Mc

Xa

b

c

4d

AA

KMDCPG

1

1

cpepaab

a

0h

Journal of Chromatography A, 1296 (2013) 93– 103

Contents lists available at SciVerse ScienceDirect

Journal of Chromatography A

jou rn al hom epage: www.elsev ier .com/ locate /chroma

icrofluidic lab-on-a-chip derivatization for gaseousarbonyl analysis

iaobing Panga,c, Alastair C. Lewisb,∗, Milagros Ródenas-Garcíad

Department of Chemistry, University of York, Heslington, York YO10 5DD, UKNational Centre for Atmospheric Science, University of York, Heslington, York YO10 5DD, UKKey Laboratory of Wetland Ecology and Environment, Northeast Institute of Geography and Agroecology (NEIGAE), Chinese Academy of Sciences,888 Shengbei Road, Changchun 130102, ChinaInstituto Universitario Centro de Estudios Ambientales del Mediterráneo (CEAM-UMH), Valencia, Spain

a r t i c l e i n f o

rticle history:vailable online 8 May 2013

eywords:icrofluidic lab-on-a-chiperivatizationarbonyl compoundsFBHAC–MS

a b s t r a c t

We present a microfluidic lab-on-a-chip derivatization technique for the analysis of gaseous carbonylcompounds using O-(2,3,4,5,6-pentafluorobenzyl) hydroxylamine (PFBHA) as the derivatizing reagent.The novel microfluidic lab-on-a-chip derivatization technique has been developed to measure nmol permole (ppbv) mixing ratios of gaseous carbonyl compounds, which are of particular importance to atmo-spheric chemistry. The technique utilised a planar glass microreactor comprising three inlets and oneoutlet, gas and fluid splitting and combining channels, mixing junctions, and a 2.0 m long, 620 �m inter-nal diameter reaction microchannel. The microreactor integrated three functions, providing: (1) a gasand liquid mixer and reactor, (2) reagent heating, and (3) sample pre-concentration. The concentrationof derivatization solution, the volumetric flow rates of the incoming gas sample and PFBHA solution,and the temperature of the microreactor were optimised to achieve a near real-time measurement. Theenhanced phase contact area-to-volume ratio and the high heat transfer rate in the microreactor resultedin a fast and high efficiency derivatization reaction, generating an effluent stream which was ready fordirect introduction to GC–MS. Good linearity was observed for eight carbonyl compounds over the mea-surement ranges of 1–500 ppbv when they were derivatized under optimal reaction conditions. Themethod detection limits (MDLs) were below 0.10 nmol mol−1 for most carbonyls in this study, which is

below or close to their typical concentrations in clean ambient air. The performance of the technique wasassessed by applying the methodology to the quantification of glyoxal (GLY) and methylglyoxal (MGLY)formed during isoprene photo-oxidation in an outdoor photoreactor chamber (EUPHORE). Good agree-ments between GLY and MGLY measurements were obtained comparing this new technique with FourierTransform InfraRed (FTIR), which provides support for the potential effectiveness of the microfluidic

asure

technique for gaseous me

. Introduction

.1. Derivatization

Derivatization is a technique used to transform a chemicalompound into a new compound, called a derivative, which hasroperties more amenable to a particular analytical method. Gen-rally derivatization is employed to realise two main goals: (1) toermit separation and analysis of species which are not directly

menable to analysis due to inadequate volatility or stability,nd (2) to enhance the analysis by improving chromatographicehaviour or detectability [1]. In separation science derivatization

∗ Corresponding author. Tel.: +44 01904 322522; fax: +44 01904 322516.E-mail addresses: pangxbyuanj@gmail.com, pangxbyuanj@163.com (X. Pang),

lly.lewis@york.ac.uk (A.C. Lewis).

021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.chroma.2013.04.066

ments.© 2013 Elsevier B.V. All rights reserved.

has been commonly employed in gas chromatography (GC) [2–5],high performance liquid chromatography (HPLC) [6–10] and inGC/LC-mass spectrometry (MS) [11–13]. The foremost reason foremploying analytical derivatization in GC is to impart volatilityor thermal stability to compounds which are polar, have poorvolatility or poor thermal stability [14]. In HPLC analysis the chro-mographic derivatization can impart or increase UV or fluorescenceproperties to analytes which have weak absorption or fluores-cence emission, enabling detection following HPLC. In MS analysisderivatization is applied for structure confirmation and to obtainderivatives where the spectra contains a molecular ion signal [15].

A good derivatizing reagent should produce the desiredchemical modification of the compound(s) of interest, and be

reproducible, efficient, and non-hazardous. For GC, there are threebasic types of derivatization reactions: silylation, acylation, andalkylation [1]. Silylating reagents react with those compounds con-taining active hydrogen functional groups such as OH, COOH,

9 atogr.

mraswvdhw

stsrTpklp

1

tctfVCptai[cAatuddehr[

taedwtacdc

toah(3dt

4 X. Pang et al. / J. Chrom

SH, NH2, NH, POH, SOH, and they are the most com-on type of derivatization reagents used in GC [14,16]. Acylating

eagents react with highly polar multi-functional compounds suchs amino acids or carbohydrates [17,18]. Alkylating reagents sub-titute the active hydrogens on amines and acidic hydroxyl groupsith aliphatic or aliphatic-aromatic groups and is ideal for con-

ersion of organic acids into esters [19,20]. Some other types oferivatization reagents have been used to obtain oxime [2,21],ydrazone [22–24], methylation [25], and cyclic derivatives [26]ith other functional groups.

Analytical derivatization requires suitable containers for theamples and derivatization reagents and vessels for the deriva-ization reactions [1]. The means of transferring and measuringamples and reagents and means of heating, mixing and evapo-ating the samples are also particular to derivatization processes.hose apparatus used for analytical derivatization are rather com-lex and the processes somewhat labour intensive. Therefore, theey to making major improvements in derivatization methods mayie in apparatus which can realise simple and automated samplereparation prior to analysis.

.2. Derivatization of carbonyl compounds

Carbonyl compounds (aldehydes and ketones), contain the func-ional group (C O), and are a ubiquitous class of volatile organicompound (VOCs) in the atmosphere [27,28]. They are emit-ed directly from incomplete combustion of biomass and fossiluel and formed indirectly by atmospheric photo-oxidation ofOCs emitted from anthropogenic and natural sources [28–30].arbonyl compounds (carbonyls) are of critical importance in tro-ospheric chemistry as a major source of free radicals, and arehe precursors to ozone [27,31], peroxyacyl nitrates (PANs) [27]nd secondary organic aerosol (SOA)[28]. Carbonyls are commonndoor air pollutants which may be hazardous to human health32]. Formaldehyde (HCHO) for example is classified as a humanarcinogen that causes nasopharyngeal cancer by the Internationalgency for Research on Cancer (IARC)[33]. Acetaldehyde (CH3CHO)nd acrolein (CH2 CHCHO) are also listed as possibly carcinogenico humans by IARC. In recent years carbonyl concentrations in manyrban ambient air and indoor air environments have increasedue to increasing emissions from vehicular exhausts and interiorecoration materials [29,34,35]. In urban air and indoor micro-nvironments, carbonyl mixing ratios can vary from tens of ppbv toundreds of ppbv [29,36,37] while in remote clean oceanic areas orural sites carbonyls vary from less than one ppbv to tens of ppbv30,38,39].

