Evaluation of OH and HO 2 concentrations and their budgets during photo-oxidation of 2-methyl-3-butene-2-ol (MBO) in the atmospheric simulation chamber SAPHIR

. Several field studies reported unexpected large concentrations of hydroxyl and hydroperoxyl radicals (OH and HO 2, respectively) in forested environments that could not be explained by the traditional oxidation mechanisms which largely underestimated the observations. These environments were characterized by large concentrations of biogenic volatile organic compounds (BVOC) and low nitrogen oxide concentration. In isoprene-dominated environments, models developed 20 to simulate atmospheric photochemistry generally underestimated the observed OH radical concentrations. In contrast, HO 2 radical concentration showed large discrepancies with model simulations mainly in non-isoprene dominated forested environments. An abundant BVOC emitted by lodgepole and ponderosa pines is 2-methyl-3-butene-2-ol (MBO), observed in large concentrations for studies where the HO 2 concentration was poorly described by model simulations. In this work, the photooxidation of MBO by OH was investigated for NO concentrations lower than 200 pptv in the atmospheric simulation 25 chamber SAPHIR at Forschungszentrum J(cid:252)lich. Measurements of OH and HO 2 radicals, OH reactivity ( k OH ), MBO, OH precursors and organic products (acetone and formaldehyde) were used to test our current understanding of the OH-oxidation mechanisms for MBO by comparing measurements with model calculations. All the measured trace gases agree well with the model results (within 15%) indicating a well understood mechanism for the MBO oxidation by OH. Therefore, the oxidation of MBO cannot contribute to reconcile the unexplained high OH and HO 2 radical concentrations found in previous 30 field studies. gas concentrations. The model was implemented with specific chamber-related properties. First, a dilution rate was applied to all the trace gases present in the model to account for the dilution from the replenishing flow. The background production of HONO, HCHO and acetone known to occur in the sunlit chamber (Rohrer et al., 2005; Karl et al., 2006), was parametrized by an empirical function that depends on temperature, relative humidity and solar radiation. Source strengths were adjusted to match the time series of HCHO and acetone during the zero-air phase, when the chamber 40 was the only source for these species (~ 0.3 ppbv h -1 ). These chamber sources also impacted the OH reactivity measured during the zero-air phase. Ideally, after accurately accounting for the chamber sources, the OH reactivity should be well agreement between modelled and measured HO 2 radical also the agreement in the HO 2 budget increases. 40 The good agreement observed in the HO 2 budget, although partly relying on modelled species concentrations, indicates that the HO 2 production, within this chamber experiment, can be explained by alkoxy radical decomposition, photolysis of formaldehyde, and the chamber specific Y source.


Introduction
The hydroxyl radical (OH) is the most important daytime oxidant in the troposphere and its concentration affects the fate of many pollutants thus having a direct impact on the formation of ozone (O 3 ), oxygenated volatile organic compounds (OVOCs) and as such, influencing particle formation and climate. Understanding the OH radical formation and destruction 35 paths is therefore critical.
