Article pubs.acs.org/JPCA
Vapor Pressure of Hexamethylene Triperoxide Diamine (HMTD) Estimated Using Secondary Electrospray Ionization Mass Spectrometry Matthew J. Aernecke, Ted Mendum, Geoff Geurtsen, Alla Ostrinskaya, and Roderick R. Kunz* Lincoln Laboratory, Massachusetts Institute of Technology, 244 Wood Street, Lexington, Massachusetts 02420, United States ABSTRACT: A rapid method for vapor pressure measurement was developed and used to derive the vapor pressure curve of the thermally labile peroxidebased explosive hexamethylene triperoxide diamine (HMTD) over the temperature range from 28 to 80 °C. This method uses a controlled flow of vapor from a solid-phase HMTD source that is presented to an ambientpressure-ionization mass spectrometer equipped with a secondary-electrosprayionization (SESI) source. The subpart-per-trillion sensitivity of this system enables direct detection of HMTD vapor through an intact [M + H]+ ion in real time at temperatures near 20 °C. By calibrating this method using vapor sources of cocaine and heroin, which have known pressure−temperature (P− T) curves, the temperature dependence of HMTD vapor was determined, and a Clausius−Clapeyron plot of ln[P (Pa)] vs 1/[T (K)] yielded a straight line with the expression ln[P (Pa)] = {(−11091 ± 356) × 1/[T (K)]} + 25 ± 1 (error limits are the standard error of the regression analysis). From this equation, the sublimation enthalpy of HMTD was estimated to be 92 ± 3 kJ/mol, which compares well with the theoretical estimate of 95 kJ/ mol, and the vapor pressure at 20 °C was estimated to be ∼60 parts per trillion by volume, which is within a factor of 2 of previous theoretical estimates. Thus, this method provides not only the first direct experimental determination of HMTD vapor pressure but also a rapid, near-real-time capability to quantitatively measure low-vapor-pressure compounds, which will be useful for aiding in the development of training aids for bomb-sniffing canines.
■
INTRODUCTION Trace chemical detection remains one of several means to screen for the presence of threats and/or contraband, and vapor detection plays a key role in the operation of these sensing systems. The majority of deployed chemical detection systems are based on ion mobility spectrometry (IMS) and sample particulate residues on surfaces by means of a swipe. The surface swipe is then inserted into the detection system where it is flash heated to vaporize the condensed-phase material that was collected. The vapor-phase material is then transferred into the ionization region of the IMS instrument, where it is ionized and subsequently analyzed. Whereas IMS-based explosives detection involves indirect vapor detection as an intermediate step inside the instrument, there are other methods that employ direct vapor detection in the ambient environment; the most notable of these systems are canines. Thus, given the importance of vapor detection across chemical detection systems, it is important to understand the magnitude and temperature dependence of a compound’s vapor pressure. Knowing the vapor source term helps in understanding, testing, and predicting sensor performance and can be useful when developing nonexplosive training aids for canines. The vapor pressures of most explosive compounds of importance to the explosives detection community have already been measured, and recently, they were compiled into two comprehensive reviews.1,2 © 2015 American Chemical Society
One notable omission from these collections of thermochemical data is hexamethylene triperoxide diamine (HMTD). This omission is not due to lack of effort; rather, HMTD can be difficult to measure due in part to its instability combined with its “very low” vapor pressure as determined from empirical observations. Studies aimed at characterizing the thermal stability of HMTD revealed a kinetic rate constant for decomposition of 10−3 s−1, 3 orders of magnitude higher than that of triacetone triperoxide (TATP, 10−6 s−1), with trimethylamine and carbon dioxide observed as the primary HMTD decomposition products at temperatures of 99.999% pure nitrogen; drying gas flow, 5 L/min; nebulizer pressure, 15 psi; fragmentor voltage, 100 V; and electrospray needle voltage, 4000 V. Vapor Generation and Delivery. Test vapor streams were generated by flowing nitrogen carrier gas over a solid sample of source material (in this work, either cocaine, heroin, or HMTD) that had been immobilized and dispersed onto a solid support, namely, a vapor generator.24 Individual vapor generators for cocaine, heroin, HMTD, and a blank were constructed by cutting a section of glass tube (3.9 mm i.d.) to ∼7 cm in length and packing ∼5 cm of glass wool inside the tube. Then, 100 μL of cocaine standard solution (1 mg/mL in acetonitrile), 100 μL of heroin standard solution (1 mg/mL in acetonitrile), or 100 μL of HMTD stock solution (1 mg/mL in acetonitrile, prepared from synthesized material) was added to 11516
DOI: 10.1021/acs.jpca.5b08929 J. Phys. Chem. A 2015, 119, 11514−11522
Article
The Journal of Physical Chemistry A Table 1. Published Vapor Pressure Data for Cocaine-Free Base Psat (20 °C) ref
temperature range (°C)
10−7 Torr
pptv
equation
coefficients
21−41
0.88 ± 0.3
116 ± 40
log10 P = A − B/T
Ziegler et al.30
not provided
1.2 ± 0.3
157 ± 40
ln P = slope × (1/T)
Hilpert et al.29 Dindal et al.31 commercial source
−10−40
0.6
79
not provided
17−42
2.96 ± 1.48a
389 ± 195
ln(n/V) = A + BT + error
illicit source
16−42
1.38 ± 0.74a
181 ± 98
ln(n/V) = A + BT + error
Lawrence et al.