Given their crucial role in tropospheric chemistry and the nega-ive health effects on humans, atmospheric carbonyls are thereforen important class of species for monitoring and regulation. How-ver, some of carbonyls are difficult to directly detect by GC or HPLC,ue to their low volatility and thermal stability for GC analysis, oreak absorption in the UV for HPLC analysis [21]. Chemical deriva-

ization, therefore, is commonly employed to enable GC or HPLCnalysis for carbonyls through improving their volatilities or UVharacteristics. In recent decades many researchers have studiederivatization reagents, sampling techniques and optimal reactiononditions.

Oximes and hydrazones are the most commonly used deriva-ives for carbonyl derivatizations though there are four typesf derivatives: oximes, hydrazones, trimethylsilyl enol ethersnd acetal/ketal derivatives [1]. O-(2,3,4,5,6-pentafluorobenzyl)ydroxylamine (PFBHA) [2,40], pentafluorophenyl hydrazine

PFPH) [22,6,41], 2,4-dinitrophenylhydrazine (DNPH) [10,42],-methyl-2-benzothiazolinonehydrazone (MBTH) [43,44],ansylhydrazine (DNSH) [23] and N,O-bis(trimethylsilyl)-rifluoroacetamide (BSTFA) [45] are the most common

A 1296 (2013) 93– 103

derivatization reagents reported in existing literatures. Theircorresponding derivatives are the oximes forming by PFBHA andhydrazones by the DNPH, DNSH, MBTH and PFPH derivatizations.Some derivatization reagents are specific for formaldehyde byUV/vis spectrometer analysis such as pararosaniline [46,47],chromotropic acid [48,49], 4-amino-3-hydrazino-5-mercapto-4H-1,2,4-triazole (AHMT) [50], acetylacetone [51].

The sample preparation for chemical derivatization analysis ofcarbonyls in general consists of trapping gaseous carbonyls on solidabsorbents [22,23] or filters [52,53] coated with the derivatizationreagent, or dissolving gaseous samples into a derivatization solu-tion using an impinger [2], denuder [53,54], or scrubber [24,43,55].The resulting derivatives can subsequently be desorbed by anappropriate solvent, evaporated, re-dissolved and then separatedand detected by GC–FID, GC–MS or HPLC–UV. Such methods haveacceptable sensitivity, good reproducibility and are versatile andcarbonyl specific. They have the considerable advantage of beingable to quantify multiple carbonyl compounds simultaneously.Although derivatization methods are state-of-the-art for the anal-ysis, the drawbacks of previously reported methods are longsampling times, typically 1–2 h, with a lengthy procedure of deriva-tization and solvent extraction/evaporation following sampling ofambient air, which can be labour intensive and unattractive forcontinuous analysis. Solid phase microextraction (SPME) on-fibrederivatization techniques have been reported to reduce samp-ling times to 10 min or less [7,56–59], but the detection rangeof SPME techniques is reduced due to a relatively small amountof derivatization reagent absorbed on the fibre [60–62]. Further-more, the preparation processes for PFBHA-coated SPME are stillrelatively cumbersome. A continuous automated HPLC system hasbeen reported using a single silica microcartridge based on DNPHderivatization [63]. Despite this diversity of approaches in the liter-ature, there remains a pressing need for the development of a rapidand highly convenient carbonyls detection method which has lowMDLs and high specificity. There are substantial benefits thereforein any methodology which can use the universality of a GC–MSmethod whilst eliminating the bench chemistry required for sam-ple preparation. The key therefore to making major improvementsin carbonyl measurements may lie in automated sample prepara-tion.

1.3. Microfluidic lab-on-a-chip for atmospheric measurement

Microfluidic lab-on-a-chip is highly integrated compact minia-turisation approach which may have complex and specificfunctions. It has emerged as a powerful technology over the lasttwo decades in the life sciences, medicine, cell biology, materialsscience, analytical and synthetic chemistry, physics, fluid mechan-ics and engineering disciplines. Notably, microfluidic systems havea much smaller footprint, faster reaction rates that help shortenthe time to result, reduced sample consumption which can lead tolower cost, multiplexing capability and the ability to perform highthroughput experiments. Although microfluidics have been com-monly employed in life sciences and other scientific disciplines,the technology has not been widely used in environmental moni-toring, especially for gaseous atmospheric measurement. Researchon microfluidics for atmospheric measurements has really onlyemerged in the last ten years.

Some recent examples are given here: a microfluidic elec-trochemical sensor was developed to measure on-line aerosoloxidative properties coupled directly to a particle-into-liquid sam-pler system. The oxidative activity was found to be in good

correlation with aerosol concentration with a time resolution ofnear 3 min [64]. The same group developed further microflu-idic paper-based analytical devices for aerosol oxidative activitybased on the same chemical mechanism [65]. A microfluidic

atogr.

aunbneewspstntptc3zAttsotddmtsi

1c

d

itedTawtarws

Flr

X. Pang et al. / J. Chrom

erosol-to-hydrosol sampler converted analytes from gas to liq-id phase and was integrated with an antibody modified siliconanowire sensor to realise near real-time detection of airborneiological aerosols [66]. One microfluidic device for gaseous ammo-ia (NH3) analysis utilised an annular gas/liquid laminar flow toxtract and concentrate NH3 from air in nearly 100% extractionfficiency and had on-site monitored NH3 continuously for threeeeks in a cleanroom [67]. An on-chip miniaturised distillation

ystem measured gaseous sulfur dioxide (SO2) from wine androvided a compact and reliable tool for SO2 concentration mea-urement to wine producers on site, rather than shipping sampleso laboratories [68]. A microchip has been designed for measuringitrogen dioxide (NO2) in the atmosphere by gas absorption into ariethanolamine flow (containing fluorescent reagents) through aorous glass layer plate, chemical reaction and fluorescence detec-ion to enable ppbv limit of detection [69]. A microfluidic deviceonsisting of concentration and detection cells was fabricated on a

cm × 1 cm Pyrex plates for the optical detection of airborne ben-ene, toluene, ethylbenzene and xylenes (BTEX) at ppm levels [70].n air-cooled cold trap channel was integrated in the above men-

ioned microfluidic device for BTEX analysis to prevent dilution ofhe desorbed gas from adsorbent. The detection sensitivities wereignificantly improved with detection limits (0.05 ppm) about 2rders of magnitude lower than the previous work without coldraps [71]. Microfabricated planar glass gas chromatography waseveloped on a 95 mm × 95 mm microchip comprising of gas flui-ic interconnections, injection zone and capillary GC columns. Theicrofabricated GC integrated with a modified lightweight pho-

oionization detector showed good analytical performance for amall set of VOCs and sub ng detection sensitivity to monoaromat-cs [72].

.4. Microfluidic lab-on-a-chip derivatization for gaseousarbonyls

In this paper, we will present a microfluidic lab-on-a-chiperivatization technique for a range of gaseous carbonyl analysis.

