Report

Gas phase observation and microwave spectroscopic characterization of formic sulfuric anhydride

See allHide authors and affiliations

Science  03 Jul 2015:
Vol. 349, Issue 6243, pp. 58-61
DOI: 10.1126/science.aaa9704

An unexpected gaseous sulfur species

Sulfuric acid plays a central role in both industrial and atmospheric contexts. As such, the behavior of SO3 mixtures in gas phases has been studied for over a century. In gas-phase experiments on wet SO3 and formic acid, Mackenzie et al. discovered a previously unrecognized covalent adduct: formic sulfuric anhydride, or HC(O)OSO3H. The combination of microwave spectroscopy and theoretical calculations reveals its structural properties. The compound may play a role in the nucleation of atmospheric aerosols by serving as an intermediate to H2SO4 formation.

Science, this issue p. 58

Abstract

We report the observation of a covalently bound species, formic sulfuric anhydride (FSA), that is produced from formic acid and sulfur trioxide under supersonic jet conditions. FSA has been structurally characterized by means of microwave spectroscopy and further investigated by using density functional theory and ab initio calculations. Theory indicates that a π2 + π2 + σ2 cycloaddition reaction between SO3 and HCOOH is a plausible pathway to FSA formation and that such a mechanism would be effectively barrierless. We speculate on the possible role that FSA may play in the Earth’s atmosphere.

There is an extensive literature on the chemistry of sulfur oxides and their derivatives. The area is rich in fundamental science and finds applications ranging from industrial chemistry to laboratory synthesis. Sulfur compounds are also active species in the atmosphere (1), and in particular, the oxides and oxyacids are important players in the formation of atmospheric aerosol (2). Here, we present a microwave spectroscopic study of the anhydride derived from formic and sulfuric acids, produced in a supersonic jet containing HCOOH and SO3.

The present work was stimulated by a series of studies concerned with the formation of sulfuric acid in the atmosphere. The acid, which can form via both gas-phase and aqueous-phase processes, is generated in the gas phase by the oxidation of SO2 to SO3, which is subsequently hydrated to give H2SO4SO3 + H2O → H2SO4 (1)Both theoretical and experimental studies of this reaction indicate that viable mechanisms in the gas phase involve a facilitator molecule. For example, using ab initio theory, Morokuma and Muguruma (3) showed that the addition of a second water molecule to H2O⋅⋅⋅SO3 substantially lowers the activation barrier for Eq. 1, and indeed, kinetic data show a second-order dependence on water concentration (4). More recently, computational work by Hazra and Sinha has indicated that the conversion is essentially barrierless within the complex HCOOH⋅⋅⋅H2O⋅⋅⋅SO3 (5). Upon considering the concentration of formic acid, one of the most common atmospheric volatile organic compounds (VOCs), these authors argued that this pathway may be important for the formation of sulfuric acid. Furthermore, this mechanism terminates in the formation of the hydrogen-bonded complex H2SO4∙∙∙HCOOH, which may be a preliminary step in a nucleation process. Indeed, the involvement of organic compounds in new particle formation has become a central topic in atmospheric particle research, and it now has been shown that organics participate not only in particle growth (68) but in nucleation as well (915). Carboxylic acids are abundant in the atmosphere (16), and thus, their primary interactions with sulfur-containing atmospheric species are of great interest.

Previous work in our laboratory has characterized a variety of atmospheric molecular complexes—including SO3∙∙∙H2O (17), H2SO4∙∙∙H2O (18), and HNO3∙∙∙(H2O)n=1–3 (19)—in a supersonic jet by using microwave spectroscopy. Such systems, in general, are important to study because of the roles they play as intermediates in chemical reactions and precursors to atmospheric aerosol. Although conditions in the jet do not mimic those in the atmosphere, they can produce the same species, albeit under conditions amenable to microwave spectroscopy (in a collisionless environment that is cold enough to ensure population of only the lowest rotational energy levels). Spectral analysis yields accurate information about gas-phase molecular and electronic structure, which in turn provides an important touchstone for computational studies. Therefore, inspired by Hazra and Sinha, and by the recent research suggesting the involvement of organic acids in aerosol formation, we set out to investigate the complexes such as H2SO4∙∙∙HCOOH and SO3∙∙∙HCOOH by means of microwave spectroscopy in a supersonically expanded mixture of SO3, H2O, and HCOOH seeded in argon (Ar). What we found was entirely unexpected.