Measurements of OH radicals in environments characterized by low NO concentrations, pristine conditions and isoprene being the most abundant measured BVOC (Carslaw et al., 2001;Tan et al., 2001;Lelieveld et al., 2008;Whalley et al., 2011;Wolfe et al., 2011a), as well as in environments characterized by mixed emissions from biogenic and anthropogenic sources Lu et al., 2012;Lu et al., 2013;Tan et al., 2017), have shown a significant underestimation of 40 observed OH concentrations by state-of-the-art models. In addition, the analysis of the OH budget using only measured Atmos. Chem. Phys. Discuss., https://doi.org /10.5194/acp-2018-134 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 14 March 2018 c Author(s) 2018. CC BY 4.0 License. species obtained by comparing all known OH radical sources together with the OH radical loss rate has demonstrated that the discrepancy with model simulations is due to a large missing OH radical source . Theoretical studies have proposed new OH recycling paths which, contrary to traditional mechanisms, do not require NO for the regeneration of HO x from RO 2 radicals. The proposed mechanism involves unimolecular reactions of specific isoprene peroxy radicals (RO 2 ) (Dibble, 2004;Peeters et al., 2009;Nguyen et al., 2010;Peeters and Müller, 2010;Silva et al., 2010;Peeters and 5 Nguyen, 2012;Peeters et al., 2014). Laboratory (Crounse et al., 2011) and chamber studies (Fuchs et al., 2013) have confirmed this mechanism and have helped constraining its atmospheric impact. Contemporary, other trace gases have been investigated as the results from the isoprene studies show that OH recycling through isoprene-RO 2 , alone, is not sufficient to explain the OH concentrations observed in the field. Chamber and laboratory studies on methacrolein (MACR) Fuchs et al., 2014), methyl vinyl ketone (MVK) (Praske et al., 2015), isoprene hydroxy hydroperoxide (D' Ambro 10 et al., 2017) and glyoxal (Feierabend et al., 2008;Lockhart et al., 2013), important products from the oxidation of isoprene by OH, also have shown new OH recycling paths as predicted by theory (Peeters et al., 2009;da Silva, 2010;Setokuchi, 2011;da Silva, 2012). Further laboratory studies also have discovered OH radical recycling in the bimolecular reaction of HO 2 with acyl peroxy radicals which was previously considered to be a radical termination reaction only Groß et al., 2014;Praske et al., 2015). These results underline the need to carefully investigate the OH 15 radical budget, at least for the most abundant volatile organic compounds, to test our current knowledge.
In a similar way, the HO 2 radical concentration measured in several field campaigns performed in forested areas have shown discrepancy of measurements with model calculations highlighting an incomplete understanding of the chemistry involving formation and loss paths of HO 2 radicals. In some environments, the model tends to overestimate the measured HO 2 concentration  while in others there is the tendency to underestimate the measurements (Kubistin et al., 20 2010;Wolfe et al., 2011b;. It has been recently shown that HO 2 radical measurements performed by laser induced fluorescence (LIF) via conversion of HO 2 into OH after reaction with nitrogen oxide (NO) are likely affected by an interference that originates from organic peroxy radicals (Fuchs et al., 2011;Whalley et al., 2013;Lew et al., 2018). Therefore some of the discrepancies observed in previous studies may be partly caused by inaccurate HO 2 radical measurements. Nevertheless, recent studies where this peroxy radical interference is accounted for 25 still showed discrepancy with model calculations (Griffith et al., 2013; These recent studies, performed in environments where isoprene was not the dominant measured BVOC, were all characterized by poor agreement between modelled and measured results for both OH and HO 2 concentrations, with the measurements being up to a factor of 3 higher than the model results. Good agreement was observed between modelled and measured OH radicals when the model is constrained to the HO 2 radical measurements. These studies have concluded that there is a missing HO 2 30 source for environments where the dominant measured BVOC are monoterpenes and 2-methyl-3-butene-2-ol (MBO) . Corresponding photo-oxidation studies have been performed for -pinene in the SAPHIR chamber at Forschungszentrum Jülich. Consistent with field studies, a significant (up to a factor of 2) model underprediction of both OH and HO 2 concentrations was observed when -pinene was oxidized by OH under low-NO conditions (< 300 pptv) (Kaminski et al., 2017). The observed discrepancies in the chamber could be explained by additional HO 2 35 production, for which Kaminski et al. (2017) proposed a mechanism involving unimolecular radical reactions and photolysis of oxygenated products.
