Article pubs.acs.org/JPCA
Measurement of Vapor Pressures and Heats of Sublimation of Dicarboxylic Acids Using Atmospheric Solids Analysis Probe Mass Spectrometry Emily A. Bruns, John Greaves, and Barbara J. Finlayson-Pitts* Department of Chemistry, University of California, Irvine, Irvine, California 92697-2025, United States ABSTRACT: Vapor pressures of low volatility compounds are important parameters in several atmospheric processes, including the formation of new particles and the partitioning of compounds between the gas phase and particles. Understanding these processes is critical for elucidating the impacts of aerosols on climate, visibility, and human health. Dicarboxylic acids are an important class of compounds in the atmosphere for which reported vapor pressures often vary by more than an order of magnitude. In this study, atmospheric solids analysis probe mass spectrometry (ASAPMS), a relatively new atmospheric pressure ionization technique, is applied for the first time to the measurement of vapor pressures and heats of sublimation of a series of dicarboxylic acids. Pyrene was also studied because its vapor pressures and heat of sublimation are relatively well-known. The heats of sublimation measured using ASAP-MS were in good agreement with published values. The vapor pressures, assuming an evaporation coefficient of unity, were typically within a factor of ∼3 lower than published values made at similar temperatures for most of the acids. The underestimation may be due to diffusional constraints resulting from evaporation at atmospheric pressure. However, this study establishes that ASAP-MS is a promising new technique for such measurements.
■
aerosols.11−22 Dicarboxylic acids are primarily formed through photooxidation reactions,13,14,16,17,20 although they can also be directly emitted from both anthropogenic and biogenic sources.12,16 Vapor pressure measurements of low volatility compounds have been made using several techniques, including tandem differential mobility analyzers (TDMA),23−29 effusion cell techniques,30−33 electrodynamic balance techniques34 and temperature programmed desorption (TPD).35−40 Reported vapor pressures of dicarboxylic acids typically vary by 1−2 orders of magnitude.23−37,40 Measured heats of sublimation also vary significantly.23−37,40−43 The variations illustrate the challenges of making these determinations and the need for new approaches for such measurements. We report here the first application of atmospheric solids analysis probe mass spectrometry (ASAP-MS) to the measurement of the vapor pressures and heats of sublimation of dicarboxylic acids. We have previously demonstrated the utility of this technique for the study of aerosol composition.44 The 2D data sets generated by ASAP-MS, consisting of mass spectra as a function of temperature, can be used to directly measure
INTRODUCTION Aerosols in the atmosphere are known to negatively impact human health,1−3 degrade visibility,4 and affect climate.5 The wide range of aerosol composition and properties, and their constant emission into and transformation in the atmosphere, makes complete characterization of their effects challenging. In the atmosphere, aerosols can be classified as either primary or secondary.6 Primary aerosols are those that are emitted directly into the atmosphere. Secondary aerosols are formed when gaseous compounds react in the atmosphere to form products with sufficiently low volatility to form new particles, or to be taken up into or onto pre-existing particles.7,8 Knowledge of the properties of individual aerosol components both by themselves and in mixtures is critical for understanding aerosol processes and impacts. A particularly important property is vapor pressure, which is needed to predict the potential contribution of compounds to the formation and growth of airborne particles. Atmospheric models typically incorporate the assumption that particles grow via equilibrium partitioning of gas phase compounds onto preexisting particles, with a partitioning coefficient that is inversely proportional to vapor pressure.9,10 However, vapor pressures of low volatility compounds are very challenging to measure because the compounds of interest often stick to surfaces and can undergo phase changes during analysis. Dicarboxylic acids are widely dispersed in our atmosphere and often have sufficiently low vapor pressures to be found in © 2012 American Chemical Society
Special Issue: A. R. Ravishankara Festschrift Received: October 18, 2011 Revised: March 18, 2012 Published: March 20, 2012 5900
dx.doi.org/10.1021/jp210021f | J. Phys. Chem. A 2012, 116, 5900−5909
The Journal of Physical Chemistry A
Article
