Adsorption and Hydrolysis of Alcohols and Carbonyls on Ice at

Article pubs.acs.org/JPCA

Adsorption and Hydrolysis of Alcohols and Carbonyls on Ice at Temperatures of the Upper Troposphere Angela Symington, Lay May Leow, Paul T. Griffiths, and R. Anthony Cox* Centre for Atmospheric Science, Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K. S Supporting Information *

ABSTRACT: The uptake of gaseous ethanol, 1,1,1-trifluoroethanol, acetone, chloral (CCl3CHO), and fluoral (CF3CHO) on ice films has been investigated using a coated-wall flow tube at temperatures 208−228 K corresponding to the upper troposphere (UT), with a mass spectrometric measurement of gas concentration. The uptake was largely reversible and followed Langmuir-type kinetic behavior, i.e., surface coverage increased with the trace gas concentration approaching a maximum surface coverage at a gas phase concentration of Nmax ∼ (2−4) × 1014 molecules cm−3, corresponding to a surface coverage of ∼30% of a monolayer (ML). The equilibrium partition coefficients, KLinC, were obtained from the experimental data by analysis using the simple Langmuir model for specific conditions of temperature and concentration. The analysis showed that the KLinC depend only weakly on surface coverages. The following expressions described the temperature dependence of the partition coefficients (KLinC) in centimeters, at low coverage for ethanol, trifluoroethanol, acetone, chloral, and fluoral: KLinC = 1.36 × 10−11 exp(5573.5/T), KLinC = 3.74 × 10−12 exp(6427/T), KLinC = 3.04 × 10−9 exp(4625/T), KLinC = 7.52 × 10−4 exp(2069/T), and KLinC = 1.06 × 10−2 exp(904/T). For acetone and ethanol the enthalpies and entropies of adsorption derived from all available data showed systematic temperature dependence, which is attributed to temperature dependent surface modifications, e.g., QLL formation. For chloral and fluoral, there was an irreversible component of uptake, which was attributed to hydrate formation on the surface. Rate constants for these surface reactions derived using a Langmuir−Hinshelwood mechanism are reported.

■

INTRODUCTION The lifetime of trace gases in the troposphere is determined by the most rapid sink, and it has been established that adsorption on ice surfaces may contribute significantly to removal via precipitation scavenging and vertical redistribution of those trace gases that partition strongly to ice.1 To evaluate this loss process for oxygenated organic compounds (OVOC), knowledge of the partition coefficients of the gaseous oxygenated organics to ice particles at upper tropospheric temperatures is needed. This paper describes an extension of our recently reported studies of the uptake of oxygenated organics with ice surfaces2 to investigate how the chemical properties of the molecule affect ice-partitioning parameters. A wide range of oxygenated organics are known to be present in upper troposphere (UT) where condensed water is present mainly in the ice phase. The equilibrium partition coefficients, defined as the number of molecules adsorbed per surface unit area in the presence of a given gas phase concentration, (KLinC) had been measured for less than 20 OVOC at the outset of this work. Ethanol3−5 and acetone6−11 adsorption on ice have been most widely studied at UT temperatures because these molecules are abundant in the UT, with mixing ratios >100 pptv (∼1 × 109 cm−3) for ethanol and >1000 pptv (∼1 × 1010 cm−3) for acetone.12 Although there was reasonable agreement between laboratories in measured KLinC and, to some extent, for ΔHads, there was very poor agreement between ΔSads values. Therefore, one aim of this © 2012 American Chemical Society