Rotational spectra were observed by means of Fourier transform microwave (FTMW) spectroscopy. Two methods were used: conventional cavity-type FTMW spectroscopy and the newer broadband, chirped-pulse FTMW technique. The cavity spectrometer in our laboratory has been described elsewhere (20). Briefly, molecules enter a tuned microwave cavity and are coherently excited by a 1- to 2-μs pulse of radiation. The resulting free induction decay is heterodyne-detected, digitized, and Fourier transformed so as to produce a signal in the frequency domain. Uncertainties in spectral transition frequencies are typically on the order of 2 to 3 kHz. In our broadband spectrometer (21), which applies the methods developed by Pate and coworkers (22), the cavity is eliminated, and the radiation for sample excitation is generated by up-conversion of a 0.2- to 3.2-GHz chirped pulse to the microwave spectral region of interest. A 20-W amplifier boosts the power before irradiation of the molecular sample, which is accomplished via a microwave horn. The resulting free induction decay is received by an identical companion horn, down-converted for digitization, and Fourier transformed so as to produce a 3-GHz-wide segment of the microwave spectrum. Linewidths are typically ~90 kHz, and the apparatus is less sensitive than that of the cavity spectrometer but it allows for rapid location and identification of spectra. In this work, spectra were initially recorded with the broadband spectrometer and subsequently remeasured at high resolution by using the cavity system.

A key feature of this experiment was the molecular source (21), which consisted of a supersonic expansion of SO3 in Ar, to which a flowing stream of formic acid and water vapor was added. The Ar carrier gas was seeded with SO3 by means of entrainment upon passage over solid, polymerized SO3. A gaseous mixture of HCOOH and H2O was added by bubbling Ar through 88% aqueous formic acid and injecting it into the expansion (Fig. 1A). This configuration provides “on-the-fly” mixing of reactive reagents during the first few tens of microseconds of the supersonic expansion.

Fig. 1 Experimental setup and results.

(A) A diagram of the nozzle source. HCOOH/H2O/Ar was introduced a few millimeters downstream of the start of the expansion through a 0.41-mm inner-diameter hypodermic needle. (B) Stick spectrum of transitions observed by using the chirped-pulse spectrometer and confirmed on the cavity system from an Ar, SO3, H2O, and HCOOH mixture for the 6- to 12-GHz region. Unassigned transitions that were independent of HCOOH are not shown. The relative intensities were maintained from the original chirped-pulse data [average of 50,000 free induction decays (FIDs)] and are comparable for broadband spectra within the same spectral region (for example, 6 to 9 and 9 to 12 GHz, demarcated by the break in the spectrum). Red lines represent FSA transitions. Transitions assigned to Ar-SO3, Ar-HCOOH, and HCOOH were left in the spectrum for comparison with FSA transitions and are marked in blue, pink, and green, respectively. The strongest transition observed for each is labeled on the figure. Black lines are unassigned. (C) Portion of the 9- to 12-GHz spectrum showing two FSA transitions highlighted in red.

Transitions from a variety of known species—including Ar∙∙∙SO3, H2O∙∙∙SO3, H2SO4 (formed from SO3 + H2O), H2SO4∙∙∙H2O, Ar∙∙∙HCOOH, and H2O∙∙∙HCOOH—were observable, as were numerous lines of unknown origin (Fig. 1B). Among these was a set of strong transitions whose pattern made them readily identifiable as the a-type spectrum of an asymmetric rotor, and a preliminary least-squares fit to a Watson A-reduced Hamiltonian gave residuals of <5 kHz. The appearance of these transitions required the presence of HCOOH, and our initial hypothesis was that they were due either to the H2SO4∙∙∙HCOOH or SO3∙∙∙HCOOH weakly bound complex. However, theoretical predictions of the rotational constants for both species by using the density functional M06-2X (table S1) clearly indicated that neither is the carrier of the observed spectrum. Indeed, the computed rotational constants of H2SO4∙∙∙HCOOH as well as those of several weakly bound isomers of SO3∙∙∙HCOOH (table S1 and fig. S3) differ by up to ~1 GHz and ~200 MHz, respectively, from those fitted from experimental data.

