Distinct Temperature Trends in the Uptake of Gaseous n-Butylamine on Two Solid Diacids

Uptake coefficients of n-butylamine (BA) on solid succinic (SA) and glutaric acids (GA) from 298 to 177 K were measured using a newly combined Knudsen cell temperature-programmed desorption apparatus. The uptake coefficients on SA increase monotonically from (1.9 ± 0.5) × 10–4 at 298 K to 0.14 ± 0.05 at 177 K (errors represent 2σ statistical errors, overall errors are estimated to be ±60%). This is consistent with a surface reaction mechanism to form solid aminium carboxylate. In contrast, the uptake coefficients on GA increase from 0.11 ± 0.04 at 298 K to 0.25 ± 0.04 at 248 K but then decrease to 0.030 ± 0.010 at 177 K. This unusual trend in temperature dependence of the uptake coefficient is due to formation of an ionic liquid (IL) layer upon the surface reaction of BA with GA, leading to a competition between the rate of desorption of BA and the rates of diffusion and reaction within the IL. Overall, the kinetic multi-layer model of aerosol surface and bulk chemistry (KM-SUB) satisfactorily reproduces these unique trends. This work provides mechanistic insight and predictive capability for the temperature-dependence of reactive uptake processes involving multiple phase changes upon surface reaction.


Text S1:
Correction of I0 and Ir for background n-butylamine (BA) signals and uptake on empty sample cup at low temperatures.Amines, including BA, adsorb readily on surfaces and desorb slowly on pumping.To correct I0 and Ir for background BA signals, the flow of BA was shut off after a steady state was reached on exposure of the diacid to the BA and the decay was followed with time.The residual signal after 5 min was used as the background signal, Ib, to subtract from Ir and the subsequent I0 when the sample lid was closed off and the BA flow restored (Figure S1).This was done for each value of Ir and the subsequent I0.

Figure S1. Correction of I0
and Ir for the background from BA. Temperature was 298 K, the sample mass of glutaric acid was 0.0255 g and BA initial concentration was 4 ×10 11 molec cm -3 .Orifice area was 0.33 cm 2 .
The system was designed to cool the bottom of the sample cup and hence control the temperature of the sample.The walls of the sample cup that connected it to the main chamber were made of thin stainless steel whose thermal conductivity was small.However, over the course of a number of hours required for the experiments, some cooling of the walls occurred, resulting in uptake of BA at temperatures below 298 K. Figure S2 shows the decrease in BA signal when the empty sample cup was cooled.A small decrease (~10%) was observed down to ~170 K but then there was a dramatic drop at lower temperatures.As a result, measurements were restricted to T > 175 K, and corrections were applied to I0 for this uptake.Preparation of ionic liquids (ILs) and viscosity measurements.Glutaric acid (GA, 2.5 g) was dissolved in 300 mL nanopure water (18.2MΩ cm) and then mixed with either 1.87 or 3.74 mL of liquid BA to give salts with 1:1 or 2:1 mole ratios of BA to GA.The water was removed by evaporation using a rotovap (Wheaton, SPIN-VAP) at 20 kPa and 70~75 o C. The resulting ionic liquids were pumped under vacuum of < 1 Pa for 8 hr to remove the remaining water.Viscosity was measured using the falling sphere viscometer technique. 1The dried liquid was placed in a 5 mL graduated cylinder and a metal sphere of known density (  = 4.2 × 10 3 kg m -3 ) and radius (r = 0.51 mm) was allowed to fall through the liquid.The viscosity of the liquid (µ) is given by Eq S1, 1 where g is the gravitational constant,   is the density of the ionic liquid (  =1.1 × 10 3 kg m -3 ), 2 and v is the velocity of the sphere falling through the liquid.Cooling baths were prepared in order to measure viscosities at selected temperatures.These included an ice/water bath for 273 K, an ice/CaCl2 water solution for 238 K, dry ice/acetone for 195 K, and liquid nitrogen for 77 K.The graduated cylinder containing the liquids and the sphere were equilibrated with each cooling bath for 10 min before each measurement.The measurements were carried out with the liquid submerged in the bath.
The viscosities were measured at several temperatures from 195 K to 298 K; visual recordings of the motion of the material were also made (see Web Enhanced Objects).Both the 1:1 and 2:1 BA:GA salts are liquid at room temperature and become more and more viscous with decreasing temperatures.
At temperature below around 238 K, both salts become semi-solids, 3 and in liquid nitrogen (77 K), both salts form solid with cracks due to rapid cooling.This increase in the viscosities plays a significant role in driving the decrease in uptake coefficients at decreasing temperatures below 248 K in the GA case (Figure 4a).
The KM-SUB model was not able to match the experimental observations for the uptake on GA using the measured temperature-dependent viscosities (Figure S3c).This discrepancy is possibly due to IL not obeying the Stokes-Einstein equation or proton hopping where protons instead of GA molecules diffuse in the layer of IL by hopping between neighboring glutarate ions. 4Another possible reason is that the micro-viscosity of IL formed on particles could be significantly smaller than the measured bulk viscosity. 5    (1.9 ± 0.4) × 10  b Uncertainties represent 2σ statistical errors.Overall error includes possible systematic errors estimated to be ± 60% due to uncertainties associated with sample surface areas of the polydisperse powders and the correction to I0 and Ir from the BA background.
c Averaged from 19 points summarized in Table S2.

Figure S2 .
Figure S2.Ratio of the BA signal intensity on opening to the empty sample cup (Ir,e) to that

Figure S3 .
Figure S3.KM-SUB predicted temperature dependence (lines) of (a) reaction rate coefficients, (b) desorption rate coefficients and (c) viscosities.Circles in (c) are the measured values.

Figure S4 :
Figure S4: Viscosity as a function of time and depth at 298 K predicted by the model for the (a) lowest and (b) highest BA concentrations at which experiments were conducted.Note that steplike changes are an artifact of the model and are caused by having multi-layers in KM-SUB.Contour lines are in units of Pa s.
is 0.011 cm 2 and [BA]0 is 4.0 × 10 11 molec cm -3 .bUncertainties represent 2σ statistical errors.Overall error includes possible systematic errors estimated to be ± 60% due to uncertainties associated with sample surface areas of the polydisperse powders and the correction to I0 and Ir from the BA background.c SA mass is 0.2801 g.

Table S1 .
Parameters used in the KM-SUB model.Note that gas constant R = 0.0083 kJ mol -1 K - 1 .T is in units of Kelvin.

Table S2 .
Uptake coefficients (γ ± 2σ) for BA on glutaric acid (GA) as a function of sample mass at 298 K. a Uncertainties represent 2σ statistical errors.Overall error includes possible systematic errors estimated to be ± 60% due to uncertainties associated with sample surface areas of the polydisperse powders and the correction to I0 and Ir from the BA background. b

Table S3 .
Uptake coefficients (γ ± 2σ) for BA on succinic acid (SA) as a function of sample mass at 298 K. a

Table S4 .
Uptake coefficients (γ ± 2σ) for BA on glutaric acid (GA) as a function of BA initial concentration ([BA]0) at 298 K. a a Orifice area is 0.33 cm 2 .