Anal. Chem. 1909, 61 343-349 I
343
Characterization of Benzodiazepine Drugs by Ion Mobility Spectrometry A. H.Lawrence Trace Vapour Detection Section, National Aeronautical Establishment, National Research Council of Canada, Ottawa, Ontario K 1 A OR6, Canada Chemlcal ionization Ion moblilty spectrometry (CI-IMS) was used to characterize a number of benzodiazeplnes. I n almost every example studled, the positive ion mobility spectrum consisted of a slngle Ion peak corresponding to [MI' or [MH]'. With some compounds, e.g., oxazepam, iorazepam, and chlordiazepoxide, fragment lons were noted that serve as good markers for the identlficatlon of these chemicals. Reduced mobility constants ( K O )for the most slgnlflcant peaks were calculated, and all Ions produced were mass-anaiyzed by InJectloninto a quadrupole mass spectrometer. The results of this study point to the potentlai of IMS as a qualltatlve tool for the ,rapld detection (analysls time < I O 8 ) and reliable ldentiflcatlon of benzodiazepines. Preilmlnary results on the application of dlgltal slgnai processing and a secondderlvative aigorlthm to partlaily Overlapping IMS peaks are presented, and potentlal Improvements are discussed.
The benzodiazepines are an important class of sedativehypnotic drugs that are used widely to treat anxiety and insomnia. Moreover, these drugs are the predominant compounds encountered in cases of patients admitted to hospital with drug overdose (1). Furthermore, some benzodiazepines, particularly diazepam, are manufactured in clandestine laboratories with the intent of supplying them to the illicit drug market. The detection and rapid identification of benzodiazepines in various biological (blood, urine, etc.) and nonbiological (tablets, containers, syringes, etc.) specimens is therefore of significant importance in a number of clinical, forensic, and law enforcement applications. Analytical methods currently available for the identification and determination of benzodiazepines in various matrices include radioimmunoassay (RIA) (2), gas chromatography (GC) (3), and high-performance liquid chromatography (HPLC) (4).However, RIA gives no indication of the identity of the benzodiazepine detected, and GC and HPLC analyses require several minutes. In addition, GC analyses are further complicated by the thermal instability of some of the compounds, particularly chlordiazepoxide and oxazepam, and HPLC analyses often require a carefully selected mobile phase. Ion mobility spectrometry (IMS), also known as plasma chromatography, is a relatively new analytical technique that resolves ionic species on the basis of the differences in their mobilities through a gas under an applied electrostatic field ( 5 ) . The operation of an IMS is analogous to the operation of a time-of-flight mass spectrometer (TFMS), the difference being that the TFMS operates under vacuum whereas the IMS operates at or near atmospheric pressure. Consequently, IMS avoids the excessive hardware associated with vacuum technology and has been miniaturized into a compact detector alarm system for field use (6, 7). Another advantage of IMS operation at atmospheric pressure is good ionization efficiency, thus leading to high sensitivity, which can reach the subpart-per-billion level (8). In chemical ionization ion mobility spectrometry (CI-IMS), the sample, in vapor form, is introduced into a reaction chamber by means of a carrier gas (N2,air). Ionization of
reactive trace impurities in the carrier gas, such as water and ammonia, is brought about by energetic electrons released from a 63Niradioactive source with the formation of a number of positive and negative reactant ions, e.g., (H20),H+ and (H20),0,. These ions transfer their charge through a series of complex ion/molecule reactions (6)to the trace species of interest, and positively and/or negatively charged ions characteristic of the sample are produced. At the same time, a shutter grid periodically gates (typically 24-ms cycles) a small sample of the ions into a drift cell. The ions are then moved down the cell under the influence of a strong electric field and through the countercurrent of a drift gas, and consequently, they separate into individual pulses due