ARTICLE IN PRESS

1352-2310/$ - se

doi:10.1016/j.at

�Correspondfax: +5516 360

E-mail addr

Atmospheric Environment 41 (2007) 8924–8931

www.elsevier.com/locate/atmosenv

Dissolved organic carbon in rainwater:Glassware decontamination and sample preservation and

volatile organic carbon

M.L.A.M. Camposa,�, R.F.P. Nogueirab, P.R. Damettob,J.G. Franciscob, C.H. Coelhoa

aDepartamento de Quımica, Faculdade de Filosofia, Ciencias e Letras de Ribeirao Preto, Universidade de Sao Paulo, Av Bandeirantes, 3900,

14040-901 Ribeirao Preto, SP, BrazilbUNESP—Universidade Estadual Paulista, Instituto de Quımica, CEP 14800-900, Araraquara, SP, Brazil

Received 26 February 2007; received in revised form 2 August 2007; accepted 9 August 2007

Abstract

The efficiency of different methods for the decontamination of glassware used for the analysis of dissolved organic

carbon (DOC) was tested using reported procedures as well as new ones proposed in this work. A Fenton solution bath

(1.0mmol L�1 Fe2+ and 100mmol L�1 H2O2) for 1 h or for 30min employing UV irradiation showed to combine

simplicity, low cost and high efficiency. Using the optimized cleaning procedure, the DOC for stored UV-irradiated ultra-

pure water reached concentrations below the limit of detection (0.19 mmolCL�1). Filtered (0.7 mm) rain samples

maintained the DOC integrity for at least 7 days when stored at 4 1C. The volatile organic carbon (VOC) fraction in the

rain samples collected at two sites in Sao Paulo state (Brazil) ranged from 0% to 56% of their total DOC content.

Although these high-VOC concentrations may be derived from the large use of ethanol fuel in Brazil, our results showed

that when using the high-temperature catalytic oxidation technique, it is essential to measure DOC rather than non-

purgeble organic carbon to estimate organic carbon, since rainwater composition can be quite variable, both

geographically and temporally.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Glassware decontamination for DOC; Rainwater DOC storage; Wet deposition; Carbon flux; Non-purgeble organic carbon

1. Introduction

The removal process of the atmospheric dissolvedorganic carbon (DOC) by rain prior to its oxidationto carbon dioxide is of great importance whenevaluating the global carbon biogeochemical cycle.

e front matter r 2007 Elsevier Ltd. All rights reserved

mosenv.2007.08.017

ing author. Tel.: +55 16 3602 4368;

2 4838.

ess: lcampos@ffclrp.usp.br (M.L.A.M. Campos).

It has been estimated that the organic carbonfraction may account for 80% of the global fluxof rainwater carbon (Willey et al., 2000) and it canbe of greater bioavailability to the oceanic biotathan that derived from river water (Avery et al.,2003).

In urban areas, vehicular emissions are animportant source of carbon input to the atmo-sphere, which include volatile organic carbon(VOC) species carcinogenic to humans, as well as

.

ARTICLE IN PRESSM.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–8931 8925

precursors for the formation of photo-oxidants(Vasconcellos et al., 2005; Guo et al., 2007).Different types of vehicular fuels may dominatethe species of VOC found in the urban atmosphere(Colon et al., 2001), and therefore rain water DOCmeasurements should account for such watersoluble volatile compounds. DOC concentrationsin rainwater from urban areas or from placesimpacted with biomass burning can reach relativelyhigh values (Willey et al., 2006; Williams et al.,1997; Lara et al., 2001), while in remote areas theycan be as low as 10 mmolCL�1 (Kieber et al., 2002).In deep seawater, DOC concentrations are about30–40 mmolCL�1 (Aminot and Kerouel, 2004).These low concentrations of organic carbon innatural waters enhance the importance of analyticalaccuracy and the care needed to avoid contamina-tion through all the sampling and analytical steps.The samples’ contact with the atmosphere, thesampling bottles, and the careless handling duringcollection and analysis are a few examples ofpotential contamination for DOC in natural waters(Spyres et al., 2000). Sharp et al. (2002) reported aninternational effort to elaborate an intercalibrationexercise to produce a certified reference material toassure the accuracy of the data for DOC in seawatersamples in order to overcome the uncertaintiestowards the analytical methods used in the early tomid-1990s (Sharp, 1997).

