Tuning Mobility Separation Factors of Chemical ... - ACS Publications

Article pubs.acs.org/ac

Cite This: Anal. Chem. XXXX, XXX, XXX-XXX

Tuning Mobility Separation Factors of Chemical Warfare Agent Degradation Products via Selective Ion-Neutral Clustering Pearl Kwantwi-Barima,† Hui Ouyang,§ Christopher J. Hogan, Jr.,‡ and Brian H. Clowers*,† †

Department of Chemistry, Washington State University, Pullman, Washington 99164, United States Department of Mechanical Engineering, University of Minnesota, Minneapolis, Minnesota 55455, United States § Department of Mechanical Engineering, University of TexasDallas, Richardson, Texas 75080, United States ‡

S Supporting Information *

ABSTRACT: Combining experimental data with computational modeling, we illustrate the capacity of selective gas-phase interactions using neutral gas vapors to yield an additional dimension of gas-phase ion mobility separation. Not only are the mobility shifts as a function of neutral gas vapor concentration reproducible, but also the selective alteration of mobility separation factors is closely linked to existing chemical functional groups. Such information may prove advantageous in elucidating chemical class and resolving interferences. Using a set of chemical warfare agent simulants with nominally the same reduced mobility values as a test case, we illustrate the ability of the drift-gas doping approach to achieve separation of these analytes. In nitrogen, protonated forms of dimethyl methyl phosphonate (DMMP) and methyl phosphonic acid (MPA) exhibit the reduced mobility values of 1.99 ± 0.01 cm2 V−1s−1 at 175 °C. However, when the counter current drift gas of the system is doped with 2-propanol at 20 μL/ h, full baseline resolution of the two species is possible. By varying the concentration of the neutral modifier, the separation factor of the respective clusters can be adjusted. For the two species examined and at a 2-propanol flow rate of 160 μL/h, MPA demonstrated the greatest shift in mobility (1.58 cm2V−1s−1) compared the DMMP monomer (1.63 cm2V−1s−1). Meanwhile, the DMMP dimer experienced no change in mobility (1.45 cm2V−1s−1). The enhancement of separation factors appears to be brought about by the differential clustering of neutral modifiers onto different ions and can be explained by a model which considers the transient binding of a single 2-propanol molecule during mobility measurements. Furthermore, the application of the binding models not only provides a thermodynamic foundation for the results obtained but also creates a predictive tool toward a quantitative approach.

I

buffer gas molecule structure as well as gas temperature and pressure. Theoretically, each ion should possess a unique gasphase mobility (at a given temperature, pressure and for a given gas composition), however, because mobility is a joint, orientationally averaged property of an ion and a neutral, it is entirely possible that ions containing different chemical function groups will display the same nominal mobility. This behavior is one contributing factor in the observation of false positive rates in ion mobility systems. When viewed from a purely analytical perspective, the use of inert buffer gases provides little opportunity to influence selectivity during mobility separations; ions of similar mobility in one buffer gas often have similar mobilities in a second buffer gas.4,5 Buffer gas modifiers are hence routinely used in field applications of IMS to promote specific types of ion chemistry; unlike inert buffer gas molecules, such modifiers, when chosen appropriately, can selectively cluster with the analyte. For example,

on mobility spectrometry (IMS) is an analytical technique that largely employs the mobility coefficient of ions in the gas-phase as both a separation and an identification tool for analytes. IMS, although a relatively low resolution separation method, is frequently applied to the identification and separation of explosives, chemical warfare agents, and illegal substances in environmental or security settings, in which higher resolution mass spectrometers are either difficult or too expensive to operate. The proportionality coefficient, also known as ion mobility coefficient, K,1,2 provides a link between average velocity (vd) of an ion moves and an external electric field (E). In a drift tube ion mobility measurement, K is provided by the equation:

K = vd /E = L2 /Vtd

(1)

where L is the ion drift distance in centimeters, V is the potential drop across the drift cell in volts, and td, is the measured time it takes the ion to reach the detector in seconds. The mobility coefficient itself is linked directly to the rate of momentum transfer from buffer gas molecules to the ion (i.e., the drag force on the ion),3 and is dependent on both ion and © XXXX American Chemical Society

Received: August 28, 2017 Accepted: October 23, 2017 Published: October 23, 2017 A

DOI: 10.1021/acs.analchem.7b03518 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Simplified degradation pathway for G-type nerve agents illustrating the relationship to methylphosphonic acid. Dimethyl methyl phosphonate is used a comparatively safe simulant for laboratory studies of gas-phase behavior. For reference, R′ = CH3 and R″ = CH3.