to their different mobilities (5). The arrival of the individual ion pulses at the collector electrode produces a current pulse with an amplitude proportional to the number of ions. An electrometer measures the ion peak pattern continuously, and a complete ion drift pattern (positive or negative mobility spectrum, depending on the polarity of the applied electric field) is generated in a few milliseconds. Drift time, t (ms), and ion mobility reduced to standard temperature and pressure, KO(cm2V-'s-l ), are used as a qualitative measure of specific ions (5). Since IMS signals consist of Gaussian peaks distributed in highfrequency noise, it is desirable to average the sampled data over a number of gated cycles to obtain improved resolution and signal-to-noise ratio. For example, at a frequency of 41 Hz, signal averaging of 128 cycles lasts approximately 3 s before a spectrum is displayed. Ion mobility measurements, however, cannot provide high resolution since the peak widths are ultimately diffusionlimited. Spangler and Collins have attempted to discuss the separation efficiency of ion mobility instruments in chromatographic terms (9). They developed a relation for the height equivalent to a theoretical plate (HETP) within the drift region and reported that HETP values are less than those encountered in GC and mass resolution is less than that obtained in mass spectrometry. Furthermore, the mobility of a particular ion is not strictly dependent upon its mass but instead on its collision cross section or effective size, and attempts to relate mobilities to mass are not always accurate. One direct way to assign masses to ions giving particular mobility peaks is to interface the IMS to a quadrupole mass spectrometer (MS). Because of its sensitivity, fast response time, and moderate selectivity, IMS is specifically suitable for initial screening, fingerprinting, and the rapid identification of samples in situations where time constraints dictate a short analysis cycle. We have recently reported on the application of skin surface sampling and ion mobility spectrometry to the detection of drug residues on the hands of emergency patients suspected of drug overdose (1). At the same time, work is in progress to apply IMS to the rapid identification of individual benzodiazepines, taken in overdose quantities, in serum and urine samples. This paper describes the characterization of a number of benzodiazepines by IMS and the mass identification of the ions associated with the mobility peaks of the drugs by IMS/MS. Some initial results obtained from the application of the second-derivative technique (IO)to overlapping
0003-2700/89/0361-0343$01.50/00 1989 American Chemical Society
344
ANALYTICAL CHEMISTRY, VOL. 61, NO. 4, FEBRUARY 15, 1989
IMS peaks are also presented. EXPERIMENTAL SECTION Reagents and Chemicals. Authentic drug samples were obtained from the reference collection of the Department of Health and Welfare Canada and were used without further purification. Standard drug solutions were prepared as g/wL in methanol. Linde ultrahigh-purity air was dried by Linde molecular sieve 13X and was used for both the carrier and drift gases. Some compounds were run by using, as the carrier gas, air that had been saturated in a water bubbler at ambient temperature. Methylamine hydrochloride, 99+ % ,was obtained from Aldrich (Milwaukee, WI). Methanol was of distilled-in-glass grade (Caledon, Georgetown, ON, Canada). Sample Tubes and Injection Techniques. The sample tubes used in this study consisted of a small narrow probe (6-cm length X 0.3-cm 0.d. X 0.22-cm id., nickel tube) and contained a finemesh platinum filter. The tubes were conditioned for several minutes at 250 OC in an air stream at a flow rate of 200 mL/min. Before use an IMS trace was recorded to check the cleanliness of the filter. In all IMS experiments, 1 fiL of a standard drug solution was deposited with a syringe directly on the platinum filter of the sample tube. After evaporation of the solvent by the passage of an air flow for a few seconds, the sample tube was inserted in the heated inlet of the IMS. The benzodiazepine drug evaporated immediately and was flushed with the carrier gas stream to the ion