In our previous work, the rapid, simple and sensitive microflu-dic derivatization approach had been successfully employedo analyse gaseous formaldehyde using accelerated and highlyfficient derivatization reactions between formaldehyde and aerivatization reagent (PFBHA or PFPH) inside a microreactor [21].he Pyrex micro-reactor comprises three inlets and one outlet, gasnd fluid mixing junctions, and a 2.0 m long reaction microchannel,hich provided a zone for gas sample mixing, chemical reac-

ion and pre-concentration. The flow rates of the gas sample

nd derivatization solution and the temperature of the micro-eactor were optimised to achieve a near real-time measurementith a rapid and high efficiency derivatization step following gas

ampling. Combining the microreactor with the auto-sampler of

ig. 1. Panel a: basic layout of micro-reactor including three inputs, onength × width × thickness = 90 mm × 45 mm × 4.5 mm). Panel b: mixture and reactioneacting microchannel. (For interpretation of the references to colour in this figure legend

A 1296 (2013) 93– 103 95

a commercial GC–MS allowed for continuous sampling and anal-ysis with a time resolution of around 30 min or less. Methodcalibration showed a good linearity in the measurement rangeof 26–331 ppbv with correlation coefficients R2 = 0.988 and 0.997for PFPH and PFBHA derivatives. For low gas phase formaldehydemixing ratios (<26 ppb) the derivatization solution was repeat-edly recycled through the microchip enabling preconcentration oftheir derivatives - oxime for PFBHA and hydrazone for PFPH. Thecalibration curves for this recycling approach also showed good lin-earity from 4.0 to 26 ppbv with method detection limits (MDLs) of1.1 ppbv and 2.1 ppbv for PFBHA and PFPH derivatives. The feasibil-ity of the technique was assessed using measurements of laboratoryambient air, with formaldehyde the predominant carbonyl com-pound at the 10 ppbv level. The proof of principle experimentsdemonstrated the potential of the approach for on-line measure-ments of other carbonyls including acetaldehyde, acetone andpropionaldehyde [21]. Microfluidic derivatization is attractive formany reasons; of relevance here for field measurements are itseconomical consumption of reagents and energy. An advantageof undertaking reactions in microfluidic systems is that enhancedreaction efficiency may be achieved due to a high phase contactarea-to-volume ratio in micro-channels.

Based on the above formaldehyde analysis, the methodol-ogy of microfluidic derivatization was further improved for thedirect analysis of some other eight gaseous carbonyls at low mix-ing ratios < 10 ppbv by using a GC–quadrupole MS instead of theprevious GC–ToF (time of flight) MS. The performance of themicrofluidic derivatization technique was tested by attemptingto simultaneously determine gaseous carbonyls in the EUropeanPHOtoREactor (EUPHORE) (http://euphore.es/), a highly instru-mented, large-scale outdoor reaction chamber located in Valencia,Spain. The microfluidic derivatization technique is compared herewith FTIR instrumental observation on the temporal evolutionof carbonyls formed during the photo-oxidation of isoprene. Acomprehensive intercomparison of different techniques includ-ing PTR-MS, FTIR, offline analysis and broadband cavity enhancedabsorption spectroscopy (BBCEAS) methods is given in a separatepublication, along with modelling and mechanistic details.

2. Experimental

2.1. Microfluidic lab-on-a-chip layout

The modular glass microfluidic chip was fabricated by a spe-cialised manufacturer of micro-scaled glass devices (DolomiteCentre, UK) using hydrogen fluoride etching and thermal

bonding processes. The layout of the microfluidic chip isshown in Fig. 1a with a size of 90 mm × 45 mm × 4.5 mm(length × width × thickness), which is similar to half the size of aniPhone mobile phone. The microfluidic chip was installed with a

e output, mixing junctions, and 208 cm reacting microchannel (size: between gaseous dicarbonyls (blue flow) and PFBHA solution (red flow) in, the reader is referred to the web version of the article.)

96 X. Pang et al. / J. Chromatogr. A 1296 (2013) 93– 103

F ve sol

celrfimb6traac

tapddPtMfsnai(ataclG

2

Secp

ig. 2. Schematics of the microfluidic lab-on-a-chip derivatization system. Derivati

hip holder and chip header to allow quick connection to 1/16 in.xternal diameter tubing through its three inlets and one out-et (Fig. 1b). Sample gas and the solution of the derivatizationeagent were passed into the splitting channel (hemispherical pro-le with 620 �m internal diameter and 2.1 cm length) and thenixed through the mixing junctions. The reaction micro-channel

etween the mixing junctions and output (circular profile with20 �m internal diameter and 208 cm length) provides space andime for the reaction between gaseous carbonyls and derivatizationeagent to occur. The potential maximum operating temperaturend pressure of the chip were 300 ◦C and 30 bar, respectively, butll experiments here were conducted under much less extremeonditions (60 ◦C and less than 1.5 bar).

In this study the microfluidic chip integrates three key func-ions: (1) a gas and liquid mixer and reactor, (2) reagent heating,nd (3) sample pre-concentration. The gaseous carbonyls pre-ared in EUPHORE chamber or test gases in the chamber wererawn into the microreactor through inlet 2 and inlet 3 by a KNFiaphragm gas pump (PM20994-022, Neuberger, Germany). TheFBHA derivatization solution was introduced into the microreac-or simultaneously through inlet 1 by a peristaltic pump (Watson

arlow 205S, UK). The microreactor was fixed securely on the sur-ace of a hotplate head for direct heating. The carbonyls in the gastream reacted with PFBHA with high efficiency in the microchan-el through diffusion into the derivatization solution which formed

thin wall coated layer on the channel walls, analogous to a cap-llary chromatography stationary phase [21]. A stainless metal coil50 cm in length) was connected to the outlet of the microreactor,nd placed in an ice–salt bath with a temperature of 0 ◦C. The reac-ion solution eluting from the microreactor outlet was cooled tovoid the evaporation loss of reaction solution by hot gases. Afterold trapping the reaction solution from the stainless coil was col-ected directly into a sample vial within the auto-sampler of theC–MS (Fig. 2), for subsequent detection by GC–MS.