MBO is the dominant emission from lodgepole (Pinus Contorta) and ponderosa (Pinus Ponderosa) pines (Goldan et al., 1993;Harley et al., 1998). Its global emission is lower than that of isoprene (Guenther et al., 2012), but in forested areas within the West-US, MBO is the most abundant BVOC measured contributing most of the measured OH reactivity 40 (Nakashima et al., 2014;Ortega et al., 2014). In the atmosphere, MBO reacts primarily with OH forming two peroxy radicals that yield acetone, glycolaldehyde, 2-hydroxy-2-methylpropanal (HMPR) and formaldehyde after reaction with NO (Fantechi et al., 1998;Ferronato et al., 1998;Carrasco et al., 2006) (Fig. 1). The reaction of the peroxy radicals with HO 2 Atmos. Chem. Phys. Discuss., https://doi.org /10.5194/acp-2018-134 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 14 March 2018 c Author(s) 2018. CC BY 4.0 License. yields two different dihydroxy hydroperoxides (MBOAOOH and MBOBOOH, Fig. 1). Recent theoretical studies have described a mechanism that involves additional hydrogen shift reactions for the RO 2 that reforms OH and produces HO 2 (Knap et al., 2015;Knap et al., 2016). As the predicted unimolecular reaction rate following the hydrogen shift is, at most, 1.1 x 10 -3 s -1 (at 298 K and 1013 hPa) the study by Knap et al. (2016) concludes that in environments where the NO concentration is high (> 1 ppbv), the reaction between RO 2 and NO will be the dominant loss path for RO 2 radicals, and in 5 forested areas, where the NO concentration is lower than 0.2 ppbv, reactions with HO 2 and RO 2 will dominate the RO 2 fate.
In this study, the photo-oxidation of MBO initiated by OH radicals is investigated in the atmospheric simulation chamber SAPHIR in the presence of approximately 200 pptv of NO. The OH and HO 2 budget are analysed and a comparison with an up-to-date model is performed to test the current understanding of the oxidation processes of this important BVOC.

Atmospheric simulation chamber SAPHIR
The experiment performed in this study was conducted in the atmospheric simulation chamber SAPHIR at Forschungszentrum Jülich, Germany. The chamber allows for the investigation of oxidation processes and mechanisms of organic compounds at atmospheric conditions in a controlled environment. The SAPHIR chamber has a cylindrical shape 15 with a volume of 270 m 3 and is made of a double-wall Teflon (FEP) film that is inert and has a high transmittance for solar radiation (Bohn and Zilken, 2005). It is equipped with a shutter system that is opened during photolysis experiments allowing the natural solar radiation to penetrate the chamber. The air provided to the chamber is mixed from ultra-pure nitrogen and oxygen (Linde, > 99.99990 %). A fan in the chamber ensures a complete mixing of trace gases within two minutes. The pressure in the chamber is slightly higher than ambient pressure (~ 30 Pa) to avoid external air penetrating the 20 chamber. Due to small leakages and air consumption by instruments, a dilution rate of ~ 4 % h -1 was required during this study. More details regarding the chamber can be found elsewhere Poppe et al., 2007;Schlosser et al., 2007).

Experimental procedure
At the beginning of the experiment, only synthetic air was present after flushing during the night. Evaporated Milli-Q ® water 25 was first introduced into the dark chamber by a carrier flow of synthetic air containing until a concentration of ~ 5 x 10 17 cm -3 of water vapour was reached. Ozone produced by a silent discharge ozonizer (O3onia) was subsequently added to reach 40 pbbv in the chamber. This initial phase is defined as the dark phase. After opening the shutter system of the chamber, nitrous acid (HONO) was photochemically formed on the Teflon surface and released into the chamber. Its subsequent photolysis produced OH radicals and NO  during this so-called zero-air phase. Ozone (~ 40 ppbv) was injected to 30 keep the concentration of NO within a few hundreds of pptv. Afterwards, the MBO was injected three times at intervals of about two hours using a high-concentration gas mixture of MBO (~ 600 ppm, 98% from Merck) pre-mixed in a Silcosteel Canister (Restek) to reach ~ 4 ppbv of MBO in the chamber after each injection. Table 1 summarizes the instruments available during the experiment quoting time resolution, 1 accuracy, and precision for 35 each instrument. The concentrations of OH and HO 2 radicals were measured with the laser induced fluorescence (LIF) instrument permanently in use at the SAPHIR chamber and described previously (Holland et al., 2003;Fuchs et al., 2011).

