ANALYST, JULY 1993, VOL. 118 74 1 Supercritical Fluid Extraction and Chromatography-Mass Spectrometry of Flame Retardants From Polyurethane Foams Graham A. MacKay Laboratory of the Government Chemist (LGC), Queens Road, Teddington, Middlesex, UK TWI I OLY Roger M. Smith Department of Chemistry, Loughborough University of Technology, Loughborough, Leicestershire, UK LEI I 3TU Four chlorinated organophosphate flame retardants present in polyurethane foams were analysed by supercritical fluid chromatography with flame-ionization detection and mass spectrometry. In addition, their amenability to gas chromatography was investigated. Real samples were analysed by using these techniques after on-line extraction with supercritical carbon dioxide. On-line quantitative extraction of the flame retardants from the polyurethane foams was investigated by using an external calibration.Keywords: Flame retardant; polyurethane foam; supercritical fluid chromatography; supercritical fluid extraction; mass spectrometry Flame retardants are incorporated in polyurethane foam mattresses to prevent rapid combustion. The most widely used retardants are chlorinated organophosphates, which are often employed in conjunction with more involatile retardants such as melamine.1 The organophosphates are viscous liquids that are almost totally immiscible with water and which breakdown at temperatures above 200 "C.2 Because of their instability at high temperatures, their extraction and chromatography must be performed at temperatures below 200 "C. The temperature required for their elution by gas chromatography (GC) often also causes thermal decomposition.Indeed, when analysed by gas chromatography-mass spectrometry (GC-MS) it was found that the two retardants with the highest relative molecular mass (Thermolin 101 and Amgard V6) did not elute at all. Their total immiscibility with water and lack of a suitable chromophore make reversed-phase liquid chromato- graphy (LC) difficult. It was found at the Laboratory of the Government Chemist (LGC) that selectivity was difficult to achieve in normal-phase LC as all the flame retardants tended to elute with the mobile phase front, although LC tests have previously been developed for the analysis of the solid flame retardant melamine. 1 The present study has examined the use of supercritical fluid extraction (SFE) and supercritical fluid chromatography (SFC) as a non-destructive test for all the flame retardant components of a foam.Because assays can be carried out under mild conditions, SFC has been proposed as the ideal method for the analysis of thermally labile compounds.3 Because of the temperatures that are commonly employed for carrying out SFC, elution of all the compounds was achieved without any signs of breakdown. Previously, SFE and SFC have been successfully performed on-line for the quantitative analysis of a number of other additives from polymers4-5 and both SFC-Fourier transform infrared spectrometry6 and SFC-mass spectrometry (SFC-MS)73X have been used for the identification of the components. The four flame retardants employed in this study are all chlorinated organophosphates and in common use as additives in foam mattresses.They are normally assayed by their retardant activity in a flame test.9 Experimental The mobile phase was carbon dioxide (SFC grade, Air Products). The standards were obtained from Albright & Wilson. Samples were from the Consumer Hazards sub- division, LGC. Supercritical fluid chromatography with flame-ionization detection (SFC-FID) and SFC-MS were performed on a Lee Scientific Series 600 chromatograph equipped with a 10 m X SO pm i.d., 0.25 pm film thickness, 30% biphenyl polysiloxane column