study was to investigate possible reasons for the scatter in ΔSads and ΔHads that are in excess of known experimental uncertainties. Second, at the outset of this work partitioning parameters on ice had only been measured for one fluorinated OVOC on ice at UT temperatures, namely trifluoracetic acid, which was reported in our previous paper.2 Perfluorinated compounds are present at much lower mixing ratios than their aliphatic analogues but their concentrations are known to be increasing. Therefore, the second aim was to assess if KLinC derived for aliphatic compounds can be used as an approximation for their fluorinated analogues. Halogenated species 2,2,2-trifluoroethanol (trifluoroethanol), 2,2,2-trifluoroacetaldehyde (fluoral), and 2,2,2-trichloroacetaldehyde (chloral) were chosen because, taken together with data for the carboxylic acids from our previous paper,2 they allow KLinC for C2 and C2-perfluorinated molecules at different oxidation level to be compared. The C2 and C2-perfluorinated aldehydes were compared by using the reported values for CH3CHO of Petitjean et al.13 In addition to adsorbing on the ice surface, fluoral and chloral may react to form covalent hydrates (gem diols) in the condensed Special Issue: A. R. Ravishankara Festschrift Received: November 14, 2011 Revised: January 30, 2012 Published: January 30, 2012 5990

dx.doi.org/10.1021/jp210935b | J. Phys. Chem. A 2012, 116, 5990−6002

The Journal of Physical Chemistry A

Article

Uncertainty in both Nmax and KLinC is low when KLinC derived from linearized Langmuir analysis is comparable to the value derived from the slope in the linear region of the isotherm. To extrapolate data over a wider range temperature range in atmospheric modeling, the temperature dependence of KLinC is required. The experimentally derived parameter KLinC (cm) is related to K, the dimensionless equilibrium constant, by (E4):

phase. Halogenated diols are the stable product of the respective aldehydes and water at room temperature, (R1): CX3CHO + H2O ←⎯⎯→ CX3C(H)(OH)2 kf ,kr

(R1)

They can be isolated as stable solid (fluoral hydrate) or liquid (chloral hydrate) at room temperature, and if formed at UT temperatures, they would be expected to remain in the condensed phase, introducing an additional potential sink for gas phase halogenated aldehydes. Therefore, another aim of this study was to investigate if chloral and fluoral uptake behavior consistent with hydrate formation could be observed.

K = KLinC

⎛ ΔG 0 ⎞ N S K ≡ exp⎜⎜ ads ⎟⎟ ≡ ads [A] V ⎝ RT ⎠

EXPERIMENTAL PROTOCOL AND ANALYSIS METHODS Adsorption profiles of ethanol, acetone, trifluoroethanol, fluoral, and chloral on ice were measured using the coated-wall flow tube coupled to a mass spectrometer detection as described in our earlier paper on uptake of carboxylic acids and outlined in Supporting Information. Trace gases diluted in He were introduced into the flow tube via a sliding injector, which allowed control of their exposure to the ice film. In this work the flow tube pressure, Ptot, was 2.4 hPa and the volumetric flow rate, Ftot, was 100 sccm. Surface coverages, as a function of gas phase concentration, were measured from integrated adsorption peak areas and used to obtain the Langmuir equilibrium constant, KLang, the partition coefficient, KLinC, and the maximum surface coverage Nmax at temperatures (208−228) K. Adsorption isotherms were used to describe the relationship between the gas phase concentration of species A, [A], at a given temperature and the surface coverage, Nads. The Langmuir model was chosen because Nads appears to saturate at coverages approaching 1 monolayer (ML) for a range of inorganic and organic species at upper tropospheric temperatures and in the ice stability region of the trace gas−water phase diagram. KLang, describes the partitioning between the condensed and gas phases and at equilibrium, Nads is related to KLang by (E1): NmaxKLang[A] (1 + KLang[A])

where is the Gibbs free energy of adsorption and S/V is the surface to volume ratio of the gas adsorbed at the surface (∼1.7 × 0 107 cm−1, Kemball and Rideal14). K can be used to derive ΔGads and, therefore, ΔHads and ΔSads, the enthalpy and entropy of adsorption. ΔHads and ΔSads can be calculated using a Van’t Hoff plot (ln(K) vs 1/T, (E6)): ln(K ) = −

ΔHads ΔSads + RT R

(E6)

