MS of Explosives with

Jun 14, 2018 - *Phone: +49-172-2524747; Fax: +49-228-732596; E-mail: ... in Proteins by 193 nm Ultraviolet Photodissociation (UVPD) Mass Spectrometry...
24 downloads 0 Views 1MB Size
Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Hyperfast Flow-Field Thermal Gradient GC/MS of Explosives with Reduced Elution Temperatures Jan Leppert,† Martin Härtel,‡ Thomas M. Klapötke,‡ and Peter Boeker*,†,§ †

Institute of Agricultural Engineering, University of Bonn, Nussallee 5, D-53115 Bonn, Germany Department of Chemistry, Energetic Materials Research, Ludwig-Maximilians University of Munich, Butenandtstr. 5−13 (Haus D), D-81377 Munich, Germany § HyperChrom SA, 121 Avenue de la Faiencerie, Luxembourg, Luxembourg 1511 Downloaded via KAOHSIUNG MEDICAL UNIV on June 28, 2018 at 18:20:30 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



S Supporting Information *

ABSTRACT: Hyperfast GC/MS below 60 s of measurement time has been used for the measurement of explosives. The new flow-field thermal gradient GC (FF-TG-GC) utilizes a modified transport process of the explosives at lowered temperatures. In combination with the focusing effect of the gradient, high-resolution chromatograms are obtained even in very short time intervals. The reduction of the elution temperature by applying a thermal gradient along the chromatographic column is demonstrated by the simulation of the migration of analytes through the column. The simulation shows an interesting effect of the difference between maximum temperature and elution temperature of analytes during their separation with the spatial gradient. The results show the benefit of the gradient elution both from a modeling perspective and by measurements of explosives with low limits of detection (LOD) in the range from 0.1 to 20 μg/mL (0.5 to 150 pg of analyte mass on column). Results were compared to state-of-the-art vacuum outlet GC/MS as a reference method. A correlation between the reduction of elution temperatures and lower LODs are found for thermal labile nitrate ester explosives (EGDN, NG, ETN, and PETN), while no significant influence of the reduced elution temperature on LODs of more stable explosives, like DNT and TNT, was found. time.9 Shorter columns,3,4,6 higher column flow, and hydrogen as carrier gas are reducing the void time, while a slower temperature program reduces the heating rate. Standard GC methods for the detection of traces of explosives already use a fast chromatography with short columns of 6 m10,11 and use a high flow of 5 to 16 mL/min together with an ECD as detector. Mass spectrometers, besides special setups like supersonic MS,5,6,12 are not designed for such high flows because of the limitation of the vacuum pumps to maintain a vacuum inside the MS. Slower temperature programs with lower heating rates lead to lower elution temperatures but at the expense of time. Meanwhile, hydrogen as carrier gas can lower the elution

G

as chromatography (GC) is a widely used analytical technique for the analysis of volatile components, but some restrictions limit its application.1,2 The analytes must be able to be vaporized without decomposition and should be stable throughout the separation on the GC column. Thermally unstable components, like explosives, pesticides (e.g., the organic insecticide pyrethrine3,4), steroids5,6 and drugs,5,6 can be problematic to analyze by GC. To transport these components in the GC system, the temperatures should be as low as possible, but also, the residence time of the analytes inside the GC should be as short as possible. The elution temperature, the temperature of the GC oven at the retention time of the analyte, is the critical temperature for thermally unstable components in the chromatographic separation, besides the temperatures of the injector and detector. Different ways to lower the elution temperature are available.3,4,6−8 Most of them reduce the heating rate per void © XXXX American Chemical Society

Received: February 26, 2018 Accepted: June 14, 2018 Published: June 14, 2018 A

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

and θchar is the characteristic thermal constant. The characteristic temperature Tchar is the temperature at which the retention factor k = 1 or the mobility μ = 0.5, which is equal to an even distribution of the analyte between stationary and mobile phase. A raise in the temperature T by θchar is equal to a decrease of the retention factor k by the factor e ≈ 2.72. Figures 1 to 3 show the calculation of the migration of decane (ΔCp = 76.602 J mol−1 K−1, Tchar = 106.437 °C, and