a
28
A = 13 ± 0.5 B = 5884 ± 147 slope = −15293 ΔHsub = 127.2 kJ/mol not provided A = −56.0 B = 0.106 error = 0.142 A = −70.0 B = 0.152 error = 0.084
95% confidence.
loss of hydrogen peroxide from m/z 179. The fragments observed in the mass spectrum could arise from thermal decomposition of the source material or from an acid-catalyzed decomposition of HMTD in the ionization source due to the presence of formic acid in the spray solvent. In an effort to determine the source of this decomposition, the temperature of the HMTD vapor generator was ramped (from 35 to 80 °C), and the ratios of the intensities of the primary fragments (m/z 179, 145, and 88) to that of the HMTD [M + H]+ ion were determined. These ratios remained constant over the temperature range of 35−50 °C. This observation supports the idea that fragmentation is occurring either in the ionization source or through collision-induced dissociation (CID) in the mass spectrometer and is not due to thermal degradation in the vapor generator. Given this fact, the declustering potential on the mass spectrometer inlet was adjusted to maximize the signal of the [M + H]+ ion at m/z 209 relative to those of the fragment peaks at m/z 179, 145, and 88. At temperatures higher than 50 °C, thermal degradation did begin to take place in the vapor generator, and the intensities of m/z 74 and 60 increased. These ions correspond to dimethylformamide and trimethylamine, two breakdown products of HMTD that have been previously reported.3,4 Calibration of the Instrument Response to HMTD Vapor. The absence of published experimental vapor pressure data for HMTD precluded a direct calibration of the instrument with HMTD vapor. Thus, both cocaine and heroin vapors were used instead of HMTD vapor to calibrate the instrument response. Cocaine was selected for this purpose because its vapor pressure has been measured28−31, and its saturated vapor pressure at 20 °C has been reported to be 79−389 parts per trillion by volume (pptv), which is close to the value calculated for HMTD of 32 pptv.29 A summary of published vapor pressure data for cocaine-free base is presented in Table 1. Heroin was selected because its vapor pressure has also been reported,28 albeit by fewer researchers than for cocaine. Heroin also is expected to have a lower vapor pressure than HMTD, with a saturated vapor pressure at 20 °C of just 0.36 pptv. The equations used in this work to calculate the vapor pressure of both cocaine and heroin were drawn from Lawrence et al.28 because the full experimental description, results, and temperature range over which measurements were made were provided in the report and mirror those used in this work (21− 45 °C). The other cocaine results reported in Table 1 either omitted critical information (i.e., the temperature range in
switched on, and sample vapor was presented to the instrument for 1 min. This vapor presentation interval was sufficient for the signal in the mass spectrometer to reach a steady-state level. Following the vapor pulse, the carrier gas was turned off at the mass flow controller, and the valve was closed. The run finished with a second 30-s period of spray background. All instrument responses to a vapor sample at a specific temperature were acquired in triplicate. Vapor generator temperatures used, as measured on its exterior, were as follows: for cocaine, 21 ± 0.5, 25 ± 0.5, 28 ± 0.5, and 37 ± 1 °C; for heroin, 45 ± 0.5, 50 ± 1, 60 ± 1.5, 70 ± 1.5, 75 ± 1.5, and 80 ± 1.5 °C; and for HMTD, 28 ± 0.5, 37 ± 1, 45 ± 1.5, 60 ± 1.5, and 80 ± 1.5 °C. The error limits on the experimental temperature values represent the total change in the temperature of the vapor generator over triplicate measurements. The vapor generator was allowed to equilibrate for 5 min after reaching the programmed temperature set point and between vapor pulses to the mass spectrometer. The instrument responses were processed by extracting the time trace (chromatogram) for the [M + H]+ ion of cocaine (m/z 304), heroin (m/z 370), or HMTD (m/z 209) and averaging the signal intensity over the portion of the curve where the response had reached a steady state.