from Lee Scientific, connected to a frit restrictor held at 350 "C. On-line SFE-SFC was performed using a 0.5 ml extraction cell incorporated in the Lee Scientific system.The extract was trapped cryogenically and concentrated at the top of the analytical column prior to chromatography. A Kratos MS50 double-focusing mass spectrometer fitted with an electron impact (EI) source set at 350 "C was used for SFC-MS. The scanning range was between mlz 100 and 600. The SFC column was connected to the mass spectrometer with a length of SO pm i.d. capillary tubing (enclosed in stainless- steel tubing) interposed between the column and the frit.8 Gas chromatography-MS was performed on a Finnigan 4500 quadrupole mass spectrometer fitted with a S:95 phcnyl-dimethyl (substituted) polysiloxane column, 25 m X 0.25 mm i.d., 0.25 pm film thickness, from Chrompack. Both spectrometers were interfaced to a Finnigan data system. Results and Discussion Supercritical Fluid Chromatography Four common flame retardants, Amgard TCEP, Amgard TMCP, Thermolin 101 and Amgard V6 (Fig.l), were studied. They were made up into a single solution in toluene and examined using SFC-FID. The compounds were eluted with no sign of breakdown (Fig. 2). Although, in capillary SFC, compounds are usually eluted in the order of ascending relative molecular mass, Amgard TMCP, which has a higher relative molecular mass than Amgard TCEP, was eluted earlier. This effect was reflected in the relative ease of extraction of the two compounds. At low pressures [lo0 atm (10.13 MPa)] Amgard TMCP was extracted with greater ease than Amgard TCEP. This was supported by further work, which showed that at 80 atm (8.11 MPa) Amgard TMCP was extracted slowly whereas Amgard TCEP was not extracted at all.Apparently the solubility of the retardant in the super- critical fluid is governing the order of elution in the chromato- graphy and the lower relative molecular mass compound is being retained.742 A - a - h~ -- I I I ANALYST, JULY 1993, VOL. 118 Amgard TMCP I (CI- CHZ-CH -0)3P=O CH3 Amgard TCEP (CI-CH2CHz- 0- )3P=O Thermolin 101 CI - CH2CH2 - 0 O-CH2CH2-CI CI-CH2CH2-0 ‘0- CH2CH2-CI >I-,- CH2-CH2- 0 - Amgard V6 O-CH2CH2-CI ;HZ-cI / CI - CH2CH2- 0 >:-O-CH2- C- CH2-0- P, CI- CH2CH2- 0’ I CH2-CI Fig. 1 Structures of the four flame retardants employed in this work 16 k II I m Amgard TMCPl 2 AmgardV6 (T) r; Amgard TCEP Thermolin, 101 I I I I I 1 0 10 20 30 Ti me/m in Fig. 2 SFC trace of the four flame retardants.Condition: column, 30% biphenyl polysiloxane; oven temperature, 100 “C; pressure programme, 100 atm (10.13 MPa) for S min then increased from 100 to 300 atm (10.13-30.40 MPa) at 5 atm (0.51 MPa) min-1; detection, FID at 350 “C On-line Qualitative Extraction In order to analyse polyurethane foams for the presence of the fire retardants, they would normally have to be extracted with a solvent and the retardants isolated and individually identi- fied. If there is a mixture of retardants in the foam, identification of each component will be difficult because of the lack of selectivity of conventional forms of chromato- graphy for these compounds. On-line SFE-SFC was examined as a method of resolving this problem. Three different foams, which contain among them the four most common flame retardants, were analysed by SFE-SFC.These were: ‘Safegard’, which contained Amgard TMCP and Amgard V6; ‘HG35S’, which contained Thermolin 101; and ‘FHR30H’ which contained Amgard TCEP. Usually the flame retardants are present in the foams at levels of 1-3% and, therefore, only a small amount of the foam was required; about 0.2 mg was sufficient to obtain a representative chromatogram. In addition, because of their relatively easy