We performed additional calculations near the global minimum energy configuration of SO3∙∙∙HCOOH with starting configurations at progressively shorter intermolecular separations. These eventually revealed a new chemically bonded species in which the acidic proton of the formic acid transfers to the SO3 and a new sulfur-oxygen bond is formed (tables S2 and S3). We performed frequency calculations in order to verify the authenticity of the potential energy minimum, and the density functional theory (DFT) results were confirmed with MP2 calculations. The structure of this molecule, formic sulfuric anhydride (FSA), is shown in Table 1. This structure has substantial dipole moment components along each of its three inertial axes (μa, μb, and μc = 3.19, 0.48, and 0.97 D, respectively; μTotal = 3.37 D), and with this information, b-type and c-type transitions were easily predicted and located. With a, b, and c-type transitions recorded, we performed a final fit, which yielded the rotational constants listed in Table 2. As indicated in the table, the predicted rotational constants of FSA match the experimental values to within 0.7% (32 MHz for A and 14 MHz for B and C).

Table 1 Comparison of experimental and theoretical intermolecular distances.

View this table:
Table 2 Comparison of experimental and calculated rotational constants for FSA.

View this table:

As a final test, we predicted and observed spectra of the 34S-, 13C-, and both monodeuterated isotopologues. We recorded 34S- and 13C- isotopologue spectra in natural abundance and used isotopically enriched HCOOD or DCOOH in experiments on the deuterated species . Transition frequencies and fitted spectroscopic constants for the observed isotopologues are provided in tables S4 to S9. All observed isotope shifts were in excellent agreement with those predicted from the theoretical structure. When we used HCOOD, the deuterium was found in the H9 position, confirming the occurrence (direct or indirect) of proton transfer (23); isotopic substitutions allowed determination of several interatomic distances by using Kraitchman’s equations (24). These results are given and compared with the theoretical values in Table 1, where agreement is again seen to be excellent. The agreement between experimental and theoretical rotational constants, their isotope shifts, and the computed interatomic distances unambiguously establishes that the observed species is FSA.

The literature on compounds related to FSA appears sparse, although a few prior condensed phase studies are noteworthy. The sodium salt of acetic sulfuric anhydride has been described (25), but the parent acid (CH3COOSO3H) is reportedly unstable with respect to rearrangement or decomposition (26, 27). Dissolved salts of form [Mn+][SO3OCHO]n have also been described in a patent concerning the preparation of isoflavones in a variety of nonaqueous solvents (28). We are unaware, however, of any previous gas-phase observations of FSA or its analogs. To better understand the pathway for the formation of FSA in the gas phase, we performed calculations (21) to locate the transition state connecting it with the SO3∙∙∙HCOOH van der Waals complex (table S10). For these calculations, using the optimized MP2 geometries, single-point coupled-cluster with single, double, and perturbative triple excitations [CCSD(T)] calculations were done by using the complete basis set extrapolation scheme of Neese and Valeev with the ANO-pVDZ to ANO-pVTZ basis sets (29). The transition state (TS in Fig. 2A) was found and corresponds to a concerted π2 + π2 + σ2 cycloaddition (Fig. 2B), in which the acidic proton is transferred as the new S−O bond is formed. A similar reaction has been proposed for SO2∙∙∙HCOOH, but the formation of the resulting monomer, formic sulfurous anhydride, from SO2 and HCOOH is endothermic (30). In the current case, however, FSA is 4.4 (4.2) kcal/mol lower in energy than the SO3⋅⋅⋅HCOOH van der Waals complex, where the value in parentheses is zero-point corrected. The energy of the transition state is 2.2 (0.2) kcal/mol higher than that of SO3∙∙∙HCOOH, indicating that its conversion to FSA, with zero-point corrections, is essentially barrierless. In this light, it is not surprising that transitions of FSA were among the most prominent features in the observed spectrum. A second conformer of FSA, with the OH bond rotated 180°, was also identified from the M06-2X calculations (fig. S5). However, this local minimum lies 4.0 (3.6) kcal/mol above the global minimum of Table 1, and no spectra corresponding to this structure were identified, presumably because of insufficient population in the supersonic jet.

Fig. 2 FSA energetics.

