lable at ScienceDirect

Atmospheric Environment 89 (2014) 453e463

Contents lists avai

Atmospheric Environment

journal homepage: www.elsevier .com/locate/atmosenv

Using total OH reactivity to assess isoprene photooxidation viameasurement and model

A.C. Nölscher a,*, T. Butler b, J. Auld a,1, P. Veres a,2, A. Muñoz c, D. Taraborrelli a,L. Vereecken a, J. Lelieveld a, J. Williams a

aDepartment of Atmospheric Chemistry, Max Planck-Institute for Chemistry, Mainz, Germanyb Institute for Advanced Sustainability Studies, Potsdam, Germanyc Instituto Universitario CEAM-UMH, Paterna, Valencia, Spain

h i g h l i g h t s

� Isoprene photooxidation was examined via total OH reactivity measurements.� Isoprene and major products accounted for the observed total OH reactivity.� The commonly used MCM 3.2 reproduced the experimental findings reasonably well.� The recent isoprene chemistry in MIME showed large discrepancies to the measurements.� Subsequent modifications improved MIME to match the measured total OH reactivity.

a r t i c l e i n f o

Article history:Received 7 October 2013Received in revised form24 January 2014Accepted 12 February 2014

Keywords:Total OH reactivityIsoprene photooxidationHydroxyl radicalMaster Chemical Mechanism MCM 3.2Mainz Isoprene Mechanism Extended MIME

* Corresponding author.E-mail address: a.noelscher@mpic.de (A.C. Nölsch

1 Now at: Department of Chemistry and BiochemWindsor, Canada.

2 Now at: NOAA Earth System Research Laboratory

http://dx.doi.org/10.1016/j.atmosenv.2014.02.0241352-2310/� 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

The Tropics provide a reactive atmospheric environment with high levels of biogenic emissions, rapidlygrowing anthropogenic influence, high solar radiation and temperature levels. The major reactivebiogenic emission is isoprene which reacts rapidly with the primary daytime oxidant OH, the hydroxylradical. This key photooxidation process has recently been the focus of several experimental andcomputational studies. A novel isoprene degradation mechanism was recently proposed (MIME) sup-plementing the commonly used MCM 3.2 scheme.

This study examined the photooxidation of isoprene in the controlled conditions of the Valencia at-mospheric reaction chamber, EUPHORE (EUropean PHOtoREactor). Besides the detection of isoprene andits major oxidation products formaldehyde, methyl vinyl ketone (MVK) and methacrolein (MACR), thetotal loss rate of OH (total OH reactivity) was measured. The total OH reactivity was compared to theindividual measurements of isoprene and its oxidation products to assess the significant contributors tothe overall OH loss rate. Measured total OH reactivity showed excellent agreement to the calculationbased on individual compounds detected by a Proton-Transfer-Reaction-Time-Of-Flight-Mass-Spec-trometer (PTR-TOF-MS). On average 97% of the measured total OH reactivity could be explained byisoprene and its major oxidation products.

Total OH reactivity was also compared to various isoprene degradation schemes to evaluate knownmechanisms. TheMCM3.2 isoprenemechanism reproduced the temporal degradation of total OH reactivity(and isoprene) reasonablywell with a 57% (and 95%) agreementwithin themodel uncertainties and a linearcurve fit slope of 0.69 (and 1.02) for a model to measurement correlation. Large discrepancies betweenmodeled values and all observed compoundswere found for the recent isoprene oxidation scheme inMIME.Possiblemechanistic reasonsarediscussedand improvementsproposed. Thesubsequentlymodifiedversionof MIME differed from the measured total OH reactivity only about 12% at the end of the experiment andrepresented best the overall temporal profile (linear curve fit slope of correlation: 0.95).

� 2014 Elsevier Ltd. All rights reserved.

er).istry, University of Windsor,

, Boulder, USA.

Fig. 1. EUHPORE chamber in Valencia while opening the roof to start the experiments.

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463454

1. Introduction

Photooxidation processes in the troposphere impact the con-centration of trace gases, ozone and aerosols, and determine theformation and destruction rates of radicals (e.g. Atkinson and Arey,2003; Kanakidou et al., 2005; Hofzumahaus et al., 2009). The mostimportant oxidant in the daytime is the hydroxyl radical (OH).While sources (predominantly radiation dependent) are commonlywell described, sinks (due to all OH reactive species) are as yetpoorly constrained (Di Carlo et al., 2004; Rohrer and Berresheim,2006; Mao et al., 2012). The sum of all OH sinks, the overall lossrate of hydroxyl radicals, is termed total OH reactivity. Recently, thedirect measurement of total OH reactivity has become an importanttool for examining the atmospheric composition, oxidative capacityand ozone formation potential (Jeanneret et al., 2001; Ren et al.,2006; Yoshino et al., 2011; Nölscher et al., 2012b; Sinha et al.,2012; Edwards et al., 2013; Nakashima et al., 2014).

Globally, isoprene accounts for almost half of the emittedbiogenic reactive volatile organic compounds (VOCs) (Guentheret al., 2006). Regionally, i.e. in tropical forested areas, it can bethe most abundant VOC (e.g. Williams et al., 2001). First mentionedby Sanadze (1956) as an emission from vegetation, it is now knownthat isoprene is released through the plant’s stomata dependent onlight and temperature (Rasmussen and Jones, 1973; Tingey et al.,1980; Lamb et al., 1987). Due to its high reactivity to OH(k(298 K)¼ 1�10�10 molecule�1 cm3 s�1, IUPAC), emission rate andOH recycling potential especially over forests, it is a driving factor oftropospheric chemistry. Its impact on radical consumption and re-generation is highly debated (e.g. Lelieveld et al., 2008;Hofzumahaus et al., 2009), hence it remains the focus of manyexperimental and computational studies (Paulot et al., 2009b;Peeters and Mueller, 2010; Crounse et al., 2011; Taraborrelli et al.,2012; Berndt, 2012; Zhang and Kamens, 2012).

Isoprene and its oxidation products have been observed innumerous ambient forest field studies (Pierotti et al., 1990;Montzka et al., 1995; Williams et al., 2001; Karl et al., 2009;Eerdekens et al., 2009). Product studies via chamber photooxida-tion experiments have identified and quantified many isoprenesecondary products such as formaldehyde (HCHO), methyl vinylketone (MVK), methacrolein (MACR), methylglyoxal, hydrox-ycarbonyls and peroxyacetyl nitrates (PANs) (Tuazon and Atkinson,1990; Paulson and Seinfeld, 1992; Kwok and Atkinson, 1995;Ruppert and Becker, 2000; Liu et al., 2013) the yields of whichdepend on available NOx.

Based on these studies, detailed degradation schemes ofisoprene were developed (Paulson and Seinfeld, 1992; Carter andAtkinson, 1996; Jenkin et al., 1997; Geiger et al., 2003) whichcould be applied to atmospheric models. The Master ChemicalMechanism (MCM), constructed as a detailed protocol for atmo-spheric degradation processes of various VOCs, includes a highlyexplicit isoprene reaction scheme (Saunders et al., 2003; Pinhoet al., 2005). For the purpose of global modeling, a condensedisoprene oxidation mechanism has been developed, based on thecomplex MCM reaction schemes. The first version of the MainzIsoprene Mechanism (MIM) included 16 organic species and 44chemical reactions (Pöschl et al., 2000).

However, atmospheric field investigations in low NOx andisoprene-rich environments revealed a significant discrepancy be-tween predicted hydroxyl radical levels and the direct measure-ments of OH (Tan et al., 2001; Lelieveld et al., 2008; Hofzumahauset al., 2009; Whalley et al., 2011). Within the Tropics, coexistence ofhigh isoprene and high OH levels has puzzled scientists. Reasons forthis can be twofold: the measurement of the OHmixing ratio couldhave been impacted by interferences (e.g. Mao et al., 2012) or theapplied atmospheric model did not reproduce atmospherically

relevant photooxidation mechanisms correctly (e.g. Butler et al.,2008).

Based on theoretical treatments, novel isoprene photooxidationchannels were suggested and some have been included in atmo-spheric models (Peeters et al., 2009; Paulot et al., 2009b; Wolfeet al., 2012). Taraborrelli et al. (2012) extended MIM in order toinvestigate this novel isoprene chemistry in detail, implementing900 species and 4000 reactions into MIME (MIM Extended). Key tothe novel isoprene chemistry is the recycling of OH due to isom-erization of isoprene peroxy radicals. The rate of isomerization andhence the importance for atmospheric chemistry is highly debated.Suggested values range from extremely fast (k � 1 s�1 at 303 K,Peeters et al., 2009) to relatively slow (k z 0.002 s�1 at 295 K,Crounse et al., 2011). Please note, that Crounse et al. (2011) deter-mined the isomerization rate experimentally by considering theentire pool of peroxy radicals. Whereas Peeters et al. (2009) pro-vided isomer specific isomerization rates based on theoreticalcalculations.