Figure 2. Schematic representation of the atmospheric pressure dual gate ion mobility system coupled with a LTQ mass spectrometer. Mobility spectra may be acquired using either the integrated Faraday plate or by encoding the mobility data in the frequency domain.13 For the latter the FFT may be used to identify the mobility distributions of m/z selected ion populations. Drift gas modifiers are introduced into the counter-current drift gas using a syringe pump and a temperature controlled heated injection liner.

Bollan et al.6 demonstrated the potential of ketone (e.g., 5nonanone) drift gas modifiers to shift the mobilities of protonated ammonia and select hydrazines. Introduction of small amounts of polar volatile organic compounds into the buffer gas have also been used to change the arrival time of analyte ions. Roscioli et al.,7 found that by using acetonitrile as the buffer gas modifier, the protonated response ion of dimethyl methyl phosphonate (DMMP), exhibited a shift in mobility of ∼11%. In a separate work by the same research group, Fernandez-Maestre and co-workers8 demonstrated that the mobilities of tetramethylammonium (TMA), tributylamine (TBA), 2,4-dimethylpyridine (2,4-lutidine), serine, and valinol decreased to different extents with the amounts of ammonia introduced in the mobility spectrometer. Relative ion and dopant size, charge site location, and steric hindrance have all been postulated as factors that influence the degree of mobility shifts when using drift gas dopants but a unifying quantitative model has been elusive.6,9,10 However, recent works by Rawat et al. and Oberreit et al.11,12 established the foundations of an approach to quantitatively account for vapor-mediated mobility shifts occurring under ambient conditions. Provided the correct input parameters, their model was capable of assessing whether the observed mobility shift is due to specific or nonspecific interactions. Building upon these insights and findings, the

objectives of the current work was to further develop a model to predict vapor-induced mobility shifts and realize analytically relevant mobility separations using vapor modifiers. Toward these ends, using a set of chemical warfare agent simulants with nominally the same reduced mobility in dry nitrogen, we illustrate selective mobility shifts enabling resolution of these species in the presence of a vapor modifier (2-propanol). The simulants used were dimethyl methylphosphonate (DMMP) and methyl phosphonic acid (MPA), both of which are hydrolyzed products of the G-type nerve agents specifically GB (Sarin) nerve agents (Figure 1). Comparison of mobility shift measurements as functions of 2-propanol gas phase concentration (nondimensionalized as a saturation ratio) to a mobility shift model provides a thermodynamic foundation for interpreting the results obtained.

■

EXPERIMENTAL SECTION Reagents. Methyl phosphonic acid (MPA) and dimethyl methyl phosphonate (DMMP) are chemical warfare agent simulants; 2-propanol was used as the buffer gas modifier to assess the magnitude of the mobility shifts as a function of gasphase concentrations. Both the analytes and the gas-phase modifier listed above as well as HPLC grade methanol and 0.1% formic acid (ACS reagent grade, ≥ 97 or 98% purity)

B


Article

Analytical Chemistry were purchased from Sigma-Aldrich Chemical Co. (Milwaukee, WI, U.S.A.), to be used in electrospray based MPA and DMMP protonated ion generation. Ion Mobility Measurements. An atmospheric pressure, dual-gate drift tube ion mobility spectrometer (ExcellIMS, MA3100) coupled with a LTQ mass spectrometer (Thermo Fisher Scientific) as seen in Figure 2 was used to obtain both the mobility and the mass information that enabled the identification of the separated clusters. The operating conditions used were a liquid sample flow of 2 μL/min, an electrospray ionization voltage of 2400 V (relative to first IMS electrode); first Bradbury-Nielsen ion gate potential, 50 V; drift tube voltage, 8000 V; pressure, atmospheric pressure (∼690 Torr in Pullman WA); nitrogen flow, 1.5 L/min; drift gas, 0.5 L/min; and drift tube and nitrogen temperature, 175 °C ± 0.2. Using a custom pulled glass capillary as an electrospray source, ion current was measured both with a Faraday plate and the LTQ mass analyzer. By using the custom ExcellIMS electronics assisting the MA3100 unit, an electric field of ∼419 V cm−1 was utilized throughout the experiment. Mass-selected mobility spectra were obtained by frequency encoding the mobility data using the approach by Morrison et al.13 For the frequency modulated IMS experiments (Fourier Transform-IMS (FTIMS)) the frequency sweep was 10 kHz over the course of 8 min. Analyte Preparation and Modifier Introduction. The final concentration of the prepared mixture of analytes was 2.3 μM DMMP and 10 μM MPA. The analyte solutions were prepared in HPLC grade methanol with 0.1% formic acid. These analytes were electrosprayed into the desolvation region of the instrument through a custom ionization stage using a flow rate of 2 μL/min. As a drift gas modifier, 2-propanol was introduced into the primary drift gas inlet via a glass capillary coupled to a temperature-controlled GC injection liner at a variable flow rate from 5 μL/h to 160 μL/h using a syringe pump. Under the conditions stated above, the gas phase concentration for 2-propanol corresponds to a saturation ratio of S = 0−6.5 × 10−5. Saturation ratio is defined as the vapor pressure divided by the saturation pressure, which is equivalent to the number concentration of vapor molecules divided by the equilibrium number concentration (derived via the Clausius− Clapeyron equation). To avoid the potential for temperature gradients in the IMS drift cell, the drift gas mixing chamber was held at the same nominal temperature as the mobility drift cell. Data for all drift gas modifier conditions were performed in triplicate in a sequential order to assess the reproducibility of the observed shifts.