For some of the molecules studied here, irreversible loss from the gas phase was observed. Permanent loss of chloral or fluoral from the gas phase was attributed to gem-diol (CX3C(H)(OH)2) formation, and interpreted according to the Langmuir− Hinshelwood reaction scheme, in which both reactants are adsorbed on the surface, with the surface concentrations of the adsorbates determined by a Langmuir adsorption mechanism. In the case of hydrolysis on ice surfaces, the overall loss process can be described by surface adsorption, (R2), CX3CHO(g) + S ←⎯⎯⎯⎯→ CX3CHOH(ads) k1f , k1r

(R2)

followed by reaction with the surface H2O, R3: ks

CX3CHO(ads) + H2O(surface) → CX3C(H)(OH)2(ads) (E1)

(R3)

In the linear (low coverage) region of the isotherm, the expression for Nads simplifies to (E2):

In this case, the rate of hydrate formation, (E7), d[CX3C(H)(OH)2 ] = k s[CX3CHO(ads)][H2O(surafce)] dt

(E2)

(E7)

The product KLangNmax is defined as the partition coefficient KLinC and is equivalent to the gradient of the linear region of the isotherm. KLinC can be retrieved most accurately in the linear region of the isotherm provided the uptake can be fully resolved because no assumptions have to be made regarding lateral interactions between adsorbed molecules when the surface coverage approaches saturation. The Langmuir model assumes there are no lateral interactions. However, at these low concentrations Nmax cannot be derived because the adsorption of trace gases is not limited by the number of available sites. At higher concentration, where surface saturation induces curvature in the isotherm, Langmuir parameters KLinC and Nmax (and therefore KLang) can all be extracted from a plot of the linearized form of the Langmuir isotherm (E3): [A] [A] 1 = + Nads Nmax KLinC

(E5)

0 ΔGads

cm−2

Nads = NmaxKLang[A] = KLinC[A] cm−2

(E4)

K is defined as (E5),

■

Nads =

S V

on a solid surface is limited by [H2O(surface)] if hydrate formation is irreversible and the hydrate remains on the surface. The amount of available surface water [H2O(surface)] was calculated from (E8), [H2O(surface)] = [H2O(surface)0] −

1015 ([CX3CHO(ads)] Nmax ,A

+ [CX3C(H)(OH)2(ads)])

(E8)

where ([CX3CHO(ads)] + [CX3C(H)(OH)2(ads)]) is the integrated loss of aldehyde from the gas phase per area of surface exposed and each adsorbate is assumed to occupy 1 site on the water surface. The initial concentration on the surface [H2O(surface)0] is approximately 1 × 1015 cm−2 for the basal plane

(E3) 5991



Article

of hexagonal ice,15 and 1 adsorbate site is equivalent to 1015/NmaxA water molecules, where NmaxA is the maximum adsorbate surface coverage. The product of ksKLang for fluoral adsorption (KLinC) and reaction (ks) on ice can be obtained using the Langmuir−Hinshelwood equation (E9) by plotting of 1/α, where α is the time dependent accommodation coefficient of fluoral on ice, versus 1/[H2O(surface)], ω fluoralσfluoral 1 1 = + α αs 4k sKLang[H2O(surface)] (E9)

end of the ice surface. The gas phase ethanol signal increased due to desorption from the ice surface before stabilizing at its original level. At 170 s, the ice was re-exposed to ethanol (13.0 cm) and, at 270 s, the injector was returned to the upstream end of the ice surface. Nads, measured from integrated peak areas, for the first and subsequent exposures of ethanol to the ice surface were indistinguishable within experimental uncertainty. The ratios R, of adsorption and desorption peak areas, were similar (R = 0.9 ± 0.5 at 208 K, 1.0 ± 0.2 at 218 K, and 1.1 ± 0.2 at 228 K) indicating that uptake is fully reversible within experimental uncertainty and that there is no measurable chemical aging of the surface. Ethanol adsorption isotherms measured at 208, 218, and 228 K were plotted in Figure 2. At low concentrations, Nads was