temperature, but it could be reactive with thermally labile analytes. Therefore, these two approaches are not favorable for explosives. The vacuum outlet GC (VO-GC)13−17 is another method to reduce the elution temperature by reducing the void time. For this method, a MS detector, which operates at low pressures, is used in combination with a 0.53 mm wide bore separation column and a restrictor, for example, a short precolumn with a small internal diameter connected to the wide bore column inside the injector.16,17 With the vacuum on one side and the restrictor on the other side of the column, the pressure of the carrier gas inside the column is low. As a result, the diffusion inside the carrier gas increases, and the optimum carrier velocity increases as well. Also, the volatility of the analytes is increased at the lower pressure. The retention times for the analytes and therefore the elution temperatures are reduced. Additionally, a small column film thickness leads to chromatography with lower elution temperatures because of the reduced retention of the analytes, but smaller sample capacity can be a problem. A new way to reduce the elution temperature is the use of a negative temperature gradient over the length of the column (spatial gradient) in addition to the temperature program (temporal gradient).18−20 With the combination of a spatial and temporal gradient, the transport of an analyte over the column takes place at a nearly constant temperature. With only the temporal gradient, the temperature for the analyte rises continuously during the temperature program. In the following discussion, the term gradient is solely used for the term spatial gradient. A detailed description of how the gradient is realized can be found in an earlier publication.20

Figure 1. Temperature profiles for the simulation of migration of analytes in chromatography with gradient (blue and red line) and without gradient. The red line shows the temperature on the top of the column (injector side), the blue line shows the lower temperature at the bottom of the column (detector side), and the green line is the temperature difference between top and bottom. The black line is the corresponding temperature of the program without gradient.

θchar = 28.588 °C)21 and pentadecane (ΔCp = 109.405 J mol−1 K−1, Tchar = 184.746 °C, and θchar = 32.117 °C)21 through a SLB5 ms-column with and without gradient. Figure 1 shows the temperature programs for the chromatography with gradient and without gradient. The temperature between the injector and detector side is varying in a linear manner. At the beginning, no gradient is present, until the heating is started. The gradient increases for 15 s until it reaches its maximum of 50 °C. For 17 s, the gradient is constant. After this time, the gradient decreases for 8 s to avoid overheating at the top of the column, until the temperature of the whole column is constant. The heating rate of the program without gradient was adjusted to achieve nearly the same retention time for the analytes as in the program with gradient. Figure 2 shows the temperature of the analytes along the migration distance through the column. In this representation of the migration of the analytes, the reduction of the elution temperature due to the gradient is clearly visible. At the beginning of the temperature program, the analytes migrate faster in the program with gradient, because the heating rate on the injector side of the column is higher than on the program without gradient. After some time, the temperature of the analytes rises slowly compared to the conventional program without gradient and reaches a nearly constant temperature level. In contrast, the temperature of the analyte in the program without gradient is increasing through the whole migration process. Figure 3 shows the temperature of the analytes over time. In this representation, the blue and red solid lines (with gradient) can be interpreted as a specific temperature program for these analytes. After the 10 s of holding time, the temperature rises fast until it reaches 60 °C for decane and 130 °C for pentadecane. After this point, the respective temperatures rise



SIMULATION MODEL OF THERMAL GRADIENT GC To show the effect of the reduction of the elution temperature by the gradient used in the FF-TG-GC, the migration of analytes through a GC column is calculated according to principal theories.7 The column is divided into a large number n of equidistant subcolumns of the length Δxi. In these subcolumns, the temperature for the time t is constant, but in case of a spatial gradient, the temperatures between the subcolumns are different. The time Δti for the analyte to travel through one piece of these subcolumns i is calculated by Δti =

3 3 128Δxi2ηi(pin, − pout, ) i i 2 2 3d 2μi (pin, − pout, )2 i i

Hereby d is the column diameter, and pin,i and pout,i are the inlet and outlet pressure of the subcolumn i. Mobility μi, which is connected to the retention factor ki by μi = ki/(1 + ki), and carrier gas viscosity ηi are functions of the temperature Ti of the subcolumn i. Addition of the Δti leads to the retention time tR. For the simulation of the migration of decane and pentadecane through a chromatographic column with and without gradient, the distribution-centric three-parameter thermodynamic model21 with kref = 1 is used. ΔCp ij ΔCp T yz T − T T ln k = jjj ln + char zzz char + j R z T R T θ char { char k

Hereby, ΔCp is the change in the isobaric molar heat capacity of the analyte in the stationary phase, R is the universal gas constant, Tchar is the characteristic temperature, B

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

during the chromatographic separation. Possible higher temperatures in the injector, transfer lines, or the detector are not included. In contrast, the dashed lines follow exactly the conventional temperature program without gradient. In the case of the program without gradient, every analyte has the same heating rate, while in the case of the program with gradient, every analyte has specific heating rate profiles. In a simplified way, these effective temperature programs start with a high heating rate, which slows down after the analyte migrates different lengths of the column to an almost isothermal elution. The calculations for the migration of analytes through a GC column were made with the numerical computing software Scilab.