Instrumentation
Recent studies have shown the possibility of interferences on the OH signal measured by LIF instruments that depend both on the chemical conditions of the sampled air and on the geometry of the different instruments (Mao et al., 2012; Rickly and Stevens, 2018). A laboratory study performed with this LIF instrument (Fuchs et al., 2016) showed only interferences for high ozone concentrations (300 ppbv -900 ppbv) together with BVOC concentrations ranging 1 to 450 ppbv which are far beyond any condition encountered in this study. Therefore, the OH radical concentration measured by the LIF instrument in this study is considered free from interferences. In addition, OH was measured by differential optical absorption spectroscopy (DOAS) (Dorn et al., 1995). Numerous intercomparisons between the LIF and the DOAS 5 instrument in the SAPHIR chamber Schlosser et al., 2009;Fuchs et al., 2012) showed very good agreement amid these two instruments giving high reliability to the OH radical measurements performed in the chamber. OH concentration measurements by DOAS in this study were on average 14% lower than those by LIF. This difference was well within the combined accuracies of measurements and was taken into account as additional uncertainty of OH concentration measurements in the analysis of this study, for which mainly OH data from the LIF instrument was used. 10 Several studies have proven that RO 2 radicals can cause an interference signal in the HO 2 radicals measured by conversion to OH after reaction with an excess of NO (Fuchs et al., 2011;Hornbrook et al., 2011;Whalley et al., 2013;Lew et al., 2017).
It was shown that a reasonable approach for avoiding the interference is to lower the concentration of NO reacting with the sampled air inside the instrument. During this study, the NO concentration (~ 2.5 x 10 13 cm -3 ) was thus adjusted to lower the interference to a minimum as described in Fuchs et al. (2011). As during the investigation of the interference from RO 2 15 originating from the oxidation of different VOCs, MBO was not tested, the amount of interference that arises from its oxidation products is not known (Fuchs et al., 2011;Whalley et al., 2013;Lew et al., 2017). An upper limit could be estimated from experiments with isoprene peroxy radicals at similar operational conditions of the instrument. The relative detection sensitivity for RO 2 radicals originated from isoprene, compared to the HO 2 signal, was among the largest of the studied peroxy radical species, with a value of 20% under the conditions of the present work (Fuchs et al., 2011). This value 20 is considered to be a reasonable estimate bias in the HO 2 radical measurements and will be considered later in the discussion.
The OH reactivity (k OH ), the inverse lifetime of OH, was measured by a pump and probe technique coupled with a timeresolved detection of OH by LIF (Lou et al., 2010;. MBO was measured by a proton-transfer-reaction time-of-flight mass spectrometer (PTR-TOF-MS, (Lindinger et al., 1998;Jordan et al., 2009)) and a gas chromatography system (GC, ) with a PTR-TOF-MS to GC ratio of 1.1 ± 0.1, and acetone by PTR-TOF-MS. 25 Formaldehyde (HCHO) was detected with a Hantzsch monitor (Kelly and Fortune, 1994), HONO with a long-path absorption photometry (LOPAP ), carbon monoxide (CO) with a reduction gas analysis instrument (RGA, ), carbon dioxide (CO 2 ), methane (CH 4 ), water vapor by an instrument applying cavity ring-down spectroscopy (CRDS, Picarro), NO, nitrogen dioxide (NO 2 ) with chemiluminescence (CL, (Ridley et al., 1992)) and O 3 by UV absorption (Ansyco). Photolysis frequencies were calculated from measurements of solar actinic radiation by a 30 spectroradiometer Bohn and Zilken, 2005).