For each compound, a 5 μL droplet was placed on the ASAPMS probe tip. A significant fraction of the solvent was observed to evaporate and the sample “beaded up” on the probe tip during a 3 min period before inserting the sample into the ASAP-MS ionization region. However, visual inspection showed that the droplet often looked liquid-like when placed in the ionization region, suggesting that some of the solvent remained. The temperature of the sample compartment is ∼40 °C, which should result in further solvent evaporation immediately upon inserting the sample into the instrument. As a test, a 5 μL droplet was placed on the probe tip and then briefly placed into an oven at 40 °C. The droplet was observed to further shrink and become solid on the probe tip surface, indicating that the remaining solvent had evaporated. The N2 gas (500 L h−1) temperature was increased stepwise in increments of ∼8 °C every minute, starting at ∼40 °C, the lowest starting temperature of the instrument. The experiment ended when all the material had desorbed from the probe tip and the signal had returned to its background level. A desorption profile was obtained by monitoring the signal from the peak, or sum of peaks, corresponding to the compound of interest in the mass spectra as a function of time. Background scans were collected following the same procedure, but without any sample on the probe tip. The entrance into the mass spectrometer was held constant at 100 °C. The instrument was operated in negative ionization mode for the dicarboxylic acids and in positive ionization mode for pyrene. For the dicarboxylic acids, an [M − H]− peak was observed, as well as a peak corresponding to [2M − H]− resulting from dimer formation, which has been previously observed for carboxylic acids.58 In some cases, a small peak corresponding to [M − H − H2O]− was observed. Pyrene exhibited only an [M + H]+ peak in the mass spectra. Prior to use, the probe was cleaned by baking at 150−200 °C. Accurate temperature readings at the probe tip surface are needed. However, the N2 gas temperatures reported by the instrument software (MassLynx, Waters) are not taken at the probe tip surface. To obtain accurate temperature measurements at this surface, measurements were made using a type K thermocouple (Model HH21, Omega). Unfortunately, the proximity of the corona discharge to the probe tip precluded thermocouple measurements while analyzing samples. Several experiments with the corona discharge voltage set to zero were performed daily to determine the thermocouple temperature readings as a function of the software reported temperatures. The average of the thermocouple measurements at each temperature was used to convert the instrument readings to the corresponding temperatures at the probe tip surface during the actual analyses. Significant adsorption of the sample to the probe tip can broaden and shift the desorption profile, which affects vapor pressure measurements. For ASAP-MS analysis, a glass melting point tube (mpt) is typically used as the probe tip. However, the sample was observed to spread out on the mpt and the desorption profiles were broad and inconsistent. This indicated significant interaction between the sample and the probe tip. Therefore, several different probe tips and coatings were tested by analyzing hexanedioic acid. Experiments were performed using: poly(ether ether ketone) (PEEK) tubing (1/16 in. o.d., Upchurch Scientific), an mpt silanized using either pure chlorotrimethylsilane (Sigma Aldrich, 97+%) or a dilute solution in toluene (EMD Chemicals, >99.99%), an mpt coated with a n-octadecyltrichlorosilane (Pfaltz and Bauer, Inc.,
evaporation rates. The kinetic theory of gases,45 which relates evaporation rates to vapor pressures, has been previously used to determine vapor pressures from data acquired using similar thermal desorption techniques35−40 and is applied here to the ASAP-MS data.
■
EXPERIMENTAL SECTION Vapor pressure measurements reported in the literature for pyrene, a polycyclic aromatic hydrocarbon,46−55 are generally in excellent agreement, making this a good benchmark for investigating the ability of ASAP-MS to measure vapor pressures. Thus, as a proof of concept, vapor pressures were first measured for pyrene (≥99.0%). Studies were then carried out on eight dicarboxylic acids, each used as received: butanedioic acid (≥99.0%), pentanedioic acid (99.0%), hexanedioic acid (≥99.0%), heptanedioic acid (98%), octanedioic acid (98%), nonanedioic acid (98%), decanedioic acid (99%), and dodecanedioic acid (99%). Pyrene and all the dicarboxylic acids were purchased from Sigma Aldrich. All the compounds were solid at room temperature. A solution of each was made by dissolving the solid in methanol (Fisher Scientific, 99.9%). Vapor pressure measurements for each compound were made using ASAP-MS (LCT Premier time-of-flight mass spectrometer and ASAP probe, Waters).44,56,57 Figure 1
Figure 1. Schematic of the ionization region of the atmospheric solids analysis probe mass spectrometer (ASAP-MS). Material desorbs from the probe tip into the atmospheric pressure ionization region as the temperature of heated N2 gas flowing over the sample is increased. Chemical ionization occurs in the presence of a corona discharge and a small vial of water. Either positive or negative ions are detected using mass spectrometry.