In this study, the OH-initiated isoprene photooxidation wasstudied in the controlled environment of the EUPHORE (EUropeanPHOtoREactor). Chamber Experiments Examining Reactivity andSpecies (CHEERS) were set up to compare the two isopreneoxidation mechanisms, MCM and MIME, with measurements ofindividual species and total OH reactivity.

The individually measured or modeled compounds were con-verted to OH reactivities, through their mixing ratios [X] and therate coefficients with OH (kxþOH), which sum up to the totalcalculated OH reactivity (Equation (1)). For the reaction of isopreneand OH, an overall decreasing total OH reactivity is expectedbecause isoprene, as the predominant contributor to the total OHsink, is converted into less reactive secondary products (Atkinsonand Arey, 2003; Karl et al., 2006; Nakashima et al., 2012).

Rtotal;calc ¼X

½X� � kXþOH (1)

In this way the total OH reactivity, calculated, modeled anddirectly measured, functioned as diagnostic tool to evaluateisoprene degradation by OH under controlled conditions, anddifferent views (models) could be compared by this measurablemetric for a composite mixture of reactive VOCs.

2. EUPHORE and instrumentation

The EUPHORE chamber in Valencia (Siese et al., 2001) providedan excellent reaction vessel for studying atmospheric

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463 455

photooxidation processes during the CHEERS campaign in autumn2011. A typical experiment started in the morning with backgroundmeasurements from the empty chamber, after flushing it forcleaning purposes. The compound of interest was injected (exceptfor the empty-chamber-background experiment) and kept insidethe closed dark chamber (for ca. 15 min) to homogeneously mixthrough the entire volume. Then the roof over the 200 m3 halfspherical Teflon chamber was opened to illuminate the chamberand start the experiment (Fig. 1). Natural irradiation initiated bothOH production from HONO emitted from the walls (Rohrer et al.,2005; Zador et al., 2006), and the photooxidation of the com-pounds inside the chamber. Experiments were planned to last for6 h and were centered on local noon.

The instrumentation provided by the EUPHORE teammonitoredchamber temperature, pressure and relative humidity, ozone, NO2,NO, HONO and CO mixing ratios. Humidity was kept low to limitthe formation of aerosol during the study. Detailed informationabout instrumentation and chamber conditions can be found in theSupplementary Information (Table S1, Fig. S1). A Proton-Transfer-Reaction-Time-Of-Flight-Mass-Spectrometer (PTR-TOF-MS, ION-ICON) followed the development of the examined VOC and theevolution of some of the reaction products (e.g. Jordan et al., 2009).Compounds with a higher proton affinity thanwater can be ionizedby the reaction with H3Oþ, accelerated in the drift tube and sepa-rated by mass using a high resolution TOF. Configured in V-modewith a mass resolution of approximately 3700 m/Dm, the instru-ment allows for example the compoundsmethylglyoxal (m/z 73.03)and methyl ethyl ketone (m/z 73.07) to be distinguished. A detaileddiscussion of the PTR-TOF-MS data analysis methods is describedelsewhere (Mueller et al., 2013, and references therein). The PTR-TOF-MS was calibrated using a standard gas mixture (Apel-Riemer Environmental, 2011) and a portable permeation source e.g.for formaldehyde (Veres et al., 2010) (Table 1).

The total OH reactivity was measured by the ComparativeReactivity Method (CRM), an instrumental technique which com-pares the reaction of a single reagent with OH alone and then in thepresence of all the reactive molecules from the chamber (Sinhaet al., 2008; Nölscher et al., 2012a). As reagent, pyrrole (C4H5N)was used and the detectionwas conducted using a quadrupole PTR-MS (pyrrole detection at m/z 68). The total OH reactivity methodwas tested by introducing known amounts of calibration gases(such as propene and isoprene). In addition, the initial closedchamber concentration of the compound introduced by sweepingan accurately measured liquid volume into the chamber was usedto examine the sensitivity and linearity for the CRM. The analysis ofthe CRM total OH reactivity data included a humidity (Sinha et al.,2010) and a chemical correction (Sinha et al., 2008). The techniqueis sensitive to NO and might underestimate the true total OHreactivity due to internal recycling of OH via NO and HO2. Sinha

Table 1PTR-TOF-MS measured mass, the corresponding compounds, overall uncertainty, detectiItalicized compounds have been used to constrain the model to EUPHORE conditions.

Mass Compound Uncertainty[%]

m/z 31 formaldehyde 6m/z 33 methanol 18m/z 45 acetaldehyde 6m/z 59 acetone 13m/z 61 acetic acid, glycolaldehyde 11m/z 69 isoprene 3m/z 71 methyl vinyl ketone,

methacrolein4

m/z 73.03 methylglyoxal 8m/z 75 hydroxyacetone 16m/z 83 methylfuran 11

et al. (2008) determined a threshold of about 10 ppbV NO beforethe total OH reactivity shows effects of interference. The appliedinstrument has been tested in a previous campaign for an effect ofNO and a threshold of 4 ppbV was observed. During the isopreneoxidation experiment NO was always measured to be below thedetection limit of the instrument (<2.7 ppbV) and no interferencewas expected. Additionally, due to the extremely high concentra-tions of isoprene the sample was diluted before the CRM. Thisfurther reduced the NO concentration inside the instrument byabout a factor of 2.

The overall uncertainty in the total OH reactivity measurementsfor this experiment was calculated as a propagation of errorsincluding the uncertainty of the gas standard (5%), the reaction ratecoefficient for the reaction of pyrrole with OH (14%), PTR-MS error(5%), flow variability (2%), and an additional combined uncertaintyoccurring in post measurement corrections (18%). These includedthe correction for dilution of the sampled air and chamber effects,that occurred during the CHEERS experiments. In total, the mea-surement uncertainty was 24%, and the limit of detection (2s of thebaseline noise) was 4 s�1. Both instruments, the PTR-TOF-MS andthe CRM, were located directly underneath the chamber floor withshort inlet lines (<4 m).

3. Modeling CHEERS

Simulations of the CHEERS experiments were performed usingthe photochemical box model MECCA (Sander et al., 2005)employing the chemical mechanisms of the MCM version 3.2 andMIME (Jenkin et al., 1997; Saunders et al., 2003; Pöschl et al., 2000;Taraborrelli et al., 2012). The version of MECCA used has beenpreviously described in Butler et al. (2011).

For the present study, the model was further modified tosimulate EUPHORE chamber conditions according to the in-structions given on the MCM website (http://mcm.leeds.ac.uk/MCM/MCMChamber.htt):

(a) reactions were added to account for processes occurring onthe chamber walls;

(b) all deposition velocities were set to zero;(c) photolysis rates were calculated based on observed J (NO2)

accounting for transmission through the chamber walls,backscatter from the chamber floor and cloud cover;

(d) and a dilution rate based on observations of SF6 was appliedto all model species.

For each experiment, the model was initialized with theobserved mixing ratio of the VOC of interest as measured by PTR-TOF-MS. A constant mixing ratio of 1.8 ppmV was assumed forCH4. Time-dependent boundary conditions representing conditions

on limit (LOD ¼ 3s of background), calibration method and calibrated compounds.

LOD[ppbV]

Calibration method,compound

0.4 Permeation tube, formaldehyde0.9 Gas standard, methanol0.2 Gas standard, acetaldehyde1.9 Gas standard, acetone2.4 Permeation tube, acetic acid0.3 Gas standard, isoprene0.1 Gas standard, MVK

0.08 Adopted calibration factor for acetone0.5 Permeation tube, hydroxyacetone0.1 Permeation tube, furan

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463456

in the chamber and sources, and sinks of species on the chamberwalls were applied to the model. These boundary conditions arebased on measurements of temperature, pressure, humidity,J(NO2), O3, CO, HONO, and NO2, as well as formaldehyde, methanol,acetaldehyde, acetone, and hydroxyacetone measured by PTR-TOF-MS (Table 1).

Measurements of HONO were only available for the backgroundexperiment due to instrument malfunction. For simulations of theisoprene oxidation experiment, a range of HONO boundary condi-tions was employed based on values of 80%e120% of the mea-surements from the background experiment.