t=

L KE

(1a)

In instances where vapor molecules may sorb and desorb from the ion during measurement, the ion spends a fraction of its time without a vapor molecule bound (t0), and a fraction with a specific number (g) of vapor molecules bound (tg). In these time intervals, it traverses distances L0 (without the vapor molecule bound) and Lg (with vapor molecules bound), with L = L0 + L1 + L2+ ··· L0 and Lg are linked to t0 and tg via an equation similar in form to eq 1, hence: L = tKE = L0 + L1 + L 2 + = t0K 0E + t1K1E + t 2K 2E + ··· + tgK g E + ···

K=

(2a)

tg t0 t t K 0 + 1 K1 + 2 K 2 + ··· + K g + ··· t t t t

(2b)

where K0 and Kg are the mobilities of the bare ion and the ion plus vapor molecule complex comprised of g vapor molecules, respectively. Provided the residence time within the drift cell is sufficient for ions to equilibrate with their surroundings, the time ratios are equivalent to P0 and Pg, which are defined as the probability an ion at a given instant in time has zero or g vapor molecules bound, respectively. Shown previously12 these ratios can be written as follows: 1

P0 = 1 + S exp

(

ΔG − kT1

)+S

2

exp

(

ΔG − kT2

) + ··· + S

g

(

exp −

ΔGg kT

) + ··· (3a)

(

S g exp − Pg = 1 + S exp

(

ΔG − kT1

)+S

2

exp

(

ΔG − kT2

ΔGg kT

)

) + ··· + S

g

(

exp −

ΔGg kT

) + ··· (3b)

where S is the saturation ratio of the vapor, and ΔGg is the Gibbs free energy change associated with the binding of g vapor molecules to an ion, under conditions where the drift gas is saturated with the vapor. Under extremely subsaturated conditions, S ≫ S2, and typically terms larger than g = 1 can be neglected (i.e., only the binding of one molecule need be considered). Combining eq 2b with eqs 3a and 3b with this restriction leads to the following: K

ΔG

( ) ( )

1 + K1 S exp − kT1 K 0 = ΔG K0 1 + S exp − kT1

■

THEORETICAL CONSIDERATIONS The shift in an ion’s mobility measured in the presence of vapor which can bind transiently to the ion has been described theoretically by Oberreit et al.11,12,14 Although these prior studies focused on differential mobility analyzers,15 a nanoparticle ion mobility spectrometer, 16 and a transverse modulation ion mobility spectrometer,17 the same derivation applies to small molecule measurements in drift tubes. Here, we develop similar equations to these prior works to model 2propanol uptake by ions, but do not force the free energy change upon binding to fit a specific functional form, as has been done previously. For a drift tube ion mobility spectrometer, an ion’s arrival time is linked to the drift length (L), its observed mobility (K), and the electric field strength (E) via the equation:

(4)

On the basis of the Mason-Schamp equation,18 the ratio

K1 K0

can be expressed as follows: Ω 0μ11/2 K1 = K0 Ω1μ01/2

(5)

where Ω0 and μ0 refer to the collision cross section and the reduced mass for the bare ion and the subscript “1” denotes these quantities for the ion-vapor molecule complex (i.e., g = 1). Combining eqs 4 and 5, and noting the link between arrival time and mobility for linear drift tube ion mobility spectrometer, leads to the following: C