As discussed by Thomas et al. (1995), problems inthe analytical method, and more particularly blankproblems, may explain most of the differences inDOC concentrations measured in seawater byvarious groups using high-temperature catalyticoxidation. The importance of better understandingthe global carbon cycle demands an analytical effortto assure that the reported data represent the actualstate in the study area.

This work aims to identify a simple and efficientmethod to decontaminate glassware used for DOCanalysis, reports a rainwater storage study, andshows the importance of accounting for volatileorganic compounds when choosing the analyticalmethod to measure organic carbon species inrainwater.

2. Materials and methods

2.1. Sites description, sampling and sample treatment

The rain samples were collected in two Braziliancities, Ribeirao Preto (RP; 2111004200S and

4714802400W) and Araraquara (A; 2114703700S and4811005200W), located in Sao Paulo state, being80 km apart, at approximately 313 km from thestate capital (Sao Paulo City) and 350 km from theAtlantic coast. Ribeirao Preto has a population of550,000, and Araraquara 196,000. The region iseminently agricultural, where sugar cane productionand processing is dominant. Rainwater sampleswere collected based on events using a wet-onlyhome-made automatic collector, placed at a heightof 1.5m at the ‘‘Universidade of Sao Paulo’’,campus Ribeirao Preto, located in a very low-trafficsuburban area. The rain sampler at the ‘‘Universi-dade Estatual Paulista’’, campus Araraquara,was a wet only automatic precipitation samplerUNS 130/E—Eigenbrodt placed on a building roofat a height of ca. 15m. The polyethylene funnelfrom the automatic collector was covered withaluminium foil, and the original drain and flaskswere replaced by glass ones.

A glass funnel (23 cm diameter) was attached toan amber glass bottle by a moulded PTFE screwtop, and the whole sampling system was replaced bya clean set after each rain event. For the largemajority of the events (except for the night events),the samples were immediately collected, stored in afridge (4 1C) and filtered within a maximum of 24 h.The filtration was performed in a laminar flow hoodfollowing clean protocols during the whole proce-dure (Campos et al., 2002).

The filtration system consisted of a glass syringe,a stainless steel filtration unit (25mm diameter;Millipore), and a 0.7-mm pore size glass fibremembrane (GF/F; Millipore-AP40) previouslybaked in a muffle furnace at 550 1C for 5 h. Themembrane filters were rinsed with ca. 20mL ofultra-pure water (Milli-Q Simplicity system, Milli-pore) before use to remove possible contaminants.When the rain volume allowed, the first samplealiquot (5–20mL) was used to rinse the membraneand subsequently the pre-cleaned amber storingflask.

2.2. Glassware contamination

To evaluate the efficiency of different proceduresto eliminate DOC residues from glassware, a set ofborosilicate beakers (25mL) were contaminatedwith a 0.01mol L�1 citric acid solution (a model oforganic carbon compound; Tue-Ngeun et al., 2005)prior to treatment. The contamination was per-formed by completely filling the beakers with the

ARTICLE IN PRESSM.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–89318926

citric acid solution, disposing of the solution after24 h, and then letting the vessels to dry upside downon an aluminium foil. Typically, sets of threebeakers were submitted to a specific cleaningprocedure, after which, each of them was filled upto the top with ultra-pure water. The beakers werethen capped with PVC film, stored in closed plasticcontainers and let to rest at room temperature for atleast 24 h before measuring the residual DOC in thewater. When the cleaning procedure consisted ofsubmersing the beakers in a cleaning solution, theflasks were retrieved from the solution and im-mediately filled up with ultra-pure water, cappedand let to rest as described above.

2.3. Methods and instrumental blank

Organic carbon was determined by high-tempera-ture catalytic oxidation using a Shimadzu TOC5000A total organic carbon analyzer. The standardsfor organic carbon were prepared from reagentgrade potassium hydrogen phthalate in ultra-purewater, while the inorganic carbon standards wereprepared from a mixture of anhydrous sodiumcarbonate and sodium hydrogen carbonate.