■
RESULTS AND DISCUSSION Mass Spectrum of HMTD Vapor. The headspace of HMTD was initially characterized in the mass spectrometer by presenting a gas sample generated from a vapor generator heated to 80 °C. This temperature provided a strong instrument response. A background-subtracted positive-polarity mass spectrum of HMTD vapor at this temperature is shown in Figure 1. The major background peaks subtracted were those attributable to the spray solvent, consisting of 12 different ion clusters of methanol and/or water ([CH3OH]x[H2O]y)H+ with x = 0−7 and y = 1−5 with the five most abundant being x = 3− 7, y = 1−2), in addition to several of the above solvent peaks containing formic acid and two peaks attributed to phthalate contamination. In Figure 1, the [M + H]+ ion for HMTD is observed at m/z 209.0764, and this ion is accompanied by HMTD associated fragments at m/z 179.0660, 145.0604, 88.0400, 74.0608, and 60.0453. These fragments correspond to those previously observed in the mass spectrum of HMTD using the DART (direct analysis in real time) ionization source.25−27 The ion at m/z 179 corresponds to a loss of formaldehyde from the [M + H]+ ion, and that at m/z 229 is a water/methanol adduct of m/z 179. The ion at m/z 145 is a 11517
DOI: 10.1021/acs.jpca.5b08929 J. Phys. Chem. A 2015, 119, 11514−11522
Article
The Journal of Physical Chemistry A Ziegler et al.30 or the model and fit coefficients in Hilpert et al.29) or their published values deviated significantly from those reported previously. Thus, the values of Lawrence et al.28 were used in this report, where, in addition to the cocaine coefficients listed in Table 1, the heroin vapor pressure coefficients of A = 16.2 and B = 7549 were reported for the vapor pressure equation log10 P = A − B/T. In addition to their low vapor pressures, cocaine and heroin both readily form [M + H]+ protonated adducts when ionized at atmospheric pressure (as does HMTD). Cocaine’s proton affinity is 930.1 kJ/mol as calculated by ab initio methods.33 The proton affinities for heroin and HMTD have not been published; however, we estimate them both to be greater than 940 kJ/mol by reviewing the proton affinities of similar compounds that have either one or two proton-accepting tertiary amines contained in a ring structure, a feature common to both heroin (one tertiary amine) and HMTD (two tertiary amines). Selected proton affinity values for such compounds are listed in Table 2.34 Based on these high values of proton
Figure 2. SESI/Q-TOF MS raw data for the temperature-dependent measurements for cocaine, heroin, and HMTD. The solid straight lines are to guide the eye.
Table 2. Published34 Proton Affinities for Non-Peroxide Tertiary Azacyclo Compounds That Contain ProtonAccepting Nitrogen Groups Similar to Heroin and HMTD name
molecular formula
proton affinity (kJ/mol)
1,6-diazabicyclo[4.4.4]tetradecane 1,5-diazabicyclo[3.3.3]undecane 1,4-diazabicyclo[2.2.2]octane 2,6-t-butyl piperidine 1-azabicyclo[2.2.2]octane 1-methyl-2,6-t-butyl piperidine
C12H24N2 C9H18N2 C6H12N2 C13H25N C7H13N C14H27N
947 972 985 948 971 1001
Figure 3. Cocaine and heroin data from Figure 2 replotted by converting the temperature scale to partial pressure using the Antoine coefficients from Lawrence et al.28 This established the relationship between raw counts and vapor pressure that formed the basis for the estimation of the HMTD vapor pressure.