Article


Table 1. Summary of the Measured and Calculated Collision Cross Sections, as Well as the Reduced Masses (in Nitrogen) of the Ions of Interest ION methylphosphonic acid-H+ methylphosphonic acid-2-propanol-H+ dimethyl methylphosphonate-H+ dimethyl methylphosphonate-propanol-H+

experimental CCS

theoretical CCS

reduced mass (Da)

± ± ± ±

91.79 123.6 95.05 121.4

21.73 23.76 22.88 24.32

91.80 119.1 89.10 112.2

0.07 0.2 0.07 0.3

Table 2

a Temperature of the experiment was performed at 250 °C.26 bTemperature of the experiment was performed at 180 °C.23 cReplicates measurements were made as outlined in the Experimental Section a standard deviation of 0 in the hundredths place.

(

ΔG

(

Ω1μ0

structure.20 We therefore limit calculations to these single structures, as other local energy minimum structures do not yield substantially different collision cross sections from one another. To compute collision cross sections, we utilized the collision cross section calculation package IMOS (v1.06b, http://www.imospedia.com/imos/).3 In IMOS, each atom was modeled as a sphere with its radius equivalent to its van der Waals radius. Colliding gas molecules were modeled as spheres with radii of 0.3 nm.21 Collision cross sections were calculated using both the diffuse and elastic scattering methods at the measurement temperature (175 °C), with the ion-induced dipole considered between the excess proton and the impinging nitrogen gas molecules (with a polarizability of 1.7 A3). As discussed in prior work,20,22 the elastic and diffuse scattering models give rise to distinct collision cross section predictions from one another. The elastic model assumes that the ion is “frozen” during collision with a gas molecule and its vibrational and rotational energy remains constant (as does the gas molecules rotational and vibrational energy). The diffuse model assumes complete equilibration of the ion internal energy and gas molecule translational energy during each collision. They therefore are the upper (diffuse) and lower (elastic) limits for the true collision cross section, respectively. In the absence of isopropanol vapor, elastic scattering calculations yielded collision cross sections of 83.22 A2 for [MPA+H]+ and 95.05 A2 for [DMMP+H]+, while the diffuse scattering calculations yielded collision cross sections of 114.3 A 2 for [MPA+H] + and 116.8 A 2 for [DMMP+H] + . Experimentally inferred collision cross sections were 91.80 A2 for [MPA+H]+ and 89.10 for [DMMP+H]+ (with the difference in reduced mass leading to near identical mobility coefficients for these ions). For [MPA+H]+, we fit calculations to experiments by modeling collisions as 27.6% diffuse, i.e., the collision cross section is modeled as a linear combination of the elastic collision cross section and diffuse collision cross section, with a weighting factor of 0.724 for elastic scattering and 0.276 for diffuse scattering. Such weighting yields a collision cross section of 91.79 A2 for [MPA+H]+, in near perfect agreement

)

1 + S exp − kT t = Ω μ 1/2 ΔG ti 1 + 0 11/2 S exp − kT

)

(6)

where ti is the initial arrival time measured in the absence of vapor, eq 6 provides a link between the experimental observed drift time, the collision cross section of a bare ion and an ionvapor molecule complex, and the Gibbs free energy change associated with vapor molecule binding. With from theory, plots of

t ti

Ω 0μ11/2 Ω1μ01/2

calculated

as a function of S hence yield, ΔG.

To estimate the collision cross section ratio, Ω1/Ω0, we predicted local energy minimum structures for [MPA+H]+ and [DMMP+H]+ with and without a single 2-propanol molecule bound. Structural predictions were made using the Gaussian 09 software package (Gaussian Inc.,Wallingford, CT, U.S.A.) using density functional theory with the B3LYP density functional and the 6-311g basis set. Energy minimization was carried out for the four test structures that were constructed manually and informed by prior literature configurations.19 All resulting structures, displayed in the Supporting Information (SI Table S2) along with their atomic coordinates, had positive frequencies, indicating they are true local minima. In addition to the [MPA+H]+, [DMMP+H]+, and 2-propanol adducts, we obtained two local energy minimum structures for the protonated DMMP dimer; these two structures were also examined by Namazian et al.19 Although these authors utilized the G3(MP2) functional, the relative ordering in energies of these two dimer cluster structures is in line with our calculations. Density functional theory calculations yielded only a single structure, which is not necessarily a global minimum. Generally, it has been found that collision cross sections calculated for model structures are relatively insensitive to subtle changes in structure predictions, i.e., typically the collision cross section is only affected by large scale, conformational changes in ion D