Although the methods used here are standardprotocols, the analytical character of this workdemands their description. Initially, a 200 mLsample aliquot was injected in the combustion tubeto measure the total carbon (TC) concentration. TCis the sum of inorganic carbon (IC) and totalorganic carbon (TOC). The latter comprises parti-culate carbon (PC) and DOC, which, on its turn,includes the VOC species. In this work, TOC isequivalent to DOC because the rain samples werefirstly filtered, and the synthetic compounds usedwere completely dissolved. After TC was deter-mined, a new aliquot was acidified (H3PO4 25%,v/v) and purged with a CO2-free carrier gas toremove and measure the CO2 produced by theinorganic species. Although VOC is eliminatedduring the purging procedure, only CO2 can bedetected by the infrared sensor and, therefore,measured. The difference between TC and IC allowsfor the calculation of the concentration of all thedissolved organic species, that is, of DOC.

In order to estimate if the rain samples hadappreciable concentrations of VOC, a samplealiquot (typically 5mL) was acidified (HCl2.0mol L�1; pH 2) and purged during 10min beforean aliquot was introduced in the combustion tube.In this way, only the non-purgeble organic carbon

(NPOC) that remained soluble in the sample wasdetected and measured. The difference betweenDOC and NPOC allowed for the estimation ofVOC, also denominated purgeble organic carbon(POC).

Sample aliquots were injected three times in theinstrument for each of the carbon species measured.The relative standard deviation was p2% forconcentrations of 167 mmolCL�1 (2 ppm), which islower than those reported by other studies (e.g.,Decesari et al., 2005). The instrumental variabilitywas daily tested using hydrogen phthalate standard(250 mmolCL�1) stored at 4 1C. When the varia-bility was 47% new standards were prepared and anew analytical curve was built.

It is essential to report the protocol for blankestimation and data correction for the accuratedetermination of DOC concentrations in sampleswith a low organic carbon content (Spyres et al.,2000). The catalyst bed used was Pt supported onalumina or on quartz wool to improve sensitivity. Inthe case of the high-sensitivity catalyst, the instru-mental blank (also called blank system or blankcheck) was performed by a special procedureallowed by the Shimadzu 5000A equipment. Theprogramme is set to inject twice 2mL of ultra-purewater into the system in order to remove anyremaining carbon within the equipment. Then, anew aliquot of ultra-pure water is injected into theTC combustion tube to produce a ‘‘carbon freewater’’ which is recovered in a pre-cleaned trap foranalysis. This ‘‘carbon free water’’ is then re-injectedinto the system using the same volume and protocolas for the samples (200 mL), and after five injectionsthe average area is recorded. This last procedure isrepeated 10 times and the last mean value obtainedis considered to be the ‘‘instrumental blank’’.During this work, the instrumental blank wasperiodically checked and the obtained area wastypically between 3% and 8% of the area of a42 mmolL�1 standard. This value was subtractedfrom those obtained for the samples. The analyticalcurve was shifted to zero to subtract the organiccarbon present in the ultra-pure water used toprepare the standards.

3. Results and discussion

3.1. Study for glassware decontamination

In order to set a reference value to evaluate thedecontamination procedures tested, ultra-pure

ARTICLE IN PRESSM.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–8931 8927

water freshly retrieved from the deionizer systemwas used as a blank control during the course of thisstudy. At this point, the glassware decontaminationprocedure was not yet defined, and a rigoroushandling care was not crucial.

In order to define the ‘‘maximum’’ contaminationlevel, DOC concentrations were determined in fivesets of beakers contaminated with citric acid, whichwere not submitted to any decontamination treat-ment. While the elevated average DOC residualconcentration (Table 1, procedure ‘‘A’’) shows theneed to establish a vigorous treatment to efficientlyclean the flasks, the relatively homogeneous con-tamination (RSD ¼ 16.3%) allows for comparisonamong the different decontamination procedures.

The first commonly used cleaning proceduretested was to soak the contaminated beakers in abath of sodium persulphate 10% for 1 h at 60 1C(Kaplan, 1994). The instrumental peaks producedwere badly shaped with noticeably enlarged areas.This is probably because persulphate residuesreacted with some organic material adsorbed onthe catalyst, leading to its oxidation and theproduction of carbon dioxide. In this study, theresidual DOC concentration was measurable in onlytwo of the treated beakers (Table 1, procedure ‘‘B’’).