Model calculations
A zero-dimensional box model using chemical mechanistic information from the Master Chemical Mechanism, MCM version 3.3.1 (Jenkin et al., 1997;Saunders et al., 2003) downloaded via website: http://mcm.leeds.ac.uk/MCM was used to 35 calculate radical and trace gas concentrations. The model was implemented with specific chamber-related properties. First, a dilution rate was applied to all the trace gases present in the model to account for the dilution from the replenishing flow.
The background production of HONO, HCHO and acetone known to occur in the sunlit chamber Karl et al., 2006), was parametrized by an empirical function that depends on temperature, relative humidity and solar radiation.
Source strengths were adjusted to match the time series of HCHO and acetone during the zero-air phase, when the chamber 40 was the only source for these species (~ 0.3 ppbv h -1 ). These chamber sources also impacted the OH reactivity measured during the zero-air phase. Ideally, after accurately accounting for the chamber sources, the OH reactivity should be well represented by the model, but it is commonly the case that there is still the need for an OH reactant equivalent to ~ 1.0 s -1 of OH reactivity. This unexplained reactivity is parametrized with a co-reactant Y added to the model, which converts OH to HO 2 in the same way as CO does (Fuchs et al., 2012;Fuchs et al., 2014;Kaminski et al., 2017). The concentration of Y was adjusted to match the observed OH reactivity during the zero-air phase of the experiment and was kept constant throughout the experiment. The uncertainty of the OH reactivity in this experiment was ± 0.6 s -1 determined by the uncertainty in the 5 instrumental zero (1.5 s -1 ) of the OH reactivity instrument. This uncertainty was applied in sensitivity runs of the model, but had a minor effect on the results discussed here.
Because of the unknown chemical nature of the background reactivity that dominates the loss of OH radicals for the zero air phase of the experiment, agreement between measured and modelled radical concentrations cannot be expected during this initial phase. Therefore, no model calculation is shown for this part of the experiment. For completeness, the model included the reaction of MBO with O 3 although this reaction contributed on average 3% to the reactivity of MBO which was dominated by the reaction with OH radicals.
3 Results and discussion 20 Figure 2 shows the times series for the trace gases measured during the MBO experiment compared to the model including the sensitivity runs for the uncertainty introduced by the zero OH-reactivity value. At the beginning of the experiment, during the dark phase, formation of radicals was not expected as the roof was closed and only water vapour and ozone were added. The reactivity of 1.7 s -1 observed during this phase was due to desorption of trace gases from the walls of the chamber 25 that could be observed during the humidification process. Some of these trace gases are HONO, HCHO and acetone as seen from their slow but steady increase. Immediately after opening of the roof, there was production of OH, HO 2 radicals and NO x from the photolysis of HONO and HCHO. After the injection of MBO, the OH reactivity was dominated by the reaction with MBO (~ 70% for all three MBO injections) and its oxidation products contributed significantly to the OH reactivity, up to 40%, the more of the MBO reacted away. Good agreement between modelled and measured concentrations, well within 30 the accuracy of the different instruments, could be observed for the majority of the species when MBO was oxidized by OH in this experiment. Formation of both measured major products from the oxidation of MBO, formaldehyde and acetone, was well reproduced by the model (averaged measurement to model ratio of 1.00 ± 0.02 for both). The modelled OH fitted the observation with an average measurement to model ratio of 1.0 ± 0.2 and the agreement between modelled and measured HO 2 was, although less good compared to the OH, still satisfying (0.9 ± 0.1). The MBO decays due to its reaction with OH 35 radicals were slightly over-predicted by the model (average observed to model ratio of 1.3 ± 0.2) in accordance with the measured decline of OH reactivity. This is in agreement with the PTR-TOF-MS measurement. However, results did not change significantly if the model was constrained to measured MBO concentrations.

