Structure-dependent reactivity of Criegee intermediates studied with spectroscopic methods

Jim Jr-Min Lin *ab and Wen Chao ab
aInstitute of Atomic and Molecular Sciences, Academia Sinica, Taipei 10617, Taiwan. E-mail:
bDepartment of Chemistry, National Taiwan University, Taipei 10617, Taiwan

Received 12th May 2017

First published on 25th August 2017

Criegee intermediates are very reactive carbonyl oxides that are formed in reactions of unsaturated hydrocarbons with ozone (ozonolysis). Recently, Criegee intermediates have gained significant attention since a new preparation method has been reported in 2012, which employs the reaction of iodoalkyl radical with molecular oxygen: for instance, CH2I + O2 → CH2OO + I. This new synthesis route can produce Criegee intermediates with a high number density, which allows direct detection of the Criegee intermediate via various spectroscopic tools, including vacuum UV photoionization mass spectrometry, absorption and action spectroscopy in the UV and IR regions, and microwave spectroscopy. Criegee intermediates have been thought to play important roles in atmospheric chemistry, such as in OH radical formation as well as oxidation of atmospheric gases such as SO2, NO2, volatile organic compounds, organic and inorganic acids, and even water. These reactions are relevant to acid rain and aerosol formation. Kinetics data including rate coefficients, product yields and their temperature and pressure dependences are important for understanding and modeling relevant atmospheric chemistry. In fundamental physical chemistry, Criegee intermediates have unique and interesting features, which have been partially revealed through spectroscopic, kinetic, and dynamic investigations. Although previous review articles have discussed Criegee intermediates, new data and knowledge on Criegee intermediates are still being accumulated. In this tutorial review, we have focused on structure-dependent reactivity of Criegee intermediates and various spectroscopic tools that have been utilized to probe the kinetics of Criegee intermediates.

image file: c7cs00336f-p1.tif

Wen Chao (left) and Jim Jr-Min Lin (right)

Jim Jr-Min Lin received his PhD degree under the guidance of Yuan T. Lee at National Taiwan University, studying reactive scattering with crossed molecular beams. He then joined Institute of Atomic and Molecular Sciences (IAMS), Academia Sinica. He is currently a research fellow in IAMS and an adjunct professor in the Department of Chemistry, National Taiwan University. His research interests include reactive scattering using molecular beams, photodissociation, spectroscopy, and kinetics that are relevant in fundamental understanding and atmospheric chemistry. Recent research focus is on the spectroscopy and kinetics of Criegee intermediates.

Wen Chao obtained his Bachelor's and Master's degrees in Chemistry at National Taiwan University. He studied the kinetics of Criegee intermediates using time-resolved UV absorption spectroscopy under the direction of Jim Jr-Min Lin. He finished his Master's program in June 2017 and is currently a draftee of Substitute Military Service of Education in Taiwan.

Key learning points

(1) Generation of Criegee intermediates: ozonolysis of alkenes versus photolysis of diiodo precursors.

(2) Various spectroscopic tools to probe Criegee intermediates: advantages and limitations.

(3) Structure-dependent reactivity of Criegee intermediates.

(4) Anti-type Criegee intermediates react quickly with water dimer: a good example of a water-catalyzed reaction.

(5) Syn-type Criegee intermediates may have fast intramolecular hydrogen atom transfer via tunneling: a remarkable agreement between laser excitation and thermal measurements.


Criegee intermediate formation and preparation

Unsaturated hydrocarbons are emitted into our atmosphere in large quantities from both human and natural sources. Unsaturated hydrocarbons (represented by a simple C[double bond, length as m-dash]C bond herein) react with ozone (ozonolysis) to form a carbonyl molecule and a highly reactive carbonyl oxide. The formation of carbonyl oxide was first postulated by Rudolf Criegee; thus, the carbonyl oxide structure is also known as Criegee intermediate (CI). Criegee intermediates are strong oxidants and play important roles in atmospheric chemistry (see Fig. 1), such as in the formation of OH radicals, aerosol and acid rain.1–5 Note that most atmospheric alkenes and Criegee intermediates are generated near the Earth's surface; thus, low-altitude atmosphere is more relevant in this discussion.
image file: c7cs00336f-f1.tif
Fig. 1 Schematic showing a few key reactions involving Criegee intermediates.

Recently, Criegee intermediates have gained considerable attention and there are already comprehensive review articles on their physical chemistry.3,4 In this Tutorial Review, we would like to discuss various spectroscopic methods that have been utilized to probe the kinetics of simple Criegee intermediates and two types of Criegee intermediate reactions—reactions with water vapor and thermal decomposition. These two types of reactions are chosen because new experiments have demonstrated that their reaction rates depend strongly on the structure of the Criegee intermediate, and based on current knowledge, they are the main decay pathways of Criegee intermediates in the atmosphere.

This Tutorial Review is organized as follows. First, a brief overview of the preparation, spectroscopy, and kinetics of simple Criegee intermediates will be given in the Introduction. Then in the section “Methods to probe Criegee intermediate”, the pros and cons of various detection tools in probing the kinetics of Criegee intermediates will be analyzed and commented. Finally, in the section of “Highlights in kinetics studies”, structure-dependent reactivity of Criegee intermediates will be discussed in depth, followed by the Outlook showing our future perspective and Concluding remarks.

Due to their high reactivity, the concentration of Criegee intermediates produced by ozonolysis is low, prohibiting their direct detection. Steady-state approximation may be applied to estimate the concentration as below.

image file: c7cs00336f-t1.tif

image file: c7cs00336f-t2.tif
Here, [CI]ss is the steady-state concentration of Criegee intermediate, kozo is the rate coefficient of ozonolysis reaction, ϕCI is the yield of stabilized Criegee intermediate, and kdecay is the effective rate coefficient for Criegee intermediate removal. The reported value of ϕCI at 1 atm pressure ranges from less than 0.1 for cyclohexene1 to about 0.4 for ethene6 and up to almost 1 for trans-5-decene.1 (ϕCI is smaller at lower pressure due to decomposition of hot Criegee intermediate). Depending on the substitution groups of the reactive C[double bond, length as m-dash]C bond, the order of magnitude of kozo ranges from 10−18 cm3 s−1 for O3 + C2H4 to 10−15 cm3 s−1 for O3 + (CH3)2C[double bond, length as m-dash]C(CH3)2 (see NIST Chemical Kinetics Database,, for rate coefficients). For simplicity, here we use “cm−3” as the unit for molecular number density instead of the more formal “molecule cm−3”. Criegee intermediates are short-lived; the decay rate kdecay includes thermal decomposition (kth) and reactions of Criegee intermediate with O3 (kO3), alkene (kene), and even the Criegee intermediate itself (kself).
kdecay = kth + kO3[O3] + kene[alkene] + 2kself[CI] + ⋯

Since kdecay increases with increasing [O3] and [alkene] and kozo is not very large, it is difficult to obtain high [CI]ss from ozonolysis. The possible reaction products include carbonyl compounds, which can further react with Criegee intermediates to produce secondary ozonides.1,2

In the atmosphere, the concentrations of O3 and alkenes (in ppb) are significantly lower than those used in laboratory ozonolysis experiments (in ppm), resulting in lower [CI]. With current technology, it is not practical to measure [CI] in the atmosphere. Instead, [CI] can only be estimated from kinetics modeling of its formation and decay (similar to the above steady-state approach, but more sophisticated). Therefore, understanding of relevant rate coefficients and product yields is crucial.

In 2012, there was a breakthrough in the preparation of Criegee intermediates. Welz et al.7 utilized the reaction of iodoalkyl radical with O2 to prepare Criegee intermediates; for example,

CH2I + O2 → CH2OO + I
The iodoalkyl radical can be easily obtained by UV photolysis of the corresponding diiodoalkane precursor,
CH2I2 + → CH2I + I

In this photolysis method of synthesis, Criegee intermediates can be prepared in much higher concentrations than is possible in ozonolysis due to a combination of several factors. (i) The photolysis of the diiodoalkane precursor can be very fast (on the order of 10−8 s) and efficient (photolysis yield may easily reach a few percent) if an intense pulsed laser (often excimer laser at 248, 308, or 352 nm or YAG laser at 266 or 355 nm) is used. (ii) Criegee intermediates do not react with O2; thus, high [O2] can be employed to speed up the conversion from iodoalkyl radical to Criegee intermediate (e.g., the reaction CH2I + O2 → CH2OO + I may be completed within a few microseconds with an O2 pressure around 10 Torr).8 (iii) The yield of CH2OO from CH2I + O2 is high (up to 0.78 at a pressure of 20 Torr). This yield is found to be pressure dependent. At a low pressure of less than 20 Torr, some of the internally excited CH2OO may decompose; at high pressures, formation of the CH2IOO adduct competes with CH2OO formation. Nonetheless, a substantial yield ranging from 0.8 to 0.3 can be realized for pressures ranging from 20 Torr to 760 Torr.8 The high CH2OO yield at low pressure and the adduct formation at high pressure may be attributed to the small energy difference between CH2I + O2 and CH2OO + I (nearly thermoneutral), so that the products (Criegee intermediate and I atom) have less excess energy and the Criegee intermediate can easily be stabilized by collisions with buffer gas molecules.3

[CH2OO] as high as 1014 cm−3 can be easily prepared in a laboratory. However, the lifetime τ of Criegee intermediate is short when its concentration is high (τ ∼ 100 μs for [CI]0 ≈ 1 × 1014 cm−3versus τ ∼ 20 ms for [CI]0 ≈ 1 × 1011 cm−3) due to a number of fast reactions, including reactions of the Criegee intermediate with iodine atoms (e.g., CH2OO + I → CH2O + IO) and itself (e.g., 2CH2OO → 2CH2O + O2).8,9


With the above mentioned photolysis synthesis, direct detection of several simple Criegee intermediates has been realized using a number of methods, including vacuum UV photoionization mass spectrometry (VUV-PIMS),7,10,11 microwave (MW) rotational spectroscopy,12,13 infrared (IR) absorption14–16 and action17–19 spectroscopy, and UV absorption20–23 and depletion22–27 spectroscopy. These spectroscopic data have provided (i) detailed information about the molecular structure, bonding, and electronic properties and (ii) methods to probe Criegee intermediates. The main results are summarized below.