Another type of treatment that is broadly used isbaking the glassware for 5–6 h at 450–550 1C(Kaplan, 1994; Sharp et al., 2002). In this study,the average concentration for residual DOC was

Table 1

Average residual DOC concentrations in the water rested in the

beakers after different decontamination procedures, S.D. and

number of samples (beakers) measured (n)

Procedure DOC7S.D.

(mmolCL�1)

Ultra-pure water—control 4.3671.83 (n ¼ 10)

(A) No decontamination 156725.4 (n ¼ 15)

(B) Persulphate bath 10%: 1 h, 60 1C 215716.3 (n ¼ 2)

(C) Calcination: 5 h, 550 1C 27.7713.2 (n ¼ 6)

(D) Calcination: 5 h, 550 1C, rinsing 22.570.82 (n ¼ 3)

(E) Nitric acid bath 10%: 1 h 11.272.66 (n ¼ 12)

(F) Fenton1 bath (1 h) 12.475.65 (n ¼ 9)

(G) Fenton2 bath (1 h) 6.8372.02 (n ¼ 12)

(H) Fenton2 bath: UV 30min 8.3370.80 (n ¼ 3)

Ultra-pure water—stored in cleaned glass

flask

1.0370.25 (n ¼ 3)

Ultra-pure water—UV-irradiated prior

storage

o0.19

The ‘‘control’’ is the ultra-pure water freshly withdrawn from the

deionizer system.

surprisingly high—six times higher than the control(Table 1, procedure ‘‘C’’). Rinsing the contaminatedbeakers before baking them did not improve theresults (Table 1, procedure ‘‘D’’). Calcinationappeared to be highly inefficient for heavilycontaminated borosilicate glassware, since such alarge DOC contamination is unlikely to happenduring the cooling process and/or the resting timewith ultra-pure water. The most efficient of thepublished methods tested here was soaking thebeakers for 1h in a nitric acid 10% (v/v) bath(Thomas et al., 1995; Sharp et al., 2002; Tue-Ngeun etal., 2005), although the decontamination achieved wasnot yet satisfactory when compared with the ultra-pure water used as control (Table 1; procedure ‘‘E’’).

New decontamination procedures were tested inthis work based on the Fenton and photo-Fentonprocesses which are well known for performing arapid and efficient oxidation of organic matter dueto the formation of hydroxyl radicals arising fromthe decomposition of hydrogen peroxide in thepresence of Fe (II) (Pignatello et al., 2006; Nogueiraet al., 2007). Initially, different sets of contaminatedbeakers were submerged for 1 h in freshly preparedsolutions named ‘‘Fenton1’’ (0.5mmol L�1 Fe2+

(Fe(NH4)2(SO4)2 � 6H2O) and 20mmol L�1 H2O2;pH 2.5) and ‘‘Fenton2’’ (1.0mmol L�1 Fe2+ and100mmol L�1 H2O2; pH 2.5). The beakers werewithdrawn from the cleaning solution and filledwith ultra-pure water, which was analysed forresidual DOC after resting for ca. 24 h. The increasein the concentration of Fe2+ and H2O2 solutionssignificantly improved the decontamination process,even when compared with the HNO3 treatment(t-test; P ¼ 0.05; Table 1, procedures ‘‘F and G’’).The use of two germicide lamps (15W each) abovethe Fenton2 solution for 30min did not furtherdecrease the final carbon residue (t-test; P ¼ 0.05;Table 1, procedure ‘‘H’’). Several other tests wereperformed using both Fenton solutions, such as:increasing the immersion time of the contaminatedbeakers, increasing the UV-irradiation time andrising the bath temperature to 60 1C. None of theseprocedures, including a bath with 100mmol L�1

H2O2 only, achieved DOC average concentrationsbetter than 6.8 mmolCL�1 (results not shown).

It is noteworthy that none of the treatmentsshown in Table 1 were able to reach the DOCbackground level found in the control ultra-purewater. This could be due to the cleaning solutionsthemselves, as the beakers were not rinsed aftersubmersion in those media. At this point atmo-

ARTICLE IN PRESSM.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–89318928

spheric contamination was also possible, as theexperiments were performed in a common labbench.

