Isoprene chemistry under upper-tropospheric conditions

Abstract

Isoprene (C₅H₈) is the non-methane hydrocarbon with the highest emissions to the atmosphere. It is mainly produced by vegetation, especially broad-leaved trees, and efficiently transported to the upper troposphere in deep convective clouds, where it is mixed with lightning NO_x. Isoprene oxidation products drive rapid formation and growth of new particles in the tropical upper troposphere. However, isoprene oxidation pathways at low temperatures are not well understood. Here, in experiments at the CERN CLOUD chamber at 223 K and 243 K, we find that isoprene oxygenated organic molecules (IP-OOM) all involve two successive ${\mathrm{OH}}^{\bullet }$ oxidations. However, depending on the ambient concentrations of the termination radicals ( ${{\mathrm{HO}}_2}^{\bullet },{\mathrm{NO}}^{\bullet }$ , and ${\mathrm{NO}}_2^{\bullet }$ ), vastly-different IP-OOM emerge, comprising compounds with zero, one or two nitrogen atoms. Our findings indicate high IP-OOM production rates for the tropical upper troposphere, mainly resulting in nitrate IP-OOM but with an increasing non-nitrate fraction around midday, in close agreement with aircraft observations.

Fig. 1. Schematic of OH∙ initiated isoprene oxidation under daytime upper-tropospheric conditions.

Isoprene is oxidised by ${\mathrm{OH}}^{\bullet }$ to form a peroxy radical, ${\mathrm{ISOPOO}}^{\bullet }$ (C₅H₉O₃), which can be terminated by ${{\mathrm{HO}}_2}^{\bullet },{\mathrm{NO}}^{\bullet }$ or ${\mathrm{NO}}_2^{\bullet }$ to form first-generation intermediates ISOPOOH (isoprene hydroxy hydroperoxide, C₅H₁₀O₃), IHN (isoprene hydroxy nitrate, C₅H₉NO₄), and IHPN (isoprene hydroxy peroxy nitrate, C₅H₉NO₅), respectively. The isoprene backbone contains two C=C bonds, therefore, the first-generation intermediates react with ${\mathrm{OH}}^{\bullet }$ to form peroxy radicals, which are terminated by ${{\mathrm{HO}}_2}^{\bullet },{\mathrm{NO}}^{\bullet }$ or ${\mathrm{NO}}_2^{\bullet }$ to produce six distinct second-generation species (C₅H₁₂O₆, C₅H₁₁NO₈, C₅H₁₀N₂O₁₀, C₅H₁₀N₂O₉, C₅H₁₀N₂O₈ and C₅H₁₁NO₇), each with hydroperoxide groups, nitrate groups and peroxynitrate groups depending on which radicals led to termination. The red, blue and purple arrows represent, respectively, ${\mathrm{OH}}^{\bullet }$ then ${{\mathrm{HO}}_2}^{\bullet },{\mathrm{OH}}^{\bullet }$ then ${\mathrm{NO}}^{\bullet }$, and ${\mathrm{OH}}^{\bullet }$ then ${\mathrm{NO}}_2^{\bullet }$ reactions. This mechanism augments existing isoprene mechanisms to include peroxynitrates, a potentially important upper-tropospheric pathway. There are additional isomers that exist beyond those shown, and they follow similar chemical pathways.

Fig. 2. Composition and yield of IP-OOM at 223 K.