σfluoral is the adsorption area per fluoral molecule, ωfluoral is the mean speed of a fluoral molecule at the experimental temperature, and αs is the surface accommodation coefficient of a fluoral molecule on the surface. α is measured from the adsorption profile from the ratio of the gas phase concentration exiting the flow tube before exposure, S(0), and at time t, S(t), during exposure (E10) and (E11): k=

a=

v ⎛ S(0) ⎞ ln⎜ ⎟ z ⎝ S(t ) ⎠

(E10)

2kr ω fluoral

(E11)

k is the rate of loss of fluoral from the gas phase, z the exposure length, and r is the flow tube radius. If surface accommodation in (E9) is rapid compared to the reaction, (E9) simplifies to (E12): ω fluoralσfluoral 1 = α 4k sKLang[H2O(surface)] (E12)

■

RESULTS 1: PHYSICAL ADSORPTION ON ICE Ethanol. Figure 1 shows a mass spectrometer adsorption profile for repeated exposure of ethanol to an ice surface. At

Figure 2. Ethanol adsorption isotherms at 208 K (blue), 218 K (green), and 228 K (red). Fits using Nmax derived from linearized Langmuir plots at 208 K and KLinC from the linear regions of the isotherm.

linearly dependent on [CH3CH2OH]. At higher concentrations, the dependence of coverage on [CH3CH2OH] became nonlinear and appeared to approach saturation, consistent with the Langmuir adsorption model. Although Nmax (as defined by the linearized Langmuir plots) appeared to decrease at higher temperatures, Nmax at 208 K was chosen as the temperature independent value to plot the isotherms because the coverages closest to saturation were reached. The values of Nmax at 208 K and KLinC(T) obtained from the slope in the linear region of the isotherm were used to plot the isotherm fits in Figure 2. To determine the magnitude of the uncertainties associated with the Langmuir model assumptions, KLinC was derived from the slope of the adsorption isotherm at low surface coverages and was compared to the value obtained from linearized Langmuir plots (Figure 3). Low coverage was defined as the linear region of the plot, i.e., the region where increasing the concentration range included in the analysis resulted in no change in reported KLinC within experimental uncertainty. The values of KLang, Nmax, and KLinC, derived from both analyses are recorded in Table 1. Retrieved KLinC did not appear to have strong coverage dependence. KLinC as a function of temperature was parametrized as

Figure 1. Adsorption profiles for the first exposure and second exposure of ethanol ((1.1 ± 0.1) × 1011 cm−3) to ice at a flow rate of 200 sccm, T = 208 K, and PTot = 2.4 hPa.

45 s, the injector was withdrawn on a time scale of about 1 s to expose a length of 13.0 cm of ice. Gas phase ethanol detected by the mass spectrometer fell to a nonzero value due to adsorption of ethanol on the ice surface. The signal returned to the initial level, indicating that the surface coverage had reached equilibrium. At 127 s, the injector was moved to the upstream

KLinC = 1.36 × 10−11 exp(5573.5/T ) cm

(E13)

ΔHads and ΔSads were calculated from Van’t Hoff plots using K, the dimensionless equilibrium constant, to give ΔHads = −54 ± 11 kJ mol−1 and ΔSads = −86 ± 50 J mol−1 K−1. The Van’t Hoff 5992



Article

approximately 1 order of magnitude smaller than for ethanol at the same temperature and concentration. Adsorption isotherms for acetone are shown in Figure 5. Adsorption parameters derived from the linear region of the

Figure 3. Linearized Langmuir plots for ethanol on ice at 208 K (blue), 218 K (green), and 228 K (red).

