Anal. Chem. 1989, 6 1 , 145-148
145
Qualitative Supercritical Fluid Chromatography/Fourier Transform Infrared Spectroscopy Study of Methylene Chloride and Supercritical Carbon Dioxide Extracts of Double-Base Propellant M. Ashraf-Khorassani and L. T. Taylor* Department of Chemistry, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061 -0212 Supercritical fluld chromatography has been conducted on a series of CH,CI2 extracts of double-base propellant with online Fourier transform Infrared (FT-IR) spectroscopy detection. DI-n-propyl adipate, triacetin, 2-nltrodlphenylamhe, and nltroglycerln were separated and detected. Several minor components were detected In both “bad” and “good” propellant. Due to the low concentration of these mlnor components in CH,Ci,, FT-IR was not able to detect them. Slmliar propellant samples were extracted with supercritical fluid CO, and chromatographed under the same conditlons as the CH2C12extract. Since In supercritlcal fluld extractlons (SFE) the solvent Is the same as the mobile phase employed in the chromatography system, there Is no interference from the solvent peak in the chrornatographlc technique. Due to this fact, a larger sample was Injected on the column and better detection was obtalned with FT-IR for low concentrations of mlnor components. Slnce most of the nonpolymeric lngredlents In double-base propellants contain nitro or nitrate functlonallty, a poor response was observed with the flame ionization detector (FID) (e.g. nitroglycerin or lower nitrated giycerlns). However, the nitro and nltrate functlonal groups have strong absorbance In the Infrared region. With SFESFC and F I D as a detector, the “bad” and “good” propellants were not dlstlnguished, but with FT-IR as a detector a malor chromatographlc peak was detected In the “bad” propellant due to the presence of water.
INTRODUCTION Double-base propellants consist of nitrocellulose and another nitrated ester, usually nitroglycerin. In addition, a number of additives such as stabilizers, plasticizers, oxides, and metals, which seek to stabilize the propellant and control its burning, are incorporated. After a period of time the nitroglycerin and stabilizers are degraded. In aged double-base propellant the determination of nitroglycerin and stabilizers in the presence of these degradation products has been a problem. Traditional methods for isolation of nonpolymeric components in propellant systems have involved dissolution of the more readily soluble components via Soxhlet extraction (1). After extraction of the sample for 12-72 h (depending on the solvent), extracted compounds are analyzed via chromatography. For example several workers have utilized thin-layer chromatography (TLC) combined with various spectroscopic methods (UV, IR) (2,3). These techniques possess several disadvantages such as (a) poor quantitation, (b) different developers and visualization methods used which are time-consuming,and (c) large amounts of sample required ( 4 ) . Gas chromatography (GC) is another method that has been used for analysis of plasticizer and stabilizer in both single and double-base propellants because of its speed, sensitivity, and simplicity. This technique, however, is limited by the fact that many explosive compounds are incompatible 0003-2700/89/036 1-0 145$01.50/0
with the usual GC conditions because of their inherent thermal instability and involved derivatization (5, 6) process. An alternative method to TLC and GC is high-performance liquid chromatography (HPLC). This technique is ideal for separation of thermally unstable compounds (7); however, the lack of efficiency to separate a complex mixture and also the incompatibility of this technique with spectroscopic detectors (MS, FT-IR) have limited analysis of explosives. Supercritical fluids offer several attractive features as a solvent for both extraction and chromatography. It has been demonstrated that greater diffusivity and lower viscosity afforded by supercritical fluids relative to conventional liquids can yield more efficient extraction and faster chromatography with higher efficiency. Many supercritical fluids are inert, nontoxic, pure, inexpensive, readily available, and easily removed from the resulting extract. Another advantage of supercritical fluids is the ability to control solvent power by applying a different pressure and temperature. Many authors have reported the extraction of one compound out of a matrix using low density fluid, while retaining the compound of interest for later extraction using a more dense fluid (8). Several articles have been reported regarding analysis of thermally labile compounds via supercritical C 0 2using both capillary and packed columns with different types of detection systems (9-11).