a Normalised fraction of C₅ IP-OOM (${\mathrm{n}}_{\mathrm{C}}=5,{\mathrm{n}}_{\text{O}}^{eff}>3$) based on differing oxidation levels versus ${\mathrm{OH}}^{\bullet }$ concentration. Isoprene is a diene, thus two ${\mathrm{OH}}^{\bullet }$ additions can occur to form saturated products with −OH, −OOH, −ONO₂ and −O₂NO₂ functionality. Within this oxidation process, unsaturated products are also produced. These compounds have either retained a C=C bond, isomerised to contain a carbonyl group, C=O, or formed an epoxide, C-O-C. Saturation levels are determined by the hydrogen and nitrogen content: Saturated: ${\mathrm{n}}_{\mathrm{N}}$ = 0, ${\mathrm{n}}_{\mathrm{H}}$ = 12; ${\mathrm{n}}_{\mathrm{N}}$ = 1, ${\mathrm{n}}_{\mathrm{H}}$ = 11; ${\mathrm{n}}_{\mathrm{N}}$ = 2, ${\mathrm{n}}_{\mathrm{H}}$ = 10. Unsaturated: ${\mathrm{n}}_{\mathrm{N}}$ = 0, ${\mathrm{n}}_{\mathrm{H}}$ = 10; ${\mathrm{n}}_{\mathrm{N}}$ = 1, ${\mathrm{n}}_{\mathrm{H}}$ = 9. Other refers to the remaining C₅ IP-OOM, which includes radicals, H-abstraction products and products from nitrate and ozone oxidation. Circles and squares represent runs with and without NO_x, respectively. The orange dashed line fits the data in the form Saturated fraction = 0.32 × log[${\mathrm{OH}}^{\bullet }$] − 1.81, the other fits are listed in Table S3. b total measured IP-OOM (grey) with (circles) and without NO_x (squares) and ${\mathrm{IP}}_{\mathrm{0N}}$ (red cross), ${\mathrm{IP}}_{\mathrm{1N}}$ (pink cross) and ${\mathrm{IP}}_{\mathrm{2N}}$ (blue cross) versus reacted IP-OOM precursors, see SI section C for individual x-axis definitions. IP-OOM: ${\mathrm{n}}_{\mathrm{C}}>$ 3, ${\mathrm{n}}_{\mathrm{O}}^{\mathit{\text{eff}}}>$ 3; ${\mathrm{IP}}_{\mathrm{0N}},{\mathrm{IP}}_{\mathrm{1N}}$ and ${\mathrm{IP}}_{\mathrm{2N}}$ are all subsets of IP-OOM, where ${\mathrm{n}}_{\mathrm{N}}$ = 0, ${\mathrm{n}}_{\mathrm{N}}$ = 1 and ${\mathrm{n}}_{\mathrm{N}}$ = 2, respectively. The dashed lines indicate that the yield of IP-OOM is between 50 and 100%, and between 25 and 50% for each ${\mathrm{IP}}_{\mathrm{0,1,2N}}$. Using linear regression, the average yield of IP-OOM is 68%. For the individual subsets, specific IP-OOM precursors can be used to determine more accurate yields as described in SI section C. Using these definitions, the average yields are 36%, 31%, 42% for ${\mathrm{IP}}_{\mathrm{0N}},{\mathrm{IP}}_{\mathrm{1N}}$ and ${\mathrm{IP}}_{\mathrm{2N}}$, respectively.

Fig. 3. Molecular composition for different radical dominant cases.

Mass defect (difference from integer mass) versus mass-to-charge for three radical distribution cases: a HO₂ case, b NO case and c NO₂ case. The data points are coloured by the number of oxygen atoms in the molecule and the outlines are coloured by the number of nitrogen present: N = 0  → black, N = 1  → red, N = 2  → blue. The symbol area is proportional to the log of the normalised species concentration. For concentrations $\ge 10^8{\mathrm{cm}}^{\mathsf{-3}}$, symbols are faded to the background (C₄ H₆O for all cases and C₅H₁₀O₃ for HO₂ case). Only concentrations above $5\times 10^5{\mathrm{cm}}^{\mathsf{-3}}$ are shown in accordance with the highest limit of detection. This is a combination of three mass spectrometers applying NO₃−, Br⁻ and NH₄+ ionisation methods.

Fig. 4. Second-generation products as a function of radical ratios.