Having defined 1 h of submersion in the Fenton2solution as the most efficient and simple treatmentfor glassware decontamination, this procedure wasthereafter used for all the routine work in ourlaboratories followed by thoroughly rinsing theflasks with ultra-pure water. Any iron residues thatmay have remained in the storing flasks areprobably too low to oxidize organic carbon fromsamples stored in amber flasks at low temperatures.When the samples’ volume allowed, the flasks werealso rinsed with an aliquot of the filtered samplebefore storage.

3.2. Blank tests

To evaluate the possible DOC contaminationfrom the lab atmosphere, six aliquots of ultra-purewater were transferred to pre-cleaned beakerscapped with aluminium foil and were maintainedon the lab bench for approximately 12–27 h. TheDOC concentrations measured were within thevalues found for the control water showed in Table 1.

The DOC average concentration fell to1.03 mmolCL�1 (Table 1) when ultra-pure waterwas stored in a properly cleaned and sealed amberglass flask, having the aliquots retrieved in a laminarflow hood using clean protocols for trace analysis(Campos et al., 2002). Although this concentrationis better than some reported values (e.g.,2.2 mmolCL�1; Avery et al., 2006), it is still higherthan desired (Sharp et al., 2002). Therefore, anotheraliquot of ultra-pure water was UV-irradiated incapped quartz tubes for 4 h using a home-madereactor (Campos et al., 2001) and stored. The DOCconcentrations obtained were below the determinedlimit of detection of 0.19 mmolL�1 (3 S.D.; Table 1),indicating the presence of residual carbon from thewater purification system, which did not have adevice for UV photo oxidation.

3.3. Storage study for preserving DOC in rainwater

Some authors adopt acidification as a method topreserve samples for DOC analysis, including theaddition of H3PO4 and storage at 4 1C for up to 15months (Wiebinga and de Baar, 1998; Lara et al.,2001). However, the drawbacks of acidifying thesamples include the possibility of contamination,volatilization of dissolved carbon and hydrolysis of

organic matter. Furthermore, the use of HgCl2should be avoided as it can deactivate the platinumcatalyst (Spyres et al., 2000; Kaplan, 1994).

As the chemical characteristics of the DOCcontent can vary according to the type of sampleand atmospheric inputs, a rainwater preservationstudy was conducted in order to assure the qualityof the data reported. Rain samples freshly colletedat the Araraquara site were filtered (0.7 mm),immediately analysed and then stored for furtheranalysis.

Preliminary tests performed with potassiumhydrogen phthalate standards (250 mmolCL�1)stored at 4 1C showed that even in the presence ofpotential contaminants such as aqueous solutions oforganic compounds, DOC concentrations remainedconstant for weeks by sealing the flasks mouth witha PTFE tape, an insert plug and a screw cap.Although those solutions were discarded, the flasksealing procedure was maintained, and thereafterclean protocols for trace analysis where used(Campos et al., 2002). No measurable contamina-tion was observed in the blank tests and the storagestudy performed.

Fig. 1 shows the behaviour of DOC concentra-tions for several rainwater samples for up to 21 daysof storage using the optimized cleaning procedure.For two of the samples an abrupt loss of DOC of�60% was observed after 10–11 days of storage,while for the others the loss was much smaller andsmoother with time, or even not significant up to 21days (Fig. 1). The analytical variability for thesamples that had a high enough volume to allow forthe analysis of three sub-samples at each timeinterval was from 3% to 15%. Although it isexpected some bacterial activity after filtration at0.7 mm (Kaplan, 1994), for the type of samplesanalysed in this work DOC integrity can bemaintained for at least 1 week at 4 1C, agreeingwith the findings reported by Willey et al. (2000). Itis unlikely that significant amounts of VOC couldhave been lost during the storage period at 4 1C orduring the short period that the samples stayed atroom temperature before analysis, but these possi-bilities cannot be completely dismissed.

3.4. Volatile organic carbon

In order to evaluate whether VOC could be animportant component of organic carbon in the rainfrom the two sites studied, DOC as well as NPOCwere measured in 16 rain samples (Table 2). The

ARTICLE IN PRESS

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

day number

DO

C (

µm

olC

L-1

)

17/12/03

09/01/04

04/10/04

14/10/04

11/11/04

17/11/04

16/12/04

Fig. 1. Concentrations of DOC in rainwater from Araraquara according to the number of days of storage using the optimized procedure.