Model comparison
The major uncertainties in this measurements-model comparison are introduced by the uncertainty of the zero measurement of the OH reactivity data and by the possible interference of RO 2 radicals in the measured HO 2 signal. The first mostly 40 affects the agreement between measured and modelled results for the OH reactivity itself. Modelled OH and HO 2 radicals are Atmos. Chem. Phys. Discuss., https://doi.org /10.5194/acp-2018-134 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 14 March 2018 c Author(s) 2018. CC BY 4.0 License. also partially affected but the uncertainty introduced is lower than the accuracy of radical measurements while the remaining modelled species are not influenced. As mentioned in the instrumental description section, 20% is the upper limit for the interference from RO 2 radicals that could be expected from MBO on the HO 2 signal, based on the experiments performed with isoprene (Fuchs et al., 2011), for the conditions the instrument was run. The HO 2 concentration obtained from the model when accounting for this RO 2 interference would be, on average, only 8% larger (~ 5.5 x 10 7 cm -3 ) than the HO 2 5 concentration without any RO 2 interference for the periods in the experiment where the MBO was present in the chamber.
This value is lower than the accuracy of the HO 2 measurement itself and has an insignificant impact on the other trace gases.

Model comparison including hydrogen shift reactions
In a recent theoretical work from Knap et al. (2016), hydrogen shift reactions (H-shift) in the peroxy radicals originated after 10 photooxidation of four different methyl-buten-ol isomers were investigated. The 1,4, 1,5 and 1,6 H-shift reactions were studied and the rate coefficients at ambient temperature and pressure were given. For the photooxidation of the MBO isomer under investigation in this study, predicted products are OH and HO 2 radicals, 2-hydroxy-2-methylpropanal (HMPR), acetone, and glycolaldehyde ( Fig. 1). Also band -epoxides are proposed as possible products although, due to their extremely slow unimolecular rate coefficients (Fig. 1), they are insignificant. As a sensitivity study, the three H-shift 15 reactions, excluding the branching towards the epoxides, were included in the MCM 3.3.1 model as shown in Fig. 1, using the upper limit rate coefficients at 1013 hPa and 298 K as calculated by Knap et al. (2016). In the model, the H-shift reactions proceed directly to the final stable products with no formation of any intermediate. As expected from the low reaction rates for these reactions, their addition to the MBO degradation scheme has a very small impact with a change of less than 1% on any of the trace gases modelled in our chamber study bringing no improvement in the already good 20 agreement between measurements and model calculations. This is consistent with the study by Knap et al. (2016) where they concluded that H-shift reactions are not relevant for the oxidation scheme of MBO even for low NO conditions (< 50 pptv) where the reaction with HO 2 remains the dominant loss process for the MBO-RO 2 radicals. This is also expected as MBO contains only one double bond and the fast H-shift reactions observed for isoprene and methacrolein are favoured by the formation of conjugated double bonds in the stable radical co-products (Peeters and Nguyen, 2012). 25

OH and HO 2 radicals budget analysis
The calculation of the experimental OH budget helps identifying possible missing OH sources, trusting on the correctness of the measured OH concentration and OH reactivity. The total experimental OH loss rate, L OH , is given by the product of the OH concentration and the OH reactivity and, as the OH radical is assumed to be in steady-state, it should be equal to the total 30 OH production rate (P OH ) (Eq. 1). P OH includes the OH production rate from known sources, P OH Meas (Eq. 2), plus other possible sources. L OH can be compared with P OH Meas, which can be calculated from the measured data.  Figure 3 shows the comparison between P OH Meas and the total experimental OH loss, L OH . The averaged ratio between P OH Meas and L OH is 0.9 ± 0.1 (1 ). A small deviation from unity, which would indicate a missing OH source contributing at most 20% to the total OH production, is obtained. Nevertheless, if the errors of the different measurements are taken into 5 account, this deviation becomes not significant. For example, the total error of the total experimental OH loss is ~ 25% to which the errors of the measured traces gases, mainly of the HO 2 radicals (16%) and of the rate coefficients (~ 10%) used to calculate the P OH Meas, should be added. From these considerations, the experimental OH budget can be considered closed and no additional OH sources aside the ones considered in Equation 2 is needed to explain the OH radicals loss. Figure 3 also depicts the total modelled OH production P OH Mod. This is included in the analysis of the experimental OH 10 radical budget to understand how well the OH-formation paths in the model can describe the measurements. The averaged ratio between P OH Mod and L OH provides a value of 1.0 ± 0.1 (1 ). The good agreement observed between P OH Mod and L OH indicates that the model is able to correctly represent the OH radical sources. The averaged difference between P OH Mod and P OH Meas is (2.3 ± 1.9) x 10 6 cm -3 s -1 . A large part of the difference, ~ 1.5 x 10 6 cm -3 s -1 , is due to additional OH radical sources included in the model and not considered in the experimental OH production, e.g. RO 2 (CH 3 CO 3 and HOCH 2 CO 3 ) 15 reacting with HO 2 forming OH. The additional small discrepancy (~ 0.8 x 10 6 cm -3 s -1 ) is due to the differences observed for HO 2 and ozone between measurements and model calculations.