.2. Materials and apparatus

All chemicals unless otherwise stated were purchased from

igma–Aldrich. Solvents including acetonitrile, methanol, ethanol,thyl acetate, 2-propanol, hexane (HPLC grade) were pur-hased from Fisher (UK). The derivatization reagent O-(2,3,4,5,6-entafluorobenzyl) hydroxylamine (PFBHA) (99%) was used

ution was collected directly from micro-reactor and ready for GC–MS analysis.

directly in this study without any further purification. The flowrates of the sample carbonyl gases and other tested gases fromthe EUPHORE chamber were pumped into the microreactor usinga clean KNF diaphragm gas pump (PM20994-022, Neuberger,Germany). Flow rate of sample gas was controlled by a massflow controller (Model GFC17, Aalborg, USA), itself calibrated bya gas flow meter (Alicat Scientific, USA). Flow of the deriva-tization solution was generated and controlled by a peristalticpump, varying from 20–120 �L min−1. A magnetic stirrer hotplate(RH Basic 2 IKAMAG, UK) with a contact thermometer was usedto heat the microchip directly. Separation and detection of thecarbonyl-PFBHA derivatives were performed on a GC–MS sys-tem incorporating a Perkin Elmer (USA) Auto system XLGC and aTurbo Mass (USA) quardrupole MS equipped with a DB5 column(60 m × 0.25 mm × 1.0 �m, length × internal diameter × film thick-ness). GC conditions were as follows: the GC oven temperature wasinitially set at 80 ◦C for 2 min, programmatically ramped to 300 ◦Cwith a temperature ramp of 8 ◦C min−1 and then held at 300 ◦C for1.5 min. The solvent delay was set at 10 min to avoid possible dam-age to the MS detector. The GC injection mode was set as splitless.The temperatures of GC inlet and GC–MS transfer line were keptat 280 ◦C. The mass spectrometer was operated in scan mode witha mass range of 100–500 Da to identify the most abundant ions ofcarbonyl derivatives. The selected ion chromatograms of the mostabundant ion were used to quantify the concentration of deriva-tives in solution. In this study the most abundant ion ([C6F5CH2

•]+)with m/z = 181 Da, which is a fraction from the PFBHA molecule,was identified for all carbonyl-PFBHA derivatives.

2.3. Description of EUPHORE chamber and instrumentation

All experiments in this study were performed in the Euro-pean Photo-Reactor (EUPHORE), located in Valencia, Spain duringthe period of 28 July to 2 August 2012. EUPHORE consistsof two hemispherical FEP (fluorine ethene propene) chambersmounted on aluminium floor panels, which are described in detailelsewhere [73–75]. The FEP foil is highly transparent even to short-

wavelength sunlight. The chamber used in the present study hasa volume of 200 m3 and is protected by retractable hemispheri-cal shutters when it is not in use or the presence of sunlight isnot required. The chamber is operated at ambient temperature and

atogr.

al

ra5c(pl5caob2biTw2a

2

tsagwwawi

MmsvsGwpwtAmgasthgcSdrf

TO

X. Pang et al. / J. Chrom

pproximately atmospheric pressure using purified air, while twoarge fans provide homogeneous mixing of the chamber contents.

The chamber is equipped with a multi-pass White-Cell mir-or system for long path in situ monitoring of gaseous reactantsnd products by FTIR spectrometry. The spectrometer (NICOLET50, MCT/B-detector) is coupled to a long-path multi-reflectionell installed in the chamber for detection in the IR spectral range400–4000 cm−1). The gold-coated mirrors of the cell allow a totalath length of 616 m. Spectra are collected with 1 cm−1 reso-

ution by averaging 280 scans, resulting in a sampling time of min per spectrum. Concentrations of GLY and MGLY in the smoghamber during the experiments were determined with improvednalysis software using calibrated reference spectra collected a pri-ri [76]. GLY and MGLY were monitored in the aldehydic C Hand, within the spectral region of 2750–2900 cm−1 for GLY and920–2750 cm−1 for MGLY. For the calibration of gases measuredy FTIR, GLY was injected into the EUPHORE chamber with a syringe

n repeated trials while MGLY was cross-calibrated against DOAS.he linear range of FTIR measurement for GLY is 2.5 ppbv–3.5 ppmvith MDL 2.5 ppbv and for MGLY is 2.7 ppbv–4.2 ppmv with MDL

.7 ppbv. The accuracy is estimated to be 8% for GLY measurementnd 15% for MGLY measurement for FTIR in this study.

.4. Preparing gaseous carbonyls for calibrations

Preparing standard carbonyl gases is of foremost importanceo establish calibration curves for accurate measurements. In thistudy the standard gases of eight carbonyls including acetaldehyde,cetone, glyoxal (GLY), glycoaldehyde, hydroxyacetone, methyllyoxal (MGLY), methyl vinyl ketone (MVK), methacrolein (MACR),ere prepared in the EUPHORE chamber. Gaseous GLY and MGLYere prepared together as mixed standard whilst hydroxyacetone

nd MVK were prepared simultaneously. Glycoaldehyde and MACRere prepared in one batch while acetaldehyde and acetone were

n the same group.The detailed description on the preparation of gaseous GLY and

GLY can be seen in our another paper focusing on GLY and MGLYeasurements [77]. In brief the GLY gas was prepared from the

olid trimer-dihydrate (95%) using the methods described in pre-ious literature [78] and MGLY originated from its 40% aqueousolutions according to previous descriptions [79]. Freshly preparedLY and MGLY (green-yellowish liquid) were collected in the flaskshich were stored in liquid nitrogen before experimental use. Theure GLY or MGLY was introduced into the EUPHORE chamber,hich was full of clean air, by passing a small flow of nitrogen

hrough their cold-trap whilst allowing the trap to warm gently.fter the addition, the chamber’s contents were homogenouslyixed by the fans and stabilised for 5 min. A stable GLY and MGLY

aseous mixture inside the chamber was obtained, which was useds the calibration standard gas for the microfluidic derivatizationystem for GLY and MGLY measurement. Then the standard gas inhe chamber was sampled and derivatized in the microreactor foralf an hour. After the first sampling period the standard carbonylas was then diluted down to a lower concentration by addinglean air and the chamber contents allowed to mix thoroughly.

econd sampling and derivatizating of the standard gases were con-ucted for another half an hour. The same process was conductedepeatedly at least five times or more, to obtain enough data pointsor establishing calibration curves. Therefore, the downward steps

able 1ptimal experimental conditions for microfluidic derivatization reaction between PFBHA

PFBHA concentration Temperature (◦C) Flow rat

5 × 10−4 mol L−1 60 300

A 1296 (2013) 93– 103 97

provide a series of gas phase calibration standards of GLY and MGLYcovering the full range of GLY or MGLY concentrations accessedduring the whole experiment in this study. We used the differentconcentrations of GLY and MGLY to test our instrument’s responseand linearity, and assigned absolute values to the GLY and MGLY inthe chamber using values derived from FTIR co-measurements. Inthe EUPHORE chamber FTIR has been used to accurately measureGLY and MGLY for many years with stable performance with MDLof 2.5 ppbv for GLY and MDL of 2.7 ppbv for MGLY in an integratetime of 5 min.

The processes to prepare other gaseous carbonyls were the sameas those of GLY and MGLY. Glycoaldehyde dimer (solid phase) wasdirectly employed to produce gaseous glycoaldehyde in the cham-ber. It was added into one glass flask, which was heated by a hairdryer to help it evaporate, and introduced into the chamber bya flow of clean air. Liquid acetaldehyde (purity ≥ 99.5%), acetone(≥99.9%), MVK(≥99%), MACR(≥95%), and hydroxyacetone (≥95%)were injected into different glass flasks by a syringe and thenintroduced into EUPHORE chamber by a clean air flow. To confirmthe complete evaporation of the liquid carbonyls the flasks wereheated by a hair dryer during the introduction process. The ini-tial concentrations (unit: ppbv) of gaseous carbonyls in EUPHOREwere calculated as the volume of gaseous carbonyls into cham-ber/the chamber volume (200 m3). Their following concentrationsafter dilution in chamber were calculated according to their dilu-tion ratios.