The standard deviations shown for samples 17/12/03 and 11/11/04 are related to the measurements realized in three independent aliquots.

Table 2

Concentrations (mmolCL�1) of DOC, NPOC and VOC (calcu-

lated by the difference) in rain samples from Araraquara (A) and

Ribeirao Preto (RP)

Date (local) DOC NPOC VOC (%)

29–30/01/06 (RP) 285 297 �12.0 (�4.2)

22/03/06 (A) 61.2 62.1 �0.90 (�1.5)

25/03/06 (A) 195 197 �2.60 (�1.3)

10–11/02/06 (RP) 141 139 1.30 (0.9)

21/03/06 (A) 160 153 7.50 (4.7)

30/01/06 (RP) 276 254 21.4 (7.8)

25/03/06 (A) 179 155 23.8 (13)

16/03/06 (RP) 433 370 62.3 (14)

30/01/06 (A) 180 147 33.2 (18)

15–16/02/06 (RP) 126 90.4 35.3 (28)

17/03/06 (RP) 327 233 93.5 (29)

16/03/06 (RP) 180 121 58.0 (32)

30/03/06 (A) 49.6 33.4 16.2 (33)

25/03/06 (RP) 101 61.9 39.5 (39)

21/03/06 (A) 167 89.9 77.5 (46)

07/02/06 (RP) 796 347 449 (56)

The percentage of VOC is expressed in brackets.

M.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–8931 8929

estimated VOC concentrations were larger than theinstrumental variability accepted (7%) for 11samples, corresponding to 8–56% of the dissolvedcarbon present in the rainwater samples. There wasno significant correlation between VOC concentra-tions and either the duration of the dry periodpreceding the rain event or the rain volume (t-test;P ¼ 0.05). However, it is worth noting that the twohighest concentrations of VOC were present in thesamples with the lowest volumes—14 and 20mL—when typical rain volumes collected during the rainyseason are about 300mL. To identify the sourcesand the dominant aspects of the variability of theVOC species, a larger number of rain samples wouldbe needed.

The vehicular source of VOC in Brazil is verypeculiar, as ethanol has been used as a car fuel sincethe 1970s, and now accounts for 47.5% of allvehicular fuel used in the country (CETESB, 2006).Out of the 7.3 million vehicles registered in themetropolitan region of Sao Paulo, 65% run ongasohol (addition of �22% ethanol to the petrol),12.3% run exclusively on hydrated ethanol, and6.1% are flex fuel, running on any proportion ofethanol and gasohol. The VOC profile in SaoPaulo’s air is dominated by ethanol with highconcentrations of methanol and 1- and 2-propanol(Colon et al., 2001). Alcohol concentrations in SaoPaulo atmosphere are from 10 to 100 times higherthan those found in Los Angeles, USA, while the

C4–C9 n-aldehydes are 5–10 times higher (Colon etal., 2001). These organic species are very soluble inwater and therefore could significantly contribute tothe total concentration of DOC found in ourrainwater samples.

In 2006, the vehicular fleet in the metropolitanregion of Sao Paulo was responsible for 97% of the

ARTICLE IN PRESSM.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–89318930

358 thousand metric tons of hydrocarbons emittedper year in the region, being 53% of those vehiclesover 10 years old (CETESB, 2006). Assuming thatRibeirao Preto’s vehicular fleet is similar to that ofSao Paulo, its 256 thousand registered vehicleswould emit ca. 12,178 t of hydrocarbon a year,which could at least partially explain the highconcentrations of VOC carbon found in our rainsamples. Furthermore, the residence time of VOC—on the order of days—(Avery et al., 2006) points tolong-range transport from high-emission areas as apotential additional source of VOC in RibeiraoPreto.

Another important VOC source in the studiedregion is the incomplete sugar cane foliage burningpractice (to facilitate manual harvesting) and the useof sugar cane crushing residues as fuel. Theseprocesses occur intensively during the sugar caneharvest period (mainly between April and October),and therefore do not coincide with the samplingperiod for the VOC estimation. However, the high-DOC concentrations found in the rain samplesshown in Fig. 1 could at least be partially explainedby the intense biomass burning activities in thisregion.