The analysis of the HO 2 budget is shown in Fig. 4. Here, differently from the OH budget, the measured HO 2 loss rate, , is compared to the total modelled HO 2 production rate, ‚ . This comparison provides information on the completeness in the understanding of the HO 2 production and loss processes for the MBO photooxidation mechanism. 20 Within the model, the fifteen most important HO 2 production paths are explicitly considered and depicted in Fig. 4. The largest contribution to the HO 2 production comes from the decomposition of alkoxy radicals (46%), followed by the conversion of OH by the unknown background reactivity Y in the chamber (30%, not atmospherically relevant). Smaller contributions originate from formaldehyde photolysis (15%), and H-abstraction reaction by oxygen from the methoxy radical (CH 3 O, 7%). As most of the relevant species contributing to the HO 2 production rate such as the alkoxy radicals were not 25 measured and the background reactivity Y cannot be specified, it is not possible to calculate the production rate of HO 2 only from measured species as it was done for the OH radical budget.
The HO 2 radical is expected to be lost mainly via its reaction with NO accounting for ~ 90% of the total loss rate calculated from the model. Additional losses are HO 2 +HO 2 self-reaction, reaction with ozone and reaction with the first generation RO 2 produced from the MBO oxidation (MBOAO2 and MBOBO2, Fig. 1). Therefore, the majority of the HO 2 loss rate can be 30 obtained from measured NO, HO 2 and ozone concentrations (Eq. 3).
Here [HO 2 ], [NO], and [O 3 ] represent the measured concentrations of the trace gases, k € •€ , k € € and k € € the rate coefficient of HO 2 with NO, itself and O 3 , respectively. The measured HO 2 loss rate, L € Meas, is in good agreement with the total modelled HO 2 production rate with an average ratio of measured to modelled rates of 0.8 ± 0.1. The agreement 35 with the total modelled HO 2 production increases (average ratio of measured to modelled rates of 0.9 ± 0.1) when including in the L € Meas the modelled loss rate by the reaction of HO 2 with modelled RO 2 radicals. The largest discrepancies are observed during the first injection of MBO, because the calculated HO 2 production rate is smaller than what is obtained from the model. The main reason is the lower measured HO 2 concentration (13%) compared to the model during this period. With the increasing agreement between modelled and measured HO 2 radical also the agreement in the HO 2 budget increases. 40 The good agreement observed in the HO 2 budget, although partly relying on modelled species concentrations, indicates that the HO 2 production, within this chamber experiment, can be explained by alkoxy radical decomposition, photolysis of formaldehyde, and the chamber specific Y source.