3. Results and discussions

According to our previous research [21], PFBHA was foundto offer several advantages over PFPH as derivatization reagentincluding higher sensitivity, much better chromatographic sepa-ration, and more chemical stable derivatives, oximes. Therefore,PFBHA was chosen as derivatization reagent instead of PFPH in thisstudy.

The solvent, PFBHA concentration, microchip temperature, flowrates of gas and solution, reaction time, can all affect the reactionefficiency and reproducibility of the derivatization reaction and anoptimisation of the parameters is required to achieve the highestefficiency of derivatization from each micro-preparative system. Atthe same time to achieve a good separation, chromatographic con-ditions were optimised to the values described in Section 2.2. Wehave described previously the optimisation process and parame-ters for formaldehyde measurement [21] and have undertaken asimilar optimisation here. The detailed discussions on their effectson GLY and MGLY analysis can be seen in our another paper onGLY and MGLY measurements [77]. The optimal condition for thederivatization in microfluidic chip is show in Table 1.

3.1. Solvent selection for PFBHA

PFBHA is a water-soluble derivitisation reagent for carbonyl-containing compounds. However water is not a suitable solventto dissolve PFBHA in this study since aqueous solutions cannot beeasily injected directly into the capillary GC column. Instead, sev-

eral organic solvents, including acetonitrile, ethanol, ethyl acetate,hexane, methanol, and 2-propanol were tested as the carrier sol-vent and their influences on the measurement were studied. TheGC chromatograms of PFBHA in different solvents are shown in

and gaseous carbonyls in microreactor.

e of dicarbonyls (mL min−1) Flow rate of solution (�L min−1)

40

98 X. Pang et al. / J. Chromatogr. A 1296 (2013) 93– 103

F thyl ac thanol

FoPwMiocta

c

ig. 3. Chromatograms of PFBHA in different solvents (Panel a: acetonitrile; Panel b: ean be found in ethyl acetate solution. Acetaldehyde and acetone can be found in e

ig. 3. A pronounced peak of formaldehyde PFBHA derivative wasbserved in the GC chromatogram of the methanol solution ofFBHA while two important peaks of acetaldehyde derivativesere found in both the ethanol solution and ethyl acetate solution.ethyl ethyl ketone and butyraldehyde derivatives were observed

n the GC chromatograms of ethyl acetate solution. Those traces ofther carbonyl compounds were found in several PFBHA solventombinations, which potentially interfere with the derivatiza-

ion reaction between the target carbonyls such as formaldehyde,cetaldehyde, acetone, etc. and PFBHA.

Acetonitrile and hexane were not found to contain enougharbonyl compounds to produce any derivative peaks in the GC

C6F5CHONH2 + HCOCHO2k1k-1

F5C6CH2O

N C

C

H

N

HOCH2C

F5C6CH2ON

PFBHA Glyoxal

Pentaflu

Fig. 4. Two-step process of derivatization reactions of GLY with PFB

cetate; Panel c: ethanol; Panel d: methanol). Methyl ethyl ketone and butyraldehyde solution. Formaldehyde was contained in methanol solution.

chromatograms of PFBHA solutions. Since PFBHA dissolves con-siderably more readily in acetonitrile than in hexane due to itspolar properties, acetonitrile was chosen as the optimal solvent toprepare the derivitisation solution in this study.

3.2. GC chromatograms and MS spectra of derivatives

The reaction between carbonyls and PFBHA to form the

corresponding pentafluorobenzyl oximes is a two-step process,illustrated in Fig. 4. In the first step of the reaction, the PFBHAreacts with the carbonyls through nucleophilic attack to form anamino alcohol. The second step is dehydration of the amino alcohol,

6F5

H CH

OH

CH

HO

NHOCH2C6F5

k2

+

F5C6CH2O

N C

C

H

N

HOCH2C6F5

Aminoalcohol

orobenzyloxime

HA to form their corresponding derivative isomers – oximes.

X. Pang et al. / J. Chromatogr. A 1296 (2013) 93– 103 99

Fig. 5. GC–MS chromatogram (panel a) of PFBHA derivatives of GLY and MGLY and their EI mass spectra (panel b for GLY at 26.38 min and panel c for MGLY at 27.02 min).F ify thew in we

wbafHocoppo

ragmentation ion (m/z 181 Da), which is the most abundant ion, is chosen to quantith retention time of 26.38 min and MGLY derivative with retention time of 27.0 m

hich leads to formation of oximes [40,80]. Some isomers of car-onyl derivatives will be formed, especially for dicarbonyls, GLYnd MGLY, which contain two carbonyl groups. Two isomers areormed for GLY and four isomers are potentially formed for MGLY.owever in practice, as seen in the GC chromatograms (Fig. 5a),nly two of the four MGLY isomers are formed in significant con-entrations. The retention times of one GLY derivative and one

f the observed derivative of MGLY are so close that their twoeaks overlap and are not resolved by the current instrument. Thishenomenon has been previously reported in published studiesn derivatization of GLY and MGLY [2]. Therefore only the GLY

concentration of derivatives to obtain highest sensitivity. Only the GLY derivativere used to quantify derivative concentration of each dicarbonyl.

derivative with retention time of 26.4 min and MGLY derivativewith retention time of 27.0 min were used to quantify the concen-tration of each dicarbonyl. The MS spectra of the two isolated peaksdemonstrate that the two carbonyl groups of dicarbonyls have bothbeen derivatized with molecular ion peaks of m/z = 448 Da for GLYderivative and m/z = 462 Da for MGLY derivative (Fig. 5b and c).The ion with m/z 181 Da is identified as the most abundant ion

for the two derivatives and other carbonyl derivatives (Fig. 5b andc). Therefore, the selected ion chromatograms of the most abun-dant ion, m/z = 181 Da, were used to quantify the concentration ofderivatives to obtain high sensitivity in this study. For some other

100 X. Pang et al. / J. Chromatogr. A 1296 (2013) 93– 103

Table 2GC retention times and method detection limits (MDL) of carbonyl-PFBHA derivatives.

No. Carbonyls Linear formula MWa Structure MW-derivativesb RTc (min) MLDd (ppbv) Linear rangef (ppbv)

1 Formaldehyde HCHO 30

HH

O225 11.3 1.1e 4.0–331e

2 Acetaldehyde CH3CHO 44O

239 13.4*g, 13.2 0.70 20–200

3 Acetone CH3COCH3 58O

253 14.6 0.80 125–500

4 Methacrolein H2C C(CH3)CHO 70 O 265 16.0 0.20 7.8–31.2

5 Methyl vinyl ketone CH2 CHCOCH3 70O

265 16.2 0.40 37.5–199

6 Hydroxyacetone CH2(OH)COCH3 74O

OH269 18.9*, 18.7 0.50 37.6–201

7 Glycoaldehyde OHCH2CHO 60HO

O255 18.1*, 17.8 1.10 10–43.7

7 Glyoxal HOCCHO 58

O

O448 26.4*, 26.5 0.08 5.4–405

8 Methylglyoxal OCCH3CHO 72

O

O462 26.7*, 26.5 0.20 2.1–226

a Molecular weight.b Molecular weight of carbonyl-PFBHA derivatives.c Retention time of carbonyl-PFPH derivatives.d Method detection limit.e The results obtained from previous work [21].

oncen

ciids

3

tgtrmstc

iMomtfh1cottt

f Linear range is based on the actual measurement in this study.g The asterisk indicates the main peak of isomer used to quantify the derivative c

arbonyls including acetaldehyde, hydroxyacetone, glycoaldehyde,somers were observed in their GC chromatograms and only onesomer peak in GC was used to quantify its concentration. Theetailed GC chromatograms and MS spectra information for thetudied carbonyls are listed in Table 2.