The chemiluminescence detector used for the NO2 measure-ments is known to be affected by interferences from othernitrogen-containing compounds (e.g. Winer et al., 1974). For thisreason the NO2 boundary condition was used to constrain themodeled sum of NO2 and all PAN-like species. The identity of thenitrogen-containing molecules desorbing from the chamber wallswas unknown. To account for this uncertainty, it was assumed to bethe relatively unstable peroxyacetylnitrate, and a range of simula-tions was performed where the constraint was varied between0 and 100% of the NO2 reported by the chemiluminescencedetector.

Measured values of NO during the experiment remained belowthe instrumental detection limit of 2.7 ppbV (3s) at all times. Anupper limit for an NO concentration boundary condition in themodel was determined by comparing the modeled and measuredNO2 as a range of different concentration boundary conditions forNO. In this way we calculated an upper limit of 1.5 ppbV of NO as aconcentrations boundary condition. Higher values of NO led tomodeled values of NO2 which were inconsistent with the measuredNO2. To explore the uncertainty in themodel due to the unsensitivemeasurement of NO in the chamber, we present simulations with aconcentration boundary condition for NO of 1.5 ppbV, and with noconcentration boundary condition for NO (see Fig. 2).

Model results for each mechanism are reported as a range ofvalues based on a set of model runs done using the abovementioned ranges of boundary conditions for HONO, NO2, and NO.

4. The MCM 3.2 and MIME in detail

The principal idea and a detailed schematic of both basicmechanisms (MCM 3.2 and MIME) are presented in Fig. 3.

Fig. 2. Measurements of (A) HONO from the background experiment and (B) NO2 fromthe isoprene oxidation experiment were used as time-dependent input for the model.The boundary limits are represented by the green shading. The resulting calculated NOmixing ratio is presented in (C) as green line (just using the detected values of HONOand NO2) with green shading to indicate the model variability due to the uncertaintiesof the measurements. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

All mechanisms used in this study describe the initial isoprenereaction with OH similarly: The hydroxyl radical adds to the twodouble bonds of isoprene, in MCM 3.2 on position 1 and 4, and inMIME also on position 2 and 3 (Fig. 3). Oxygen subsequently reactswith the adduct to form peroxy radicals, which are represented inMIME via an isomer specific chemistry. According to Peeters et al.(2009) the reaction with oxygen can be easily reversed, which ispart of the more novel isoprene chemistry included in MIME.Assuming low NOx conditions, further reactions of the peroxyradical involve mainly HO2 and other peroxy radicals (RO2). Typicalproducts present in both mechanisms are HCHO, MVK, MACR,acetaldehyde, methylglyoxal, glyoxal and other carbonyls. Severalmajor products have been measured in this study, and have beenused to evaluate the model mechanism.

Additionally included inMIME is the isomerization of four of theperoxy radicals, namely the 1,5 and 1,6 H-shifts (Peeters et al.,2009). While the 1,5 H-shift is thought to immediately generateHCHO,MVK, MACR and OH, the 1,6 H-shift should produce HO2 andunsaturated hydroxyperoxideealdehydes (HPALDs).

As presented in Taraborrelli et al. (2012), HPALDs in theisoprene-OH-chemistry may act as buffer for the tropical OH con-centrations. Therefore, implemented in the isoprene degradationscheme of MIME are two reaction paths: Depending on ambient OHlevels, HPALDs either are predominantly oxidized by OH, or theyformOH through photolysis (Peeters et al., 2009;Wolfe et al., 2012).MIME explicitly treats all reaction paths to undergo further reactionincluding 1,5 and 1,6 H-shifts following Peeters and Mueller (2010).Many of the applied rate coefficients or energy barriers for H-shiftshave been calculated computationally (e.g. in Kuwata et al., 2007;Peeters et al., 2009; Nguyen et al., 2010). Experimental evidencefor HPALDs, recently reported by Crounse et al. (2011), found sub-stantially slower rates for the 1,6 H-shift than assumed in themodelmechanisms. An overall 1,6 H-shift isomerization rate constant of(k z 0.002 s�1 at 295 K) was determined from the production rateof HPALDs (assuming that all peroxy radicals form HPALDs) albeitwith a high uncertainty (about 50%). This overall isomerization rateconstant would imply an average isomer specific rate constant ofabout 0.04 s�1 (298 K, Table S3). Further chemical products havebeen studied in low NOx by Paulot et al. (2009b), who was able todetect isoprene-epoxides.

The main difference of MIME to the MCM 3.2 isoprene mecha-nism is a more detailed treatment of many reactions of isoprene andit’s secondary products. For example in the MIME isoprene chemis-try, initial OH addition is also allowed on position 2 and 3. Reversereactions between alkyl and peroxy radicals involving oxygen areimplemented as well as the complete degradation pathway of allisoprene products. Additionally, OH generation is initialized throughisomerization of various products in the mechanism.

This study presents MIME in three different versions:

(a) MIME: As described by Taraborrelli et al. (2012), this versionimplements the 1,6 H-shift with a rate constant of k¼ 0.5 s�1

at 298 K. This isomerization rate was derived theoreticallyand applied to both isoprene peroxy radical isomers.

(b) MIME*: Additional OH generation through isomerization inthe recent isoprene-chemistry of MIME was disabled toquantify its impact. All 1,5 and 1,6 H-shifts were switched offin the mechanism.

(c) MIME�: According to recent findings in the literature, andguided by an extensive set of new theoretical calculations, twomajor pathways in the degradation mechanism have beenmodified (Fig. 3). Alongwith the 1,6 H-shifts the followingwaschanged: (Ia) Two isomer specific rate coefficients for the1,6 H-shifts were calculated to be k¼ 0.09 s�1 and k¼ 0.33 s�1

(at 298 K) (Taraborrelli et al., 2012). The modeling of the

Fig. 3. Schematic of the isoprene-OH oxidation mechanism as implemented in MCM 3.2 and MIME: The conventional accepted isoprene chemistry (MCM 3.2, http://mcm.leeds.ac.uk/MCM/MCMChamber.htt) includes the addition of OH to the two double bonds on position 1 and 4, subsequential reactions with oxygen and HO2, RO2, NO or permutations. Thismechanism is pictured in black. Additional in the MIME mechanism (in red) is the OH addition to positions 2 and 3 as well as the back-reaction with O2, and the 1,5 and 1,6 H-shifts(Peeters et al., 2009; Taraborrelli et al., 2012). Through these additional reaction paths, OH can be recycled. In blue typical measured products are highlighted. Recently also theepoxides have been detected by Paulot et al. (2009b). Modifications of MIME� as described in Section 4 are highlighted in roman numbers. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463 457

Crounse et al. (2011)-results with MIME showed betteragreement with these lower isomerization rates. High levelquantum chemical calculations persisted in the marked dif-ference between the two isomer specific rate coefficients.Interestingly, the lower rate coefficient is for the most abun-dant isoprene peroxy radical (that is the one fromOH-additionto position 1 in Fig. 3). In total this reduces the yield of HPALDs.(Ib) Only 33% of the 1,6 H-shifts produce directly HPALDs, therest may form other products and eventually HPALDs, whichoverall decreases the HPALD yield (Crounse et al., 2011, addi-tion). (Ic) Products of HPALD photolysis may undergo a 1,5 H-shift which was set to be limited by a higher energy barrier(3.5 kcal mol�1 higher than the analogous 1,6 H-shift). Thisresults in less OH production per HPALD.

Additional changes to increase the overall production onMVK&MACR in the model: (II) Based on results from Liu et al.(2013) and Paulot et al. (2009b), the isoprene peroxy radicalreactions with HO2 have low yields in dry conditions. Berndt(2012) showed that MVK(MACR) production increases withhumidity. This was taken into account in MIME� by changing

reactions of isoprene peroxy radicals with HO2 to partiallygenerate OH and alkoxy radicals. This is consistent withrecent findings of Paulot et al. (2009b), Berndt (2012), and Liuet al. (2013).

5. Results

The controlled environment of the EUPHORE reaction chamberwas used for investigating the photooxidation of isoprene. Thedegradation of isoprene and the formation of major reactionproducts were monitored with time and subsequently compared tothe CRM total OH reactivity measurements and model simulations.

5.1. Oxidation of isoprene

High levels of isoprene (197 � 5 ppbV, PTR-TOF-MSdata � instrumental uncertainty) were injected into theEUPHORE chamber and photooxidation initiated. Isoprene deple-tion and the formation of products such as formaldehyde (HCHO),methyl vinyl ketone and methacrolein (detected at m/z 71 as

Table 2Initial and final levels of individually measured compounds by PTR-TOF-MS � instrumental uncertainty as well as the OH reactivity of each compound. Onthe right, the relative contribution of each compound to the total (calculated) finalOH reactivity after 6 h of oxidation.