The normalised fraction (Eq. 5) of the three cross-branch products is plotted as a function of the respective branch radical ratios. For example, a shows the double ${\mathrm{NO}}^{\bullet }$ termination product C₅H₁₀N₂O₈ (blue), the mixed product C₅H₁₀N₂O₉ (purple), and the double ${\mathrm{NO}}_2^{\bullet }$ termination product C₅H₁₀N₂O₁₀ (red) normalised to the sum of all three, plotted against the ${\mathrm{NO}}_2^{\bullet }$:${\mathrm{NO}}^{\bullet }$ ratio. b repeats the same treatment for ${{\mathrm{HO}}_2}^{\bullet }$:${\mathrm{NO}}^{\bullet }$ and the second-generation products C₅H₁₀N₂O₈ (blue), C₅H₁₁NO₇ (purple) and C₅H₁₂O₆ (red), and c presents C₅H₁₀N₂O₁₀ (blue), C₅H₁₁NO₈ (purple) and C₅H₁₂O₆ (red) as a function of ${{\mathrm{HO}}_2}^{\bullet }$:${\mathrm{NO}}_2^{\bullet }$ ratios. Circles represent data at 223 K, and triangles at 243 K. Sigmoid and Gaussian curves have been fitted through the 223 K data and the standard deviation has been shaded, to represent uncertainty in the fit. Each data point in Fig. 4 is an averaged period of a steady-state chemistry stage achieved in CLOUD and vertical dashed lines indicate the point of equal termination for each branch, 3.51, 0.41 and 0.09 for (a, b, c), respectively. All second-generation concentrations are obtained from the NO₃-CIMS.

Fig. 5. Aircraft observation and laboratory comparison of the second-generation distribution as a function HO2∙:NO∙.

Solid lines represent the experimental fits of C₅H₁₀N₂O₈ (blue), C₅H₁₁NO₇ (purple) and C₅H₁₂O₆ (red) from Fig. 4b of the CLOUD experiment, 223 K and 965 mbar. Square markers indicate atmospheric aircraft observations (CAFE) from Curtius et al., specifically T4–T9, at 213 K and 187 mbar. To adjust for the pressure difference (see Fig. S8), circles (213 K and 950 mbar) represent CAFE T4–T9 periods, shifted by a factor of two in ${{\mathrm{HO}}_2}^{\bullet }$:${\mathrm{NO}}^{\bullet }$ (CAFE (adj)). Error bars represent 1σ accuracy of the ${{\mathrm{HO}}_2}^{\bullet }$ measurement.

Fig. 6. Global simulation of upper-tropospheric IP-OOM chemistry.

a The global daytime reactivity of isoprene + ${\mathrm{OH}}^{\bullet }$. Using annually averaged daytime isoprene and ${\mathrm{OH}}^{\bullet }$ concentrations from the EMAC global model and a reaction rate coefficient $k_{\mathrm{IP}}=2.7\times 10^{\mathrm{-11}}{e}^{\frac{390}{\mathrm{T}}}{\mathrm{cm}}^3{\mathrm{s}}^{\mathrm{-1}}$. Focusing on the Amazon rainforest, isothermal-surface plots depict the reactivity of isoprene with ${\mathrm{OH}}^{\bullet }$ at three daytime periods b Sunrise, c Midday and d Sunset. The ${{\mathrm{HO}}_2}^{\bullet }$:${\mathrm{NO}}^{\bullet }$ ratio at three daytime periods e Sunrise, f Midday and g Sunset is also plotted for each daytime period. A hashed grid (masking insufficient oxidation) has been added so to highlight regions with isoprene and ${\mathrm{OH}}^{\bullet }$ levels high enough that sufficient oxidation could occur: isoprene  >2.4 $\times 10^7{\mathrm{cm}}^{\mathrm{-3}}$ (223 K, 230 mbar) and ${\mathrm{OH}}^{\bullet }$  >6 $\times 10^5{\mathrm{cm}}^{\mathrm{-3}}$ (223 K, 230 mbar). Midday is defined using the hour of highest shortwave flux at the top of the atmosphere, whereas sunrise and sunset are defined when the shortwave flux is closest to a quarter of the maximum shortwave flux. All data are annually averaged isothermal-surface plots at the temperature of 223 K from the global model (EMAC), ranging between 200 and 300 hPa and (−51.3°, 51.3°) latitude. A detailed description of the model conditions is presented in ‘Methods’. The map in the figures were made with Natural Earth.