reaction chamber of the instrument (11). In the IMS/MS experiments, all samples were introduced into the inlet of the IMS by using a clean stainless steel wire; the wire was dipped in the drug solution and the solvent air evaporated prior to sample injection. Instrumentation. The IMS data presented in this paper were obtained with a Phemto-Chem 100 ion mobility spectrometer (PCP, Inc., West Palm Beach, FL). The instrument has a membraneless inlet and has been recently described (12). The signal from the IMS was averaged with a Nicolet 1170 signal averager and/or with a compact digital signal averager developed in this laboratory (13). The analog-to-digital converter (ADC) resolution used with the Nicolet signal averager was always 12 bits, and the full-scale volt setting was kept at f4 V. The basic measurement in this study was the drift time, which can be measured to k 0 . 0 2 ms as determined experimentally, and from which the reduced mobility, KO,can be calculated according to eq 1, where d = drift length (8cm), t = drift time (s),E = electric K = - -d- 273 P a tE T 760 field gradient (V/cm), T = temperature (K), and P = pressure (Torr). Mass identification of ions giving particular mobility peaks was achieved, in a separate set of experiments, with a Phemto-Chem MMS-160 ion mobility spectrometer/mass spectrometer (IMS/MS) (14). The experimental parameters used to operate the IMS and the IMS/MS are presented in Table I. Resolution Studies. The resolution studies consisted of two sets of experiments. In the first set, the effect of the strength of the electric field on the separation of the molecular ion M+ of diazepam and the fragment ion (M - H20)+of lorazepam was investigated by varying the total voltage applied on the IMS tube. The drift times of the above ions, as well as the molecular ion of bromazepam, were measured relative to that of the [MH]' ion of the reference compound, flurazepam. Table I11 lists the various voltages tested along with the electric fields produced and the resolution data. All measurements were made at an entrance pulse duration of 0.2 ms, a cycle time of 24 ms, and a dwell time of 20 ps/data point. The total number of data points was 1024, and the number of scans that were averaged usually was 512. In the second set, digital signal processing, namely the second-derivative technique, was applied to partially overlapping IMS peaks obtained by injecting 1:l mixtures of diazepam and lorazepam, and diazepam and bromazepam, respectively. In these experiments, the dwell time was 2 ks/data point, and a 17.2-ms delay was added after the shutter grid synchronization pulse. Multiple scans of 1024 data points (2-ms drift time spectra) were stored digitally in the memory of the Nicolet 1170 signal averager. The stored spectra were then transferred to a data file on a Hewlett-Packard
Table I. Instrument Parameters parameter
value
Ion Mobility Spectrometer (Phemto-Chem 100) cell length 14 cm drift length 8 cm carrier gas (purified air) 200 mL/min drift gas (purified air) 600 mL/min inlet and drift temperature 220 oc drift voltage 12700 V dwell time 20 gs/channel 0.2 ms gate width delay time" 6 ms number of scans 256 scan time 6s Ion Mobility Spectrometer-Mass Spectrometer (Phemto-Chem MMS-160)* cell length 15 cm 5 cm drift length (between shutter grid and IMS collector) carrier gas (purified air) 100 mL/min drift gas (purified air) 500 mL/min *2700 V drift voltage inlet and drift temperature 220 "C dwell time 20 p/channel 0.2 ms gate width 4X Torr mass spectrometer pressure scanning speed 1000 amu/s "Time between gate opening and start of data collection.
* IMSiMS experiments were conducted at PCP Laboratories.
10oO minicomputer for further signal processing. A Gaussian peak and a Fortran 77 program performed the shape was assumed (9), following operations: (1) setting of a specific time window about the signal of interest (17.2-19.2 ms), (2) performing a fast Fourier transform (FFT) in order to obtain the frequency domain representation of the signal, (3) low-pass filtering the signal to attenuate the high-frequency noise, (4) calculating the second derivative in the frequency domain, and (5) performing an inverse FFT and plotting the results.