(A) Potential-energy landscape of the bimolecular complexes formed between H2O, SO3, and HCOOH (FA). Geometries were optimized by using MP2/6-311++G(3df,3pd), and single-point electronic energies were computed by using CCSD(T)/complete basis set with double and triple extrapolation [CBS(D-T)]. The red trace is uncorrected for zero-point energy (ZPE). The blue trace is ZPE-corrected with frequencies from the MP2/6-311++G(3df,3pd) calculations. Zero is defined as the sum of the monomer energies with and without ZPE corrections for the blue and red traces, respectively. The barrier to conversion from FA∙∙∙SO3 to FSA is 2.2 kcal/mol without ZPE corrections and 0.2 kcal/mol with ZPE corrections. (B) Possible formation mechanism of FSA from SO3 and HCOOH.

We also sought to establish the stability of the putative FSA precursor, SO3∙∙∙HCOOH, relative to that of other bimolecular complexes with potential roles as sulfuric acid and/or aerosol precursors. Computed energies of the complexes formed from SO3, H2O, and HCOOH are included in Fig. 2A, fig. S4, and table S10. With zero-point corrections, the SO3∙∙∙HCOOH complex is 3.6 and 4.4 kcal/mol more stable than H2O∙∙∙HCOOH and H2O∙∙∙SO3, respectively. Similar calculations for H2O∙∙∙HCOOH and H2O∙∙∙SO3 have been given elsewhere (3, 5) but are reproduced here in order to ensure comparisons at a uniform level of theory.

Both carboxylic and sulfonic acid anhydrides hydrolyze in water, and the hydrolysis of acetic sulfuric anhydride has been studied (27, 31). Thus, it is reasonable to hypothesize that the formation of SO3∙∙∙HCOOH with a barrierless conversion to FSA, followed by reaction with water and/or uptake into liquid droplets, may constitute an alternate pathway for H2SO4 production in the atmosphereSO3 + HCOOH → SO3∙∙∙HCOOH (2)SO3∙∙∙HCOOH → FSA (3)FSA + H2O(g and/or ℓ) → H2SO4(g and/or aq) + HCOOH(g and/or aq) (4)Although such a mechanism is not expected to dominate over established pathways involving direct reaction with water, it requires only the formation of a bimolecular—not a trimolecular—complex and may contribute to H2SO4 production, especially in areas with elevated concentrations of carboxylic acids. Additionally, the process described in Eq. 4 occurring in small water-containing clusters and/or liquid droplets may provide a pathway for the incorporation of volatile organic compounds into atmospheric aerosol. The mechanism may also be extendable to compounds of low volatility, which are more likely to contribute to prenucleation clusters (32). Indeed, we have performed additional calculations with larger carboxylic acids, which confirm the viability of their reaction to form FSA analogs (fig. S6). Moreover, laboratory studies that generated H2SO4 in situ have suggested that sulfur-containing species other than H2SO4 could be the initial nucleating agent (33). Under such a conclusion, FSA or its analogs could be nucleating agents themselves. Thus, although the atmospheric importance of Eqs. 2 to 4 is by no means certain, we suggest that scenarios involving FSA and larger sulfuric–carboxylic anhydrides should be explored in conjunction with models that involve trimolecular complexes such as (H2O)2∙∙∙SO3 and H2O∙∙∙HCOOH∙∙∙SO3 and in mechanisms for the early incorporation of organics in nucleation schemes. This could include, but is not limited to, mechanistic studies of the hydrolysis of sulfuric-carboxylic anhydrides in both water-containing clusters and bulk phase.

Supplementary Materials

www.sciencemag.org/content/349/6243/58/suppl/DC1

Materials and Methods

Figs. S1 to S6

Tables S1 to S10

References (3436)

References and Notes

  1. Methods and materials are available as supplementary materials on Science Online.
  2. Because formic acid inevitably contains water, experiments do not definitively establish whether the transformation of SO3 + HCOOH occurs directly or whether an additional water molecule participates. However, the theoretical results of this work suggest that the reaction is possible without such assistance.
  3. Acknowledgments: This work was supported by the National Science Foundation (grant CHE-1266320) and the Minnesota Supercomputing Institute. We are grateful to P. McMurry, T. Hoye, and M. Canagaratna for helpful conversations. We also thank W. Isley for assistance with the CCSD(T) calculations and J. Mendez and the Tektronix Corporation for invaluable aid in construction of the broadband spectrometer. The authors of this work are unaware of any conflicts of interest.
View Abstract

Navigate This Article