shows a schematic of the ASAP-MS ionization region. ASAPMS involves placing material on a probe tip, and then inserting it into the atmospheric pressure ionization region of a mass spectrometer. Heated N2 gas flows over the sample, and as the temperature of the gas is increased, material desorbs from the probe tip. A small vial of water in the ionization region provides a source of water vapor and a corona discharge (3 kV, 5 μA) results in chemical ionization. The ionization forms primarily either [M − H]− or [M + H]+ ions, for which the mass to charge ratios are determined in the mass analyzer. 5901
dx.doi.org/10.1021/jp210021f | J. Phys. Chem. A 2012, 116, 5900−5909
The Journal of Physical Chemistry A
Article
Figure 2. Examples of evaporation rate as a function of time and temperature for pyrene (a, b), octanedioic acid (d, e), and heptanedioic acid (g, h). The temperature profiles are shown in (a), (d), and (g). The vertical dashed lines in (h) and (i) correspond to the melting temperature of heptanedioic acid (378 K).60 The melting temperatures of pyrene (424 K)60 and octanedioic acid (415 K)60 are above the temperature ranges shown in (b, c) and (e, f). Clausius−Clapeyron plots of pyrene (c), octanedioic acid (f), and heptanedioic acid (i) including the least-squares fits (dashed lines) were used to calculate the heats of sublimation.
■
95%) self-assembled monolayer,59 an mpt and PEEK tubing coated with TruSheen (TruSheen Co., Inc., Henderson, NV), and an mpt and PEEK tubing coated with Kisscote (KISS Polymers, LLC, Tampa, FL). Both TruSheen and Kisscote are inert and hydrophobic nonstick polymer coatings designed to minimize interactions between a compound and the coated surface. Chattopadhyay and Ziemann36 used a Kisscote coated molybdenum vaporizer to reduce adsorption of carboxylic acids to the surface as compared to a gold coated or uncoated vaporizer. In the current study, the desorption profiles that peaked at the lowest temperature, indicating minimal adsorption to the probe tip, were obtained when using uncoated PEEK tubing and this was used in all subsequent experiments for which results are reported here. As discussed in more detail in the following section, measurements of the initial dimensions of the sample on the probe tip were necessary. After briefly placing the sample in an oven at 40 °C to simulate the ASAP-MS ionization region, the sample was observed to become solid and have a disk-like shape. The diameter of the dried disk was measured and was on the order of 0.5−1.5 mm for all samples. Using the known mass of material on the probe tip and the compound density, the volume of material on the probe tip was calculated. This was then used to calculate the height of the disk, which was 10−4 to 10−3 mm. For pyrene, a UV lamp (Model UVGL-58, UVP, Inc.) was used to induce fluorescence, which could then be observed visually to view the dried sample on the probe tip.