RESULTS AND DISCUSSION Positive Ion Results. It is well established that the ionization characteristics and the nature of the ion-molecule reactions involved in IMS depend on the type of reactant ions generated in the reaction region. The dominant positive reactant ions formed when air is used as the carrier gas are (H20),H+. However, (H20),NH4+ and (H,O),NO+ are also formed in smaller quantities, with the value of n being zero or a small integer depending on the concentration of the water vapor. The major primary ions produced in air are N2+,and the mechanism by which these ions are converted to (H20),H+ has been elucidated in IMS (5) and chemical ionization mass spectrometric (CIMS) studies (15). The reactant ions undergo a complex series of ion-molecule reactions with sample vapors to produce product ions through proton transfer and other mechanisms (eq 2), where M is the molecular species of the sample studied and B is a minor contaminant with high proton affinity (PA). Nz + eN2+ + 2e(24
-
N2+ + 2N2 N4+ + HZO
+
N4+ + N2
(2b)
H2O+ + 2N2
(2c)
+ H20 H30++ OH H30++ M H20 + MH+ H20 + BH+ H30+ + B N4+ + M 2N2 + M+ BH+ + M - B + H2 + [ M - H ] + H,O+
(2d) (2e) (2f) (2g) (2h)
ANALYTICAL CHEMISTRY, VOL. 61, NO. 4, FEBRUARY 15, 1989
Table 11. Reduced Mobility Values in Air
KO,cm2 compd
MW
nitrazepam
281
diazepam oxazepam
284.8 286.7
chlordiazepoxide
299.7
V-'s-l 1.22 1.23 1.21 1.23 1.28 1.18
(2.53) 1.19 1.19
alprazolam 308.7 bromazepam 316 lorazepam
321
triazolam flurazepam
342.8 387.8
1.15 1.24 1.19 1.22 1.13 1.03
ion mass, amu
ion formula
282 [MHl+ 280 IM1284, 286 [MI+ 285, 287 [ M - H]+ 269, 271 [ M - HzO]+ 300, 302 [MH]' 284, 286 [MH - 161' 269, 271 [ M H - 311' 32O [MeNH3]+ 298, 300 [ M - HI298, 300 [M-HI297, 299 [M- 2H]309, 311 [MH]+ 316, 318 [MI+ 320, 322 [MI+ 302, 304, 306 [ M - H,O]+ 342, 344 [ M - HI+,[ M H ] + 388, 390 [MH]+
345
lorazepam (Ib) contain an OH moiety, a good leaving group, and consequently exhibit similar dehydration behavior. Both chemicals exhibit intense [M - 18]+peaks and adjacent, weak M+ ion peaks that are not completely resolved from their fragment ion peaks (Table I1 and Figure 3b). Furthermore, the relative intensity of the M+ peak decreases as a function of time. The positive ion profiles for oxazepam and lorazepam are indicative of the type of useful fragmentation possible in IMS; although the main ion peak from lorazepam and nitrazepam is a t KO= 1.22 cm2 V-* s-l and corresponds to [M - HzO]+ ( m / z 302, 304, 306) and [MH]+ ( m / z 282), respectively, the two compounds can be differentiated upon inspection by noting the smaller molecular ion peak of lorazepam adjacent to the [M - 181' peak (Figure 3b). H
/
"CH3
a N o t confirmed by IMS/MS. Parentheses indicate the appearance of a secondary ion mobility peak.
The main drawback in IMS is that the presence of background contaminants in the sample may result in a number of competing reactions (e.g., eq 20, thus leading to the formation of several product ions and confusion in the interpretation of the data. Furthermore, a small amount of an interfering compound with high gas-phase basicity can be ionized a t the expense of the analyte (16). Benzodiazepines, however, are nitrogenous drugs and are nucleophilic in nature. Their relatively high gaseous basicity (17)reduces the number of interferences capable of capturing an appreciable fraction of the positive charge available in the reaction chamber of the IMS. Consequently, benzodiazepines give strong response signals in positive IMS (Table 11). Furthermore, these chemicals can be injected directly in the IMS, either in the base or salt form, without chemical derivatization, thus eliminating tedious and time-consuming sample preparation steps. The individual positive ion mobility spectra for benzodiazepines were simple with no appreciable fragments, and at a tube temperature of 220 "C, ion-molecule clusters with H 2 0 and Nz were not formed. The observation that can be made from the data presented in Table I1 is that, in almost every example studied, a single major ion peak with a peak width a t half height of -0.4 ms and corresponding to the molecular ion M+ or MH+ or [M - H]+ is produced. Although care was taken not to saturate the spectrometer with the test compound (saturation is defined as the total depletion of reactant ions), in some cases the molecular ion M+ was formed instead of MH+, indicating that the concentration of the species was greater than that of background water. Under these circumstances, M+ is probably produced by a chargetransfer reaction between M and Nz+or N4+(eq 2g) (18). Also, [M - H]+ ions are sometimes formed by hydride ion abstraction (eq 2h) depending on the hydride ion affinities of the BH+ ions. Similar behavior is frequently observed in CIMS (19). In IMS, as well as CIMS, it is desirable to maximize the probability of MH+ ion formation and minimize fragmentation since this ion provides molecular weight information. HowC