Comparison with previous studies
MBO was the major BVOC measured in two field campaigns which included measurements of OH and HO 2 radicals and a comparison with model calculations. The Biosphere Effects on Aerosols and Photochemisty Experiment II (BEARPEX09) campaign was performed near the Blodgett Forest Research Station in the California Sierra Nevada Mountains (Mao et al., 2012). This campaign was characterized by large MBO concentrations (daily average ~ 3000 pptv), followed by isoprene 5 (daily average ~ 1700 pptv) and monoterpenes ( -pinene, daily average 100 pptv and -pinene, daily average 70 pptv). Both measured OH and HO 2 radicals compared reasonably well with modelled calculations, in agreement with the results observed in this chamber study. Colorado Rocky Mountains (Ortega et al., 2014). Here the dominant measured BVOC was MBO (daily average ~ 1600 pptv) followed by monoterpenes (daily average ~ 500 pptv) (Kim et al., 2013). As observed for the OH radical budget within this study, during the BEACHON-ROCS campaign the calculated OH concentration from ozone photolysis and from the recycling via HO 2 plus NO reaction, divided by the measured OH reactivity, agreed with the measured OH concentration (Kim et al., 2013). No additional OH recycling paths were necessary to close the OH budget. Nevertheless, during the 15 BEACHON-ROCS campaign, the model was able to reproduce the OH radical concentration only when constrained to the HO 2 radical measurements as the model underestimated the measured HO 2 radicals up to a factor of 3 (Kim et al., 2013;. This is different from what observed in the chamber experiment discussed in this study where a good agreement can be found between modelled and measured HO 2 concentration. One difference between the two field studies is the BVOC compositions. During the BEARPEX09 campaign the 20 concentration of the measured monoterpenes relative to the concentration of MBO during daytime was smaller (6%) compared to what observed during the BEACHON-ROCS campaign (31%). Two recent studies in environment with large concentrations of monoterpenes Kaminski et al., 2017) also showed model calculations largely underestimating HO 2 radical measurement. Both studies concluded that the unaccounted HO 2 source seems to originate from monoterpene-oxidation products. The results collected by  and by Kaminski et al. (2017) together with 25 what is observed in this chamber study support that model-measurement discrepancies for HO 2 radicals in the BEACHON-ROCS campaign are not related to the MBO and its oxidation products but rather to the presence of monoterpenes and, as they were present in smaller concentrations, they would need to constitute a very efficient source of HO 2 radicals.

Summary and conclusions
A photooxidation experiment on 2-methyl-3-butene-2-ol (MBO), an important BVOC emitted by lodgepole and ponderosa 30 pines, was performed in the atmospheric simulation chamber SAPHIR. Measurements of OH and HO 2 radicals and OH reactivity together with other important trace gases were compared to results from a state of the art chemical mechanistic model (MCM v3.3.1). The overall agreement is very good: firstly, an average observed to model ratio of 1.0 ± 0.2 and 0.9 ±0.1 is found for OH and HO 2 radicals, respectively. Also the MBO decay caused by reaction with OH radicals fits the expected decay from the model (average observed to model ratio of MBO concentration 1.3 ± 0.2) and is consistent with the 35 measured OH reactivity. Moreover, the major measured products, acetone and formaldehyde, both match the model calculation with an average ratio of 1.00 ± 0.02. Addition of H-shift reactions from RO 2 radicals to the kinetic model as suggested in the literature (Knap et al., 2016) does not have a significant impact on the model results as expected from the small reaction rates (< 1.1 x 10 -3 s -1 ). The observed closure for both OH and HO 2 radical budgets indicates that their chemistry is well described by our current understanding of the MBO OH-initiated degradation processes . 40 Atmos. Chem. Phys. Discuss., https://doi.org /10.5194/acp-2018- (Mao et al., 2012;Kim et al., 2013). However there was no closure for the HO 2 budget or agreement between measurements and model results when a larger concentration of monoterpenes was also observed . This discrepancy cannot be explained by MBO photo-oxidation as a good agreement between measured and calculated concentration of HO 2 is found in this chamber study. As a large discrepancy was also observed for 5 chamber studies with b-pinene (Kaminski et al., 2017) and in environments with large monoterpenes concentrations , it is reasonable to assume that field observation for HO 2 radicals could be explained by an additional HO 2 radical source from monoterpenes-oxidation products, as proposed by Kaminski et al. (2017).