With the demonstration of supercritical fluids for both extraction and chromatography,the desirable extension would be to extract the compounds of interest into the supercritical fluid before analysis with SFC. This would be analogous to the case in HPLC where the mobile phase solvent is commonly used for dissolving the sample. The solvent for sample introduction in SFC is usually different from the mobile phase. Therefore, solvent peaks and any associated impurities are always present. Several articles have appeared regarding on-line supercritical fluid extraction (SFE) of complex mixtures followed by the separation of extracted materials using SFC (12). The advantages of this on-line simultaneous extractiop and chromatography are (a) sample preparation and the chance for sample contamination are minimized, (b) solvent absorbances that inhibit spectral interpretation when higher information detectors are used can be avoided and, (c) column life would be extended, because only solubilized materials are placed on the column. In this study we wish to report the results obtained from the analysis of nonpolymeric materials found in double-base propellants that extrude satisfactorily (”good”)and unsatisfactorily (“bad”). Also the qualitative efficiency of methylene chloride (CH2C1,) and supercritical COPare compared as solvents for extraction of double-base propellants. Detection and identification of the separated components were obtained with both flame ionization and Fourier transform infrared spectrometry.
EXPERIMENTAL SECTION Lee Scientific 501 (Salt Lake City, UT) and Suprex 200A (Pittsburgh, PA) supercriticalfluid chromatographs were utilized with a packed column. A restrictor was drawn from fused silica 0 1989 American Chemical Society
146
ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989
capillary (50 pm i.d.) tubing. A Valco injection valve with 0.1 p L of total volume was used to introduce the CHzClzextract of double-base propellant onto the column. An on-line Rheodyne Model 7125 injection valve with 10-pL sample loop was used to inject,directly a supercriticalCO, extract of double-base propellant onto the column. SFC/FT-IR data were collected with a Nicolet (Madison,WI) 5SXC spectrometer equipped with a prototype 0.6 mm i.d. X 5 mm path length high-pressure flow-through cell. The temperature of the flow cell (1.4 pL) was maintained at 32 "C. Spectral data at 8 cm-' resolution were acquired at 4 scans/file to yield a time resolution of 1 file/s. Model compounds and propellantsemployed in this study were provided by George Nauflett of the Naval Surface Warfare Center, Indian Head, MD. Model compounds were all dissolved in pesticide grade methylene chloride (T. J. Baker Chemical Co.) prior to introduction onto the column. The concentration of sample per component ranged from 200 to 500 ng/pL. A DELTABOND (KeystoneScientificCo.) cyanopropyl packed column, 25 cm X 1.0 mm i.d., 5-pm particle size, was employed for all chromatographic separations. Carbon dioxide was obtained from Scott Specialty Gases (Plumbsteadville, PA). Nonpolymeric components of two different propellants (RAD84F002-008 and RAD87B 004-003) were isolated via two extraction methods. In the first method a Soxhlet extractor with 300 mL of CH2ClZwas employed to extract 2 g of each propellant for 72 h. After extraction, 20 mL of extract was evaporated to 1 mL by aeration, and the concentrated extract was chromatographed. In the second method samples were extracted with supercritical COz. A Milton Roy (Redondo Beach, FL) sample preparation accessory equipped with extraction bomb and recirculating pump was used to extract 100 mg of the propellant for 12 h. The extraction parameters were set at 275 atm and 60 "C.
RESULTS AND DISCUSSION Soxhlet Extraction. Supercritical fluid chromatography (SFC) has been conducted on both "good" (RAD87B 004-003) and "bad" (RAD84F 002-008) double-base propellant by using a packed column with on-line sequential detection via FT-IR and FID. The components extracted into the CHzClz are
0
5
145
115
15 200
10 150
2
20
TIM
PnEssunE
Flgure 1. Separation of CH,CI, extract of "good" (RAD87) and "bad" (RAD84) double-base propellant via 100% CO, at 60 "C with DELTABOND cyanopropyl (25 cm X 1.0 mm, i.d., 5-pm particle size) packed cdumn, using FID as a detector: S = CH,CI,, 1 = di-npropyl adipate (DNPA), 2 = triacetin FA), 3 = 2-nbodiphenylamine(BNDPA), 4 = nitroglycerin. 4-
Sn
GOOD
. 00 _. I
20
. 00
RETENTIONT I M E (MIN)
320
620 DATA
920
1 h O
1i20
1620
POINTS
0 0
0
I1
II
C3H7OC(CH2),C0C,H7 DNPA
0
II
I/
0-CCH,
I
CH 3COCH 2CH
I
CH 20CCH 3
I1
0 TA
Flgure 2. Comparison of Gram-Schmidt reconstruction for CHPCIP extract of "good" (RAD87) and "bad" (RAD84) double-base propellant, 5-pm particle size packed column, using FT-IR as a detector. See Figure 1 for peak identity.