Article


Figure 3. (A) Faraday plate ion mobility spectra with (top trace) and without (bottom trace) 2-propanol as a modifier. The offset peaks were m/z identified using the instrument shown in Figure 2. The separation factors, alpha, for the peaks corresponding to DMMP and MPA illustrate that without the modifier the two species are indistinguishable. (B) Mobility separations of DMMP and MPA as a function of increasing drift gas modifier concentration. The stability of the proton-bound dimer of DMMP compared to the MPA and DMMP monomers illustrate the selective interactions between the ion populations and the modifier.

literature values for the given compound (DMMP at 1.99 ± 0.02 cm2 V−1s−1).7,23 However, noted in Table 2, an apparent discrepancy exists between the reduced mobility reported in a single source for [MPA+H]+ and the current experimental value. Although the current assessment is primarily focused on a single temperature (175 °C), repeated measurements of [MPA+H]+ consistently produced a reduced mobility value of 1.99 cm2V−1s−1. However, when examining the reduced mobility of [MPA+H]+ ion from 100 °C to 175 °C, the mobility was observed to trend from 1.85 ± 0.01 to 1.99 ± 0.01 cm2V−1s−1 with a linear increase in mobility of 0.0018 cm2CV−1s−1. Figure 3A, demonstrates the ion mobility spectra with and without the modifier for DMMP and MPA mixture The bottom trace proves that protonated forms of methyl phosphonic acid (MPA) and dimethyl methyl phosphonate (DMMP) demonstrate the nominally same reduced mobility values in nitrogen at 1.99 ± 0.01 cm2 V−1s−1 as these two analytes demonstrate an overlapping drift time of 7.04 ms ±0.01. However, a complete baseline resolution of the two species is achievable in the mobility domain when 2-propanol is introduced as a buffer gas modifier of the system, even at low flow rate of 20 μL/h. At 20 μL/h, the drift time of [MPA+H]+ and [DMMP+H]+ selectively shifted to 8.52 ms ±0.01 and 8.12 ms ±0.02 which corresponds to a mobility value of 1.65 cm2V−1s−1 and 1.73 cm2V−1s−1, respectively. This mode of separation would not have been achievable on the present instrument in pure nitrogen gas without the use of the modifier. Differential clustering of the two compounds with the neutral buffer gas modifier appears to be the origin of enhancement of these separation factors. In order to probe the degree of

with measurements. The 27.6% diffuse scattering model was also applied for collision cross section predictions of the [isopropanol+MPA+H]+ complex; this yields a collision cross section of 123.6 A2 for the complex and Ω1/Ω0 of 1.35. Meanwhile, for [DMMP+H]+, the elastic collision cross section was larger than the measurement. We hence elected to use solely elastic scattering calculations for both the [DMMP+H]+ ion and its isopropanol complex; and the complex ion had an elastic collision cross section of 121.5 with a Ω1/Ω0 ratio of 1.28. These values are summarized in Table 1 and are used in inferring the Gibbs free energy change of vapor molecule binding. Although different scattering models are used for different ions, which suggests that there is some ambiguity in the bare ion structures in the gas phase, we note that application of eq 6 in comparing theory to measurements only requires determination of Ω1/Ω0 (which can also be estimated from measurements themselves, as shown subsequently). While the computed collision cross sections themselves are highly dependent on the scattering model employed, the ratio Ω1/Ω0 is much less dependent on the scattering model, provided that Ω1 and Ω0 are calculated in an identical fashion.

■

RESULTS AND DISCUSSION Using a mixture of organophosphates, the data presented illustrate the ability of buffer gas modifiers to alter separation factors in the ion mobility domain. The positive ion mode mobility values for [DMMP+H]+, [(DMMP)2+H]+, and [MPA +H]+ were 1.99, 1.44, and 1.99 cm2 V−1s−1, respectively, as noted in Table 2. These data aligned well with the reported E