Figure 2. Temperature of analytes over the migration distance x through the column. Decane C10 is in blue, and pentadecane C15 is in red. Solid lines represent programs with gradient while dashed lines represent programs without gradient. Dots and crosses mark the positions in 5s intervals. The gray lines represent isochrones (temperatures at constant times).

MATERIALS AND METHODS

The explosive substances listed in Table 1 were synthesized by the workgroup for energetic materials of the department of chemistry of the Ludwig-Maximilians-University Munich. Purity was tested by 1H NMR and GC/MS. For dilution, acetone (99%) was purchased from Sigma-Aldrich. From the stock solution (concentrations in Table 1), serial dilutions were prepared, with resulting concentrations of Cn = Cstock · 2−n. Dilutions n = 2 to n = 12 were used for the calibration measurements to determine the limit of detection. A newly developed flow-field gas chromatograph (FF-TGGC)20 was coupled with a BenchTOF-Select Time-of-Flight Mass Spectrometer from Markes International. For easier change of the chromatographic column, some changes to the FF-TG-GC system were made. The 2.08 m long Rxi-5sil MS separation column (0.1 mm internal diameter, 0.1 μm film thickness) is threaded only inside the resistively heated stainless-steel capillary. The injector and detector are connected with the separation column by methyl deactivated fused silica capillaries (internal diameter = 0.1 mm, length = 20 and 40 cm, respectively) from Trajan. The transfer lines and the chromatographic column are connected by SilTite Mini Union connectors from Trajan. The FF-TG-GC was equipped with a split/splitless injector from a Thermo Fisher TRACE Ultra GC. A Jennings cup split liner (L = 105 mm, ID = 4 mm) from Restek was used, after tests with a gooseneck liner filled with glass wool showed a significant decreasing trend of signal response for all analytes. The split flow was set to 200 mL/min. Lower split flows

Figure 3. Temperature of analytes over time. Decane C10 is in blue, and pentadecane C15 is in red. Solid lines represent programs with gradient, while dashed lines represent programs without gradient. Dots and crosses mark the x position along the column.

slowly, reach a maximum, and even decline some °C. The maximum temperature Tmax for pentadecane is 139 °C, while the elution temperature Telu is 131 °C, or 6% below Tmax. For the earlier eluting decane, the difference is only 1.3% (Tmax = 68.5 °C, Telu = 67.6 °C). In the further discussion, maximum temperature is always the highest temperature of the analyte Table 1. Synthesized Explosives name

abbreviation

CAS no.

concentration [μg/mL]

2-nitrotoluene 3-nitrotoluene 4-nitrotoluene 2,4-dinitrotoluene 2,6-dinitrotoluene trinitrotoluene 1,3,5-trinitro-1,3,5-triazinane triacetone triperoxide 2,3-dimethyl-2,3-dinitrobutane 2,4,6-trinitrophenylmethylnitramine ethylene glycol dinitrate nitroglycerin erythritol tetranitrate pentaerythritol tetranitrate

2-MNT 3-MNT 4-MNT 2,4-DNT 2,6-DNT TNT RDX TATP DMDNB Tetryl EGDN NG ETN PETN

88-72-2 99-08-1 99-99-0 121-14-2 606-20-2 118-96-7 121-82-4 17088-37-8 3964-18-9 479-45-8 628-96-6 55-63-0 7297-25-8 78-11-5

109 124 110 131 137 148 204 97 103 327 163 195 404 341

C

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

between the top (injector side of the column) and bottom (detector side of the column) temperature. In Figure 4, in addition to chromatograms, the temperature profile for a measurement with and without a gradient is shown. Injection of the liquid solutions of the explosives were made with a CombiPal autosampler from CTC Analytics. The software Xcalibur from Thermo Fisher was used for automatic integration of the peaks of the substances. The dynamic background compensation of the ProtoTOF software from Markes was used for background subtraction of the chromatograms. The limit of detection and the limit of quantification were estimated according to DIN 3264522 with an external standard calibration curve (α = 0.01, k = 3). Five replicates of 11 different concentration levels of the analyte in Table 2 were injected. For every analyte, one m/z fragment for quantification and one for identification was chosen. The following criteria were used to include the measured signals to the calibration curve. The ratio of areas of identifier and quantifier must be equal to a target ratio with a tolerance of ±30%. The relative standard deviation of the area for the five replicates must be below 20%, and at least for four out of five replicates, a peak must be found. Beginning with the lowest concentration, for which a signal with these criteria was found, six calibration points were used to estimate the limit of detection.