Compound Initial Final

Mixingratio[ppbV]

OHreactivity[s�1]

Mixingratio[ppbV]

OHreactivity[s�1]

Relativecontribution[%]

Formaldehyde 1.5 � 0.2 0.3 61.7 � 1.3 12.9 6.4Methanol 20.3 � 4.2 0.5 95.9 � 16.4 2.1 1.1Acetaldehyde 0.1 � 0.1 0.0 15.7 � 1.3 5.8 2.9Acetone 1.0 � 0.7 0.0 9.0 � 0.8 0.0 0.0Isoprene 197.2 � 4.6 485.0 61.8 � 1.4 152.0 75.8MVK&MACR 0.3 � 0.1 0.1 46.8 � 1.0 23.1 11.5Methylglyoxal 0.0 � 0.0 0.0 1.1 � 0.1 0.4 0.2Hydroxyacetone 0.3 � 0.2 0.0 6.6 � 0.7 0.5 0.2Methylfuran 0.0 � 0.0 0.0 2.5 � 0.2 3.7 1.9

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463458

MVK&MACR), and methylglyoxal were monitored by PTR-TOF-MS(Fig. 4, top panel). Isoprene mixing ratios decreased over the 6 hreaction time to 31% of the initially introduced value. Formaldehydeand MVK&MACR, which are major products of the isoprenephotooxidation, increased significantly to 62 � 1 ppbV and47 � 1 ppbV (Table 2).

5.2. Total OH reactivity comparison and temporal trend

The total OH reactivity perspective highlights the relevance ofdetected components to the OH-initiated isoprene photooxidation.By converting each component’s mixing ratio into individual OHreactivities, highly reactive and abundant compounds are empha-sized over species that contribute only minimally to the total OHsink term. Measured total OH reactivity decreased from initially512 � 31 s�1 (10 min averaged data � standard deviation) to202 � 9 s�1 after 6 h of reaction. Initially isoprene was added as theonly reactive compound in the chamber, accounting for 485 s�1

(Fig. 4, lower panel). This is lower than the measured total OHreactivity (relative contribution: 95%) but agrees within the un-certainty of the instruments. The initial background total OHreactivity (prior to the introduction of isoprene) was found to bearound 4 s�1. This was mostly caused by CO as can be seen from themeasurements as presented in Fig. S1. However, after addition of

Fig. 4. Isoprene photooxidation experiment: Mixing ratios of isoprene (black) andoxidation products (HCHO (green), MVK&MACR (red), and methylglyoxal (blue)) weremeasured by PTR-TOF-MS and compared to simulated mixing ratios by the MCM 3.2degradation scheme (top panel). The model has been constrained to formaldehyde.Shaded areas depict the variability of the model results due to various simulationsperformed in the given uncertainty of HONO and NO2 (detailed in Section 3). Mediummodel values are presented as solid colored lines. Total OH reactivity during oxidationin the chamber was measured via CRM (dots and markers with error bars), calculatedfrom single compound measurements (PTR-TOF-MS) (black solid line) and modeled bythe MCM 3.2 (gray line and shaded area) (lower panel). Standard deviations of total OHreactivity mean values are given as error bars. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

isoprene CO accounts for less than 1% of the measured total OHreactivity which is within the uncertainty of the observations.

In Table 2 initial and final single compound OH reactivities (andmixing ratios) are summarized. Throughout the experimentisoprene dominated the total OH reactivity. After 6 h reaction time,isoprene was still the greatest contributor and accounted for 75.8%of the total (calculated) OH reactivity. Themost relevant products interms of total OH reactivity were MVK&MACR (11.5%), formalde-hyde (6.4%) and acetaldehyde (2.9%). Hence, the sum of isopreneand its most dominant reaction products accounts for 97% of thetotal measured OH reactivity. This agreement was observed be-tween the total calculated OH reactivity and the CRM measure-ments throughout the duration of the experiments as seen in Fig. 4(lower panel).

5.3. Model comparison and evaluation via measurements

Figs. 4e7 compare measured results with modeled predictions.Traditional isoprene chemistry (MCM 3.2) and the recently pro-posed degradation scheme MIME, in different versions, have beenimplemented in the photochemical box model MECCA. These re-sults are summarized in Table 3. Part A of Table 3 evaluates theagreement of measurement and model throughout the entireexperiment. The results are presented as percentage of measure-ment values that fall into the uncertainty boundaries of the model(shaded regions in Figs. 4e7). Part B of Table 3 presents a com-parison of results after 6 h reaction time as model to measurementratio (also as a percentage). Part C of Table 3 provides the slopes of alinear curve fit between medium model values and measurements(adopted from Fig. S3). These are a measure for the match of thetemporal decay.

Initialized with measured values of isoprene, the MCM 3.2reproduced the general trend of isoprene (Fig. 4, top panel) and themeasured total OH reactivity (Fig. 4, lower panel). This can be notedas well from Table 3 (and Fig. S3) as the slope of the linear fit curvebetweenmodel andmeasurement results which is 1.02 for isopreneand 0.69 for the total OH reactivity. Shaded areas in Fig. 4 indicatethe range of possible model results due to uncertainty in theboundary conditions of HONO and NO2 as was described in Section3. Within this model uncertainty, isoprene measurements andmodel agreed very well. About 95% of the measured isoprenevalues fell into the shaded model area. Final measured valuescompared well with the medium values of the model (solid line)within 7% and hence agreed within the instrumental uncertainty.

In the first two hours, isoprene was slightly overestimated bythe model. This coincided with the time that products started to

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463 459

build up in the chamber. While the model was constrained toobserved HCHO, MVK&MACR and methylglyoxal were simulatedwith a great range of possible uncertainty due to the variability inthe model constraining compounds HONO and NO2. MeasuredMVK&MACR and methylglyoxal mixing ratios agreed within themodel uncertainties about 83% and 78% for the entire experiment.Final measured MVK&MACR levels were underestimated after 6 hreaction time by about 32% when compared to medium modelvalues. Observed levels of methylglyoxal were overestimated by48% in the final model results. Measured and calculated total OHreactivity agreed within the model uncertainty at the end of theexperiment. However, after 6 h reaction time, medium modelvalues overestimated the final total OH reactivity by about 29%.

The extended isoprene degradation chemistry, implemented inMIME, was modeled in similar fashion and compared to the mea-surements in Figs. 5e7. The first presented version of MIME,significantly underestimated isoprene levels and total OH reactivityfor most of the experiment (from ca. 1 h elapsed time onwards).MIME depleted isoprene very fast, so that after 2 h reaction timeonly 6.5% of the initial isoprene was simulated in the chamber andthe model total OH reactivity was dominated by a multitude ofsecondary products. Modeled MVK&MACR and methylglyoxallevels could neither reproduce the observed levels nor the trend.Final levels of MVK&MACR were significantly underestimated by afactor of about 7. Final detected methylglyoxal was overestimatedby MIME by a factor of about 5.

MIME* was used to generate the results presented in Fig. 6. Allisomerization processes were selectively switched off. No 1,5 and1,6 H-shifts were allowed to recycle additional OH. Following thismodification, excellent agreement for isoprene occurred after 3 hreaction time until the end of the experiment. Similarly to theMCM

Fig. 5. PTR-TOF-MS (top panel) and CRM (lower panel) measurements were comparedto the MIME mechanism. Please see Fig. 4 caption for layout details which areconsistent.

3.2 results, 95% of the measured isoprene values were within themodel uncertainty, the shaded area. Final model medium valuesandmeasurements of isoprene agreed (Table 3). MVK&MACR levelswere slightly elevated in MIME*, compared to MCM 3.2, closing thegap to the measured values. About 74% of the final measuredMVK&MACR levels were accounted for by medium model resultsafter 6 h reaction time. In contrast to MIME, the modeled total OHreactivity by MIME* agreed within the uncertainties of model andmeasurements. MIME* total OH reactivity results after 6 h reactiontime showed a better match to the observations than MCM 3.2results by overestimating the measurements about 15%. The slopefor the correlation between total OH reactivity measurements andmedium model values improved slightly to 0.78 (Table 3 andFig. S3).

The results of MIME� are presented in Fig. 7. The match of allmeasured compounds (except of methylglyoxal) to MIME�

improved significantly when compared to the original version ofMIME. Similar to MCM 3.2 and MIME*, MIME� calculated too lowfinal MVK&MACR values but showed an overall agreement ofmeasurements within the model uncertainty of 73% (Table 3). Thedecay of isoprene could not be simulated correctly (slope of cor-relation curve: 1.24). While at the beginning model and measure-ment matched, towards the end of the experiment isoprene wasdepleted too much in MIME�. Overall only 42% of the isoprenemeasurements overlapped with the range of possible model re-sults. However, final values of total OH reactivity measurementsand model agreed well within 12% which was the closest match ofall tested mechanisms. Also, the decay of total OH reactivity wasreproduced best by this modified version of the mechanism whichcan be seen from the correlation fit curve slope of 0.95 (Table 3).

