic cultures of themarine-phytoplankton species Phaeocystis sp. Mar. Ecol. Proc. Ser. 97: 11–18.

Takahashi T (1961) Carbon dioxide in the atmosphere and in Atlantic Ocean water. J. Geophys. Res.66: 477–494.

Tortell PD (2005a) Small-scale heterogeneity of dissolved gas concentrations in marine continentalshelf waters. Geochem. Geophys. Geosyst. 6: Q11M04, doi:10.1029/2005GC000953.

Tortell PD (2005b) Dissolved gas measurements in oceanic waters made by membrane inlet massspectrometry. Limnol. Oceanogr. Meth. 3: 24–37.

Weiss RF (1981) Determinations of carbon dioxide and methane by dual catalyst flame ionizationchromatography and nitrous oxide by electron capture chromatography. J. Chromatogr. Sci. 19:611–616.

Wilhelm E, Battino R, Wilcock RJ (1977) Low-pressure solubility of gases in liquid water. Chem. Rev.77: 219–262.

Williams J, Poschl U, Crutzen PJ, Hansel A, Holzinger R, Warneke C, Lindinger W, Lelieveld J (2001) Anatmospheric chemistry interpretation of mass scans obtained from a proton transfer massspectrometer flown over the tropical rainforest of Surinam. J. Atmos. Chem. 38: 133–166.

Williams J, Holzinger R, Gros V, Xu X, Atlas E, Wallace DWR (2004) Measurements of organic speciesin air and seawater from the tropical Atlantic. Geophys. Res. Lett. 31: doi:10.1029/2004GL020012.

Wong PK, Wang YH (1997) Determination of the Henry’s law constant for dimethyl sulfide in seawater.Chemosphere 35, 535–544.

Zhou XL, Mopper K (1990) Apparent partition coefficients of 15 carbonyl compounds between air andseawater and between air and freshwater; Implications for air-sea exchange. Environ. Sci. Technol.24: 1864–1869.

Zhou XL, Mopper K (1997) Photochemical production of low-molecular-weight carbonyl compoundsin seawater and surface microlayer and their air-sea exchange. Mar. Chem. 56: 201–213.

high-resolution measurement of volatile organic …high-resolution measurement of volatile organic...

Documents