Welz et al.7 employed VUV-PIMS to detect the simplest Criegee intermediate CH2OO in a flow reactor. In their experiment, CH2OO was ionized by a VUV photon and detected with a mass spectrometer. By scanning the photon energy, the ionization energy of CH2OO was determined to be 10.0 eV (Fig. 2a), consistent with theoretical predictions. Taatjes et al.10 detected methyl substituted Criegee intermediate CH3CHOO with the same method and found that two conformers, anti- and syn-CH3CHOO, have different ionization energies (9.33 and 9.40 eV, respectively, Fig. 2b), which allowed them to measure conformer-specific kinetics. Remarkably, the rate coefficients of these Criegee intermediate (CH2OO, anti- and syn-CH3CHOO) reactions with SO2 and with NO2 were found to be much faster than those deduced from ozonolysis experiments (by a factor of ca. 104), suggesting Criegee intermediates play a more important role in atmospheric oxidation of SO2 and NO2 than previously thought.3

image file: c7cs00336f-f2.tif
Fig. 2 (a) Experimental photoionization efficiency spectrum of CH2OO; the calculated spectrum of dioxirane and experimental spectrum of formic acid (two possible isomers) are also included for comparison. Reproduced from ref. 7 with permission from American Association for the Advancement of Science, copyright 2012. (b) Experimental photoionization efficiency spectra of anti- and syn-CH3CHOO. The syn-CH3CHOO spectrum was obtained by adding water in the reactor to scavenge anti-CH3CHOO. The calculated spectrum of vinyl hydroperoxide (a possible isomer) is also included. Reproduced from ref. 10 with permission from American Association for the Advancement of Science, copyright 2013.

Fourier-transform microwave spectroscopy, a method with very high resolution and sensitivity, has been utilized to detect CH2OO, CH3CHOO and (CH3)2COO.12,13 Very precise rotational constants have been deduced. Furthermore, due to the ultrahigh sensitivity, direct detection of CH2OO has been realized in the ozonolysis of C2H4, even though the relative abundance of CH2OO is as low as 10−6.28

Two types of IR experiments, direct absorption and action spectroscopy, have been performed to investigate Criegee intermediates. Direct IR absorption spectroscopy has been used to detect CH2OO, anti-CH3CHOO, syn-CH3CHOO, and (CH3)2COO with a step-scan Fourier-transform infrared (ss-FTIR) spectrometer,14–16 in which early time (≤10−4 s) data contain the absorption spectra of short-lived species; this time-resolved information helps in spectral assignment.

Criegee intermediates may have multiple resonance structures (Fig. 3).3 Various resonance structures would indicate different physical properties and chemical reactivities. The observed OO and CO stretching frequencies in ss-FTIR studies14–16 are consistent with a Criegee intermediate structure that is closer to a zwitterion than a biradical. In addition, MW spectroscopy of Criegee intermediates gives precise rotational constants which are consistent with a long OO bond (1.35–1.38 Å) and a short CO bond (1.27 Å).12,13

image file: c7cs00336f-f3.tif
Fig. 3 Selected resonance structures of Criegee intermediate.

Reflecting zwitterionic character (the CO bond is similar to a double bond), the barrier to interconversion of the anti-/syn-conformers is substantial (∼38 kcal mol−1).10 Therefore the interconversion is much slower than other decay processes of Criegee intermediates. In addition, the zwitterionic structure would have a large dipole moment (ca. 4.1 Debye for CH2OO).3 Strong dipole–dipole interaction should therefore play a significant role in the reactions of Criegee intermediates with dipolar molecules.

OH radical, a product of Criegee intermediate unimolecular decomposition, has been probed with laser induced fluorescence after IR excitation of syn-CH3CHOO, syn-C2H5CHOO, and (CH3)2COO prepared in a molecular beam.17–19 By scanning the IR excitation laser frequency, an action spectrum can be obtained. The excitation energy needs to be sufficient to induce unimolecular decomposition and produce OH radicals; thus, mainly overtone and combination bands are observed using this method.

CH2OO has the same number of electrons as O3 (isoelectronic). Therefore, it is expected that CH2OO has a strong UV absorption band, similar to that of O3.3 Multiple groups have reported UV absorption spectra of CH2OO, which all show a strong and broad absorption band in the wavelength region from ca. 280 to 420 nm. However, there are discrepancies in the details of the reported spectra. Beames et al.25 used a UV laser to deplete CH2OO in a molecular beam. By monitoring the remaining CH2OO with VUV photoionization at 10.5 eV and scanning the wavelength of the depletion laser, a depletion spectrum that is equivalent to the absorption spectrum could be obtained if the dissociation quantum yield is unity. Sheps utilized a cavity-enhanced absorption technique to measure the absorption spectrum of CH2OO in a flow cell.20 His reported spectrum was red-shifted by more than 20 nm with respect to that of Beames et al.25 Sheps mentioned that the reason for this discrepancy might be that the dissociation quantum yield is much smaller than unity at longer wavelengths.20 However, Vansco et al. measured the product anisotropy in the photodissociation of CH2OO, which indicates that photodissociation is faster than rotation for the entire absorption band.27 Such rotational periods are in the picosecond time scale. If the photodissociation of Criegee intermediate is faster than picoseconds, other decay processes (e.g., fluorescence) would not compete with the photodissociation process; therefore, it is unlikely to have a small dissociation yield.

Our group measured absorption spectrum of CH2OO in a flow cell with a time-gated intensified CCD spectrometer.22 We used two methods to extract the absorption spectrum of CH2OO from the time-resolved data. The first method utilized the fact that CH2OO is a short-lived species, that is, its concentration would decay with time. The other method employed SO2 to scavenge CH2OO so that the contribution of CH2OO would be gone after a short time when enough SO2 was added. The resulting spectra from both methods showed excellent agreement with each other. Furthermore, they were consistent with the absolute absorption cross sections at 308 and 352 nm obtained by molecular beam laser depletion experiments.22 After careful evaluation and analysis, we believe our results are the most reliable and the discrepancies of other groups are probably due to uncertainties in their estimated effective laser fluence25 and absorption path length.20

Fig. 4 shows a comparison of the UV absorption spectra of CH2OO, anti-CH3CHOO, syn-CH3CHOO, and (CH3)2COO.22–24,29,30 Starting from the shortest wavelength, the peak positions are 324, 330, 340, and 360 nm for syn-CH3CHOO, (CH3)2COO, CH2OO, and anti-CH3CHOO, respectively. We can see that alkyl group substitution at the syn position causes a blue shift (transition energy increases), consistent with a picture of intramolecular hydrogen-bonding that stabilizes ground state energy but destabilizes the excited state.3,26

image file: c7cs00336f-f4.tif
Fig. 4 Comparison of the UV absorption spectra of small Criegee intermediates. The peak values are 1.2 × 10−17 cm2 (at 340 nm) for CH2OO,22 1.8 × 10−17 cm2 (at 330 nm) for (CH3)2COO,24 and roughly 1.3 × 10−17 cm2 (at 324 nm) for syn-CH3CHOO.23 There is no reliable absolute cross section measurement for anti-CH3CHOO. Note that while the oscillatory structure at the red side of the CH2OO spectrum is mostly real,22 the oscillations in the anti-CH3CHOO spectrum are mostly noise due to a lower signal-to-noise ratio.30


Due to their high reactivity (and thus short lifetime), the concentration of Criegee intermediates in the atmosphere is too low to measure and can only be estimated from their formation and removal rates. The impact of a Criegee intermediate on a given process (e.g., SO2 oxidation) depends on its concentration and the corresponding reaction rate coefficient. Therefore, kinetics data are needed in order to understand and model the related atmospheric chemistry. There are a number of investigations of reaction kinetics of Criegee intermediates, which have been reviewed recently.2–5 A selection of relevant reaction rate coefficients is listed in Table 1. The main findings are summarized below.
Table 1 Summary of selected experimental rate coefficients at room temperature. Except for the unimolecular decomposition, for which the rate coefficients are given in unit of s−1, all reaction rates are second order and their rate coefficients are given in unit of cm3 s−1. Unless noted, only rate coefficients obtained by direct detection of Criegee intermediates are shown; see ref. 3 and 4 for more thorough review
a From ozonolysis study.34 b Theoretically predicted value using an experimentally refined potential energy surface17,19 (see the “Highlights in kinetics studies” section). c For deuterated dimethyl Criegee intermediate, (CD3)2COO.11 d Upper limit of the rate coefficient, since the reaction of (CH3)2COO with iodine atoms is not considered.11
SO2 (3.9 ± 0.7) × 10−11 (ref. 7)

(3.53 ± 0.29) × 10−11 (ref. 4)

(4.1 ± 0.3) × 10−11 (ref. 20)

(3.80 ± 0.04) × 10−11 (ref. 11)

(6.7 ± 1.0) × 10−11 (ref. 10)

(2.2 ± 0.2) × 10−10 (ref. 21)