For seawater, the difference between DOC andNPOC may be negligible due to the small amount ofVOC expected for such samples (o1% of TOC;Spyres et al., 2000). However, the results presentedhere show that it is essential to include VOC specieswhen measuring DOC concentrations in rainwatersamples, as their composition is known to varygreatly (e.g., Williams et al., 1997; Avery et al.,2001, 2003; Kieber et al., 2002).

4. Conclusions

The Fenton solution bath combines simplicity,low cost and high efficiency to decontaminateglassware for DOC analysis. Using the optimizedprotocol for cleaning the flasks, DOC concentrationin ultra-pure water was found to be1.0370.25 mmolCL�1, falling to values below thelimit of detection (0.19 mmolCL�1) when the ultra-pure water was submitted to UV-irradiation prior tostorage and analysis.

The storage tests at 4 1C demonstrated that theDOC concentrations of the rain samples remainedunaltered for at least 1 week, which has theadvantage of avoiding the addition of preservativesand therefore potential contaminants. The blanktests and storage study demonstrated the good

control over potential contaminants during sampleanalysis.

To report clearly how the instrumental blank waschecked, as well as to detail the method used todetermine DOC concentrations, is of great impor-tance when dealing with samples with such a lowcontent of organic carbon as rainwater. Our resultsshowed that depending on the atmospheric chemicaldynamics it may not be acceptable to report ‘‘DOCor TOC’’ concentrations in rainwater when usingthe procedure involving acidification and purgingprior to analysis, that is, NPOC measurements.

The elevated VOC concentrations found in ourrain samples could be due to the high emissions ofhydrocarbons from the large use of ethanol andgasohol fuels compounded by the great number ofvehicles running with obsolete technology.Although Brazil is at present the only countrywhere the use of alcohol based fuels for automobilesis substantial, the appeal for reducing CO2 emis-sions from fossil fuels is increasing the demand foralcohol-based fuels worldwide. In this context, itappears that not accounting for VOC in rainsamples can lead to an important underestimationof the global rain DOC flux, which could be moresignificant in the near future.

Acknowledgements

The authors thank FAPESP for its financialsupport (research project number 03/01532-4) andCAPES and CNPq for the scholarships provided.

References

Aminot, A., Kerouel, R., 2004. Dissolved organic carbon,

nitrogen and phosphorus in the N–E Atlantic and the N–W

Mediterranean with particular reference to non-refractory

fractions and degradation. Deep-Sea Research Part I—

Oceanographic Research Papers 51, 1975–1999.

Avery, G.B., Tang, Y., Kieber, R.J., Willey, J.D., 2001. Impact of

recent urbanization on formic and acetic acid concentrations

in coastal North Carolina rainwater. Atmospheric Environ-

ment 35, 3353–3359.

Avery, G.B., Willey, J.D., Kieber, R.J., Shank, G.C., Whitehead,

R.F., 2003. Flux and bioavailability of Cape Fear River and

rainwater dissolved organic carbon to Long Bay, southeastern

United States. Global Biogeochemical Cycles 17, 1042,

doi:10.1029/2002GB001964.

Avery, G.B., Willey, J.D., Kieber, R.J., 2006. Carbon isotopic

characterization of dissolved organic carbon in rainwater:

terrestrial and marine influences. Atmospheric Environment

40, 7539–7545.

ARTICLE IN PRESSM.L.A.M. Campos et al. / Atmospheric Environment 41 (2007) 8924–8931 8931

Campos, M.L.A.M., Mello, L.C., Zanette, D.R., Sierra, M.M.S.,

Bendo, A., 2001. Construction and optimization of a low cost

reactor for the photodegradation of organic matter from

natural waters and its application for the study of copper

speciation by voltammetry. Quımica Nova 24, 257–261.

Campos, M.L.A.M., Bendo, A., Viel, F.C., 2002. Low cost

methods to purify low grade reagents and to control

contamination for trace metal determination in natural

waters. Quımica Nova 25, 808–813.

CETESB, 2006. Relatorio de qualidade do ar no estado de Sao

Paulo. /www.cetesb.sp.gov.brS.

Colon, M., Pleil, J.D., Hartlage, T.A., Guardani, M.L., Martins,

M.H., 2001. Survey of volatile organic compounds associated

with automative emissions in the urban airshed of Sao Paulo,

Brazil. Atmospheric Environment 35, 4017–4031.