Article


The mass spectrometer provides a response with regard to the precise composition of an ion from a mobility spectrum and determines the identity of protonated species; even so, transport of ions from ambient pressure to high vacuum can alter the exact identity of an ion by the removal of clustered species while preserving the core of the ion.24 In this experiment the LTQ was used to mass analyze the ions that constitute peaks in the mobility domain. Figure 5 demonstrates the mass spectrum of MPA/DMMP mixture in the presence of 2-propanol as the buffer gas modifier. The intensities of the core ions in the DMMP and MPA mixture without the modifier are shown in the bottom spectrum of Figure 5, whereas, the intensities of the cluster ion species are recorded in the upper spectrum with 2-propanol as the buffer gas modifier. The mass spectrum in Figure 5 was recorded with a modifier flow rate of 40 μL/h. At this flow rate, the evidence of the DMMP/MPA cluster ion species with 2-propanol is observed from the mass spectrum. The peaks at m/z 97 125, and 249 corresponds to the core ions [MPA+H]+, [DMMP+H]+ and [(DMMP)2+H]+. Whereas the peaks at m/z 157 and 185 corresponds to the cluster ion species of [MPA+2-propanol+H]+ and [DMMP+2propanol+H]+ respectively. An important observation is the disparity between the relative intensities of the two spectra recorded in Figure 5. Only the long-lived or stable cluster ions survive the atmospheric pressure interface (API), hence the discrepancy between the relative intensities of the arrival time distribution and the range of m/z recorded. Even though variation can be introduced as ions traverse the API region, it is reasonable to assume that the number of collisions at atmospheric pressure establish equilibrium conditions between the target ions and 2-propanol as they transverse in the drift region. Close examination of Figures 3A and 5 further support this observation. In Figure 3A, the ion mobility spectra for DMMP and MPA mixture with 2-propanol showed a single peak for DMMP cluster ions and another peak for MPA cluster ions. However, for the mass spectra in Figure 5, we see m/z for ions with different masses. The single peak in the mobility domain is due to the clustering/declustering of the analytemodifier cluster ion in equilibrium in the buffer gas at a rate appreciably high enough to produce a distinct mobility peak with a weighted average of the mobilities of the individual ions. The core ions and the cluster ion species [MPA+H]+, [MPA+2propanol+H]+ peaks were formed at the same drift time for every flow rate analyzed. The [MPA+H]+ peak was formed as a result of decomposition of the [MPA+2-propanol+H]+ species at the ion mobility and mass spectrometer interface. This is also evidently observed for [DMMP+H]+ and [DMMP+2-propanol +H]+ ion species. Both the drift time ratio and the ratio of mobility measured with modifier (Kn1) to mobility without modifier (Kn0) are plotted against the modifier (2-propanol) saturation ratio (and liquid feed flow rate) for both methyl phosphonic acid (MPA) and dimethyl methyl phosphonate monomer (DMMP) ions in Figure 6. At the maximum modifier flow rate employed (160 uL/h), the mobilities recorded for DMMP and MPA with 2propanol cluster ion species were 1.63 cm2/(V s) and 1.58 cm2/(V s) respectively, corresponding to a 20.6% mobility decrease for MPA and an 18.2% decrease for DMMP. For all examined 2-propanol saturation ratios, the mobility of MPA ions was less than that of DMMP ions, and correspondingly, MPA ions had higher drift time ratios than DMMP ions. Also apparent in Figure 6 is that at the lowest saturation ratios, the drift time ratios/mobility ratios change sharply with change in

mobility shifts that can be achieved, the concentration of the neutral buffer gas modifier was varied along the temperature of the system. In relation to the statement above, as shown in Figure 3B, as the concentration of the buffer gas increases, even at a lower flow rate of 10 μL/h the overlapping peak of [DMMP+H]+ and [MPA+H]+ begin to resolve and their drift time shift. However, [(DMMP)2+H]+ does not appear to form a cluster with 2-propanol, as its mobility coefficient is insensitive to 2-propanol concentration. Predicted structures for this dimer ion suggest this is because the excess proton is apportioned between the two DMMP molecules, effectively causing steric constraint around the proton, minimizing the selective interaction with the neutral buffer gas modifier. It should be noted that there is a small relationship between the observed arrival time distribution of the DMMP dimer and the drift gas modifier concentration. However, in the structures displayed in the SI, 2-propanol is bound in both [MPA+2propanol+H]+ and [DMMP+2-propanol+H]+ to the excess proton (H−OH bond), which is not available for 2-propanol in the dimer ion. The frequency encoded mobility spectra acquired in Figure 4 using the LTQ system in Figure 2 was employed in tandem with the information from the Faraday plate to aid in the identification of the separated ion population in Figure 4.

Figure 4. Mass selected mobility spectra obtained via the FT-IMMS experiment. When no modifier is present a clear overlap of the massselected mobility populations of DMMP and MPA is observed. However, as the flow rate is increased the drift times of two analytes differentially shift. These data serve to demonstrate the approach by which the peaks observed from the Faraday plate experiments are assigned to a specific ion population. F


Article


Figure 5. Composite mass spectrum for the separations of DMMP and MPA. Comparing the relative intensities between the mass and mobility domains illustrates that the clusters, though stable in the mobility domain, do not fully survive the atmospheric pressure interface. Comparing the relative intensities of the MPA and DMMP clusters with 2-propanol (m/z 125 and 185, respectively) between the Faraday plate (Figure 3) data and those observed in the m/z domain highlight this discrepancy.