Table 1. Summary of Ethanol Partitioning Parameters Derived from Isotherm Analysis T (K)

KLinC (cm) linear region

KLinC (cm) linearized Langmuir

KLang (10−13 cm3) linearized Langmuir

Nmax (1014 cm−2) linearized Langmuir

208 218 228

128 ± 12 52 ± 1 11.4 ± 0.6

160 ± 10 60 ± 2 11.6 ± 0.6

34.3 13.0 N/A

4.6 ± 0.2 5.0 ± 0.2 N/A

Figure 5. Isotherms for the adsorption of acetone on ice at 208 K (blue downward facing triangles represent data obtained from adsorption peaks and upward facing triangles data obtained from desorption peak areas) and 218 K (green).

isotherm and over the full concentration range using the linearized Langmuir analysis were indistinguishable within experimental uncertainty. The linearized Langmuir methods provided an estimate of Nmax for acetone on ice at 208 K only; it cannot be determined at 218 K because the highest reported Nads is far from saturation (7 × 1012 cm−3). This behavior has been seen previously close to the boundaries of hydrate or solution stability regions.20 Indeed, the bulk phase diagram of ethanol + water, which indicates that ethanol hydrates are stable at temperatures below −75 °C (198 K) at sufficiently high gas phase ethanol concentrations (Figure 19).

may arise from differences in the range of coverages studied. For example, Peybernés et al.5 included uptake data at surface coverages greater than 1 ML and analyzed this data using a BET model for multilayer adsorption, from which the Langmuir parameters have been estimated by IUPAC.16 This study reported Nads far from saturation. Uncertainty in reported ΔHads and ΔSads, obtained from intercepts of Van’t Hoff plots, exceeded experimental uncertainty. The main difference between the values of ΔHads and ΔSads reported by different studies was the temperature range over which the parameters had been derived, with both ΔHads and ΔSads becoming increasingly negative at higher temperature ranges. Figure 18 illustrates how ΔHads and ΔSads changed with

Figure 19. Phase diagram for the ethanol−water mixtures at temperatures relevant for this work (Ohkubo et al.).

The gradual, rather than step change, in ΔHads as a function of temperature suggests that the surface phase transitions were not sharp first-order changes from surface physisorbed species on solid, bulk terminated ice to bulk hydrate or QLL formation. Instead, at temperatures approaching the phase boundary the structure may become increasingly QLL or hydrate like. Other processes that could also contribute to an entropy increase at low temperatures include surface restructuring, increased release of water to the gas phase, or contact ion pair formation. Sokolov and Abbatt4 showed that at 228 K, the alcohols are not full solvated on adsorption on ice surfaces because the partitioning of hexanol to the ice surface is greater than for the lower alcohols. Partitioning of volatile alcohols to liquid water shows the opposite trend due to decreasing solubility with alkyl chain length. Theoretical calculations by Peybernés et al.,5 are

Figure 18. ΔHads (black) and ΔSads (red) for ethanol adsorption on ice, both dependent on the range of temperatures over which data are measured and linearly dependent on the mean temperature of this range.

the mean experiment temperature (T = (Tmax − Tmin)/2 (K)) from which the values of ΔHads and ΔSads were derived. ΔHads and ΔSads obtained from Van’t Hoff plots were parametrized as (E22) and (E23): ΔHads = −(1.8 ± 0.1)T + (347 ± 23) kJ mol−1

(E22)

ΔSads = −(7.9 ± 0.3)T + (1661 ± 73) J mol−1 K −1 (E23)

By setting (E22) to equal zero, the temperature at which ΔHads = 0 kJ mol−1 was obtained (=190 K) and the 5999



Article

Table 9. Acetone Adsorption Parameters at 208 and 218 Ka

this work (208−218 K) Winkler9 (198−223 K)

Peybernés5 (193−223 K) Bartels-Rausch7 (203−223 K) Behr10 (VDI) (190−220 K) a

T (K)

KLinC (cm)

208 218 208

14 ± 2 5.2 ± 0.2 15.1

218 208 218 208 218 208 218

5.1 12.9 3.4 11.0 2.85 32.6 13.9

concn range (1010 cm−3) 50−600 50−250 0.77−4 (L) 0.77−3600 0.77−3600 5.4−6200 5.4−6200

Adsorption and Hydrolysis of Alcohols and Carbonyls on Ice at

Recommend Documents