(2.4 ± 0.3) × 10−11 (ref. 10)

(2.9 ± 0.3) × 10−11 (ref. 21)

(1.3 ± 0.13) × 10−10 (ref. 29)

(2.2 ± 0.1) × 10−10 (ref. 11)

NO2 (7+3−2) × 10−12 (ref. 7) (2 ± 1) × 10−12 (ref. 10) (2 ± 1) × 10−12 (ref. 10) <5 × 10−12 (ref. 11)

(2.1 ± 0.3) × 10−12[thin space (1/6-em)]c (ref. 11)

H2O <4 × 10−15 (ref. 7)

(3.2 ± 1.2) × 10−16[thin space (1/6-em)]a (ref. 34)

(1 ± 0.4) × 10−14 (ref. 10)

(2.4 ± 0.4) × 10−14 (ref. 21)

(1.3 ± 0.3) × 10−14 (ref. 30)

<4 × 10−15 (ref. 10)

<2 × 10−16 (ref. 21)

<1.5 × 10−16 (ref. 29)
(H2O)2 (7.4 ± 0.6) × 10−12 (ref. 33)

(4.0 ± 1.2) × 10−12 (ref. 4)

(4.4 ± 0.3) × 10−11 (ref. 30) <1.3 × 10−13 (ref. 29)
H2S (1.7 ± 0.2) × 10−13 (ref. 36)
HCOOH (1.1 ± 0.1) × 10−10 (ref. 31) (5 ± 3) × 10−10 (ref. 31) (2.5 ± 0.3) × 10−10 (ref. 31)
CH3COOH (1.3 ± 0.1) × 10−10 (ref. 31) (2.5 ± 0.6) × 10−10 (ref. 31) (1.7 ± 0.5) × 10−10 (ref. 31)
HCl (4.6 ± 1) × 10−11 (ref. 32)
HNO3 (5.4 ± 1) × 10−10 (ref. 32)
CH3CHO (9.5 ± 0.7) × 10−13 (ref. 4)
(CH3)2CO (2.3 ± 0.3) × 10−13 (ref. 4)

(2.03 ± 0.15) × 10−13 (ref. 37)

(CF3)2CO (3.0 ± 0.3) × 10−11 (ref. 4)

(3.33 ± 0.27) × 10−11 (ref. 4)

Ethene (7 ± 1) × 10−16 (ref. 37)
Propene (1.8 ± 0.3) × 10−15 (ref. 37)
Isobutene (1.4 ± 0.3) × 10−15 (ref. 37)
1-Butene (1.5 ± 0.3) × 10−15 (ref. 37)
Isoprene (1.5 ± 0.1) × 10−15 (ref. 38)
Self-reaction (8 ± 4) × 10−11 (ref. 8)

(7.4 ± 0.6) × 10−11 (ref. 11)

(4 ± 2) × 10−10 (ref. 4)

(6.0 ± 2.1) × 10−11 (ref. 4)

<(6.0 ± 1.1) × 10−10[thin space (1/6-em)]d (ref. 11)
Unimolecular decomposition 0.19 ± 0.07[thin space (1/6-em)]a (ref. 34) 122b (ref. 17) 361 ± 49 (ref. 9)

305 ± 70 (ref. 11)

276b (ref. 19)

(1) Criegee intermediates [CH2OO, anti-CH3CHOO, syn-CH3CHOO, (CH3)2COO] react very fast with SO2.7,10,11,20,21,29 The reported rate coefficients are in the range of 2 × 10−11 to 2 × 10−10 cm3 s−1.

(2) Criegee intermediates [CH2OO, anti-CH3CHOO, syn-CH3CHOO] react fast with NO2.7,10,11 The reported rate coefficients range from 2 × 10−12 to 7 × 10−12 cm3 s−1.

(3) Criegee intermediates [CH2OO, anti-CH3CHOO, syn-CH3CHOO] react very fast with organic and inorganic acids.31,32 The reported rate coefficients are 1 × 10−10 to 5 × 10−10 cm3 s−1.

(4) For Criegee intermediate reactions with water vapor, there is a strong structure dependence. CH2OO and anti-CH3CHOO react fast with water vapor,6,10,30,33 mostly with water dimer. On the other hand, the reactions of water vapor with syn-CH3CHOO and with (CH3)2COO are too slow to measure.7,10,29,30

(5) For thermal (unimolecular) decomposition of Criegee intermediates, there is a strong structure dependence. The rate coefficients of the thermal decomposition of syn-CH3CHOO and (CH3)2COO are on the order of 102 s−1 at 298 K.9,11,17,19 H-atom tunneling is the main bottleneck for this thermal decomposition pathway, consistent with the observed isotope effect.9 This reaction path does not exist for CH2OO and anti-CH3CHOO, for which thermal decomposition is much slower.5,34

(6) For Criegee intermediates containing 4 or more carbon atoms, there are as yet no studies utilizing direct detection with photolysis preparation scheme.

Although there is a large variation in water content in the atmosphere, typical [H2O] in ambient air is on the order of 1017 cm−3 and 1014 cm−3 for [(H2O)2] (see Table 2); both [H2O] and [(H2O)2] are much higher than the concentrations of other potential reactants, like SO2, NO2, etc. (for example, 40 ppbv corresponds to ca. 1012 cm−3). Thus, the reactions with water and water dimer would be the main decay pathways for CH2OO, anti-CH3CHOO and similar anti-type Criegee intermediates.6,10,30,33 On the other hand, thermal decomposition of (CH3)2COO, syn-CH3CHOO and similar syn-type Criegee intermediates is faster than their reactions with water vapor and SO2.9,11,17,19

Table 2 Effective (apparent) rate coefficient for CH2OO reaction with water vapor (monomer and dimer) as a function of relative humidity at 298 K and its value relative to the rate coefficient for CH2OO reaction with SO2 estimated with the best-available bimolecular rate coefficients of CH2OO reactions with water monomer (kw), water dimer (kw2) and SO2 (kSO2).4 All values are calculated using Keq = [(H2O)2]/[H2O]2 = 2.1 × 10−21 cm3, keff = kw2[(H2O)2] + kw[H2O], kw2 = 7.4 × 10−12 cm3 s−1 (ref. 33), kw = 3.2 × 10−16 cm3 s−1 (ref. 34) and kSO2 = 3.8 × 10−11 cm3 s−1 (ref. 4)
Relative humidity (%) [H2O]/cm−3 [(H2O)2]/cm−3 (keff/[H2O])/cm3 s−1 (keff/[H2O])/kSO2
0.5 4 × 1015 3.3 × 1010 3.8 × 10−16 1.0 × 10−5
10 7.7 × 1016 1.2 × 1013 1.5 × 10−15 3.9 × 10−5
30 2.3 × 1017 1.1 × 1014 3.8 × 10−15 1.0 × 10−4
50 3.8 × 1017 3.0 × 1014 6.2 × 10−15 1.6 × 10−4
70 5.4 × 1017 6.0 × 1014 8.5 × 10−15 2.3 × 10−4
90 6.9 × 1017 9.9 × 1014 1.1 × 10−14 2.9 × 10−4

In Table 1, quite a few rate coefficients approach or even exceed 1 × 10−10 cm3 s−1, indicating these reactions are as fast as the collision limit (almost every collision event leads to reaction).11 These extremely fast reactions include the reactions of Criegee intermediates with SO2, water dimer (for anti-type Criegee intermediates), and organic/inorganic acids. Similar to typical collision-limit reactions, such Criegee intermediate reactions are barrierless (otherwise the reaction barrier would significantly reduce the rate at room temperature), consistent with quantum chemistry calculations.5 In addition, the involved reactants have large dipole moments. Long range dipole–dipole interactions, which would help to bring two reactant molecules together, may result in a large reaction cross section when there is no significant potential energy barrier. Theoretical investigations show that the reactants first form a strongly bonded pre-reactive complex, and then the system goes through a submerged barrier (its potential energy is lower than that of the reactants at infinite separation) to form products with large exothermicity.5 On the other hand, for some Criegee intermediate reactions, the reaction barrier is significant, resulting in smaller rate coefficients.5,30,37,38

The most intriguing finding is that the reaction of Criegee intermediates with water vapor (monomer10,34,35 and dimer6,30,33) and the unimolecular reaction9,11,17,19 exhibit a very strong structure dependence, whereas Criegee intermediate structure has a much weaker impact on the reaction rates of Criegee intermediates with SO2, NO2, and organic acids, as well as the self-reactions (the variations in the rate coefficients are less than a factor of ∼3). Theoretical investigations indicate that a large part of these reactions (CI + SO2, CI + organic acids) are barrierless;5 hence, the substitution group may play a minor role in the reaction rates.

Various spectroscopic tools have been utilized to detect Criegee intermediates in their kinetics experiments. Before we go to the “Highlights in kinetics studies” section, in which reactions with water and unimolecular reactions of Criegee intermediates will be discussed in depth, we would like to examine the advantages and limitations of various experimental techniques, which would also influence kinetics investigations.

Methods to probe Criegee intermediates

Vacuum UV photoionization mass spectrometry (VUV-PIMS)

This method utilizes a wavelength-tunable VUV light source from a synchrotron radiation facility to record photoionization efficiency spectrum (ion yield as a function of photon energy) of small Criegee intermediates including CH2OO,7anti- and syn-CH3CHOO,10 and (CH3)2COO.11 In most cases, Criegee intermediate has a lower ionization energy than those of its more stable isomers; therefore, this method can selectively ionize a Criegee intermediate at a photon energy higher than the ionization energy of the Criegee intermediate but lower than those of other isomers. In addition, the ionization energy of anti-CH3CHOO is about 0.07 eV lower than that of syn-CH3CHOO, permitting selective detection of anti-CH3CHOO (Fig. 2b).