showed a high background of the solvent acetone throughout the whole chromatogram, which lead to a reduced sensitivity of the ToF-MS because of ion suppression. The reason for the long tailing of the solvent are the used connectors. In these connectors, the two capillaries are placed head to head with low dead volume, but parts of the solvent diffuse into the connector and a reservoir of the solvent is formed, from which solvent diffuses back slowly into the column. The temperatures of the injector, transfer lines, and detector were carefully optimized to get the highest signals for the thermolabile analytes. The optimized temperatures are found to be 180 °C for the injector, 185 °C for the transfer ovens, 205 °C for the heating clamps, and 225 °C for the ToF-MS. The residence time in the transfer lines in total is less than 0.5 s. The chromatograph was operated in the constant pressure mode, at 290 kPa, which corresponds to a column flow of 1.6 mL/min at the starting temperature of 35 °C and 0.7 mL/min at the highest temperature of 320 °C of the program. Six different temperature programs were used, three with a thermal gradient (active fan) and three with fast conventional temperature-programmed runs. The heating rates were varied, whereby the heating rates of the measurements without gradient were adjusted in such a way that retention times similar to those obtained with gradient were achieved (retention time locked). The starting temperature was 35 °C (hold time = 10 s), and the end temperature was 320 °C (hold time = 10 s) in all cases. Table 2 shows the heating times and



RESULTS AND DISCUSSION Thermal Decomposition. An important parameter for thermal labile components is the decomposition temperature Tdec. It was determined by thermal analysis with a heating rate of 5 °C/min at the workgroup for energetic materials of the department of chemistry of the Ludwig-Maximilians University of Munich. Table 3 shows the elution temperatures Telu of the analytes for the six temperature programs in comparison to the onset decomposition temperatures Tdec.17 Also, in brackets, the maximum temperatures Tmax of the analytes during the chromatographic separation are listed. As shown in the simulation of the migration of analytes through the column, the elution temperature Telu and the maximum temperature Tmax are not equal in the case of the programs with gradient.

Table 2. Settings for the Temperature Programs program 1a 1b 2a 2b 3a 3b

heating time

heating rate

max. gradient

[s]

[°C/s]

[°C/min]

[°C]

80 66 40 33 20 17

3.6 4.4 7.2 8.7 14.4 16.9

216 264 432 522 864 1014

50 70 54 -

the corresponding heating rates of the six temperature programs. It also shows the maximum of the difference

Figure 4. Chromatograms and temperature profile of the measurements with (2a) and without gradient (2b). The temperature profile is represented by the measured temperature at the beginning of the column Ttop (red, injector side of the column), the measured temperature at the end of the column Tbottom (blue, detector side of the column), and the calculated difference ΔT of these two temperatures (green). The analytes are labeled as followed: 1−EGDN, 2−TATP, 3−2-MNT, 4−DMDNB, 5−3-MNT, 6−4-MNT, 7−NG, 8−2,6-DNT, 9−2,4-DNT, 10−ETN, 11− TNT, 12−PETN, 13−RDX, 14−Tetryl. D

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 3. Elution Temperatures Telu and Decomposition Temperatures Tdec (Onset)a name

Telu [°C] 1a

EGDN TATP 2-MNT DMDNB 3-MNT 4-MNT NG 2,6-DNT 2,4-DNT ETN TNT PETN RDX Tetryl

84.5 93.7 101.2 104.1 105.8 108.1 124.2 134.6 142.6 149.5 158.7 165.6 177.7 197.2

(85.3) (95.2) (103.4) (106.6) (108.4) (110.9) (128.2) (139.2) (147.5) (154.5) (163.7) (170.5) (182.4) (201.3)

Telu [°C]

Telu [°C]

Telu [°C]

1b

2a

2b

94.9 105.8 114.4 118.5 120.2 123.1 142.6 156.4 166.2 173.7 185.7 193.2 207.6 230.0

101.2 (104.1) 110.4 (114.0) 119.0 (123.2) 123.1 (127.5) 124.2 (128.7) 127.7 (132.4) 140.9 (146.5) 152.4 (158.8) 161.0 (168.0) 166.8 (174.2) 177.1 (185.2) 184.0 (192.4) 199.0 (207.9) 220.8 (229.5)

113.9 124.8 135.7 140.3 142.6 144.9 164.5 178.3 188.6 194.9 208.2 216.2 233.5 259.9

Telu [°C] 3a 126.5 136.9 148.4 151.8 153.5 156.4 170.2 184.6 192.1 197.8 209.3 215.1 239.2 263.4

(128.4) (139.2) (151.0) (154.5) (156.3) (159.2) (173.4) (188.2) (195.8) (201.6) (213.3) (219.1) (243.0) (266.0)

Telu [°C]

Tdec [°C]17

3b 140.3 151.8 166.8 170.2 172.5 175.4 191.5 209.9 218.5 225.4 239.2 246.1 278.3 301.3

--232 227 243 234 167 --170 306 157 208 190

a

In brackets, the estimated maximum temperature Tmax is listed.