elution from the SFC column, their solubility in carbon dioxide was expected to be sufficiently high to permit easy extraction. Each analysis consisted of sub-sampling a small amount of foam and placing the sub-sample in a 0.5 ml extraction cell. Extraction was then performed for 5 min at 300 atm (30.40 MPa) and 60 “C.The carbon dioxide depressurizes to concentrate the sample at the top of the capillary column. The sample is then chromatographed [Fig. 3(a)-(c)], using the same conditions as in Fig. 2. Q, . m Q m cc 16 0 Thermolin 101 1 . I l6 I (D c9 7 - hl Amgard TCEP Supercritical fluid extraction occurs mainly at the surface and, therefore, the extraction matrix either needs to be permeable in order to allow transport of the additives or needs to have a high surface area.’” It appears that the polyurethane foams possess both properties and as a result enabled easy and representative extraction of the flame retardants to be carried out. When compared with direct injection, however, the peak shapes were found to have deteriorated as a result of the analyte being introduced by decompression at the start of the column. SFC-MS of the Flame Retardants In order to confirm the identity of the retardants, on-line mass spectra of the compounds were obtained (Fig.4). The programme employed for SFC-MS ensured that the pressure was kept to a maximum of 250 atm (25.33 MPa). This enabled a better sensitivity to be obtained than with the programme used for SFC-FID as a higher SFC pressure would reduce the vacuum in the mass spectrometer.8 The source temperature was held at 350 “C in order to minimize the formation of carbon dioxide clusters.8 The peaks showed no deterioration in shape from the SFC-FID separation. This shows that the interface was sufficiently efficient to transport the analyte into the source. The spectra [Fig.5(a)-(d)] were similar to those obtained by probe analysis of standards on the same instrument. However, in each instance there were small differences in the relative abundances of some of the ions. In its development, the SFC-MS system had shown a tendency for a suppression of the higher relative molecular mass ions including the molecular ion when compared with the lighter ions.8 This distortion was thought to be because of the higher source pressures encountered with the SFC instrument when compared with the pressures found in either GC-MS or probe-MS. This effect has been found previously in SFC-MS by Cousin and Arpino.”ANALYST, JULY 1993, VOL. 118 13:~’ 187 205 161 743 249 329 39 1 223 267 SFE-SFC-MS of the Flame Retardants A combination of SFE and SFC-MS led to the possibility of rapid identification of the contents of a foam by SFE-SFC- MS.The extraction conditions employed in Fig. 3, for the analysis of the three foams by SFE-SFC, were combined with the chromatographic conditions employed for SFC-MS in Fig. 4. The peak shapes obtained were of approximately the same efficiency as those observed using SFE-SFC (Fig. 3) but I 200 250 300 350 400 450 500 mlz I I I I I I I I I I 13:50 17:17 20:45 24:12 27:40 31:07 34:35 Time/min : s Fig. 4 Reconstructed ion current of SFC-MS of four flame retar- dants. Conditions: isobaric at 100 atm (10.13 MPa) for 5 min followed by a pressure programme to 250 atm (25.33 MPa) at S atm (0.51 MPa) min-I with the pressurc remaining isobaric at 250 atm (25.33 MPa) for 5 min; isothermal at 80°C for 25 min followed by a negative temperature programme, 50°C at 5 “C min-l with the temperature remaining at SO “C for 9 min 100.0 50.0 8 E o 9 100.0 - > w .- + .- .