Figure 6. Comparison of the observed ion mobilities for the target analytes with and without the modifier. While the strict definition of mobility may not hold for a heterogeneous drift gas, the plotted ratio may also be related to the separation factor. The predicted mobility shifts for DMMP and MPA illustrate the reasonable agreement with the experimentally observed values. For clarity the flow rate of the drift gas modifier is converted to saturation ratio.

saturation ratio, while at larger saturation ratios, the drift time ratios/mobility ratios remain somewhat constant. This is anticipated based on the model (eq 6) developed in the preceding section; if the shift in drift time is brought about only by the transient clustering of a single isopropanol molecule, then the maximum drift time would be achieved when the

isopropanol molecules remains bound throughout the entirety of the mobility measurement, and the drift time ratio maximum shift would be (Ω1μ01/2/(Ω0μ11/2)). This ratio is estimated using collision cross section calculations to be 1.29 for MPA and 1.24 for DMMP. These estimations are in excellent agreement with measurements, as the largest drift time ratio G



■

observed for MPA is 1.24 and for DMMP it is 1.22. We list experimentally inferred collision cross sections for [MPA+2propanol+H]+ and [DMMP+2-propanol+H]+ in Table 1. These ratios are based on the assumption that at the highest saturation ratios examined, a single 2-propanol molecule remains bound to each ion during mobility measurement, but that a second 2-propanol molecule negligibly clusters to ions. In addition to comparison of the drift time ratio, comparison to eq 6 enables inference of ΔG, the Gibbs free energy change associated with binding of 2-propanol at the measurement temperature. These inferred free energy changes are subject to model assumptions (i.e., only one vapor molecule can bind), and we note that further validation studies need to be performed to ensure that IMS with vapor-modifiers can be used regularly for free energy measurements. Best fit curves, determined by minimizing the square error between measured drift time ratios and eq 6 predictions are plotted in Figure 6, and correspond to −0.546 eV (12.69 kcal/mol) and −0.519 eV (12.07 kcal/mol) for MPA and DMMP, respectively. The values we determine are similar in magnitude to hydration free energies for a variety of metal cations (which typically fall in the 10−20 kcal/mol range for the first water molecule) and are slightly higher than the magnitudes of the hydration free energies for organic cations (in the 5−10 kcal mol range near 300 K).25 Although the systems examined in these prior studies are chemically quite distinct from those in the present study, that the inferred free energy changes fall in-line with previously determined ion-neutral clustering energies does support application of the developed model in IMS analysis. Furthermore, prior measurements of hydration show that the magnitude of the free energy change for additional neutral molecule binding is smaller than the free energy change for binding of the first molecule; this supports neglecting the binding of multiple neutral molecules (in the low saturation ratio range probed here).

Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.7b03518. Table S1 giving the measurement results for MPA and DMMP mixture ions with 2-propanol as the drift gas modifier and Table S2 giving the resulting structures of the targeted analyte species along with their atomic coordinates (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (B.H.C.). ORCID

Christopher J. Hogan Jr.: 0000-0001-7655-4980 Brian H. Clowers: 0000-0002-5809-9379 Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS P.K.B. and B.H.C. would like to acknowledge the Army Research Office (Award# W911NF1510619). REFERENCES