In principle, all species can be ionized by VUV photons; thus, VUV-PIMS is considered a universal detection method. With the ability to distinguish isomers by their different ionization energies, VUV-PIMS is particularly powerful for detecting reaction products, even those that are unknown. In addition, the array detector of the mass spectrometer can monitor multiple species simultaneously.

Due to the high intensity of the synchrotron radiation light source and high detection sensitivity of mass spectrometry (single ion counting), the overall sensitivity of the VUV-PIMS method is high, which allows for monitoring of low number densities of Criegee intermediates. Although the number density of Criegee intermediates has not been reported in the VUV-PIMS experiments, a long lifetime of up to 18 ms has been observed for CH2OO,31 suggesting the number densities of iodine atoms and CH2OO are low. Based on the relationship between lifetime and [CH2OO]0 observed in our UV absorption experiment,9 [CH2OO]0 in the VUV-PIMS experiment can be estimated to be less than 1011 cm−3 (this value should be considered as an upper limit, because wall loss is the dominant pathway at low-pressure conditions used in the VUV-PIMS experiments).31

Despite its utility at low pressures, it is not easy to apply the VUV-PIMS method to kinetics studies at ambient pressure. Because a typical mass spectrometer requires high vacuum (<10−4 Torr) for its operation, it is not easy to design a high-pressure reactor and couple it with a low-pressure mass spectrometer while maintaining good time resolution (10−5 s) for kinetics studies. The pressure used for the reported VUV-PIMS experiments7,10,11 was about 4 Torr (recent similar experiments were able to use a higher pressure of about 40 Torr).39 On the other hand, detection with electromagnetic waves (UV, IR, MW) does not require low pressure. However, high collision rates at high pressures would broaden spectral linewidth. The effect of the pressure broadening will be discussed in the “IR absorption” section below.

UV depletion

Upon absorption of a UV photon, a Criegee intermediate molecule would dissociate to produce an O atom (mostly in its 1D excited state, O(1D):O(3P) ∼ 4[thin space (1/6-em)]:[thin space (1/6-em)]1 for CH2OO photolysis at 335 nm) and a carbonyl species.27 Using an intense UV laser, Criegee intermediates in a molecular beam can be significantly depleted by photodissociation.22–27 This photodepletion method offers a way to probe the interaction strength between Criegee intermediate molecules and UV radiation. The fraction of photodepleted molecules depends on the laser photon fluence, absorption cross section and dissociation quantum yield, as shown in the following equations.
image file: c7cs00336f-t3.tif
Here N is the number of molecules in the laser interaction volume, σ is the absorption cross section, ϕ is the dissociation quantum yield, F is the laser fluence in photons per unit area, N0 is the initial value of N, and δN is the change in N due to the change in F. As mentioned in the Introduction section, ϕ = 1 for UV photodissociation of Criegee intermediates, similar to the case of O3.3,27

By comparing Criegee intermediate depletion fraction with that of a reference molecule for which the cross section is well established or by calibrating the laser fluence, the absolute absorption cross section can be determined. The resulting peak cross sections have been reported to be (1.23 ± 0.18) × 10−17 cm2 at 340 nm for CH2OO,22 (1.27 ± 0.11) × 10−17 cm2 at 328 nm for CH3CHOO (mostly syn-CH3CHOO),23 and (1.75 ± 0.14) × 10−17 cm2 at 330 nm for (CH3)2COO,24 all of which are comparable to the peak cross section of 1.14 × 10−17 cm2 at 255 nm for O3.

For other spectroscopic transitions (VUV, IR, MW), it is not easy to obtain absolute values of the cross sections; hence, these values are still unavailable in the literature. Given absolute UV absorption cross section, the absolute number density of Criegee intermediate can be deduced from the observed absorbance (Abs) and experimental optical path length (L) using the Beer–Lambert law.

UV absorption

The detection limit for the concentration of Criegee intermediates measured by UV absorption can be estimated using the following equation,
image file: c7cs00336f-t4.tif
Here I and I0 are the probe light intensities with and without the Criegee intermediate, respectively. For typical absorption experiments, L ranges from 4 to 40 m.20–23 If Abs is about 10−3 for a reasonable signal, the corresponding [CI] is about 1011 cm−3 (for L = 10 m).

In principle, UV absorption experiments are not significantly affected by experimental conditions such as temperature, pressure, and humidity, provided that the windows of the reactor are well purged (otherwise, undesirable photochemistry may take place on the window surfaces).

The UV absorption spectra of Criegee intermediates are very broad (Fig. 4) due to their short excited-state lifetimes and broad Franck–Condon envelope.3 Pressure broadening thus has almost no effect on the UV absorption spectra. Therefore, the direct detection of Criegee intermediates at high pressures (up to 760 Torr) mostly utilizes UV set up. The ability to detect Criegee intermediates at near ambient pressure is important for several reasons. First, if a kinetics measurement requires a high partial pressure of a reactant, the total pressure cannot be lower than the partial pressure. Second, and more importantly, the low pressure rate may differ significantly from the rate at ambient pressure. Finally, the pressure dependence of reaction kinetics provides information about reaction mechanism.

The pressure-dependent rate coefficients for the reactions of SO2 with CH2OO and with (CH3)2COO are plotted in Fig. 5.29 The rate coefficient at pressures less than 100 Torr is significantly smaller than that at ambient pressure for the (CH3)2COO + SO2 reaction. Chhantyal-Pun et al. have also observed a similar pressure dependence.11

image file: c7cs00336f-f5.tif
Fig. 5 Pressure-dependent rate coefficients for the reactions of SO2 with CH2OO (blue squares) and with (CH3)2COO (red circles) measured by UV absorption spectroscopy at 298 K.29

Because the UV absorption bands are very broad and overlap with each other (Fig. 4), it is inefficient to distinguish different Criegee intermediates using UV spectroscopy. Although the peak positions of the UV spectra of anti-type and syn-type Criegee intermediates may differ by up to 36 nm, it is still not easy to fully separate these two conformers using UV spectroscopy alone. Fortunately, anti- and syn-conformers often have significantly different reactivities (for example, anti-conformers react with water much faster than syn-conformers), which enables investigation of their kinetics.10,21,30


The very high spectral resolution of Fourier transform microwave (FT-MW) spectroscopy makes it a powerful tool for the detection of unstable species. Because Criegee intermediates have very large dipole moments (4.1 Debye for CH2OO),3 their pure rotational transitions (MW spectroscopy) have large signal intensities. Virtually any species with a modest electric dipole moment can be detected using FT-MW. For example, Endo and coworkers have detected a hydrogen-bonded complex of CH2OO and H2O.40 In addition, if the molecular dipole moment is known, FT-MW can determine the population of the spectral carrier. For example, the population ratio of anti-CH3CHOO to syn-CH3CHOO has been estimated to be 1[thin space (1/6-em)]:[thin space (1/6-em)]5 in experiments based on theoretical values of the dipole moments.13

However, most FT-MW experiments are designed for spectroscopic purposes, not for measuring kinetics, in which temperature T, pressure P and reaction time are not very well defined. Nonetheless, Suits' group has applied chirped-pulse MW spectroscopy to kinetics studies in a uniform flow generated with a Laval nozzle (P and T can be well characterized).41 This technique exhibits better performance at low P and T (P ≤ 0.1 Torr; T ∼ 22 K). It would be difficult to perform MW spectroscopy under high-pressure conditions because collisions would attenuate the line intensities of the rotational transitions.41

IR absorption

IR absorption induces vibrational transitions of a molecule, which reveal its structural signatures and/or fingerprints. A step-scan Fourier-transform infrared (ss-FTIR) spectrometer14–16 has a spectral resolution of 0.3 or 1 cm−1 and is able to resolve anti/syn-conformers of CH3CHOO,15 which is a useful advantage of this method. However, the sensitivity of ss-FTIR method is not high. As a result, high concentrations of Criegee intermediates are needed, resulting in a short lifetime of the Criegee intermediate (ca. 10–50 μs) due to self-reaction and reaction with byproducts (mostly iodine atoms in the case of photolysis synthesis).8 The short lifetime limits the time window to probe the kinetics, making it difficult to investigate slow reactions.

The sensitivity can be improved if the spectral linewidth is reduced. For the same transition strength, a narrower linewidth would lead to a taller peak (while the peak area roughly remains constant) and thus higher detection signal. Typically, detection sensitivity is proportional to peak height rather than peak area. While it is impractical to further increase the resolution of the ss-FTIR spectrometer (this requires a great increase in the number of mirror steps), using a high-resolution IR laser is a promising way to achieve higher spectral resolution and detection sensitivity.