- w - 0) CT 50.0 0 I99 125 I 157 100 120 140 showed a lower efficiency than the peak shapes of the solvent-injected standards obtained by SFC-MS (Fig.4). However, the spectra obtained were still representative of the standard flame retardants. Even Amgard V6, which elutes at the highest SFC pressure (at this point the mass spectrometer is at its least sensitive), gave a spectrum that was compatible with the probe and SFC standards. Fig. 6 shows the SFE-SFC-MS profile of ‘Safegard’; the presence of Amgard TMCP and Amgard V6 in the extract is evident. Quantitative SFE-SFC Solvent extracts of the foams, in toluene, were examined by SFC and were compared with the standard solution to give the concentrations of the retardants (Table 1).In order to obtain the best reproducibility of injection volume, the smallest injection loop (60 nl) was used. The injection timing was set at 10 s, which would be sufficient for the loop to be fully flushed three times. It had been found that partial loop injection using a split time was not very reproducible, giving a reproducibility of, at best, 8%. The reproducibility of the injector system used here was 3.8%. This method gave calibration graphs (Fig. 7 ) with a good correlation. This study was then extended to on-line extraction-chro- matography, which has been used by several workers for the quantitative analysis of polymer additives.12713 However, in each of these instances packed chromatography was employed and the calibration was performed by adding a standard ‘spike’ to the polymer in the extraction vessel. This approach is more difficult with capillary columns as the volumes and concentrations that would need to be employed are considerably less. Other workers have even suggested that on-line quantitative analysis is not feasible as the extraction times are too long14 and erratic results can be obtained. 100.0 50.0 0 160 180 200 220 240 260 280 100 205 50.0 107 125 21 3 150 200 250 300 350 400 325 (4 223 I 107 I1 !5 205 136 161 1 I 359 i I 1 100 120 140 160 180 200 220 240 :OO 150 200 250 300 350 400 450 500 mlz Mass spectra from SFC scparation of retardants.(a) Amgard TMCP, ( b ) Amgard TCEP, (c) Thermolin 101 and (d) Amgard V6 Fig. 5ANALYST, JULY 1993, VOL. 118 744 100.0 s 2 - w C 3 0 c .4! 50 -0 4- 2 4- v) c 0 a, [r 0 31 4 Amgard TMCP , AmgardV6 100 200 300 400 500 600 mlz I I I I I I I 0 6.15 12:30 18:45 25:OO 31:15 3 7 : 3 0 Timelmin : s Fig. 6 SFE-SFC-MS of 'Safegard' showing the presence of both Amgard TMCP and Amgard V6. Chromatographic conditions as in Fig. 5 . Extraction conditions as in Fig. 4 Table 1 Levels of flame retardants in foams as found by using solvent extraction and SFC determination Foam Flame retardant Level (%)* 'Safegard' Amgard TMCP 1.4 k 0.1 'Safegard' Amgard V6 1.4 k 0.2 'HG35S' Thermolin 101 2.3 k 0.2 FHR30H' Amgard TCEP 0.9 k 0.1 * -t standard deviation (n = 4). 40 1 3 0 5 10 15 20 25 Mass of retardant/l0-7 g Fig.7 Calibration graphs for the flame retardants in SFE-SFC separation. Chromatographic conditions as in Fig. 2. A, Amgard TCEP; B. Amgard TMCP; C. Thermolin 101; and D, Amgard V6 A small weighed amount of each foam was placed in an extraction cell and extracted. The extract was trapped cryogenically at the top of the capillary column. After the extraction was complete, chromatography was performed on the extract. The peak area of the FID response was measured and compared with the calibration graphs obtained by injection. From the mass of retardant extracted from the foam under different conditions, the concentration could be calcu- lated (Tables 2-5). Extraction of Amgard TMCP From 'Safegard' For the extraction of Amgard TMCP from 'Safegard' (Table 2), a Concentration of the retardant that was close to the expected level of 1.4% was obtained after extraction for 10 Table 2 Quantitative SFE-SFC of Amgard TMCP from 