The maximum temperature Tmax is not available from measurements. Therefore, Tmax was calculated by the simulation of alkanes with retention times in the range of retention times of the explosives. The recorded temperatures of the six temperature programs were used. The difference of Tmax and Telu (see Figure S1 in the Supporting Information) of the alkanes were used to interpolate Tmax of the explosives according to their retention times. For temperature programs with gradient, Tmax is between 1 and 9 °C higher than Telu. The difference depends on the height of the gradient, for programs 1a and 3a, with the maximum gradient around 50 °C; the highest differences are 4 to 5 °C, while for program 2a, with the maximum gradient of 70 °C, the highest differences are 9 °C. For programs without gradient, the two temperatures are the same. In most cases, the elution and maximum temperatures are below the decomposition temperatures. For NG, the elution temperature is higher in two temperature programs. For RDX, the decomposition temperature is exceeded by the elution temperature in four cases, and in five cases, the maximum temperature is higher than the decomposition temperature. For ETN, the elution temperature is higher in four temperature programs, and the maximum temperature is higher in five programs. For PETN and Tetryl, the elution temperature exceeds the decomposition temperature in all cases. For NG, ETN, and PETN, the decomposition temperatures are lower than the injector and transfer temperatures. A partial decomposition in the injector is likely. Lower temperatures than the decomposition temperature are not high enough to completely vaporize the sample. The influence of the gradient is most obvious for PETN. In the fastest program, 3b, PETN could not be measured; it decomposes totally with the higher elution temperature of 246 °C, while all other substances are detectable for all programs. Also, in the chromatograms, like in Figure 4, the signal of PETN is higher in programs with gradient, where the analytes elute at lower temperatures. The Tetryl used in the solutions for the calibration could not be detected as Tetryl in any cases. Instead 2,4,6-trinitro-Nmethylaniline (CAS 1022−07−7, abbreviation TNA) was found. The NO2 group bonded to the nitrogen is replaced by a hydrogen. The mass fragments m/z = 46, 181, and 241, which

are characteristic for Tetryl, could not be observed, while the fragments m/z = 194 and 242 could be measured in a similar ratio as in the spectra of TNA in the NIST database.23 With the variation of the temperatures (injector, transfer, detector, and temperature programs) no change in the observed spectra could be found. Therefore, it is not possible to determine where the decomposition from Tetryl to TNA takes place. Reduction of Runtime while Maintaining Maximum Temperature. The spatial gradient allows the achievement of comparable or even lower maximum temperatures with much lower retention times (Figure 5). The benefit of the gradient is

Figure 5. Measured elution temperatures over retention times of the 14 analytes (different shapes and color) for the six temperature programs. For the programs 1a, 2a, and 3a, the maximum temperature during the separation is some degrees higher (see Table 3).

apparent in comparison of the slower temperature programs without gradient with the faster temperature program with gradient. The elution temperatures for the analytes in measurement 1b (no gradient) are nearly the same as in measurement 2a (gradient), while the runtime for the chromatogram is reduced significantly. A similar reduction of maximum temperatures is possible by using aforementioned methods, like reducing the column length. The gradient is an additional method to reduce the maximum temperature. In the six presented temperature programs, we combined the gradient and reduced heating rates to reduce the maximum temperature during separation. E