0 ' CHON
1
4"
CH20N \"
N DPA
U
NG
generally believed to be di-n-propyl adipate (DNPA), triacetin (TA),nitroglycerin (NG), and 2-nitrodiphenylamine (NDPA). Other components that result from the breakdown of TA, NDPA, NG, and DNPA in the propellant are also expected. Initially, a separation of three standards (TA, DNPA, and NDPA) was developed with 100% COz. Next, under similar conditions, the separation of each CH2C12extract sample was obtained, with both flame ionization and FT-IR detection, Figures 1 and 2. No major difference was observed in chromatograms of "good" and "bad" propellant with the same detector. Each of the four major components was separated and subsequently identified by using FT-IR as a detector. Several minor components were observed in FID traces of both sample extracts but due to their low concentration the iden-
tification of these components via FT-IR was not achieved. Infrared spectra of components eluting in the first three major peaks of each separation coincided to spectra that we obtained on the three standards (i.e. DNPA, TA, and NDPA, respectively, Figure 3). A fourth major peak that has a relatively weak response to FID but high absorbance in the IR region was determined to be NG (Figure 4). An authentic sample of NG was not available, but the spectrum corresponds quite well to that previously reported for NG in the solid state (13). Bands a t 1276 and 1669 cm-' correspond to the asymmetric and symmetric stretching vibrational modes of the NOz group. h o , several low intensity bands were observed in the aliphatic C-H stretching region (2900 cm-I). The 1017-cm-' band is probably due to the C-0 stretch. A fifth peak (detected via FT-IR) appeared just after the solvent but was not observed in the FID trace. The coelution of solvent and this component in the FID trace may be attributed to band broadening in the transfer line since the FT-IR preceded the FID in our study. The IR spectrum of the peak component (labeled A in Figure
ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989
147
v,
mm
uo
U O
WAVENUMBER
m
?
m
m
m
f.
_I
,
PK.1
? . P I 2
,
? , P I 3
y1
I5
1 5
145
150
ZOO
P
5 145
W 150
2b
20
25
30 TIYE
400
400 IRESSURI
25 400
30
TlUt
id0
wzssum
Q
69
B
m fo. u WAVENUMBER
WAVENUMBER
URVENUMBER
Figure 3. Comparison of on-line SFC/FT-IR spectra of components eluting during separation of CHpCI, extract of double-base propellant with spectra obtahed during elution of singly injected components. See Figure 1 for chromatographic conditions. 125
t-
I1
PK.4
I
4 II
I(
Figure 5. Separation of 1O-pL supercritical extract of “good” (RAD87) and “bad” (RAD84)double-base propellant. See Figure 1 for chromatographic conditions and the Experimental Section for extraction conditions: 1 = DNPA, 2 = TA, 4 = NO, W = unknown, X = unknown, Y = unknown: Z = unknown.
‘n’ ‘n ‘n
2 1
WAVENUMBER
RETENTION T I M E ( M I N ) 0
Figure 4. On-line SFC/FT-IR spectrum of peak 4 from separation of double-base propellant. For chromatographic conditions, refer to Figure
31tO
1.