Very recently, our group has utilized a high-resolution (<0.004 cm−1) quantum cascade IR laser to detect the absorption of CH2OO near 1286 cm−14 band).42 High-resolution spectroscopy not only allows us to determine spectroscopic constants with a high level of precision but also greatly improves detection sensitivity. Compared to a lower-resolution experiment (FWHM 0.3 cm−1),14 the sensitivity for monitoring a non-overlapping line would increase by a factor of 102 when pressure broadening is insignificant (<10 Torr). However, even with this improvement, it is still challenging to probe CH2OO at low concentrations due to its weaker IR transition strength (compared to strong UV transition). Fortunately, this vibrational band has a very congested Q branch, which has a much higher intensity than those of the P- and R-branch lines due to overlapping of several transitions.14,42 Furthermore, since the intense peaks of the Q-branch are already broad, the effect of pressure broadening is not as strong as that of a single spectral line. In Fig. 6, it can be seen that the narrow spectral peaks quickly lose their features, while the broad peaks retain considerable intensity as pressure increases. We have demonstrated that sensitive detection of CH2OO (down to 1 × 1012 cm−3, signal-to-noise ratio >10, with 4 m absorption path length) can be realized at a total pressure of 30 Torr (N2 or O2 buffer gas).42

image file: c7cs00336f-f6.tif
Fig. 6 Experimental high-resolution IR spectra of CH2OO (Q branch, ν4 band) showing pressure broadening. The total pressures (N2 buffer gas) are noted in the figure.42

High-resolution IR spectroscopy not only improves detection sensitivity, but also provides accurate molecular structural parameters. In principle, this method is immune to spectral contamination and could selectively monitor one Criegee intermediate in a mixture of multiple Criegee intermediates.

IR action

Typical absorption experiments detect a small decrease in light intensity, and the sensitivity is limited by the stability of the apparatus. This limitation can be lifted if a product of the photoabsorption process can be detected. This can be a zero-background experiment since there is no product when no absorption takes place, and the detection sensitivity can be greatly enhanced. The spectrum can be obtained by scanning the excitation wavelength while monitoring the product. This detection scheme is called action spectroscopy. Very successful examples of IR action spectroscopy of Criegee intermediates have been demonstrated by Lester's group, who detected the OH product by laser induced fluorescence.17–19 We will discuss the investigation of unimolecular decomposition using this method in the following section.

Highlights in kinetics studies

Thermal decomposition

OH radical has been observed as a product of unimolecular decomposition of several Criegee intermediates including syn-CH3CHOO,17syn-C2H5CHOO,18 and (CH3)2COO.19 Lester's group has detected the OH product by laser induced fluorescence after exciting Criegee intermediates in a molecular beam using an IR laser (via combination bands or overtones). The corresponding IR action spectra show significant line broadening, which is attributed to fast intramolecular vibrational relaxation (IVR) on a picosecond time scale.17–19

By varying the excitation-probe delay time, the time profile of OH formation can be measured. The results show that OH formation rate (3 × 105 to 5 × 107 s−1) depends strongly on excitation energy (4043 to 5971 cm−1). Since OH product formation is much slower than IVR, complete energy randomization is expected, suggesting that a statistical model would work well for describing this process provided that the potential energy surface is sufficiently accurate. The bottleneck of the unimolecular decomposition is the tunneling of an H atom from the α-carbon to the terminal O atom.9,17,19 Interestingly, although decomposition of syn-CH3CHOO has the highest reaction barrier (17.1 kcal mol−1),17 it is faster than decompositions of (CH3)2COO (barrier = 16.2 kcal mol−1)19 and syn-C2H5CHOO (barrier = 16.5 kcal mol−1)18 at the same total energy. This seemingly counterintuitive result can be explained by the fact that syn-CH3CHOO has the lowest density of vibrational states (thus less dilution of vibrational energy); hence, the microcanonical rate is the fastest.

The observed OH formation rate as a function of total energy can be used to fine tune parameters of the potential energy surface like barrier height and barrier thickness, which permits more accurate prediction of thermal decomposition rates. Room-temperature decomposition rates have been predicted to be 122, 279, and 276 s−1 for syn-CH3CHOO, syn-C2H5CHOO and (CH3)2COO, respectively,17–19 which are fast enough to make thermal decomposition the main decay pathway for these Criegee intermediates in the troposphere.

The unimolecular decay of thermalized (CH3)2COO at various temperatures (283–323 K) has been measured with UV absorption.9 Due to fast reactions of (CH3)2COO with other radical species including iodine atoms, OH radicals and even (CH3)2COO itself, for which the concentrations are mutually proportional in the preparation process [(CH3)2CI2 + → (CH3)2CI + I; (CH3)2CI + O2 → (CH3)2COO + I], the lifetime of (CH3)2COO is shorter at higher initial concentrations. To deduce the rate coefficient of the unimolecular process, which should be independent of reactant concentrations, the observed first-order rate coefficient of (CH3)2COO must be extrapolated to zero [(CH3)2COO]0 to remove the contributions of bimolecular processes for which pseudo first-order rate coefficients depend on reactant concentrations that are roughly proportional to [(CH3)2COO]0. Fortunately, Criegee intermediates have very large UV absorption cross sections, which allow the detection of Criegee intermediates at low concentrations. Two kinetics models have been used in the analysis. A simplified model only considers first-order and pseudo-first-order reactions as shown below.

image file: c7cs00336f-t5.tif

kobskth + krad[radicals]0; [CI](t) = [CI]0ekobst
where kth is the thermal decomposition rate coefficient and krad and [radicals] represent the effective rate coefficient and overall concentration of radical species including iodine atoms, OH radicals and (CH3)2COO. In the photolysis process, the initial concentrations of these radicals are proportional to [(CH3)2COO]0, i.e., [radicals]0 ∝ [(CH3)2COO]0. As a result, kobs is larger at higher [(CH3)2COO]0. By measuring kobs at various [(CH3)2COO]0, we can obtain kth by extrapolating kobs to zero [(CH3)2COO]0. It should be noted that if the precursor sample contains an impurity such as I2, more radicals (e.g., I atoms) would be generated by the photolysis laser pulse and kobs would become larger. To remove this contribution, the sample concentration is varied and kobs is extrapolated to zero sample concentration (the impurity concentration is proportional to sample concentration in our experiment).9

Here, approximating [radicals] as a constant value appears to be satisfactory since most of the decay traces of [(CH3)2COO](t) can be fit well to a single exponential function. Nonetheless, we use a more complete model that includes a second-order term to assess the above approximation, as shown in the following expressions.

[radicals] ≅ [radicals]fix + α[CI]

image file: c7cs00336f-t6.tif
Here [radical]fix is a fixed value that does not vary during the reaction and is proportional to [CI]0 and α[CI] represents the change in [radicals] during the reaction, which effectively introduces a second-order term to the decay kinetics of [CI]. As expected, the model including the second-order term fits the experimental data slightly better; see Fig. 7 for an example of decay trace with a high second-order contribution.9 The key finding is that both models give essentially the same value of kth, indicating that the kinetics analyses are robust as long as quality of data is good enough for extrapolation to zero [CI]0.9

image file: c7cs00336f-f7.tif
Fig. 7 Comparison of two kinetics models that differ in treatment of the second-order contribution for the fitting of the concentration profile of (CH3)2COO at a high [(CH3)2COO]0 of about 9 × 1011 cm−3.9 The fitting of the blue line uses only first-order decay kinetics; the red line includes both first-order and second-order kinetics. The difference between the two models becomes smaller at lower [(CH3)2COO]. Panel (a) shows a time range of 0–1.5 ms; panel (b) shows 0–5 ms.

The thermal decomposition rate of (CH3)2COO is determined to be 360 s−1 at 298 K (also see Table 1) and shows a strong temperature dependence. Fitting the Arrhenius equation to thermal decomposition rates at different temperatures yields an activation energy of 5.8 kcal mol−1.9 Here, due to the tunneling mechanism, the value of activation energy is significantly lower than the barrier height.

It is important to note that the two types of experiments described above (IR action and UV absorption) probe the same system but at very different energy distributions: IR excitation of the Criegee intermediate in a molecular beam (collision-free) gives a very well-defined energy,19 while the thermalized condition used in the UV absorption experiment yields a very broad Boltzmann distribution.9 Nonetheless, the observed rates from both types of experiments agree well with theoretical predictions, which essentially use the same potential energy surface as demonstrated in Fig. 8, indicating the suitability of the theoretical method. Moreover, direct evidence of H-atom tunneling has been provided by a similar thermal experiment with deuterium isotope substitution, in which very significant reduction in decomposition rates was observed for (CD3)2COO.9

image file: c7cs00336f-f8.tif
Fig. 8 Comparison between experimental and theoretical thermal decomposition rates of (CH3)2COO. Data points with uncertainty are measurements by Smith et al. (blue circles)9 and Chhantyal-Pun et al. (red triangle).11 Lines are master equation modeling of the unimolecular decay at the high-pressure limit. The solid line is the original Eckart model from ref. 19; the dash-dotted line is the adjusted Eckart model that has been refined to fit the microcanonical rates obtained by IR excitation at 4043 to 5971 cm−1; the dashed line is a model that neglects tunneling.19 The theoretical thermal decomposition rate of (CD3)2COO from ref. 43 (with a slightly different potential energy surface) is plotted with the dotted line.

Facile unimolecular decomposition of Criegee intermediates is expected only for structures with H-atoms on the α-carbon at the syn-position. The corresponding transition state has a 5-membered ring structure (Fig. 9a). For anti-conformers or similar structures (CH2OO, anti-CH3CHOO, etc.), the H-atoms are far away from the terminal O atom and theory predicts a different pathway for the lowest-energy unimolecular process involving formation of a CO bond between the terminal oxygen and the central carbon to form a dioxirane structure (Fig. 9b).5,9 The barriers are calculated to be 21.7 and 20.5 kcal mol−1 for (CH3)2COO and CH2OO, respectively,5 which are significantly higher than those for the H-atom transfer in syn-type Criegee intermediates (17.1 and 16.2 kcal mol−1 for syn-CH3CHOO and (CH3)2COO, respectively).17,19 Together with highly inefficient O-atom tunneling, the predicted unimolecular rates are quite slow (∼0.3 s−1 for CH2OO),5 and consistent with a previous ozonolysis experiment in which thermal decomposition rate of CH2OO was measured to be ca. 0.2 s−1.34

image file: c7cs00336f-f9.tif
Fig. 9 Transition-state geometries of the two lowest-energy unimolecular pathways of (CH3)2COO: (a) hydrogen atom transfer channel to form a COOH functional group and (b) OO bending channel to form dioxirane.9

Reaction with water vapor

To determine the atmospheric fate of Criegee intermediates, one must consider their reactions with water vapor because water vapor is the third most abundant gas (1.6% v/v at relative humidity RH = 50% and 298 K) in typical ambient air. Due to the overwhelming number density of H2O (versus ppmv or ppbv for other trace gases), its effect could be important even when the involved reaction rate coefficients are small.