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

analytes, the LOD values are comparable in the range of 0.4 to 19 pg on column (mean = 5.5 pg). Starting with ETN, the LODs of the later eluting analytes are higher in the range from 8 to 170 pg (mean = 50 pg). This corresponds to the vapor pressure of the substances. The first nine eluting analytes have vapor pressures (at 298.15 K) from 18 Pa (2-MNT)24 to 0.04 Pa (2,4-DNT),17 while starting with ETN (p = 6e−4 Pa)17 and TNT (p = 9e−4 Pa),17 the vapor pressure drops by two magnitudes and more for the later eluting analytes, PETN (p = 1.6e−6 Pa),25 RDX (p = 4.4e−7 Pa),25 and Tetryl (p = 8.7e−7 Pa).25 Peak broadening of later eluting analytes is also a reason for higher LODs. Especially RDX and Tetryl suffer from a noticeable peak broadening, most likely due to the used SilTite Mini connectors and the relative low temperatures of the transfer lines. The higher LODs for DMDNB in temperature programs 2a and 2b in comparison to the LODs obtained with the other four programs are explainable by a coelution of DMDNB with 1-dodecene. In the slower temperature programs 1a and 1b, DMDNB elutes before the 1-dodecene; in the faster programs 3a and 3b, DMDNB elutes after 1-dodecene. Most of the fragments, except the fragment with m/z = 100, are part of the spectra of both substances, and therefore, the ratio of the identifier and quantifier are not inside the specified tolerance. While not one temperature program has all the lowest or highest LOD values of the analytes, most of the lowest LOD values are found in measurements with gradient (5 for 1a, 5 for 2a, 3 for 3a). Only one of the lowest LODs is measured in a program without a gradient, which is RDX in program 2b. On the other side, the highest LODs are measured predominantly, in 11 of 14 cases, in programs without gradient (2 for 1b, 5 for 2b, 3 for 3b), while three substances have the highest LOD in a program with gradient (1 for 2a, 2 for 3a). The lowest calculated LODs of the specific explosives are around 1 order of magnitude below the values estimated with a VO-GC/MS method17 (see Table 4), while the highest estimated LODs are higher by a factor of 5 or less compared to the VO-GC/MS method only for the low volatile explosives up to 3-MNT. An exception is EGDN, where the lowest LOD is nearly the same as the LOD in the compared method. Higher volatile explosives, starting at 4-MNT, show lower

An advantage of the gradient is the peak focusing effect, which makes it possible to keep the resolution while reducing the maximum temperature. The peak capacities, calculated as the sum of the resolutions of the neighboring analytes from EGDN to Tetryl for the six temperature programs are 29.7 ± 1.7 (1a), 27.7 ± 1.6 (1b), 27.3 ± 1.7 (2a), 26.0 ± 1.5 (2b), 27.8 ± 1.7 (3a), and 26.3 ± 1.6. For the programs with gradient, these values are higher than without gradient, but not significantly (t test, α = 5%, n = m =5). Other methods for the reduction of maximum temperatures, like the reduction of the column length or higher flow rates, lead to reduced resolutions. Limit of Detection. The limit of detection (LOD) and the limit of quantification (LOQ) were determined by external calibration. The ratio between the LOQ and LOD is on average 3.6. In the further discussion, only the LOD is used. The LODs for the 14 analytes and the six chromatographic programs are shown in Table 4. For the first nine eluting Table 4. Comparison of LODs Determined with the FF-TGGC/MS with LODs Determined with a VO-GC/MSa 1a

1b

2a

2b

3a

3b

VO-GC/ MS (SIM)17

3.6 1.4 1.4 0.7 3.2 2.6 3.5 3.5 4.5 44.7 25.7 7.6 17.2 61.8

8.1 9.0 2.2 2.0 4.6 6.5 5.7 2.4 7.1 34.9 37.3 9.6 19.0 106

6.5 5.3 1.0 21.5 3.1 5.6 4.0 1.3 3.3 29.3 11.1 15.0 62.1 85.4

9.9 0.9 1.7 18.2 16.2 12.0 5.2 3.1 4.8 43.6 38.8 77.6 13.1 168

10.4 0.4 12.8 0.9 2.5 6.0 6.7 3.4 4.9 41.7 9.3 37.1 58.5 74.4

10.3 1.7 3.4 1.0 4.0 3.7 18.7 4.2 5.1 97.2 13.7 -67.9 70.9

2.8 7.4 9.4 9.2 8.3 27.9 25.8 15.1 34.9 1200 164 170 110 --

name EGDN TATP 2-MNT DMDNB 3-MNT 4-MNT NG 2,6-DNT 2,4-DNT ETN TNT PETN RDX Tetryl a

LOD in pg on column. The lowest LODs are marked in boldface.

Table 5. Pearson Correlation and Kendall Correlation between LOD and Maximum Temperature on the Separation Column Tmax and between ln(LOD) and Tmaxa name

Pearson

sig.

EGDN TATP 2-MNT DMDNB 3-MNT 4-MNT NG 26-DNT 24-DNT ETN TNT PETN RDX Tetrylb

0.8546 −0.4569 0.5096 −0.1065 0.0803 0.1059 0.8131 0.5313 0.0508 0.7418 −0.3109 0.7744 0.5914 0.1089

Pearson

sig.

(ln(LOD), Tmax)

(LOD, Tmax) 0.030 0.362 0.302 0.841 0.880 0.842 0.049 0.278 0.924 0.091 0.549 0.124 0.216 0.837

0.8118 −0.4779 0.6322 −0.1104 0.0580 0.1665 0.8785 0.4622 0.1180 0.7225 −0.3500 0.8966 0.5341 0.1087

Kendall

sig.