.3. Method calibrations

The method calibration curves for the microfluidic derivatiza-ion technique were established based on measuring a series ofaseous carbonyl mixtures of different concentration prepared inhe EUPHORE chamber under the optimal conditions of micro-eactor shown in Table 1. Only the GLY and MGLY standard gasixtures of different concentrations in the chamber were also

imultaneously determined by FTIR and other carbonyl concen-rations were calculated based on the known mass amount ofarbonyls injected into the chamber.

As Fig. 6 shows, all studied carbonyls show a good linearityn their measurement ranges and the calibration curves of GLY,

GLY, acetone and MVK show a better linearity than those ofther carbonyls with their R-squared values larger than 0.99. Theethod detection limits (MDLs) for the microfluidic derivatiza-

ion method are estimated to be 0.08 ppbv for GLY, 0.20 ppbvor MGLY, 0.20 ppbv for MACR, 0.40 ppbv for MVK, 0.50 ppbv forydroxyacetone, 0.70 ppbv for acetaldehyde, 0.80 ppbv for acetone,.10 ppbv for glycoaldehyde, 0.2 ppbv for MGLY, which are cal-ulated from the baseline signal plus 3 times standard deviation

f the S/N of the blank sample in the GC chromatograms. Sincehe acetone, acetaldehyde and glycoaldehyde can be observed inhe blank samples the MDLs for the three carbonyls are higherhan others. The comparisons of the microfluidic derivatization

tration.

technique and other methods reported in previous literatureson GLY and MGLY measurement are listed in Table 2. Signifi-cant experimental work has emerged on GLY and MGLY in thelast decade due to their crucial roles in atmospheric secondaryorganic aerosol (SOA) formation. The detection limit of our tech-nique is comparable with most spectrometric methods [81–83]and a little higher than the other derivatization method [24,80].However, the short time resolution, simple treatment processand automatization of microfluidic derivatization are balancingadvantages.

3.4. Application and comparisons with FTIR

To test the performance of the microfluidic derivatizationtechnique for time resolved measurements, the technique wasemployed to determine gaseous GLY and MGLY concentrationsduring an experiment in the EUPHORE chamber to study the oxi-dation mechanism of isoprene under low NOX (NO < 2 ppbv, NO213–21 ppbv) conditions. The microfluidic derivatization techniquewas compared with the FTIR technique for GLY and MGLY measure-ments, which are shown in Fig. 7.

Compared with the FTIR technique (Fig. 7), the microfluidictechnique displays similar results for both MGLY and GLY mea-surements over most of the reaction period. The only inconsistencybetween the two techniques is that there are two higher con-centration data points of MGLY (12.4 ± 3.9 and 7.4 ± 1.4 ppbv)measured by the microfluidic technique than those (6.9 ± 0.55 and

4.0 ± 0.3 ppbv) by FTIR during the period of 240–280 min. The GLYconcentrations measured by microfluidic technique are consistentwith those detected by FTIR during the experimental period as awhole (Table 3).

X. Pang et al. / J. Chromatogr. A 1296 (2013) 93– 103 101

Fig. 6. Calibration curves of GLY, MGLY and acetone (panel a) measured by the microfluidic derivatization technique on the optimal conditions and calibration curves ofacetaldehyde, hydroxyacetone, glycoaldehyde, MVK, and MACR. (panel b). GLY, MGLY, acetone and MVK show a better linearity than those of other carbonyls with theirR-squared values larger than 0.99.

0

4

8

12

16

0 50 100 150 200 250 300 350 400 450

Co

nce

ntr

ati

on

(p

pb

V)

Reacti on Time (mi nute)

FTIR GLY

FTIR MGLY

Microflui dic GLY

Microflui dic MGLY

Fig. 7. Comparison of temporal evolutions of GLY and MGLY measured by microflu-idic derivatization technique and FTIR during isoprene photo-oxidation under lowNOx conditions in the EUPHORE chamber. Error bars for FTIR imply 8% accuracy(uncertainty) for GLY measurement and 15% accuracy for MGLY measurements byFTIR; Error bars for microfluidic derivatization technique indicate 14.7% accuracy(uncertainty) for GLY measurement and 22% accuracy for MGLY by this technique.

Table 3Comparison microfluidic derivatization technique with other methods in linearrange and method detection limit (MLD).

Analytical technique GLY MGLY Ref.

MLD (ppbv) MLD (ppbv)

Microfluidic 0.08 0.2 This studyBBCEAS 0.05 0.4 [84]Mad LIP 0.02 – [85]IBBCEAS 0.13 – [81]LED-CE-DOAS 0.03 0.255 [82]

DOAS 0.15 – [83]PFBHA-mist chamber 0.003 0.01 [80]

4. Conclusions

A sensitive and relatively rapid microfluidic lab-on-a-chip deriv-itization technique for gaseous carbonyls has been developed basedon microfluidic derivitisation of carbonyls and pre-concentration of

reaction solution in a glass micro-reactor followed by GC–MS anal-ysis. The microchannels of the reactor enhance the phase contactarea-to-volume ratio, and allow direct heating of the on-chip reac-tion. This combination of small volumes and higher temperatures

1 atogr.

ccpircdcmtildfmnbf

A

mStftR(mpoioiatEgtt(f

R

[[[[

[[

[[[[[[

[[[[

[[[

[[

[[[

[[

[[

[

[

[[

[[[[

[

[

[[[

[

[[[[[[[

[

[

[

[

[[

[

[

[[[

[[[

[

[

02 X. Pang et al. / J. Chrom

reates a much faster and higher efficiency derivitization betweenarbonyls and PFBHA, without requiring any additional unwieldyreparative steps. Effluent leaving the microreactor can be directly

njected into a standard bench-top GC–MS without further prepa-ation. Calibration curves show a good linearity for all studiedarbonyls over the 1–200 ppb measurement range. The methodetection limits (MDLs) are around 0.10 ppbv for most studiedarbonyls, which implies this technique, can be used to directlyeasure the carbonyls in ambient air. The technique was employed

o determine GLY and MGLY during controlled isoprene oxidationn the EUPHORE smog chamber. The agreement of temporal evo-utions of GLY and MGLY measured by microfluidic lab-on-a-chiperivatization technique and FTIR provides some initial evidenceor the effectiveness of the technique in practical applications. The

icrofluidic derivitisation technique provides significant opportu-ities to build on the specificity of derivatizing carbonyl methodsut at low cost, without hands-on preparative steps and allows foraster reaction times and on-line coupling with GC–MS.