For the simplest Criegee intermediate CH2OO, reactivity towards water was previously a controversial issue. Earlier studies on ozonolysis of C2H4 indicated that CH2OO reacts with water, although there is a large scatter in the reported rate coefficients (10−17–10−12 cm3 s−1). Such experiments often rely on a final product analysis with an assumed reaction mechanism to determine relative rates. For example, the ratio of the rate coefficient for CH2OO + H2O to that for CH2OO + SO2 has been reported to be 2.3 × 10−4 for [H2O] ≤ 4 × 1015 cm−3 and 8.3 × 10−4 for [H2O] ≤ 2 × 1017 cm−3.6 Based on these values, the CH2OO + SO2 reaction would not compete with CH2OO + H2O because the typical mixing ratio of water vapor is roughly 106 times that of SO2.

However, several recent investigations utilizing the more efficient preparation of CH2OO by reaction of CH2I with O2 did not observe any significant reaction of CH2OO with water. Stone et al. monitored the H2CO reaction product by laser-induced fluorescence and placed an upper limit of 9 × 10−17 cm3 s−1 on the rate coefficient of CH2OO + H2O.44 Ouyang et al. monitored NO3 production in the reaction of CH2OO + NO2 in the presence of water vapor and determined the ratio of the rate coefficient for CH2OO + H2O (2.5 × 10−17 cm3 s−1) to that for CH2OO + NO2 (7 × 10−12 cm3 s−1 adapted from Welz et al.)7 to be 3.6 × 10−6.45

More recently, Berndt et al. measured the yield of H2SO4 in the ozonolysis of C2H4 at various SO2 and H2O concentrations.6 Due to the extremely high sensitivity of their detection method (chemical ionization mass spectrometry), they can use low concentrations of the precursors ([O3] ≈ 1011 cm−3; [C2H4] ≈ 1013 cm−3) to simplify reaction mechanism. The main involved reactions are listed below.

O3 + C2H4 → H2CO + CH2OO

CH2OO + SO2 → H2CO + SO3

CH2OO + nH2O → Products

SO3 + H2O → H2SO4
They found that the CH2OO + nH2O reaction is significant, relative to CH2OO + SO2 and exhibits second-order kinetics with respect to [H2O]; this suggested that two H2O molecules are involved in the reaction, and water dimer was assigned as the corresponding reactant.

Since water monomer and dimer exist in equilibrium, [(H2O)2] = Keq [H2O]2 and the ratio of dimer to monomer is proportional to [H2O]. The reaction of CH2OO with water dimer may not be obvious at low water concentrations and could be easily neglected. Therefore, one has to be careful when comparing the results from different experiments that used different ranges of [H2O] and [(H2O)2]. Berndt et al. demonstrated that the reaction of CH2OO with water vapor is dominant for atmospherically relevant water concentrations, which is in agreement with a series of former ozonolysis studies,6 but inconsistent with more recent studies by Stone et al.44 and by Ouyang et al.45 who used CH2I2 photolysis to generate CH2OO.

Our group directly measured the time profile of [CH2OO] using UV absorption.33 With this method, we were able to use water concentrations as high as RH = 85% at 298 K in the reaction cell. The results showed a fast reaction of CH2OO with water vapor. Importantly, a clear quadratic dependence on water concentration is observed for the CH2OO decay rate, indicating two water molecules participate in the reaction. Welz et al., who first reported the novel synthesis of CH2OO by CH2I2 photolysis, also investigated the reaction of CH2OO with water, but failed to observe any effect of added water vapor in the CH2OO decay.7 This can be explained with the water dimer mechanism—their highest water concentration was ca. 3.1 × 1016 cm−3 (RH = 4% at 298 K), which is not high enough to produce a significant concentration of water dimer. This limitation of water partial pressure (∼1 Torr) was due to the limit of total pressure (4 Torr) in their PIMS apparatus.

To explain the results of Stone et al. and Ouyang et al., it is important to consider that they did not probe CH2OO directly; instead, they monitored the presumed reaction products (H2CO or NO3) in their CH2OO reaction systems.44,45 However, the reaction of CH2OO with water vapor may not produce H2CO as a nascent product, and the reaction of CH2OO with NO2 may not produce NO3 directly. Instead, adduct formation (CH2OO + H2O → HOCH2OOH; CH2OO + NO2 → CH2OONO2) is more likely.39,46 Therefore, the formation of H2CO or NO3 cannot be used as a proxy for the kinetics of CH2OO reaction with water vapor.

The reaction of CH2OO with water monomer has been found to be quite slow at ambient conditions34 and the reaction barrier is significant (for room-temperature reaction, see Fig. 10).35,46,47 On the other hand, the CH2OO–(H2O)2 complex is very stable due to hydrogen bonding and the zwitterionic character of CH2OO. The energy of the transition state of the water dimer reaction is also low. Consistent with theoretical calculations, our experiment shows this reaction is faster at lower temperatures and exhibits an activation energy of ca. −8 kcal mol−1.33 Theory has also predicted that water trimer is more reactive towards CH2OO than water dimer.46 However, the population of water trimer in the atmosphere is too low (10−3 relative to dimer concentration) to have any significant effect.

image file: c7cs00336f-f10.tif
Fig. 10 Diagram of relative potential energy for reactions of small Criegee intermediates with water monomer (left) and water dimer (right). Energetics are calculated by Lin et al. at the level of QCISD(T)/CBS//B3LYP/6-311+G(2d,2p) and vibrational zero-point corrections by B3LYP/6-311+G(2d,2p).35,47

The concentration ratio of water dimer to water monomer is about 10−3 for typical ambient air (see Table 2) and water dimer is much more abundant than most other trace gases including SO2 and NO2 ([SO2] = 1 × 1012 cm−3, assuming 40 ppbv). As a result, the water dimer reaction predominates in CH2OO decay under typical tropospheric conditions. However, under extreme conditions, the water monomer reaction may compete with the dimer reaction. One such condition is at very low water concentration, such that the water dimer concentration is relatively low. Berndt et al. have studied the CH2OO reaction with water vapor at [H2O] ≤ 1 × 1015 cm−3 (RH ≤ 0.13% at 298 K) and reported the monomer rate coefficient to be (3.2 ± 1.2) × 10−16 cm3 s−1 at 297 K.34 Another condition is at high temperature. As temperature increases, the water dimer reaction slows down, while the water monomer reaction speeds up. Our group has performed a kinetics study at higher temperatures up to 358 K and found the monomer rate coefficient to be 7.3 × 10−16 cm3 s−1 at 358 K.35

Because the water dimer reaction predominates in the reaction of CH2OO with water vapor, the competition of SO2 against water vapor for reaction with CH2OO depends on relative humidity. Here we use the best-available bimolecular rate coefficients of CH2OO reactions with water monomer (kw), water dimer (kw2) and SO2 (kSO2) and the notation, keff = kw[H2O] + kw2[(H2O)2] and keff/[H2O] = kw + kw2[(H2O)2]/[H2O] in order to compare with earlier experiments that did not recognize water dimer as the main reactant and used water monomer concentration to report effective rate coefficients; the results are listed in Table 2. It can be seen that earlier ozonolysis studies6 overestimated the value of keff/[H2O] relative to kSO2.

Water dimer is assumed to be the main reactant in the reaction of CH2OO with water vapor. However, we cannot rule out other reaction pathways that also involve two water molecules, e.g., CH2OO(H2O) + H2O → CH2OO(H2O)2 → product + H2O. Quantum chemistry calculations show that the nascent product is hydroxylmethyl hydroperoxide (HO–CH2–OOH) and the additional H2O acts as a catalyst.46 This reaction can serve as an important model for water-catalyzed reactions, which have recently gained considerable attention. Theoretical studies have shown that for a number of reactions involving H-atom transfer, the participation of an additional water molecule can serve as a relay for H-atoms and significantly lower reaction barrier.48 It should be noted that not only the energy effect (barrier height) but also the entropy effect must be considered because the addition of H2O to the transition-state structure would significantly reduce entropy and thus would lead to a reduced concentration of such a structure and a slower overall rate. To make a significant contribution to the reaction rate, the lowering of barrier height by addition of H2O has to be large enough to overcome the cost of the entropy term (recall that ΔG = ΔHTΔS). This explains why syn-CH3CHOO and (CH3)2COO react slowly with water dimer while their transition-state energies are lower than those of the separated reactants (see Fig. 10).