DATA ANALYSIS The desorption profile generated from the ASAP-MS data can be used to directly calculate the evaporation rate (Et), in molecules s−1: Et = N0 ×
It Itotal
(1)
N0 is the total number of molecules initially on the probe tip, It is the signal at a given time (counts s−1), and Itotal is the integrated area under the entire desorption profile in counts. To remove background contributions to It , the average signal in the background scan at each temperature step for the mass of interest was subtracted from It. The average contribution of the background signal to the total signal at the maximum desorption was less than 1% for the dicarboxylic acids and ∼10% for pyrene. N0 was calculated from the known mass placed on the probe tip. Figure 2a,d,g show the evaporation rate as a function of time, calculated from eq 1, for pyrene, octanedioic acid, and heptanedioic acid. The temperature profile as a function of time is also shown in Figure 2a,d,g. Averaging the evaporation rates over the time period at each temperature gives evaporation rate as a function of temperature, as shown in Figure 2b,e,h. The first five seconds at each temperature were not used in the average, as the signal had not stabilized. For pyrene and octanedioic acid, all desorption occurred below their respective melting temperatures of 424 5902
dx.doi.org/10.1021/jp210021f | J. Phys. Chem. A 2012, 116, 5900−5909
The Journal of Physical Chemistry A
Article
⎛ N t I dt ⎞ ⎜ 0 ∫0 t ⎟ nt = N0 − ⎜ Itotal ⎟ ⎝ ⎠
and 415 K.60 This was not the case for all of the compounds studied. For example, the melting temperature of heptanedioic acid is 378 K.60 This occurred in the middle of the temperature ramp, as shown in the plots of evaporation rate as a function of time and temperature in Figure 2g,h. When the compound melted prior to the end of the experiment, the experiment was continued until all material desorbed and the signal returned to the background level. While this was necessary to accurately determine Itotal, only data acquired below the melting temperature were used to calculate vapor pressures and heats of sublimation (ΔHsub). A modified version of Langmuir’s model of evaporation,45 based on the kinetic theory of gases, was used to relate the evaporation rate to vapor pressure: Pt =
Et 2πRTM αNA SA t
The Clausius−Clapeyron equation was used to determine the heat of sublimation (ΔHsub): ln PT = −
(2)
(3)
where h was calculated from the known volume and measured diameter of the sample prior to evaporation. As the sample evaporated, rt was calculated from the volume of material remaining on the probe (Vt) and h using eq 4. rt =
Vt πh
ΔHsub +C RT
(6)
where C is the constant of integration. The ΔHsub for each compound was determined by plotting the natural log of the vapor pressure from eq 2 averaged over each temperature step, PT, as a function of inverse temperature and fitting the linear portion of the plot. As was the case for ET, the first five scans at each temperature were not used in the vapor pressure average, as the signal had not stabilized. The linear region was determined visually, as was done in previous TPD studies.35−40 It begins when the signal is greater than the background and continues until approximately the temperature of maximum evaporation, or the melting temperature of the sample. Typical Clausius−Clapeyron plots are shown in Figure 2c for pyrene, in Figure 2f for octanedioic acid, and in Figure 2i for heptanedioic acid, along with the leastsquares fits (dashed lines) to the linear portions. Only the vapor pressure measurements within the temperature range used in the least-squares fits are reported. For heptanedioic acid, experiments with differing numbers of molecules initially placed on the probe tip were performed. The heats of sublimation agreed within 5% and the average vapor pressures agreed within 20−40% when the initial sample size was varied. The error on each measured vapor pressure corresponds to two standard deviations (2s) calculated from propagation of random errors. The sample standard deviation (s) is used rather than the population standard deviation (σ) because less than 20 measurements were made in each case.64 These errors were within a factor of 2 of the errors associated with replicate experiments, with the exception of dodecanedioic acid where it was difficult to obtain reproducible data. The largest contributor to the random error was from the signal intensity relative to the noise. The signal intensity (I) was taken as I = S − B, where S is the total number of counts per second in the signal and B is the total number of counts per second in the background. The error (σI) in I was calculated as σI = (σS2 + σB2)1/2, where σS and σB are the standard deviations in S and B, respectively. The standard deviations were obtained assuming that the peak to peak noise at each temperature step was 5 standard deviations.64 The error in I decreased relative to the total sample signal as the amount of material desorbing increased. It was typically ∼15% of the total sample signal, but at the lowest temperature, where the sample signal was low, it was as high as 100% of the total signal. Other errors taken into account arise from thermocouple temperatures, from volumes used to make the solution, from volume of the sample placed on the probe tip and from measurements used in determining the surface area. The standard deviation was 0.6 K on the thermocouple measurements and the errors in the measured volumes and diameters resulted in a standard deviation of