cknowledgements

The authors acknowledge support from the Natural Environ-ent Research Council (NERC) and the Engineering and Physical

ciences Research Council (EPSRC) for support of the instrumen-ation used in this paper. Xiaobing Pang acknowledges supportrom the Chinese National Natural Science Foundation (41175110),he CAS/SAFEA International Partnership Program for Creativeesearch Teams and the National Centre for Atmospheric ScienceNCAS) for travel and logistical support. The EUPHORE experi-

ents were carried out as part of the NERC funded “Pho-SOA”roject (NE/H021108/1). We gratefully acknowledge the supportf Andrew Rickard and Maria Baeza Romero in enabling our partic-pation in the PhoSOA experiments at EUPHORE, and for allowingur use of the chamber calibration experiments to enable thisnstrument development. EUPHORE instrumentation and methodsre partly funded by the Spanish Ministry of Science and Innova-ion, through INNPLANTA project PCT-440000-2010-003 and theuropean Community’s Seventh Framework Program under therant agreement no. 228335 (Eurochamp2). The Instituto Universi-ario CEAM-UMH is partly supported by GeneralitatValenciana, andhe projects GRACCIE (Consolider-Ingenio 2010) and FEEDBACKSPrometeo - Generalitat Valenciana). We thank Prof. Lucy Carpenteror kindly providing laboratory and instrumental supports.

eferences

[1] D.R. Knapp, Handbook of Analytical Derivatization Reactions, Wiley Inter-science, New York, 1979.

[2] J. Yu, H.E. Jeffries, R.M. Le Lacheur, Environ. Sci. Technol. 29 (1995) 1923.[3] C.F. Poole, A. Zlatkis, Anal. Chem. 52 (1980) 1002A.[4] W.V. Ligon, H. Grade, Anal. Chem. 63 (1991) 255.[5] S.-M. Wang, R.J. Lewis, D. Canfield, T.-L. Li, C.-Y. Chen, R.H. Liu, J. Chromatogr. B

825 (2005) 88.[6] S.S.H. Ho, J.Z. Yu, Environ. Sci. Technol. 38 (2003) 862.[7] M.C. Prieto-Blanco, P. López-Mahía, P. Campíns-Falcó, J. Chromatogr. A 1218

(2011) 4834.[8] C.-E. Banos, M. Silva, Talanta 77 (2009) 1597.[9] J.S. Herrington, Z.-H. Fan, P.J. Lioy, J. Zhang, Environ. Sci. Technol. 41 (2006) 580.10] X. Zhou, G. Huang, K. Civerolo, J. Schwab, Environ. Sci. Technol. 43 (2009) 2753.11] Y. Shi, B. Bajrami, X. Yao, Anal. Chem. 81 (2009) 6438.12] X. Yao, Anal. Chem. 83 (2011) 4427.13] H. Shen, A.P. McNichol, L. Xu, A. Gagnon, B.G. Heikes, Anal. Chem. 81 (2009)

6310.14] J.L. Little, J. Chromatogr. A 844 (1999) 1.15] V. Zaikin, J. Halket, A handbook of derivatives for mass spectrometery, IM Pub-

lications, Chichester: England, 2009.16] J.M. Halket, V.G. Zaikin, Eur. J. Mass Spectrom. (Chichester, Eng.) 9 (2003) 1.

17] A.R. Katritzky, H.Y. He, K. Suzuki, J. Org. Chem. 65 (2000) 8210.18] V.G. Zaikin, J.M. Halket, Eur. J. Mass Spectrom. (Chichester, Eng.) 9 (2003) 421.19] S. Sechi, B.T. Chait, Anal. Chem. 70 (1998) 5150.20] J.M. Halket, V.V. Zaikin, Eur. J. Mass Spectrom. (Chichester, Eng.) 10 (2004) 1.21] X. Pang, A.C. Lewis, Anal. Methods 4 (2012) 2013.

[

[

A 1296 (2013) 93– 103

22] X. Pang, A.C. Lewis, J.F. Hamilton, Talanta 85 (2011) 406.23] J. Zhang, L. Zhang, Z. Fan, V. Ilacqua, Environ. Sci. Technol. 34 (2000) 2601.24] Y.N. Lee, X. Zhou, Environ. Sci. Technol. 27 (1993) 749.25] J.G. Herman, J.R. Graff, S. Myöhänen, B.D. Nelkin, S.B. Baylin, Proc. Natl. Acad.

Sci. 93 (1996) 9821.26] V.G. Zaikin, J.M. Halket, Eur. J. Mass Spectrom. (Chichester, Eng.) 10 (2004) 555.27] R. Atkinson, Atmos. Environ. 34 (2000) 2063.28] B.J. Finlayson-Pitts, J.N. Pitts, Chemistry of the Upper and Lower Atmosphere:

Theory, Experiments, and Applications, Academic Press, London, 2000, pp. 107.29] X. Pang, X. Lee, Atmos. Environ. 44 (2011) 2098.30] H.B. Singh, L.J. Salas, R.B. Chatfield, E. Czech, A. Fried, J. Walega, M.J. Evans, B.D.

Field, D.J. Jacob, D. Blake, B. Heikes, R. Talbot, G. Sachse, J.H. Crawford, M.A.Avery, S. Sandholm, H. Fuelberg, J. Geophys. Res. 109 (2004) D15S07.

31] X. Pang, Y. Mu, Y. Zhang, X. Lee, J. Yuan, Atmos. Environ. 43 (2009) 2142.32] X. Pang, Y. Mu, Atmos. Environ. 41 (2007) 1819.33] IARC, IARC Monographs (International Agency for Research on Cancer, Lyon,

France) 83, 2004.34] X. Pang, Y. Mu, J. Yuan, H. He, Atmos. Environ. 42 (2008) 1349.35] X. Pang, X. Shi, Y. Mu, H. He, S. Shuai, H. Chen, R. Li, Atmos. Environ. 40 (2006)

7057.36] X. Pang, Y. Mu, Atmos. Environ. 40 (2006) 6313.37] Y. Feng, S. Wen, Y. Chen, X. Wang, H. Lü, X. Bi, G. Sheng, J. Fu, Atmos. Environ.

39 (2005) 1789.38] Y. Feng, S. Wen, X. Wang, G. Sheng, Q. He, J. Tang, J. Fu, Atmos. Environ. 38

(2004) 103.39] Y.J. Mu, X.B. Pang, J.N. Quan, X.S. Zhang, J. Geophys. Res. Atmos. 112 (2007),

http://dx.doi.org/10.1029/2006JD008211, D22302.40] S.S.H. Ho, J.Z. Yu, Anal. Chem. 74 (2002) 1232.41] J. Li, Y.L. Feng, C.J. Xie, J. Huang, J.Z. Yu, J.L. Feng, G.Y. Sheng, J.M. Fu, M.H. Wu,

Anal. Chim. Acta 635 (2009) 84.42] X. Zhou, K. Mopper, Environ. Sci. Technol. 24 (1990) 1482.43] K. Toda, K.-I. Yoshioka, K. Mori, S. Hirata, Anal. Chim. Acta 531 (2005) 41.44] A.P. Altshuller, L.J. Leng, Anal. Chem. 35 (1963) 1541.45] T.E. Kleindienst, T.S. Conver, C.D. McIver, E.O. Edney, J. Atmos. Chem. 47 (2004)

79.46] R.R. Miksch, D.W. Anthon, L.Z. Fanning, C.D. Hollowell, K. Revzan, J. Glanville,

Anal. Chem. 53 (1981) 2118.47] L.T. Gibson, W.J. Kerr, A. Nordon, J. Reglinski, C. Robertson, L. Turnbull, C.M.