Decesari, S., Facchini, M.C., Fuzzi, S., McFiggans, G.B., Coe,

H., Bower, K.N., 2005. The water-soluble organic component

of size-segregated aerosol, cloud water and wet depositions

from Jeju Island during ACE-Asia. Atmospheric Environ-

ment 39, 211–222.

Guo, H., So, K.L., Simpson, I.J., Barletta, B., Meinardi, S.,

Blake, D.R., 2007. C1–C8 volatile organic compounds in the

atmosphere of Hong Kong: overview of atmospheric proces-

sing and source apportionment. Atmospheric Environment

41, 1456–1472.

Kaplan, L.A., 1994. A field and laboratory procedure to collect,

process, and preserve freshwater samples for dissolved

organic carbon analysis. Limnology and Oceanography 39,

1470–1476.

Kieber, R.J., Peake, B., Willey, J.D., Avery, G.B., 2002.

Dissolved organic carbon and organic acids in coastal New

Zealand rainwater. Atmospheric Environment 26, 3557–3563.

Lara, L.B.L.S., Artaxo, P., Martinelli, L.A., Victoria, R.L.,

Camargo, P.B., Krusche, A., Ayers, G.P., Ferraz, E.S.B.,

Ballester, M.V., 2001. Chemical composition of rainwater and

anthropogenic influences in the Piracicaba River Basin,

Southeast Brazil. Atmospheric Environment 35, 4937–4945.

Nogueira, R.F.P., Trovo, A.G., Silva, M.R.A., Oliveira, M.C.,

Villa, R.D., 2007. Fundaments and environmental applica-

tions of Fenton and photo-Fenton processes. Quımica Nova

30, 400–408.

Pignatello, J.J., Oliveros, E., Mackay, A., 2006. Advanced

oxidation processes for organic contaminant destruction

based on the Fenton reaction and related chemistry. Critical

Reviews in Environmental Science and Technology 36, 1–84.

Sharp, J.H., 1997. Marine dissolved organic carbon: are the old

values correct? Marine Chemistry 56, 265–277.

Sharp, J.H., Carlson, C.A., Peltzer, E.T., Castle-Ward, D.M.,

Savidge, K.B., Rinker, K.R., 2002. Final dissolved organic

carbon broad community intercalibration and preliminary

use of DOC reference materials. Marine Chemistry 77,

239–253.

Spyres, G., Nimmo, M., Worsfold, P.J., Achterberg, E.P., Miller,

A.E.J., 2000. Determination of dissoved organic carbon in

seawater using high temperature catalytic oxidation techni-

ques. Trends in Analytical Chemistry 19, 498–506.

Thomas, C., Cauwet, G., Minster, J.F., 1995. Dissolved organic-

carbon in the equatorial Atlantic-ocean. Marine Chemistry

49, 155–169.

Tue-Ngeun, O., Sandford, R.C., Jakmunee, J., Grudpan, K.,

McKelvie, I.D., Worsfold, P.J., 2005. Determination of

dissolved inorganic carbon (DIC) and dissolved organic

carbon (DOC) in freshwaters by sequential injection spectro-

photometry with on-line UV photo-oxidation. Analytica

Chimica Acta 554, 17–24.

Vasconcellos, P.C., Carvalho, L.R.F., Pool, C.S., 2005. Volatile

organic compounds inside urban tunnels of Sao Paulo City,

Brazil. Journal of the Brazilian Chemical Society 16,

1210–1216.

Wiebinga, C.J., de Baar, H.J.W., 1998. Determination of the

distribution of dissolved organic carbon in the Indian sector

of the Southern Ocean. Marine Chemistry 61, 185–201.

Willey, J.D., Kieber, R.J., Eymin, M.S., Avery Jr., G.B., 2000.

Rainwater dissolved organic carbon: concentrations and

global flux. Global Biogeochemical Cycles 14, 139–148.

Willey, J.D., Kieber, R.J., Avery, G.B., 2006. Changing chemical

composition of precipitation in Wilmington, North Carolina,

USA: implications for the continental USA. Environmental

Science and Technology 40, 5675–5680.

Williams, M.R., Fisher, T.R., Melack, J.M., 1997. Chemical

composition and deposition of rain in the central Amazon,

Brazil. Atmospheric Environment 31, 207–217.