Fig. 6. PTR-TOF-MS (top panel) and CRM (lower panel) measurements were comparedto the modified MIME* mechanism. Parts of the novel isoprene chemistry wereswitched off: The 1,5 and 1,6 H-shifts were disabled in this model run. Please see Fig. 4caption for layout details which are consistent.

200

150

100

50

0Isop

rene

,HC

HO

,MV

K&

MA

CR

[ppb

V]

6543210

Elapsed time [hrs]

20

15

10

5

0

Methylglyoxal [ppbV

]

measurements (PTR-TOF-MS) model (MIME°)

600

500

400

300

200

100

0

Tota

l OH

reac

tivity

[s-1

]

6543210

Elapsed time [hrs]

measurements - raw data (CRM) measurements - 10 min averages (CRM)

calculation (PTR-TOF-MS) model (MIME°)

Fig. 7. PTR-TOF-MS (top panel) and CRM (lower panel) measurements were comparedto a newly proposed version of MIME - MIME� in which the 1,6 H-shift isomerizationrate is isomer specific and lower, the HPALD yield of 1,6 H-shifts is <1, the 1,5 H-shift ofHPALD products is limited and overall more MVK&MACR is produced. Please see Fig. 4caption for layout details which are consistent.

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463460

However, MIME� mediummodeled values showed an overall offsetto the measured total OH reactivity.

Between the different isoprene photooxidation mechanismstested the overall loss of isoprene peroxy radicals varied signifi-cantly. As can be seen from Fig. 3 reactions are possible with RO2,HO2 and NO. Additionally, isomerization via H-shifts (1,5 and 1,6)and ring closure (not explicitly presented in Fig. 3) depletes theperoxy radicals which are formed during the isoprene oxidation.

Table 3Three criteria for the comparison of model and measurements: A) Agreement ofmeasurements and model results within the model uncertainty for the entireexperiment. The percentage of measured values that fall within the model vari-ability is given here. B) Comparison of the measurements to modeled final resultsafter 6 h reaction time. Model to measurement ratios were calculated and given aspercentage. C) Slope for linear curve fit between model and experimental output(Fig. S3) as measure for agreement in the temporal trend.

Total OH reactivity Isoprene MVK&MACR Methylglyoxal

AMCM 3.2 57% 95% 83% 78%MIME 0.01% 11% 32% 8%MIME* 54% 95% 94% 47%MIME� 23% 42% 73% 11%BMCM 3.2 129% 107% 68% 148%MIME 35% 0.4% 14% 517%MIME* 115% 100% 74% 213%MIME� 88% 62% 71% 433%CMCM 3.2 0.69 1.02 0.35 e

MIME e e e e

MIME* 0.78 1.05 0.4 e

MIME� 0.95 1.24 0.5 4.87

Table S2 in the Supplementary Information presents the differentpartitioning of peroxy radical losses for all tested mechanisms. TheMCM 3.2, which does not implement chemistry including isomer-ization, calculated NO as major loss term for medium and upperlimit NO conditions (60e92%). Due to the high uncertainty in themeasured NO levels (see Fig. 2), also low levels of NO (<100 pptV)were possible. In this case most peroxy radicals were lost by thereaction with HO2 (56%).

MIME included possible loss processes via the 1,5 H-shift, 1,6 H-shift and ring closure isomerization. As a result, almost 50% of theperoxy radicals were lost to these isomerization reactions. The1,6 H-shift was generally themost dominant isomerization reactionand a sink for about 22e39% of the peroxy radicals. Thus, the 1,6 H-shift was as important for the RO2 loss as the reaction with HO2(17e41%). The results in Fig. 5 and Table 3 show that this mecha-nism did not reflect the observations. Therefore two modifiedversions of MIME have been tested. In MIME* the 1,5 and 1,6 H-shifts were switched off. As a result, the peroxy radical loss pro-cesses were very similar to the traditional chemistry as imple-mented in MCM 3.2. MIME� includes several substantialmodifications as were described in Section 4. In this mechanism,less peroxy radicals were lost to isomerization (in total 14e38%)and less OH was produced. The major loss reaction for low andmedium NO levels (<300 pptV) was with HO2.

6. Discussion

6.1. Chamber observations of total OH reactivity

During CHEERS, the measurement of isoprene and its majorsecondary products via PTR-TOF-MS could explain the entiremeasured total OH reactivity within the experimental uncertainty.In other words, missing or unaccounted for OH reactivity which hasbeen found in field measurements (e.g. Di Carlo et al., 2004;Nölscher et al., 2012b; Edwards et al., 2013; Nakashima et al.,2014) was not observed.

The presented experiment was conducted in high isoprene andwithout the addition of NO (1.5 ppbV). This is in contrast to themajority of chamber experiments, in which high NO mixing ratios(about several hundred ppbV) have been used e.g. Galloway et al.(2011) or Paulot et al. (2009a). However, ambient conditions e.g.for pristine tropical rain forest provide conditions with even lowervalues of NO (<100 pptV).

Dominant for total OH reactivity throughout the 6 h reactionwas isoprene. Less than 25% of the final total OH reactivity wascaused by oxidation products, out of which MVK&MACR were themost important compounds besides HCHO. This finding can becompared to recently reported total OH reactivity measurementsfrom an isoprene oxidation experiment in which the reaction wasinitiated in elevated NOx levels. Nakashima et al. (2012) used LIFbased total OH reactivity measurements and FT-IR (Fourier Trans-form Infrared Spectroscopy) results, and found a fraction of almost40% missing OH reactivity. The gap between measurement andcalculation increased in the course of the reaction at different rates.During the first hour of reaction, the reported missing total OHreactivity increased rapidly (although being within the error of themeasurement) when isoprene and NO were present at very highlevels (816 and 195 ppbV respectively). Afterwards, for the rest ofthe photooxidation, missing OH reactivity stayed rather constant.The authors argue, that different products formed in the presenceof high and low NO (present at the beginning and the end of theexperiment, respectively). Nakashima et al. (2012) concluded, thatproducts formed in high NO conditions caused missing OH reac-tivity, but the ongoing reaction in decreased NO conditions did notadd any further missing OH reactivity. This is in agreement with the

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463 461

results presented here, which did not show missing OH reactivitywhen initially no NOx was added. It seems that in the presence ofNO the different composition of isoprene oxidation productsinduced missing OH reactivity, e.g. owing to many nitrate-compounds that were not measured during the Nakashima et al.(2012)-experiment.

6.2. Isoprene photooxidation mechanisms for CHEERS

While measured and calculated total OH reactivity agreed well,the comparison of model and measurements revealed large dis-crepancies between the applied model degradation schemes. Theisoprene oxidation scheme implemented in the MCM 3.2, showedgood agreement to the observations. The recent chemistry in MIMEfailed for the presented study.

MIME has additional, chemical OH sources which are visible inFig. 5 (and Fig. S2). High OH levels rapidly destroy isoprene andlater the isoprene oxidation products. Compared to the MCM 3.2,MIME generated OH levels were about an order of magnitudehigher and reachedmaximumvalues ahead of the peak in radiation(which typically was observed at 3 h reaction time). Possibly, in theabsence of available rate measurements, the rates for the isomeri-zation steps have been estimated too high. Since these isomeriza-tions occur several times in the course of the degradation scheme,they add to each other, and form unrealistically high OH (seeFig. S2). Estimated from the temporal decay of isoprene OH con-centrations range from 5.8� 104 to 8.9� 105 molecule cm�3. MIMEreaches peak medium peak values of 9.3 � 106 molecule cm�3.Consequently, isoprene was rapidly depleted leading to a fast decayof total OH reactivity which in the end of the experiment wasdominated by oxidation products. These compounds were neitherobserved by the total OH reactivity measurement nor by the PTR-TOF-MS detecting the individual compounds.

Two further versions of MIME have been tested in order tobetter reproduce the observed results from the CHEERS experi-ment. Main objectives were to reduce the overall OH yield and toimprove the model-measurement match for MVK&MACR. In oneversion (MIME*) the 1,5 and the 1,6 H-shift have been selectivelyswitched off. Overall this mechanism showed excellent agreementto the measurements. The detailed treatment of the isoprenechemistry apart from OH recycling via isomerization improvedespecially the final results of total OH reactivity, isoprene andMVK&MACR after 6 h reaction time when compared to the tradi-tional chemistry of the MCM 3.2 (Table 3).