Watt, A. Cheung, W. Johnstone, Anal. Chim. Acta 623 (2008) 109.48] P.E. Georghiou, C.K. Ho, Can. J. Chem. 67 (1989) 871.49] C.E. Bricker, W.A. Vail, Anal. Chem. 22 (1950) 720.50] K. Kawamura, K. Kerman, M. Fujihara, N. Nagatani, T. Hashiba, E. Tamiya, Sens.

Actuators B: Chem. 105 (2005) 495.51] M. Risholm-Sundman, A. Larsen, E. Vestin, A. Weibull, Atmos. Environ. 41 (2007)

3193.52] M. Jang, R.M. Kamens, Environ. Sci. Technol. 35 (2001) 3626.53] R. Ortiz, K. Enya, K. Sekiguchi, K. Sakamoto, Atmos. Environ. 43 (2009) 382.54] B. Temime, R.M. Healy, J.C. Wenger, Environ. Sci. Technol. 41 (2007) 6514.55] Y. Komazaki, Y. Narita, S. Tanaka, Analyst 123 (1998) 2343.56] J.A. Koziel, J. Noah, J. Pawliszyn, Environ. Sci. Technol. 35 (2001) 1481.57] P.A. Martos, J. Pawliszyn, Anal. Chem. 70 (1998) 2311.58] F. Reisen, S.M. Aschmann, R. Atkinson, J. Arey, Environ. Sci. Technol. 37 (2003)

4664.59] W.K. Carlton, B. Gump, K. Fugelsang, A.S. Hasson, J. Agric. Food Chem. 55 (2007)

5620.60] E. Gomez Alvarez, M. Vazquez Moreno, S. Gligorovski, H.M. Wortham, Talanta

88 (2012) 252.61] E. Gomez Alvarez, E. Borras, J. Viidanoja, J. Hjorth, Atmos. Environ. 43 (2009)

1603.62] E. Gomez Alvarez, J. Viidanoja, A. Munoz, K. Wirtz, J. Hjorth, Environ. Sci. Tech-

nol. 41 (2007) 8362.63] M. Aiello, R. McLaren, Environ. Sci. Technol. 43 (2009) 8901.64] Y. Sameenoi, K. Koehler, J. Shapiro, K. Boonsong, Y. Sun, J. Collett, J. Volckens,

C.S. Henry, J. Am. Chem. Soc. 134 (2012) 10562.65] Y. Sameenoi, P. Panymeesamer, N. Supalakorn, K. Koehler, O. Chailapakul, C.S.

Henry, J. Volckens, Environ. Sci. Technol. 47 (2013) 932.66] F. Shen, M. Tan, Z. Wang, M. Yao, Z. Xu, Y. Wu, J. Wang, X. Guo, T. Zhu, Environ.

Sci. Technol. 45 (2011) 7473.67] S. Hiki, K. Mawatari, A. Aota, M. Saito, T. Kitamori, Anal. Chem. 83 (2012) 5017.68] W.-J. Ju, L.-M. Fu, R.-J. Yang, C.-L. Lee, Lab Chip 12 (2012) 622.69] Y. Takabayashi, M. Uemoto, K. Aoki, T. Odake, T. Korenaga, Analyst 131 (2006)

573.70] Y. Ueno, T. Horiuchi, T. Morimoto, O. Niwa, Anal. Chem. 73 (2001) 4688.71] Y. Ueno, T. Horiuchi, O. Niwa, Anal. Chem. 74 (2002) 1712.72] A.C. Lewis, J.F. Hamilton, C.N. Rhodes, J. Halliday, K.D. Bartle, P. Homewood,

R.J.P. Grenfell, B. Goody, A.M. Harling, P. Brewer, G. Vargha, M.J.T. Milton, J.Chromatogr. A 1217 (2011) 768.

73] A. Munoz, T. Vera, H. Sidebottom, A. Mellouki, E. Borrás, M. Ródenas, E.Clemente, M. Vázquez, Environ. Sci. Technol. 45 (2011) 1880.

74] M. Karl, C. Dye, N. Schmidbauer, A. Wisthaler, T. Mikoviny, B. D’Anna, M. Múller,E. Borrás, E. Clemente, A. Munoz, R. Porras, M. Ródenas, M. Vázquez, T. Brauers,

Atmos. Chem. Phys. 12 (2012) 1881.

75] B.R. Klotz, S.R. Sørensen, I. Barnes, K.H. Becker, T. Etzkorn, R. Volkamer, U. Platt,K. Wirtz, M. Martín-Reviejo, J. Phys. Chem. A 102 (1998) 10289.

76] M. Ródenas, INTROP Report (2008) 27770. Avaiable fromhttp://www.ceam.es/pag/contam e archivos/MRodenasINTROPReport.pdf

atogr.

[

[

[

[

[

[[

X. Pang et al. / J. Chrom

77] X. Pang, A.C. Lewis, A.R. Rickard, M.T. Baeza-Romero, T.J. Adams, S.M. Ball, M.J.S.Daniels, I.C.A. Goodall, P. Monks, S. Peppe, M.R. García, P. Sánchez, A. Munoz,Atmos. Meas. Tech. Discuss (2013) (in press).

78] K.J. Feierabend, J.E. Flad, S.S. Brown, J.B. Burkholder, J. Phys. Chem. A 113 (2009)

7784.

79] M. Gurnick, J. Chaiken, B. Thomas, J.D. McDonald, J. Chem. Phys. 74 (1981)99.

80] R.S. Spaulding, R.W. Talbot, M.J. Charles, Environ. Sci. Technol. 36 (2002)1798.

[

[

A 1296 (2013) 93– 103 103

81] R.A. Washenfelder, A.O. Langford, H. Fuchs, S.S. Brown, Atmos. Chem. Phys. 8(2008) 7779.

82] R. Thalman, R. Volkamer, Atmos. Meas. Tech. 3 (2010) 1797.83] R. Volkamer, L.T. Molina, M.J. Molina, T. Shirley, W.H. Brune, Geophys. Res. Lett.

32 (2005) L08806.84] M.J.S. Daniels, S.R.T. Neil, J. Patel, E. Paterson, M. Rodenas, S.M. Ball, Atmos.

Meas. Tech. (2013), submitted for publication.85] A.J. Huisman, J.R. Hottle, K.L. Coens, J.P. DiGangi, M.M. Galloway, A. Kammrath,

F.N. Keutsch, Anal. Chem. 80 (2008) 5884.