(1) Albritton, D. L.; Miller, T. M.; Martin, D. W.; McDaniel, E. W. Phys. Rev. 1968, 171, 94. (2) Revercomb, H. E.; Mason, E. A. Anal. Chem. 1975, 47, 970−983. (3) Larriba, C.; Hogan, C. J. J. Phys. Chem. A 2013, 117, 3887−3901. (4) Davidson, K. L.; Bush, M. F. Anal. Chem. 2017, 89, 2017−2023. (5) Ruotolo, B. T.; McLean, J. A.; Gillig, K. J.; Russell, D. H. J. Mass Spectrom. 2004, 39, 361−367. (6) Bollan, H. R.; Stone, J. A.; Brokenshire, J. L.; Rodriguez, J. E.; Eiceman, G. A. J. Am. Soc. Mass Spectrom. 2007, 18, 940−951. (7) Roscioli, K. M.; Lamabadusuriya, M. R.; Harden, C. S.; Midey, A. J.; Wu, C.; Siems, W. F.; Hill, H. H., Jr. Int. J. Ion Mobility Spectrom. 2014, 17, 43−53. (8) Fernández-Maestre, R.; Wu, C.; Hill, H. H., Jr. J. Chil. Chem. Soc. 2014, 59, 2398−2403. (9) Fernández-Maestre, R.; Wu, C.; Hill, H. H., Jr. Int. J. Mass Spectrom. 2010, 298, 2−9. (10) Gan, T. H.; Corino, G. Anal. Chem. 2000, 72, 807−815. (11) Rawat, V. K.; Vidal-de-Miguel, G.; Hogan, C. J. Analyst 2015, 140, 6945−6954. (12) Oberreit, D.; Rawat, V. K.; Larriba-Andaluz, C.; Ouyang, H.; McMurry, P. H.; Hogan, C. J. J. Chem. Phys. 2015, 143, 104204. (13) Morrison, K. A.; Siems, W. F.; Clowers, B. H. Anal. Chem. 2016, 88, 3121−3129. (14) Li, C.; Hogan, C. J., Jr. Aerosol Sci. Technol. 2017, 51, 653−664. (15) Rus, J.; Moro, D.; Sillero, J. A.; Royuela, J.; Casado, A.; EstevezMolinero, F. E.; de la Mora, J. F. Int. J. Mass Spectrom. 2010, 298, 30− 40. (16) Oberreit, D. R.; McMurry, P. H.; Hogan, C. J., Jr. Aerosol Sci. Technol. 2014, 48, 108−118. (17) Vidal-de-Miguel, G.; Macía, M.; Cuevas, J. Anal. Chem. 2012, 84, 7831−7837. (18) Mason, E. A.; McDaniel, E. W. Transport Properties of Ions in Gases; Wiley: New York, 1988; Vol. 6. (19) Namazian, M.; Coote, M. L. J. Chem. Thermodyn. 2008, 40, 1116−1119. (20) Ouyang, H.; Larriba-Andaluz, C.; Oberreit, D. R.; Hogan, C. J. J. Am. Soc. Mass Spectrom. 2013, 24, 1833−1847. (21) Larriba, C.; Hogan, C. J.; Attoui, M.; Borrajo, R.; Garcia, J. F.; de la Mora, J. F. Aerosol Sci. Technol. 2011, 45, 453−467.

■

CONCLUSION Using 2-propanol as a drift gas modifier we demonstrate the capacity to induce selective shifts in gas-phase mobility of up to 20% compared to nitrogen for two chemical warfare agent simulants, DMMP and MPA. Without the modifier these two species exhibit the same nominal reduced mobility value but are fully resolved when using the drift gas modifier at 20 μL/h and counter-current nitrogen flow rate of 1.5 L/min. Compared to many of the previous efforts demonstrating mobility shifts using drift gas modifiers, the present effort demonstrates the practical benefits on a mixture of ion populations. Moreover, the inclusion of quantitative mobility shift model establishes the foundations to extend this approach to a wider range of gasphase systems. Specifically, accounting for the shifts in drift time that occur as a function of drift gas modifier concentration allows a direct relationship to be drawn between the equilibrium state of the ion and the ion-modifier cluster. When including computationally derived values of the ionneutral collision cross section for the modifier cluster, the predicted mobility shifts are well aligned with experimental result. The quantitative aspects of this model enable a range of future experiments that include predictive alteration of the mobility separation factors and may even prove useful when probing the thermodynamics of ion-neutral clustering within an ion mobility drift cell. H


Article

Analytical Chemistry (22) Larriba-Andaluz, C.; Fernández-García, J.; Ewing, M. A.; Hogan, C. J.; Clemmer, D. E. Phys. Chem. Chem. Phys. 2015, 17, 15019− 15029. (23) Ewing, R. G.; Eiceman, G. A.; Harden, C. S.; Stone, J. A. Int. J. Mass Spectrom. 2006, 255, 76−85. (24) Eiceman, G.; Karpas, Z. Ion Mobility Spectrometry; Taylor & Francis Group: Boca Raton, 2005. (25) Keesee, R. G.; Castleman, A. W., Jr. J. Phys. Chem. Ref. Data 1986, 15, 1011−1071. (26) Asbury, G. R.; Wu, C.; Siems, W. F.; Hill, H. H., Jr. Anal. Chim. Acta 2000, 404, 273−283.

I


Tuning Mobility Separation Factors of Chemical ... - ACS Publications

Recommend Documents