In the second version (MIME�) several changes have beenimplemented and tested. In accordance with recent findings andnew theoretical calculations (detailed in Section 4), the mechanismhas been explicitly modified to produce less OH and increase theyield of MVK&MACR. These improvements have been achieved (1)by including lower 1,6 H-shift isomerization rates which are spe-cific for the two isomers, (2) by reducing the HPALD yield of the1,6 H-shift to less than 1, (3) by limitation of the 1,5 H-shift ofHPALD products and (4) an overall increase in the MVK&MACRproduction. A list of recently published isomerization rates is givenin the Supplementary Table S3. Differences as can be noticed i.e. toFuchs et al. (2013) may be due to differences between the modelmechanisms. In contrast to Fuchs et al. (2013) MIME� more realis-tically represents the HPALD yield from 1,6 H-shifts (likely beingless than 1) and the conservation of OH reactivity due to HPALDproducts. MIME� significantly improved compared to the firsttested version of MIME. Of all tested mechanisms it matched bestthe final total OH reactivity and its overall decay. It is the onlymechanism that captured the temporal degradation of total OHreactivity correctly, however with an offset to the measurements.

From the underestimation of the final MVK&MACR levels inboth the MCM 3.2 and MIME it seems that isoprene products weredescribed incorrectly in the model. Indeed the MCM has beenconstructed based onmostly high NOx chamber experiments whichtypically show higher yields for MVK&MACR than low NOx studies(e.g. Galloway et al., 2011; Ruppert and Becker, 2000). Recently, anumber of studies conducted isoprene oxidation experiments aswell in lower NOx conditions (Crounse et al., 2011; Fuchs et al.,2013). For example Fuchs et al. (2013) had maximum 300 pptV ofNO in the chamber with initial isoprene concentrations up to11 ppbV.

Alternatively, the underestimation of final MVK&MACR levelscould be possibly caused by biased measurements via the PTR-TOF-MS. For low NOx conditions recent findings Liu et al. (2013) suggestan interference on the m/z 71 which is used for the detection ofMVK&MACR. They propose that isoprene products, such as hydro-peroxides, might form ions inside the instrument which occur onexactly the same m/z. Indeed, parallel observations of MVK andMACR via FT-IR in the EUPHORE chamber during CHEERS, showedslightly lower levels than the PTR-TOF-MS. However, final valueswere still underestimated by the MCM 3.2 by about 27% whencomparing to the FT-IR results.

The model has been constrained to formaldehyde measure-ments made by PTR-TOF-MS. Since the proton affinity of formal-dehyde is close to that of water, humidity may cause non-linearitywhich has to be taken into account via accurate calibrations inhumid conditions. During CHEERS, measurements were calibratedusing a permeation source, and conducted in low relative humidity.Nevertheless, simultaneous FT-IR measurements showed slightlylower values (about 10%), which result in slightly lower modeledOH levels, hence higher final MVK&MACR mixing ratios (about40 ppbV). In contrast to this, isoprene was simulated correctly bythe MCM 3.2 with the given (PTR-TOF-MS measured) HCHO level.Applying the FT-IR monitored HCHO to the MCM 3.2 no significantimpact on the results has been found. Also, the measured total OHreactivity could be explained through the individual PTR-TOF-MSdetected compounds throughout the entire experiment. Thisclosure of the OH sink budget suggests that the measurement viaPTR-TOF-MS was not impacted significantly by interferences andinstrumental issues for the major products MVK&MACR and HCHO.

7. Summary and conclusions

Total OH reactivity measurements and calculations showed thatin this study all important contributors were monitored. Theisoprene degradation scheme MCM 3.2 showed 95% and 57%overlap to the measurements of isoprene and total OH reactivitywithin its variability. Discrepancies of 32% occurred for the finalresults of the major products MVK and MACR.

The recent isoprene oxidation scheme included in the previ-ously published version of MIME (Taraborrelli et al., 2012) was notable to simulate this chamber experiment. Modeled OH levels wereabout one order of magnitude too high, hence the isopreneoxidation in the model was too fast resulting in only 11% overlap ofmeasured and modeled isoprene mixing ratios. When the 1,5 andthe 1,6 H-shifts, which are responsible formost of the additional OHformation, were switched off, excellent agreement between modeland measurement was found for isoprene (95% overlap within theuncertainties). Modeling of the total OH reactivity also improvedwith 54% of the measurements falling into the model uncertaintyboundaries which is comparable to the MCM 3.2 findings. This isconsistent with new theoretical data on the rate of peroxy radicalH-shifts that indicate that the original Peeters et al. (2009) isom-erization rate was overestimated.

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463462

A more explicit treatment of the chemistry following 1,6 H-shifts and implementation of further recent findings about peroxyradical reactions in the MIME mechanism significantly improvedthe modeled yields of OH and MVK and MACR. Overall themeasured total OH reactivity was reproduced best in shape (linearcurve fit slope: 0.95) and final value (model accounted for 88% ofobservation) by this refinement of MIME.

This first CHEERS study emphasizes the need for future in-vestigations on isoprene photooxidation. Furthermore, it highlightsthe usefulness of total OH reactivity as a diagnostic tool for oxida-tion schemes. The direct measurement of total OH reactivity couldevaluate the individual measurements of isoprene and its oxidationproducts as well as different isoprene oxidation schemes.

Acknowledgments

We are grateful for the technical and logistical support by theEUPHORE team and Thomas Klüpfel, for the critical pre-review byJohn Crowley and the two anonymous referee comments. ThisEuropean project was funded by Transnational Access Activities(E2-2011-04-07-0057) from the 7th European Union FrameworkProgramme under Grant Agreement no. 228335 (EUROCHAMP-2).The Instituto Universitario CEAM-UMH is partly supported byGerneralitat Valenciana.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2014.02.024.

References

Atkinson, R., Arey, J., 2003. Atmospheric degradation of volatile organic compounds.Chemical Reviews 103 (12), 4605e4638.

Berndt, T., 2012. Formation of carbonyls and hydroperoxyenals (hpalds) from the ohradical reaction of isoprene for low-NOx conditions: influence of temperatureand water vapour content. Journal of Atmospheric Chemistry 69 (4), 253e272.

Butler, T., Lawrence, M., Taraborrelli, D., Lelieveld, J., 2011. Multi-day ozone pro-duction potential of volatile organic compounds calculated with a taggingapproach. Atmospheric Environment 45 (24), 4082e4090.

Butler, T.M., Taraborrelli, D., Brühl, C., Fischer, H., Harder, H., Martinez, M.,Williams, J., Lawrence, M.G., Lelieveld, J., 2008. Improved simulation of isopreneoxidation chemistry with the echam5/messy chemistry-climate model: lessonsfrom the gabriel airborne field campaign. Atmospheric Chemistry and Physics 8(16), 4529e4546.

Carter, W.P.L., Atkinson, R., 1996. Development and evaluation of a detailed mech-anism for the atmospheric reactions of isoprene and NOx. International Journalof Chemical Kinetics 28 (7), 497e530.

Crounse, J.D., Paulot, F., Kjaergaard, H.G., Wennberg, P.O., 2011. Peroxy radicalisomerization in the oxidation of isoprene. Physical Chemistry Chemical Physics13, 13607e13613.

Di Carlo, P., Brune, W.H., Martinez, M., Harder, H., Lesher, R., Ren, X., Thornberry, T.,Carroll, M.A., Young, V., Shepson, P.B., Riemer, D., Apel, E., Campbell, C., 2004.Missing oh reactivity in a forest: evidence for unknown reactive biogenic vocs.Science 304 (5671), 722e725.

Edwards, P.M., Evans, M.J., Furneaux, K.L., Hopkins, J., Ingham, T., Jones, C., Lee, J.D.,Lewis, A.C., Moller, S.J., Stone, D., Whalley, L.K., Heard, D.E., 2013. Oh reactivityin a south east asian tropical rainforest during the oxidant and particlephotochemical processes (op3) project. Atmospheric Chemistry and Physics 13(18), 9497e9514.

Eerdekens, G., Ganzeveld, L., Vilà-Guerau de Arellano, J., Klüpfel, T., Sinha, V.,Yassaa, N., Williams, J., Harder, H., Kubistin, D., Martinez, M., Lelieveld, J., 2009.Flux estimates of isoprene, methanol and acetone from airborne ptr-ms mea-surements over the tropical rainforest during the Gabriel 2005 campaign. At-mospheric Chemistry and Physics 9 (13), 4207e4227.