One example of a water-catalyzed reaction is the hydration of SO3 to form H2SO4. This reaction is catalyzed by adding one additional H2O molecule; theoretical calculations also show that the barrier height for the formation of H2SO4 from the SO3–(H2O)n complex drops from 28 kcal mol−1 for n = 1 to 11 kcal mol−1 for n = 2.49 In addition, theory suggests that the hydration of formaldehyde to form diol is also catalyzed by water, and the very significant barrier height (ca. 42 kcal mol−1) is reduced to only 0.8 kcal mol−1 when two water molecules are present.48 While there are a number of systems that have been investigated theoretically,48 the related experimental investigations are quite sparse. The detailed kinetics studies of CH2OO reaction with water dimer offer a clear example to support the mechanism of water catalysis, which was initially proposed by theory.46

Compared to CH2OO, anti-CH3CHOO shows slightly higher reactivity towards water vapor.30 Both water monomer and water dimer react with anti-CH3CHOO with the relative contribution depending on the temperature and relative humidity.30 However, syn-CH3CHOO and (CH3)2COO show little reactivity towards water vapor.29,30 Even at high RH of up to 85%, we do not observe any reaction of (CH3)2COO with water vapor.29 Theoretical calculations indicate that an alkyl group at the syn position would prevent the water reactant from getting close to the reactive central carbon atom, thus leading to reduced reactivity.30,46

We have demonstrated that the reaction of (CH3)2COO with SO2 is faster than that of CH2OO by a factor of 3.29 It is interesting that the substitution groups have different effects (opposite trends) on the reactivity of Criegee intermediates towards SO2 and towards water vapor. This can be rationalized by quantum chemistry calculations, showing that SO2 reactions with Criegee intermediates are barrierless,5 regardless of the substitution groups. This finding has an important implication that some Criegee intermediates (depending on structure) may survive high humidity conditions and accumulate to a high enough concentration to oxidize atmospheric SO2 or other trace gases.

Currently, only the kinetics of a few simple Criegee intermediates (CH2OO, anti/syn-CH3CHOO and (CH3)2COO) have been widely investigated.4 The available results show that there is strong structure dependence in (i) the reactivity towards water vapor (facile only for anti-type Criegee intermediates) and (ii) unimolecular decomposition (fast only for syn-type Criegee intermediates). It should be noted that there are Criegee intermediates with more complicated structures. Their reactivity is yet to be studied.


A number of important future research directions have been pointed out in earlier review articles, including photoelectron spectroscopy, crossed molecular beam scattering, visible spectroscopy (via A ← X transition), product yields, and conformer-resolved reactions.3,4 Here we will highlight some directions from our perspective.

Product detection and yields

Most kinetics experiments of Criegee intermediates detect only the Criegee intermediates, not the nascent products. The identification and yields of the products as well as their fates are important in assessing the impact of Criegee intermediates in atmospheric chemistry, as emphasized in previous review articles. However, such experimental measurements are challenging and there are only a couple of examples available in the literature. Using the VUV-PIMS method, Taatjes et al. have observed hydroxyacetone (CH3C(O)CH2OH) as a stable end product from reactions of (CH3)2COO in a flow tube.50 Possible reaction pathways include unimolecular isomerization via hydrogen atom transfer and –OH group migration as well as self-reaction of Criegee intermediates. Using the same method, Caravan et al. observed a species that is consistent with the exact Criegee–NO2 mass in the reactions of NO2 with both CH2OO and CH3CHOO, suggesting the adduct is the major reaction product, while they did not detect any NO3 product and reported the NO3 yield to be minor (<30% in the reaction of NO2 + CH3CHOO).39

These and similar works highlight the importance of product detection and quantification, since a number of kinetics investigations have followed the formation of a postulated product. For example, Ouyang et al. followed NO3 formation in the kinetics of CH2OO reactions in the presence of NO2.45 If NO3 is not formed directly in the reaction of CH2OO with NO2, the kinetics analysis may need revision.

The VUV-PIMS method requires a synchrotron light source facility, which has limited accessibility. It may be desired to combine the high sensitivity of FT-MW detection (which requires low pressure) with pin-hole sampling of a flow reactor (similar to the case of VUV-PIMS) to achieve sensitive product detection for reactions under a medium pressure (10–100 Torr) range. We cannot see any reason for not doing this, provided that experts in both sub-fields (FT-MW and flow reactor) can work together. Hopefully, this would open a new door in reaction kinetics.

A significant number of kinetics experiments are conducted under pressures much lower than 1 atm. Product yields often change with pressure and temperature. Thus, it is important to measure pressure and temperature dependence, which will help in understanding reaction mechanisms and estimating the impact on atmospheric chemistry.

More complicated Criegee structures

Until now, only a few simple Criegee intermediates (CH2OO, anti/syn-CH3CHOO, anti/syn-C2H5CHOO and (CH3)2COO) have been directly detected in laboratory. Photolysis method requires different precursors (RR′CI2) to make different Criegee intermediates, but only a limited selection of these precursors are available. Studies of the above simple Criegee intermediates have demonstrated strong structure-dependent reactivity. More complicated structures deserve future investigation; for example, in ozonolysis of isoprene or pinenes, which are abundant in forest areas, Criegee intermediates with a vinyl substitution group or a ring structure may be produced. Little is currently known about these complicated Criegee structures.

Theory with uncertainty assessment

High-level quantum chemistry and statistical rate theory have been of great help in interpreting experimental observations and making predictions. However, the uncertainties associated with theoretical results may also lead to some confusion. In a number of cases, the theoretically predicted rates of Criegee intermediate reactions do not agree with each other or with experiments. For example, the reported theoretical rate coefficients of CH2OO reaction with ozone span a wide range of 10−16–10−12 cm3 s−1.5 The discrepancy may come from different reaction pathways or uncertainties in the calculated barrier heights. In addition, for the reaction of CH2OO with H2O monomer, early theoretical values range from 6 × 10−18 to 3 × 10−15 cm3 s−1 for the rate coefficient at 298 K.5

Theoretical studies have successfully predicted the strong structure dependence of the reactions of water with Criegee intermediates. The reaction of water monomer with an anti conformer of Criegee intermediate is much faster than that with a syn conformer by 3 to 4 orders of magnitude.5,46 Interestingly, we can only find one theoretical study showing the importance of water dimer in the reactions of water vapor with anti-type Criegee intermediates;46 this work had not received due attention until experiments demonstrated the fast reaction of water dimer in 2015.33

While reaction rate calculations alone may not give us a definitive value for a rate coefficient with sufficiently low uncertainty, precise experimental data can help point out possible problems of (or increase our confidence in) certain theoretical approaches. The predictive power of theoretical study can be improved by careful comparison (calibration) with experimental data.

Since direct probing has only been possible for a very limited number of Criegee intermediates (CH2OO, CH3CHOO, (CH3)2COO, and C2H5CHOO), theory is still necessary to yield insight into the properties of more complicated Criegee intermediates that also form in the atmosphere.

Concluding remarks

Various spectroscopic tools have been utilized not only to investigate the physical properties of Criegee intermediates, but also to probe Criegee intermediates in kinetics systems to determine their reactivity. Advantages and limitations of each tool in kinetics studies have been discussed in depth. Two important types of Criegee intermediate reactions, unimolecular decomposition and reaction with water dimer, are highlighted in this Tutorial Review.

The self-reactions of Criegee intermediates are rather fast. Therefore, to measure the kinetics of a Criegee intermediate reaction with another species, it is desirable to detect the Criegee intermediate at a low enough concentration to remove interference caused by self-reaction. In addition, byproducts in the reaction system (e.g., iodine atoms from photolysis of the precursor) often react quickly with Criegee intermediates. As a result, the lifetime of a Criegee intermediate is longer at lower concentrations. A longer lifetime is required to enable the investigation of slower reactions. In order to obtain results that are relevant to atmospheric reactions, the ability to monitor Criegee intermediates at near ambient pressure is also important.

Ozonolysis reaction systems in laboratory studies, in which the monitored products may not come directly from Criegee intermediates, can be rather complicated. However, it is still possible to obtain useful information if the reaction mechanism is well controlled and certain key reaction rate coefficients and yields are known. For example, Berndt et al. have determined the ratio of the rate coefficients for CH2OO reactions with water dimer and with SO2;6 when a reliable rate coefficient for CH2OO reaction with SO2 is used, the deduced rate coefficient for CH2OO reaction with water dimer agrees well with the result from direct kinetics experiments.33 Another example is the ratio of rate coefficients for (CH3)2COO unimolecular decomposition and reaction with SO2: the ratio reported from ozonolysis experiments agrees reasonably well with that from direct kinetics studies (see ref. 9 for a detailed comparison).

Due to the limited availability of suitable precursors, the photolysis synthesis method can only be used to prepare a few simple Criegee intermediates. More knowledge is required regarding larger and more complicated Criegee intermediates that also form in the atmosphere to understand their roles in atmospheric chemistry. Ozonolysis studies and theoretical investigations can provide valuable information, provided that careful calibrations have been performed with similar reaction systems involving simple Criegee intermediates. In other words, reliable kinetics results, which are only available for simple Criegee intermediates, may be used to guide research on larger Criegee intermediates.

The impact of Criegee intermediates in the atmosphere is still an open issue; a major uncertainty is in their atmospheric concentrations. Refined kinetics data would help modelers to better estimate the influence of Criegee intermediate reactions in the atmosphere.


This work is supported by Academia Sinica and Ministry of Science and Technology, Taiwan (MOST 103-2113-M-001-019-MY3). The authors thank Dr Mica Smith, Dr Kaito Takahashi, and Prof. Yuan T. Lee for insightful discussions.