'Safcgard' under different conditions Mass of foamlg Mass of retardant/g Yield (YO) 300 atm (30.40 MPa); 60 "C; 10 min- 0.0002 1 2.9 x 10-6 1.4 0.00028 4.8 x 1.7 0.00017 2.8 x 10-6 1.6 0.00029 4.4 x 10-6 1.5 0.00022 7.1 x 10-7 0.3 0.00026 6.7 x 10-7 0.3 0.000 16 1.7 x 10-7 0.1 0.00028 9.1 x 10-7 0.3 0.0001 7 1.1 x 10-6 0.6 0.00051 1.8 x 10-6 0.3 200 atm (20.27 MPa); 60 "C; 10 min- 100 atm (10.13 MPa); 60 "C; 10 min- 100 atm (10.13 MPa); 60 "C; 25 min- Table 3 Quantitative SFE-SFC of Amgard V6 from 'Safegard' under different conditions Mass of foam/g Mass of retardant/g Yield (%) 300 atm (30.40 MPa); 60 "C; 10 min- 0,00032 3.2 x 10-6 1 .0 0.00021 1.4 x 10-6 0.7 0.00029 4.3 x 10-6 1.5 0.00045 4.8 x 10-6 1.1 200 atm (20.27 MPa); 60 "C; 10 min- ~ Table 4 Quantitative SFE-SFC of Amgard TCEP from 'FHR30H' undcr different conditions Mass of foam/g Mass of retardant/g Yield (YO) 300 atm (30.40 MPa); 60 "C; 10 min- 0.00034 3.2 x 10-6 0.9 O.OO048 3.8 x 10-6 0.8 0.00032 2.5 x 10-6 0.8 0.0002s 1.2 x 10-6 0.5 0.00024 1.3 x 10-6 0.5 0.000 15 1.4 x 10-6 0.9 0.00024 1.7 x 10-7 0.1 0.00014 1.5 x 10-7 0.1 0.00017 1.8 x 10-7 0.1 0.0001 7 4.8 x 10-7 0.3 0.00033 5.9 x 10-7 0.2 0.00017 3.9 x 10-7 0.2 200 atm (20.27 MPa); 60 "C; 10 min- 100 atm (10.13 MPa); 60 "C; 10 min- 100 atm (10.13 MPa); 60 "C; 25 min- ~~~ Table 5 Quantitative SFE-SFC of Thermolin 101 from 'HG35S' under different conditions Mass of foam/g 300 atm (30.40 MPa); 60 "C; 0.00026 O.OO026 0.00023 0.00019 0.0004 1 0.00030 200 atm (20.27 MPa); 60 "C; Mass of retardantlg 0 min- 7 .8 ~ 7.4 x 10-6 6.5 x 4.7 x 10-6 8.5 x lo-" 0 min- 1.1 x 10-5 Yield (YO) 3.0 2.8 2.8 2.5 2.7 2.8 min at pressures of 200 and 300 atm (20.27 and 30.40 MPa). At these pressures and with the small amount of foam extracted the process was rapid. At lower pressures the reduced solubility of the retardant in the extraction fluid gave a reduced yield. Alternatively, it is possible that the slow flow rate through the back-pressure restrictor could also be hindering rapid extraction.ANALYST. JULY 1993, VOL. 118 745 Extraction of Amgard V6 From ‘Safegard’ The yields of Amgard V6 from ‘Safegard’, where it is present at a level of 1.470, were relatively low and erratic (Table 3) when compared with the extraction of Amgard TMCP under the same conditions.No extraction occurred at 100 atm (10.13 MPa). This could be explained by the low solubility of Amgard V6 in the supercritical fluid, However, there may be other reasons why Amgard V6 has a slow extraction rate even at the higher pressures. The transport of the retardant in the polyurethane foam may also be slower because of the bulky nature of Amgard V6. If this was true, then the extraction, even at elevated pressures, and hence elevated solvating power, would be expected to be incomplete. This is shown by the small difference in the extraction performance of carbon dioxide at 200 and 300 atm (20.27 and 30.40 MPa).It is most likely, however, that a combination of low solubility in carbon dioxide and slow transport in the foam is responsible for the poor extraction values. In order to achieve a more reliable value for the concentration of Amgard V6 in the foam, a longer extraction time would be needed. Conclusion Supercritical fluid extraction-SFC-MS can be used for the rapid qualitative analysis of flame retardants in polymers. This analysis would be difficult to perform by alternative methods. On-line quantitative analysis is possible because of the high surface area of the polyurethane foams and the reasonable solubility of the retardants in supercritical carbon dioxide, although there is reduced recovery for the less soluble retardants. Both the extraction and the chromatography of the