Fuchs, H., Hofzumahaus, A., Rohrer, F., Bohn, B., Brauers, T., Dorn, H.-P., Haseler, R.,Holland, F., Kaminski, M., Li, X., Lu, K., Nehr, S., Tillmann, R., Wegener, R.,Wahner, A., 2013. Experimental evidence for efficient hydroxyl radical regen-eration in isoprene oxidation. Nature Geoscience 6, 1023e1026.

Galloway, M.M., Huisman, A.J., Yee, L.D., Chan, A.W.H., Loza, C.L., Seinfeld, J.H.,Keutsch, F.N., 2011. Yields of oxidized volatile organic compounds during the ohradical initiated oxidation of isoprene, methyl vinyl ketone, and methacroleinunder high-NOx conditions. Atmospheric Chemistry and Physics 11 (21),10779e10790.

Geiger, H., Barnes, I., Bejan, I., Benter, T., Spittler, M., 2003. The troposphericdegradation of isoprene: an updated module for the regional atmosphericchemistry mechanism. Atmospheric Environment 37 (11), 1503e1519.

Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P., Geron, C., 2006. Esti-mates of global terrestrial isoprene emissions using MEGAN (model of emis-sions of gases and aerosols from nature). Atmospheric Chemistry and Physics 6(3181e3210), 3181e3210.

Hofzumahaus, A., Rohrer, F., Lu, K., Bohn, B., Brauers, T., Chang, C.-C., Fuchs, H.,Holland, F., Kita, K., Kondo, Y., Li, X., Lou, S., Shao, M., Zeng, L., Wahner, A.,Zhang, Y., 2009. Amplified trace gas removal in the troposphere. Science 324(5935), 1702e1704.

Jeanneret, F., Kirchner, F., Clappier, A., von den Bergh, H., Calpini, B., 2001. Total vocreactivity in the planetary boundary layer 1. Estimation by a pump and probeoh experiment. Journal of Geophysical Research 106 (D3), 3083e3093.

Jenkin, M.E., Saunders, S., Pilling, M., 1997. The tropospheric degradation of volatileorganic compounds: a protocol for mechanism development. AtmosphericEnvironment 31 (1), 81e104.

Jordan, A., Haidacher, S., Hanel, G., Hartungen, E., Maerk, L., Seehauser, H.,Schottkowsky, R., Sulzer, P., Maerk, T., 2009. A high resolution and high sensi-tivity proton-transfer-reaction time-of-flight mass spectrometer (ptr-tof-ms).International Journal of Mass Spectrometry 286 (2e3), 122e128.

Kanakidou, M., Seinfeld, J.H., Pandis, S.N., Barnes, I., Dentener, F.J., Facchini, M.C.,Van Dingenen, R., Ervens, B., Nenes, A., Nielsen, C.J., Swietlicki, E., Putaud, J.P.,Balkanski, Y., Fuzzi, S., Horth, J., Moortgat, G.K., Winterhalter, R., Myhre, C.E.L.,Tsigaridis, K., Vignati, E., Stephanou, E.G., Wilson, J., 2005. Organic aerosol andglobal climate modelling: a review. Atmospheric Chemistry and Physics 5 (4),1053e1123.

Karl, M., Dorn, H.-P., Holland, F., Koppmann, R., Poppe, D., Rupp, L., Schaub, A.,Wahner, A., 2006. Product study of the reaction of oh radicals with isoprene inthe atmosphere simulation chamber SAPHIR. Journal of Atmospheric Chemistry55, 167e187. http://dx.doi.org/10.1007/s10874-006-9034-x.

Karl, T., Guenther, A., Turnipseed, A., Tyndall, G., Artaxo, P., Martin, S., 2009. Rapidformation of isoprene photo-oxidation products observed in Amazonia. At-mospheric Chemistry and Physics 9 (20), 7753e7767.

Kuwata, K.T., Dibble, T.S., Sliz, E., Petersen, E.B., 2007. Computational studies ofintramolecular hydrogen atom transfers in the b-hydroxyethylperoxy and b-hydroxyethoxy radicals. The Journal of Physical Chemistry A 111 (23), 5032e5042.

Kwok, E.S., Atkinson, R., 1995. Estimation of hydroxyl radical reaction rate constantsfor gas-phase organic compounds using a structure-reactivity relationship: anupdate. Atmospheric Environment 29 (14), 1685e1695.

Lamb, B., Guenther, A., Gay, D., Westberg, H., 1987. A national inventory of biogenichydrocarbon emissions. Atmospheric Environment (1967) 21 (8), 1695e1705.

Lelieveld, J., Butler, T.M., Crowley, J.N., Dillon, T.J., Fischer, H., Ganzeveld, L.,Harder, H., Lawrence, M.G., Martinez, M., Taraborrelli, D., Williams, J., 2008.Atmospheric oxidation capacity sustained by a tropical forest. Nature 452, 737e740.

Liu, Y.J., Herdlinger-Blatt, I., McKinney, K.A., Martin, S.T., 2013. Production of methylvinyl ketone and methacrolein via the hydroperoxyl pathway of isopreneoxidation. Atmospheric Chemistry and Physics 13 (11), 5715e5730.

Mao, J., Ren, X., Zhang, L., Van Duin, D.M., Cohen, R.C., Park, J.-H., Goldstein, A.H.,Paulot, F., Beaver, M.R., Crounse, J.D., Wennberg, P.O., DiGangi, J.P., Henry, S.B.,Keutsch, F.N., Park, C., Schade, G.W., Wolfe, G.M., Thornton, J.A., Brune, W.H.,2012. Insights into hydroxyl measurements and atmospheric oxidation in acalifornia forest. Atmospheric Chemistry and Physics 12 (17), 8009e8020.

Montzka, S.A., Trainer, M., Angevine, W.M., Fehsenfeld, F.C., 1995. Measurements of3-methyl furan, methyl vinyl ketone, and methacrolein at a rural forested site inthe southeastern united states. Journal of Geophysical Research 100, 11393e11401.

Mueller, M., Mikoviny, T., Jud, W., D’Anna, B., Wisthaler, A., 2013. A new softwaretool for the analysis of high resolution ptr-tof mass spectra. Chemometrics andIntelligent Laboratory Systems 127 (0), 158e165.

Nakashima, Y., Kato, S., Greenberg, J., Harley, P., Karl, T., Turnipseed, A., Apel, E.,Guenther, A., Smith, J., Kajii, Y., 2014. Total oh reactivity measurements inambient air in a southern rocky mountain ponderosa pine forest duringbeachon-srm08 summer campaign. Atmospheric Environment 85 (0), 1e8.

Nakashima, Y., Tsurumaru, H., Imamura, T., Bejan, I., Wenger, J.C., Kajii, Y., 2012. Totaloh reactivity measurements in laboratory studies of the photooxidation ofisoprene. Atmospheric Environment 62 (0), 243e247.

Nguyen, T.L., Vereecken, L., Peeters, J., 2010. HOx regeneration in the oxidation ofisoprene iii: theoretical study of the key isomerisation of the z-d-hydroxy-peroxy isoprene radicals. ChemPhysChem 11 (18), 3996e4001.

Nölscher, A.C., Sinha, V., Bockisch, S., Klüpfel, T., Williams, J., 2012a. Total ohreactivity measurements using a new fast gas chromatographic photo-ionization detector (gc-pid). Atmospheric Measurement Techniques 5 (12),2981e2992.

Nölscher, A.C., Williams, J., Sinha, V., Custer, T., Song, W., Johnson, A.M., Axinte, R.,Bozem, H., Fischer, H., Pouvesle, N., Phillips, G., Crowley, J.N., Rantala, P.,Rinne, J., Kulmala, M., Gonzales, D., Valverde-Canossa, J., Vogel, A., Hoffmann, T.,Ouwersloot, H.G., Vilà-Guerau de Arellano, J., Lelieveld, J., 2012b. Summertimetotal oh reactivity measurements from boreal forest during humppa-copec2010. Atmospheric Chemistry and Physics 12 (17), 8257e8270.

Paulot, F., Crounse, J.D., Kjaergaard, H.G., Kroll, J.H., Seinfeld, J.H., Wennberg, P.O.,2009a. Isoprene photooxidation: new insights into the production of acids andorganic nitrates. Atmospheric Chemistry and Physics 9 (4), 1479e1501.

A.C. Nölscher et al. / Atmospheric Environment 89 (2014) 453e463 463

Paulot, F., Crounse, J.D., Kjaergaard, H.G., Kürten, A., St. Clair, J.M., Seinfeld, J.H.,Wennberg, P.O., 2009b. Unexpected epoxide formation in the gas-phasephotooxidation of isoprene. Science 325 (5941), 730e733.