(LOD, Tmax) 0.050 0.338 0.178 0.835 0.913 0.753 0.021 0.356 0.824 0.105 0.497 0.039 0.275 0.838

0.7333 −0.3333 0.4667 0.0667 −0.0667 0.0667 0.7333 0.3333 0.3333 0.2000 −0.0667 0.6000 0.4667 0.0667

0.039 0.348 0.188 0.851 0.851 0.851 0.039 0.348 0.348 0.573 0.851 0.142 0.188 0.851

a

Nitrate esters are marked in boldface. bSignificance for two-side test. F

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Consequently, no correlation between LOD and maximum temperature could be observed.

LODs for all used programs with the FF-TG-GC/MS method than with the VO-GC/MS method. The LODs in the compared VO-GC/MS method were estimated by the same DIN 32645 method22 with the same parameters (α = 0.01, k = 3, six calibration points). Correlation between LOD and Maximum Temperature. For thermal labile substances, it is to be expected that with lower maximum temperature, the chromatographic signals are enhanced, and therefore, a better LOD is achieved. To test this hypothesis, correlations (Pearson and Kendall) between LOD and maximum temperature Tmax (linear model), respectively, between the logarithm of the LOD and the maximum temperature Tmax (exponential model) were calculated. The results are presented in Table 5. It is noticeable that the four nitrate esters show a high correlation, while there is no correlation for all of the other substances. One explanation is the thermal lability of these substances. Most of the substances that show no significant correlation are thermal stable components, like the MNTs, DNTs, DMDNB, and TATP. For other substances, the maximum temperature is far below the onset decomposition temperature, e.g., TNT, Tdec = 306 °C17 with Tmax between 164 and 239 °C, see also Table 3. The nitrate esters on the other hand show significant correlations are thermally labile. For EGDN and NG, the Pearson and Kendall correlations are significant (α = 0.05) for the linear model as well as for the exponential model. For PETN, the Pearson correlation is only significant (α = 0.05) for the exponential model. For ETN, the correlations are less significant, only to a level of α = 0.10; the Pearson correlation for the linear and exponential model is significant. The Pearson correlation is higher for NG and PETN for the exponential model, while for ETN, the difference between the Pearson correlations of the two models are low. For EGDN, the correlation for the linear model is higher than for the exponential model. For the decomposition process, an exponential model is a standard approach. The more stable nitrate ester EGDN is at the beginning of the decomposition process, where the exponential model can be approximated by a linear model, while for NG and PETN, the decomposition process is more advanced. Especially for PETN, the process advances to total decomposition with temperature program 3b. The lowest calculated LODs of EGDN, NG, and PETN are for the temperature program with the lowest maximum temperature. For RDX, a similar correlation as for NG or ETN would be expected. The maximum temperatures of RDX are only in one program, the slowest one with gradient, significantly lower than the decomposition temperature (see Table 3). But the correlations between LOD and maximum temperatures are more similar to the correlations in the cases of 2-MNT and 2,6-DNT. One explanation could be the much higher heating rates in the FF-TG-GC in comparison to the heating rate in the thermal analysis experiments for the estimation of the decomposition temperature by a factor of 40 to 200. Also, the short exposure to high temperatures in the FF-TG-GC method could be the reason for the missing influence of the maximum temperature on the LOD of RDX. The maximum temperature of Tetryl is always higher than the decomposition temperature. As described before, a decomposition of Tetryl to TNA takes place. TNA seems to be stable in the range of the observed temperatures.



CONCLUSIONS The maximum temperature of an analyte in a gas chromatographic separation can be influenced by different parameters. One additional parameter is applying a spatial temperature gradient along the chromatographic column. With the lowering of the maximum temperature, the decomposition of thermal labile substances, like nitrate esters, can be reduced. For other, more stable explosives, no influence of reduced maximum temperature was observed. Furthermore, hyperfast separation in less than 1 min, including the cool-down time, could be shown. This makes high-throughput measurements possible. An advantage of the gradient over the other methods for reduction of the maximum temperature is the conserved resolution combined with a shorter runtime. A more detailed study on the relation between the gradient, maximum temperature, and elution temperature combined with the influence on resolution is necessary. A first step in this direction is the simulation model for the thermal gradient. This model will be extended to calculate also the width of the analyte bands during the separation process. Further developments of the FF-TG-GC, like purged connectors and programmable pressure control (constant flow), and combinations with other types of injections, like cold-on-column or PTV, could lead to additional improvements for the measurement of explosives with gas chromatography.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.8b00900.