2) was found to be due to acetone, which was used to clean the Soxhlet extractor even though the extractor was dried prior to use. Supercritical Fluid Extract. Next, on-line SFE was performed on the same propellants followed by SFC employing the same conditionsas the previous separations shown in Figures 1 and 2. FID traces of both the separations are shown in Figure 5. A noticeable difference exists between chromatograms of the CH2C12extract and SFE. It would appear that either the supercritical extract is more concentrated than the CH2C12extract or the former process has solubilized a larger number of components since at least twice as many major components are detected via FID of the SFE as was detected for the CH2C12extract. No major difference was observed between FID traces obtained from SFE/SFC of “good” and “bad” double-base propellant. The differences in retention time between RAD84 and RAD87 are due to different size restrictor internal diameters used in each separation. Although flame ionization is a very sensitive mode of detection, the identification of peak components can be questionable and may be difficult. It should also be noted that since no extraneous solvent is required in the SFE/SFC experiment, all components detected via FID originate from the propellant (i.e. no solvent peak). On the basis of retention time comparisons with each restrictor DNPA and T A could be accounted for, whereas NDPA was apparently not extracted since no chromatographic peak matched the retention time of NDPA. This observation was quite surprising since NDPA elutes from a column under similar conditions employed for extraction. This type of behavior has been observed previously for extraction and separation of caffeine. For example, it has been shown that caffeine can be eluted from a column with 100% supercritical COz but cannot be extracted, unless the
PO
I5
880 lb20 1360 OFITA POINTS
’m’
I
20
350
l’l00
2’bLtO
1670
2600
‘n ‘n
880 1610 13+0 DFITFI POINTS
Ffgure 6. Comparison of Gram-Schmidt reconstruction for separation of supercritical CO, extract of “good” (RAD87) and “bad“ (RAD84) double-base propellant via 100% CO,. See Figure 1 for chromatographic condition. See Figure 5 for peak identification. supercritical C02 contains a small percentage of water (14). It is possible that NDPA binds to the propellant and is not released, unless a modifier is added to the supercritical C02. Pairs of peaks labeled Y and Z are not identifiable since none of our standards exhibited these retention times. Figure 6, like Figure 5, shows the separation of “good” and “bad” double-base propellant but now with FT-IR detection. As expected fewer components are observed via FT-IR since it is a less sensitive detector than the flame ionization detector. In order to observe several minor components, the detector response was expanded such that some peaks were off scale. Individual file spectra were examined in order to gain information about eluting materials. Three major components, NDPA, TA, and NG, were easily indentified from side-by-side comparisons of the file spectra with spectra of the individual Components. Even with more specific detection (FT-IR), no evidence for the presence of NDPA could be ascertained in the SFE. Since FT-IR and FID were accomplished sequentially (i.e. one injection), valid comparisons can be made between these two traces insofar as retention times are concerned. In that regard, peaks Y and Z demonstrated much less infrared response than flame ionization response, such
148
ANALYTICAL CHEMISTRY, VOL. 61, NO. 2, JANUARY 15, 1989
Z
’
E a
.._ 1
3210
2720
2230 1740 WAVENUMBER
!250
760 I
1 I
Figure 7. Coadded file spectrum of peak W.
369c
siao
2970
zbso
[L 1450
l$40
... 1130
520
WAVENUMBER
Figure 9. Coadded file spectrum of peak eluting at approximately 7 min in the separation of “bad” propellant. See Figure 6.
U
N 0
13 10
2fZO
2230 1f40 WAVENUMBER
lk50
$60
Figure 8. Coadded file spectrum of peak X.
that no identifiable spectra could be obtained on these components. The large FID signal, however, reflects eluting materials with much hydrocarbon character. File IR spectra of components giving rise to peaks W and X in both “good” and “bad” propellant separations were studied in hopes of identifying these compounds. The IR spectra of the component leading to peak W in both “good” and “bad” propellant were identical (Figure 7). The bands at 1663, 1617, and 811 cm-’ are typical stretching vibrational modes for nitrite functional groups (15). Also weak C-H stretching bands at 2951 cm-’ indicate the presence of aliphatic carbon. It is conceivable that a reduction of nitrate ester to nitrite ester has taken place in both types of propellant. This early eluting