Paulson, S., Seinfeld, J., 1992. Development and evaluation of a photooxidationmechanism for isoprene. Journal of Geophysical Research 97 (D18), 20,703e20,715.

Peeters, J., Mueller, J.-F., 2010. HOx radical regeneration in isoprene oxidation viaperoxy radical isomerisations. ii: experimental evidence and global impact.Physical Chemistry Chemical Physics 12, 14227e14235.

Peeters, J., Nguyen, T.L., Vereecken, L., 2009. HOx radical regeneration in theoxidation of isoprene. Physical Chemistry Chemical Physics 11, 5935e5939.

Pierotti, D., Wofsy, S.C., Jacob, D., Rasmussen, R.A., 1990. Isoprene and its oxidationproducts: methacrolein and methyl vinyl ketone. Journal of GeophysicalResearch 95, 1871e1881.

Pinho, P., Pio, C., Jenkin, M., 2005. Evaluation of isoprene degradation in the detailedtropospheric chemical mechanism, mcm v3, using environmental chamberdata. Atmospheric Environment 39 (7), 1303e1322.

Pöschl, U., von Kuhlmann, R., Poisson, N., Crutzen, P.J., 2000. Development andintercomparison of condensed isoprene oxidation mechanisms for global at-mospheric modeling. Journal of Atmospheric Chemistry 37, 29e52. http://dx.doi.org/10.1023/A:1006391009798.

Rasmussen, R., Jones, C., 1973. Emission isoprene from leaf discs of hamamelis.Phytochemistry 12 (1), 15e19.

Ren, X., Brune, W.H., Oliger, A., Metcalf, A.R., Simpas, J.B., Shirley, T., Schwab, J.J.,Bai, C., Roychowdhury, U., Li, Y., Cai, C., Demerjian, K.L., He, Y., Zhou, X., Gao, H.,Hou, J., 2006. Oh, ho2, and oh reactivity during the pmtacs-ny whitefacemountain 2002 campaign: observations and model comparison. Journal ofGeophysical Research 111 (D10S03).

Rohrer, F., Berresheim, H., 2006. Strong correlation between levels of tropospherichydroxyl radicals and solar ultraviolet radiation. Nature 442, 184e187.

Rohrer, F., Bohn, B., Brauers, T., Brüning, D., Johnen, F.-J., Wahner, A., Kleffmann, J.,2005. Characterisation of the photolytic hono-source in the atmospheresimulation chamber SAPHIR. Atmospheric Chemistry and Physics 5 (8), 2189e2201.

Ruppert, L., Becker, K.H., 2000. A product study of the oh radical-initiated oxidationof isoprene: formation of c5-unsaturated diols. Atmospheric Environment 34(10), 1529e1542.

Sanadze, G., 1956. Emission of gaseous organic substances from plants. Rep. Akad.Nauk GruzSSR 17.

Sander, R., Kerkweg, A., Jöckel, P., Lelieveld, J., 2005. Technical note: the newcomprehensive atmospheric chemistry module MECCA. Atmospheric Chemis-try and Physics 5, 445e450.

Saunders, S.M., Jenkin, M.E., Derwent, R.G., Pilling, M.J., 2003. Protocol for thedevelopment of the master chemical mechanism mcmv3 part a: troposphericdegradation of non-aromatic volatile organic compounds. Atmospheric Chem-istry and Physics 3, 161e180.

Siese, M., Becker, K.H., Brockmann, K.J., Geiger, H., Hofzumahaus, A., Holland, F.,Mihelcic, D., Wirtz, K., 2001. Direct measurement of oh radicals from ozonolysisof selected alkenes: a euphore simulation chamber study. Environmental Sci-ence and Technology 35 (23), 4660e4667.

Sinha, V., Williams, J., Crowley, J.N., Lelieveld, J., 2008. The comparative reactivitymethod e a new tool to measure total oh reactivity in ambient air. AtmosphericChemistry and Physics 8 (8), 2213e2227.

Sinha, V., Williams, J., Diesch, J.M., Drewnick, F., Martinez, M., Harder, H., Regelin, E.,Kubistin, D., Bozem, H., Hosaynali-Beygi, Z., Fischer, H., Andrés-Hernández, M.D., Kartal, D., Adame, J.A., Lelieveld, J., 2012. Oh reactivity mea-surements in a coastal location in southwestern Spain during domino. Atmo-spheric Chemistry and Physics Discussion 12 (2), 4979e5014.

Sinha, V., Williams, J., Lelieveld, J., Ruuskanen, T., Kajos, M., Patokoski, J., Hellen, H.,Hakola, H., Mogensen, D., Boy, M., Rinne, J., Kulmala, M., 2010. Oh reactivitymeasurements within a boreal forest: evidence for unknown reactive emis-sions. Environmental Science and Technology 44 (17), 6614e6620.

Tan, D., Faloona, I., Simpas, J., Brune, W., Shepson, P., Couch, T., Sumner, A.,Carroll, M., Thornberry, T., Apel, E., Riemer, D., Stockwell, W., 2001. HOx budgetsin a deciduous forest: results from the prophet summer 1998 campaign. Journalof Geophysical Research-Atmospheres 106 (D20), 24,407e24,427.

Taraborrelli, D., Lawrence, M.G., Crowley, J.N., Dillon, T.J., Gromov, S., Grosz, C.B.M.,Vereecken, L., Lelieveld, J., 2012. Hydroxyl radical buffered by isoprene oxida-tion over tropical forests. Nature Geoscience 5 (3), 190e193.

Tingey, D.T., Manning, M., Grothaus, L.C., Burns, W.F., 1980. Influence of light andtemperature on monoterpene emission rates from slash pine. Plant Physiology65 (5), 797e801.

Tuazon, E.C., Atkinson, R., 1990. A product study of the gas-phase reaction ofisoprene with the oh radical in the presence of NOx. International Journal ofChemical Kinetics 22 (12), 1221e1236.

Veres, P., Gilman, J.B., Roberts, J.M., Kuster, W.C., Warneke, C., Burling, I.R., deGouw, J., 2010. Development and validation of a portable gas phase standardgeneration and calibration system for volatile organic compounds. AtmosphericMeasurement Techniques 3 (3), 683e691.

Whalley, L.K., Edwards, P.M., Furneaux, K.L., Goddard, A., Ingham, T., Evans, M.J.,Stone, D., Hopkins, J.R., Jones, C.E., Karunaharan, A., Lee, J.D., Lewis, A.C.,Monks, P.S., Moller, S.J., Heard, D.E., 2011. Quantifying the magnitude of amissing hydroxyl radical source in a tropical rainforest. Atmospheric Chemistryand Physics 11 (14), 7223e7233.

Williams, J., Pöschl, U., Crutzen, P.J., Hansel, A., Holzinger, R., Warneke, C.,Lindinger, W., Lelieveld, J., 2001. An atmospheric chemistry interpretation ofmass scans obtained from a proton transfer mass spectrometer flown over thetropical rainforest of Surinam. Journal of Atmospheric Chemistry 38, 133e166.

Winer, A.M., Peters, J.W., Smith, J.P., Pitts Jr., J.N., 1974. Response of commercialchemiluminescent NO-NO2 analyzers to other nitrogen-containing compounds.Environmental Science & Technology 8, 1118e1121.

Wolfe, G., Crounse, J., Parrish, J., Clair, J., Beaver, M., Paulot, F., Yoon, T., Wennberg, P.,Keutsch, F., 2012. Photolysis, oh reactivity and ozone reactivity of a proxy forisoprene-derived hydroperoxynals (hpalds). Physical Chemistry ChemicalPhysics 14, 7276e7286.

Yoshino, A., Nakashima, Y., Miyazaki, K., Kato, S., Suthawaree, J., Shimo, N.,Matsunaga, S., Chatani, S., Apel, E., Greenberg, J., Guenther, A., Ueno, H.,Sasaki, H., ya Hoshi, J., Yokota, H., Ishii, K., Kajii, Y., 2011. Air quality diagnosisfrom comprehensive observations of total oh reactivity and reactive tracespecies in urban central tokyo. Atmospheric Environment 49 (0), 51e59.

Zador, J., Turanyi, T., Wirtz, K., Pilling, M., 2006. Measurement and investigation ofchamber radical sources in the European photoreactor (euphore). Journal ofAtmospheric Chemistry 55, 147e166.

Zhang, H., Kamens, R., 2012. The influence of isoprene peroxy radical isomerizationmechanisms on ozone simulation with the presence of NOx. Journal of Atmo-spheric Chemistry 69 (1), 67e81.