Table of the parameter settings for the simulation of the migration of analytes through a GC column, six tables containing the retention time, LODs, and LOQs of 14 analytes for the six different temperature programs, and a figure containing the differences Tmax − Telu from a simulation (PDF)

AUTHOR INFORMATION

Corresponding Author

*Phone: +49-172-2524747; Fax: +49-228-732596; E-mail: [email protected]. ORCID

Jan Leppert: 0000-0001-8857-8103 Martin Härtel: 0000-0001-8247-5314 Thomas M. Klapötke: 0000-0003-3276-1157 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This study was financed by the German Federal Ministry of Education and Research within the project ChemAir (FKZ 13N12580) and the German Research Agency (DFG) in the project BO/51 2-1. We thank Restek for the generous support and providing of materials. G

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry



REFERENCES

(1) Grob, R. L., Barry, E. F., Eds. Modern Practice of Gas Chromatography, 4th ed.; John Wiley & Sons: New York, 2004. (2) Poole, C. F., Ed. Gas Chromatography, 1st ed.; Elsevier: Oxford, 2012. (3) Harynuk, J.; Marriott, P. J. Anal. Chem. 2006, 78, 2028−2034. (4) Bicchi, C.; Brunelli, C.; Galli, M.; Sironi, A. J. Chromatogr. A 2001, 931, 129−140. (5) Dagan, S.; Amirav, A. J. Am. Soc. Mass Spectrom. 1996, 7, 737− 752. (6) Fialkov, A. B.; Gordin, A.; Amirav, A. J. Chromatogr. A 2003, 991, 217−240. (7) Blumberg, L. M. Temperature-Programmed Gas Chromatograph; Wiley-VCH: Weinheim, Germany, 2010. (8) Korytár, P.; Janssen, H.-G.; Matisová, E.; Brinkman, U. A. T. TrAC, Trends Anal. Chem. 2002, 21, 558−572. (9) Blumberg, L. M.; Klee, M. S. Anal. Chem. 1998, 70, 3828−3839. (10) United States Environmental Protection Agency. Explosives by Gas Chromatography; SW-846 Test Method 8095; 2007. (11) Stefanuto, P.-H.; Perrault, K. A.; Focant, J.-F.; Forbes, S. L. Chromatography 2015, 2, 213−224. (12) Amirav, A.; Gordin, A.; Poliak, M.; Fialkov, A. B. J. Mass Spectrom. 2008, 43, 141−163. (13) Cramers, C. A.; Scherpenzeel, G. J.; Leclerq, P. A. J. Chromatogr. 1981, 203, 207−216. (14) Leclercq, P. A.; Scherpenzeel, G. J.; Vermeer, E. A. A.; Cramers, C. A. J. Chromatogr. 1982, 241, 61−71. (15) Sapozhnikova, Y.; Lehotay, S. J. Anal. Chim. Acta 2015, 899, 13−22. (16) de Zeeuw, J.; Reese, S.; Cochran, J.; Grossman, S.; Kane, T.; English, C. J. Sep. Sci. 2009, 32, 1849−1857. (17) Härtel, M. Studies towards the gas-phase detection of hazardous materials by vapor pressure measurements with the transpiration method in combination with vacuum outlet GC/MS. Ph.D. Thesis, Ludwig-Maximilians-Universität Mü nchen, 2017; http://nbn-resolving.de/urn:nbn:de:bvb:19-213505. (18) Ghosh, A.; Johnson, J. E.; Nuss, J. G.; Stark, B. A.; Hawkins, A. R.; Tolley, L. T.; Iverson, B. D.; Tolley, H. D.; Lee, M. L. J. Chromatogr. A 2017, 1517, 134−141. (19) Tolley, H. D.; Tolley, S. E.; Wang, A.; Lee, M. L. J. Chromatogr. A 2014, 1374, 189−198. (20) Boeker, P.; Leppert, J. Anal. Chem. 2015, 87, 9033−41. (21) Blumberg, L. M. J. Chromatogr. A 2017, 1491, 159−170. (22) Rüttinger, R. Chemische Analytik; Nachweis-, Erfassungs- und Bestimmungsgrenze unter WiederholbedingungenBegriffe, Verfahren, Auswertung; DIN 32645; 2008. (23) NIST Chemistry WebBook, NIST SRD 69; National Institute of Standard and Technology, 2017; owned by Linstrom, P. (24) Bikelyte, G.; Härtel, M.; Stierstorfer, J.; Klapötke, T. M.; Pimerzin, A. A.; Verevkin, S. P. J. Chem. Thermodyn. 2017, 111, 271− 278. (25) Ö stmark, H.; Wallin, S.; Ang, H. G. Propellants, Explos., Pyrotech. 2012, 37, 12−23.

H

DOI: 10.1021/acs.analchem.8b00900 Anal. Chem. XXXX, XXX, XXX−XXX