component, of course, would not be seen in the separation of the CHzClzextract since it would have been obscured by the solvent peak given identical chromatographic parameters. Unfortunately,we do not have appropriate model compounds to test this hypothesis. Individual files taken throughout the peak indicate the elution of only one component. The spectra of coadded files across peak X (Figure 8) are strikingly similar to the spectrum of NG (Figure 4). A small difference in the position of bands, compared to the spectrum of NG, was observed. Specifically, the band at 1669 cm-’ was shifted to lower wavenumber and the band at 835 cm-’ was shifted to 844 cm-’. This compound is believed to be a lower nitrated ester of NG such as 1,3-dinitroglycerin. The absence of an authentic sample, however, makes an absolute assignment tenuous. The separation of the “bad” double-basepropellant yielded a unique peak ( t , z 7 min) with large tailing which was detected only via FT-IR, not FID. At first, it was believed this compound was highly nitrated because it is known that FID has low response to nitrated compounds (16), whereas nitrated functional groups have strong infrared absorption bands. However, after the coadded file spectrum across this peak was checked (Figure 9), several bands were observed that cannot be attributed to nitrate ester. The spectrum has two bands at 1273 and 1384 cm-’ and several weak bands in the 2000-cm-’ region. Also a strong absorption band at 1608 cm-’ was observed. After the spectrum was studied more closely, it was reasoned that all of these bands are fundamental frequencies for C02and HzO. The weaker bands due to COzat 2052,1944,
1385, and 1274 cm-’ are not totally compensated for in the background subtraction routine. The 1608-cm-‘ band can be confidently assigned to the HzO deformation mode. Spectra of the front and back of this broad chromatographic peak were examined, and no difference was observed. The absence of an FID response further corroborates the presence of H 2 0for this chromatographic signal. In conclusion, since some of the nonpolymeric materials in double-base propellant are thermally labile (Le. nitroglycerin), SFC is a more suitable method than gas or liquid chromatography for separation of these components. Also, the results suggest that for detection and identification of separated compounds,both FID and FT-IR are desirable. However, for the detection of highly nitrated compounds FT-IR or another specific detector such as the mass spectrometer would be preferred. It was further demonstrated that SFE of doublebase propellants has several advantages over the conventional liquid solvent extraction. It therefore appears that SFE/SFC provides an easy and efficient method for both extraction and separation of nonpolymeric materials in double-base propellant.
ACKNOWLEDGMENT The technical advice of George Nauflett, Naval Surface Warfare Center, Indian Head, MD, was especially helpful. LITERATURE CITED (1) Welcher, F. J. Standard M e W s of Chemical Ana/ysis, 6th Ed.; Van Nostrand; New York, 1968; Vol. 11, Part 8, p 1372. (2) Parker, R. G.; McOwen, J. M.; Cherolis, J. A. J . forensic Scl. 1975, 20, 254. (3) Macke. G. F. J . Chromatogr. 1988, 38, 47. (4) Llndblom, T.; Chemical Problems Connected Wlth Stability of Exploslves; Hansson, J., Ed.; Sundberg: 1979; Vol. 5, p 107. (5) Sopranettl, A.; Reich, H. U. Chemlcal Problems Connected Wlth Sfablllty of Explosives; Hansson, J., Ed.; Sundberg: 1979; Voi. 5, p 163. (8) Dalton, R . W.; Kohlbeck, J. A,; Bolleter, W. T. J . Chromatogr. 1970, 50, 219. (7) McAuiey, C. D. Chemical Problems Connected With Stablllty of Exploslves; Hansson, J., Ed.; Sundberg: 1976; Vol. 4, p 199. (8) Stahl, E.; Qulrin, K. W.; Glatz. A.; Gerrard, D.; Rau, G. Ber. BunsenGes Phys. Chem. 1984, 8 , 900. (9) Fjeidsted, J. C.; Kong, R. C.; Lee, M. L. J . Chromatogr. 1983, 279, 449. (10) Wright, B. W.; Smith, R. D. HRC CC, J . Hlgh Resolut. Chromatogr. Chromatogr. Commun. 1985, 9 , 73. (1 1) Ashraf-Khorassani, M.; Taylor, L. T.. submitted for publicatlon in HRC CC , J High Resolut Chromatogr . C h r m t o g r Commun (12) McNaliy, M. E.; Wheeler, J. R. J . Chfomatogr. 1988, 435, 63. (13) Pristera, F.; Halik, M.; Catelli, A,; Frederlcks, W. Anal. Chem. 1960, 32, 495. (14) McHugh, M.; Krukonls, V.; Supercrltical FluM Extraction; Butterwwths: London, 1966; p 185. (15) Silverstein, A. M.; Bassler, G. C.; Morill, T. C. Spectromehic Identifcaflon of Organic Compounds, 4th ed.; Wiley: New York, 1981; p 98. (16) Trowell, J. M. Anal. Chem. 1970, 4 2 , 1440.
.
.
.
.
RECEIVED for review August 24,1988. Accepted October 26, 1988. The financial assistance of the Naval Ordnance Station, Indian Head, MD, is gratefully appreciated.