flame retardants are more rapid for the two lighter retardants, TMCP and TCEP, because of the lack of steric bulk acting on the mass transport in the foam arid the higher solubility of these two retardants.Extraction of Amgard TCEP From ‘FHR30H’ In the extraction of Amgard TCEP from ‘FHR30H’, most of the retardant, present at a level of 0.9%, was recovered after extraction for 10 min at 300 atm (30.40 MPa) (Table 4). Approximately 50% was recovered at 200 atm (20.27 MPa) and very little at 100 atm (10.13 MPa). There was a slight increase in recovery when the extraction was carried out for 25 min. This clearly shows the effect of solubility on extraction. Even with small amounts of foam, extraction could be hindered by the poor solvating power of the fluid for a particular retardant.Extraction of Thermolin 101 From ‘HG35S’ At 300 atm (30.40 MPa) there was almost total extraction of Thermolin 101, present at a level of 2.3%, from ‘HG35S’ and, as with the other flame retardants, there was a slight decrease at 200 atm (20.27 MPa), probably because of the reduction in solubility of the retardant in the fluid. The extraction was also attempted at 100 atm (10.13 MPa) with little or no recovery. The extractions observed in theie examples appear very rapid when compared with other work on the extraction of polymers, when hours or days have been required to obtain plasticizers.1”lfJ This can be explained by the structure of the polyurethane foams, which consist of a matrix of fibres.The thickness of these fibres is approximately 30 pm, tapering to a negligible thickness at the edges. Previous work with poly- urethane foam sorbents17 has shown that the surface area is very large and the matrix is very thin; hence the transport of the extracts can be rapid,l0 particularly at pressures above 200 atm (20.27 MPa). The work performed in this study was supported by the Valid Analytical Measurement programme, which is funded by the Department of Trade and Industry, UK. I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 References Cody, M. K . , and Patterson, I . D., Fire Mazer., 1992, 15, 1. Product Data Sheets, Albright & Wilson, Warley, West Midlands. Analytical Supercritical Fluid Chromatography and Extraction , eds. Lee, M. L., and Markides, K. E., Chromatography Conferences, Provo, UT, 1990, p. 277. Cotton, N. J., Bartle, K. D., Clifford, A. A., Ashraf, S . , Moulder, R., and Dowle, C. J., J. High Resolut. Chromatogr., 1991, 14, 164. Anton, K., Menes, R., and Widmer, H. M., Chromatographia, 1988, 26, 221. Raynor, M. W., Bartle, K. D., Davies, I. L., Williams, A., Clifford, A. A., Chalmers, J. M., and Cook, B. W., Anal. Chem., 1988, 60, 427. Arpino, P. J., Dilettato, D., Nguyen. K., and Bruchet, A.. J . High Resolut. Chromatogr., 1990, 13, 5. MacKay, G. A., and Reed, G. D., J. High Kesolut. Chroma- togr., 1991, 14, 537. BS 5852, Part 2: 1982, Fire Test For Furniture. British Standards Institute, Manchester. Bartle, K. D., Clifford, A. A., Hawthorne, S. B., Langenfeld, J. J.. Miller, D. J., and Robinson, R., J . Supercrit. Fluids, 1990, 27, 143. Cousin, J . , and Arpino, P. J., J. Chromatogr., 1987, 398, 12.5. Ryan, T. W., Yoklovich, S . G., Watkins, J. C., and Levy, E. J . , J. Chromatogr., 1990, 505, 273. Ashraf-Khorassani, M., and Levy, J. M., J. High Resolut. Chromatogr., 1990, 13, 742. Bartle, K. D., Boddington, T., Clifford, A. A., Cotton, N. J., and Dowle, C. J . , Anal. Chem., 1991, 63,2371. Hirata, Y., and Okamoto, Y., J. Microcolumn Sep., 1989,1,46. Kuppers, S . , Chromatographia, 1992, 33, 434. Hawthorne, S . B., Krieger, M. S . , and Miller, D. J., Anal. Chem., 1989, 61, 736. Paper 3100243H Received January 14, 1993 Accepted February 3, 1993