the calculated vapor pressure returned R2 values of 0.990 for cocaine and 0.997 for heroin. The equation derived from the cocaine and heroin calibration curves in Figure 3 was then used to convert the measured intensity of the HMTD [M + H]+ ion into a vapor pressure, and this was done separately using both the cocaine and heroin curves as references. These data were then compiled into a Clausius−Clapeyron plot by plotting the inverse temperature (in inverse kelvin) versus the natural logarithm of the calculated HMTD vapor pressure, as shown in Figure 4. The dashed line in Figure 4 is the theoretical P−T curve for HMTD as determined using a quantum mechanical continuum solvation model.32 However, it is important to note that the
affinity, the proton exchange from the methanol/water/formic acid spray solvent to these analytes should be strongly thermodynamically favored, and thus, these analytes should exhibit similarly high ionization probabilities. The assumption that cocaine, heroin, and HMTD ionization are all highly thermodynamically favored and thus similar (this assumption is addressed below) enables the calibration of the instrument’s response to cocaine and/or heroin vapor and then use of the equations derived from their pressure−temperature (P−T) relationships to quantify an unknown instrument response to HMTD vapor at a particular temperature. For these measurements, separate vapor generators were installed for each compound, and triplicate measurements were made on the Q-TOF/MS instrument at five temperature points within the range of 20−80 °C. The response of the instrument to each vapor was recorded by monitoring the intensity of the [M + H]+ ions for each compound. The results for each of the three compounds, plotted in the form of a Clausius−Clapeyron equation, are shown in Figure 2, and each of the plots has a value of R2 ≥ 0.99. Because the P−T relationships for both cocaine and heroin are known,28 the temperature scale in Figure 2 can be converted to a calculated partial pressure, thereby establishing a notional instrument calibration for each of the two compounds. This result is shown in Figure 3, where the y axis in Figure 2 was changed to base-10 logarithm for clarity. The most notable feature of Figure 3 is the near overlap of the data, suggesting that, at least for heroin and cocaine, the assumption of ionization probability similarities is notionally correct. The correlation between the instrument response and
Figure 4. Clausius−Clapeyron plot for HMTD. The symbols are the experimental data points derived with both the cocaine (diamonds) and heroin (circles) calibration curves from Figure 3. The dashed line is the modeled HMTD vapor pressure curve derived from ref 32. 11518
DOI: 10.1021/acs.jpca.5b08929 J. Phys. Chem. A 2015, 119, 11514−11522
Article
The Journal of Physical Chemistry A Table 3. Molecular Parameters Used to Calculate Ionization Rate Coefficients compound
molecular weight (g/mol)
ion mobility [cm2/(V s)]
polarizability (Å3)
dipole moment (D)
dipole locking constant
rate constant (×109 cm3/s)
probability of ionization (Pi) (×104)
cocaine heroin HMTD
303 369 208
1.1513 1.0513 1.50
32.2a 38.5a 17.4
2.50 3.1340 0.68b
0.125 0.1641 0.080
2.2 2.6 1.5
10.6 13.6 5.1
a
Value generated using the ACD/Labs Percepta Platform, PhysChem Module. See http://www.chemspider.com/Chemical-Structure.4575379.html (last accessed Oct 26, 2015). bEstimated from quantum chemical calculations.
Figure 5. Clausius−Clapeyron plots for HMTD. (Left) Curve fits for HMTD vapor pressure derived from data in Figure 4 using cocaine (red dashed line) and heroin (black dashed line) vapors for calibration. (Right) Same curves corrected for differences in ionization probability and [M + H]+ yield (see text for discussion). In each panel, the solid black line is the modeled HMTD vapor pressure curve derived from ref 32.
Table 3. We calculated the molecular properties in Table 3 that could not be found in the literature using quantum chemical methods [B3LYP/6-31G(d,p) level of theory], as values computed using these parameters have shown good correlation between theory and experiment.37 The net result of these calculations suggests that cocaine has a probability of ionization that is 2.1 times that of HMTD and heroin has an ionization probability that is 2.6 times that of HMTD. This is reasonable, as previous attempts to apply ADO theory to vapor ionization in electrospray clouds have concluded that vapor−vapor interaction, and not vapor−droplet interaction, is the predominant ionization mechanism.38 Additionally, cocaine decomposes into methyl ecgonine (among other compounds) at increased temperatures and when presented to our SESI ionization source.39 Importantly, the fraction of cocaine molecules detected as the cocaine [M + H]+ ion ranged from 0.20 to 0.40 across the experimental temperatures in our system and was very close to 0.25−0.31 range observed for HMTD. In contrast, the [M + H]+ peak for heroin comprised 0.6−0.8 of the total signal (factor of ∼2 times higher [M + H]+ ion yield than for HMTD). Thus, the estimated P−T data for HMTD derived from both cocaine and heroin shown in Figure 4 can be corrected using the differences in the total ionization and fractional [M + H]+ yield and those data fit using linear regression analysis to estimate corrected values of P−T over a full curve. These results are shown in Figure 5 for both the corrected and uncorrected results. Table 4 lists selected values for the five different curves shown in Figure 5. From the data in Figure 5 and Table 4, one can see that the corrected vapor pressure values derived from both cocaine and heroin are within 50% of one another and within a factor of 3 of the theoretical values over the temperature range from 10 to 40 °C. The values derived from cocaine agree with the theoretical values in ref 32 better than those derived from heroin, with results within a factor of 2 of the theoretical values over the entire temperature range of 10−80 °C. It should also be noted that previous comparisons between the theoretically derived vapor pressures from ref 32 with experimental values also varied within a factor of ∼2 (