  1. N. M. Dohanue, G. T. Drozd, S. A. Epstein, A. A. Presto and J. H. Kroll, Phys. Chem. Chem. Phys., 2011, 13, 10848–10857 RSC .
  2. C. A. Taatjes, D. E. Shallcross and C. J. Percival, Phys. Chem. Chem. Phys., 2014, 16, 1704 RSC .
  3. D. L. Osborn and C. A. Taatjes, Int. Rev. Phys. Chem., 2015, 34, 309–360 CrossRef CAS .
  4. Y.-P. Lee, J. Chem. Phys., 2015, 143, 020901 CrossRef PubMed .
  5. L. Vereecken, D. R. Glowacki and M. J. Pilling, Chem. Rev., 2015, 115, 4063–4114 CrossRef CAS PubMed .
  6. T. Berndt, J. Voigtländer, F. Stratmann, H. Junninen, R. L. Mauldin, M. Sipilä, M. Kulmala and H. Herrmann, Phys. Chem. Chem. Phys., 2014, 16, 19130–19136 RSC .
  7. O. Welz, J. D. Savee, D. L. Osborn, S. S. Vasu, C. J. Percival, D. E. Shallcross and C. A. Taatjes, Science, 2012, 335, 204 CrossRef CAS PubMed .
  8. W.-L. Ting, C.-H. Chang, Y.-F. Lee, H. Matsui, Y.-P. Lee and J. J.-M. Lin, J. Chem. Phys., 2014, 141, 104308 CrossRef PubMed .
  9. M. C. Smith, W. Chao, K. Takahashi, K. A. Boering and J. J. Lin, J. Phys. Chem. A, 2016, 120, 4789–4798 CrossRef CAS PubMed .
  10. C. A. Taatjes, O. Welz, A. J. Eskola, J. D. Savee, A. M. Scheer, D. E. Shallcross, B. Rotavera, E. P. F. Lee, J. M. Dyke, D. K. W. Mok, D. L. Osborn and C. J. Percival, Science, 2013, 340, 177 CrossRef CAS PubMed .
  11. R. Chhantyal-Pun, O. Welz, J. D. Savee, A. J. Eskola, E. P. E. Lee, L. Blacker, H. R. Hill, M. Aschcroft, M. A. H. Khan, G. C. Lloyd-Jones, L. Evans, B. Rotavera, H. Huang, D. L. Osborn, D. K. W. Mok, J. M. Dyke, D. E. Shallcross, C. J. Percival, A. J. Orr-Ewing and C. A. Taatjes, J. Phys. Chem. A, 2017, 121, 4–15 CrossRef CAS PubMed .
  12. M. Nakajima and Y. Endo, J. Chem. Phys., 2013, 139, 101103 CrossRef PubMed .
  13. M. Nakajima, Q. Yue and Y. Endo, J. Mol. Spectrosc., 2015, 310, 109 CrossRef CAS .
  14. Y.-H. Huang, J. Li, H. Guo and Y.-P. Lee, J. Chem. Phys., 2015, 142, 214301 CrossRef PubMed .
  15. H.-Y. Lin, Y.-H. Huang, X. Wang, J. M. Bowman, Y. Nishimura, H. A. Witek and Y.-P. Lee, Nat. Commun., 2015, 6, 7012 CrossRef CAS PubMed .
  16. Y.-Y. Wang, C.-Y. Chung and Y.-P. Lee, J. Chem. Phys., 2016, 145, 154303 CrossRef PubMed .
  17. Y. Fang, F. Liu, V. P. Barber, S. J. Klippenstein, A. B. McCoy and M. I. Lester, J. Chem. Phys., 2016, 145, 234308 CrossRef PubMed .
  18. Y. Fang, F. Liu, S. J. Klippenstein and M. I. Lester, J. Chem. Phys., 2016, 145, 044312 CrossRef PubMed .
  19. Y. Fang, V. P. Barber, S. J. Klippenstein, A. B. McCoy and M. I. Lester, J. Chem. Phys., 2017, 146, 134307 CrossRef PubMed .
  20. L. Sheps, J. Phys. Chem. Lett., 2013, 4, 4201 CrossRef CAS PubMed .
  21. L. Sheps, A. M. Scully and K. Au, Phys. Chem. Chem. Phys., 2014, 16, 26701–26706 RSC .
  22. W.-L. Ting, Y.-H. Chen, W. Chao, M. C. Smith and J. J.-M. Lin, Phys. Chem. Chem. Phys., 2014, 16, 10438 RSC .
  23. M. C. Smith, W.-L. Ting, C.-H. Chang, K. Takahashi, K. A. Boering and J. J.-M. Lin, J. Chem. Phys., 2014, 141, 074302 CrossRef PubMed .
  24. Y.-P. Chang, C.-H. Chang, K. Takahashi and J. J.-M. Lin, Chem. Phys. Lett., 2016, 653, 155–160 CrossRef CAS .
  25. J. M. Beames, F. Liu, L. Lu and M. I. Lester, J. Am. Chem. Soc., 2012, 134, 20045 CrossRef CAS PubMed .
  26. J. M. Beames, F. Liu, L. Lu and M. I. Lester, J. Chem. Phys., 2013, 138, 244307 CrossRef PubMed .
  27. M. F. Vansco, H. Li and M. I. Lester, J. Chem. Phys., 2017, 147, 013907 CrossRef PubMed .
  28. C. C. Womack, M.-A. Martin-Drumel, G. G. Brown, R. W. Field and M. C. McCarthy, Sci. Adv., 2015, 1, e1400105 Search PubMed .
  29. H.-L. Huang, W. Chao and J. J.-M. Lin, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 10857–10862 CrossRef CAS PubMed .
  30. L.-C. Lin, W. Chao, C.-H. Chang, K. Takahashi and J. J.-M. Lin, Phys. Chem. Chem. Phys., 2016, 18, 28189–28197 RSC .
  31. O. Welz, A. J. Eskola, L. Sheps, B. Rotavera, J. D. Savee, A. M. Scheer, D. L. Osborn, D. Lowe, A. M. Booth, P. Xiao, M. A. H. Khan, C. J. Percival, D. E. Shallcross and C. A. Taatjes, Angew. Chem., Int. Ed., 2014, 53, 4547–4550 CrossRef CAS PubMed .
  32. E. S. Foreman, K. M. Kapnas and C. Murray, Angew. Chem., Int. Ed., 2016, 55, 10419–10422 CrossRef CAS PubMed .
  33. M. C. Smith, C.-H. Chang, W. Chao, L.-C. Lin, K. Takahashi, K. A. Boering and J. J.-M. Lin, J. Phys. Chem. Lett., 2015, 6, 2708–2713 CrossRef CAS PubMed .
  34. T. Berndt, R. Kaethner, J. Voigtländer, F. Stratmann, M. Pfeifle, P. Reichle, M. Sipilä, M. Kulmala and M. Olzmann, Phys. Chem. Chem. Phys., 2015, 17, 19862–19873 RSC .
  35. L.-C. Lin, H.-T. Chang, C.-H. Chang, W. Chao, M. C. Smith, C.-H. Chang, J. J.-M. Lin and K. Takahashi, Phys. Chem. Chem. Phys., 2016, 18, 4557–4568 RSC .
  36. M. C. Smith, W. Chao, M. Kumar, J. S. Francisco, K. Takahashi and J. J.-M. Lin, J. Phys. Chem. A, 2017, 121, 938–945 CrossRef CAS PubMed .
  37. Z. J. Bura, R. M. I. Elsamra, A. Jalan, J. E. Middaugh and W. H. Green, J. Phys. Chem. A, 2014, 118, 1997–2006 CrossRef PubMed .
  38. Z. C. J. Decker, K. Au, L. Vereecken and L. Sheps, Phys. Chem. Chem. Phys., 2017, 19, 8541–8551 RSC .
  39. R. L. Caravan, M. A. H. Khan, B. Rotavera, E. Papajak, I. O. Antonov, M.-W. Chem, K. Au, W. Chao, D. L. Osborn, J. J.-M. Lin, C. J. Percival, D. E. Shallcross and C. A. Taatjes, Faraday Discuss., 2017 10.1039/C7FD00007C .
  40. M. Nakajima and Y. Endo, J. Chem. Phys., 2014, 140, 134302 CrossRef PubMed .
  41. C. Abeysekera, B. Joalland, N. Ariyasingha, L. N. Zack, I. R. Sims, R. W. Field and A. G. Suits, J. Phys. Chem. Lett., 2015, 6, 1599–1604 CrossRef CAS PubMed .
  42. Y.-P. Chang, A. J. Merer, H.-H. Chang, L.-J. Chang, W. Chao and J. J.-M. Lin, J. Chem. Phys., 2017, 146, 244302 CrossRef PubMed .
  43. C. Yin and K. Takahashi, Phys. Chem. Chem. Phys., 2017, 19, 12075–12084 RSC .
  44. D. Stone, M. Blitz, L. Daubney, N. U. M. Howes and P. Seakins, Phys. Chem. Chem. Phys., 2014, 16, 1139–1149 RSC .
  45. B. Ouyang, M. W. McLeod, R. L. Jones and W. J. Bloss, Phys. Chem. Chem. Phys., 2013, 15, 17070–17075 RSC .
  46. A. B. Ryzhkov and P. A. Ariya, Chem. Phys. Lett., 2006, 419, 479–485 CrossRef CAS .
  47. L.-C. Lin and K. Takahashi, J. Chin. Chem. Soc., 2016, 63, 472–479 CrossRef CAS .
  48. M. Kumar, A. Sinha and J. S. Francisco, Acc. Chem. Res., 2016, 49, 877–883 CrossRef CAS PubMed .
  49. T. Loerting and K. R. Kiedl, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 8874–8878 CrossRef CAS .
  50. C. A. Taatjes, F. Liu, B. Rotavera, M. Kumar, R. Caravan, D. L. Osborn, W. H. Thompson and M. I. Lester, J. Phys. Chem. A, 2017, 121, 16–23 CrossRef CAS PubMed .

This journal is © The Royal Society of Chemistry 2017