Anal. Chem. 1994,66,882-887
Feasibility of Extraction And Quantification of Fiber Finishes via On-Line SFE/FT=I R Cynthia Hume Kirschner, Sherl L. Jordan, and Larry T. Taylor' Department of Chemistry, Virginia Polytechnic Institute and State Universiv, Blacksburg, Virginia 2406 1 Paul D. Seemuth DuPont Company, Fibers Research, Kinston, North Carolina 28502
The method of on-line supercritical fluid extraction/Fourier transform infraredspectrometry (SFE/R-IR)has been applied to the analysisof fiber finisheson fibedtextile matrices. Three different fiber polymer types were examined, each requiring a different finish. Finishes ranged from a single-component poly(dimethylsi1oxane) oil to more complex multicomponent finishesthat included various surfactants,fatty acid esters and soaps,antioxidants, and oils. The three fiber types tested were polyurethane, polyamide, and aramid. Off-line extraction showed all three finishes to be over 89%extractablewith pure COz. Calibration curves were establishedfor the three finishes, with relatively low error and reasonable detection limits (Le.,
The technique would use no organic solvents and should be relatively fast and reproducible. The assessment of SFE/FT-IR has been made thus far using a simple hydro~arbon.~-~ Real world samples (such as fiber finishes) are inherently more complex. Thus, an investigationwas conducted to determine the potential of SFE/ FT-IR for fiber finishanalysis. Three fiber types with different finishes were chosen for the study. Goals of this project were first to determine whether the finishes were soluble in pure supercritical COz and, second, whether they could be quantitatively and reproducibly removed from fiber/textile matrices.
0.2-5 bg).
Many of the laboratory methods utilized in industry today are under review due to increased EPA regulation of organic solvents and solvent waste. Current fiber analysis methods typically require chlorinated solventsto remove the finish from the fiber. For example, fiber strands are extracted with tetrachloroethylene, and the resulting solution is then injected into an IR liquid cell for spectral analysis and quantification of the fiber finish. The development of a method which uses an acceptable, non-EPA-governed solvent would be beneficial (and eventually mandatory) in the fiber industry. Ideally, this new method should require no liquid solvent, but also be fairly rapid and provide reproducible quantitative/qualitative data for a given finish. The capability to separate the finish into its individual components for further analysis where necessary would certainly be advantageous as well. Many of the components used in fiber and textile finishing are low-polarity, high molecular weight compounds (i.e., waxes, surfactants, oils) and are therefore well suited to supercriticalfluid-based analysis. Despite this factor, however, literature citing the application of S F analysis to textile processing/finishing agents has been relatively This research has consisted mainly of the SFC analysisof individual components and SFE/SFC of finishes from fiber or textile matrices. On the bases of these early reports, it seemed highly probable that on-line supercritical fluid extraction/Fourier transform infrared spectrometry (SFE/FT-IR) using pure CO:, could satisfy the aforementioned new method criteria. (1 ) Analytical Supercritical Fluid Chromatography and Extraction; Lee, M. L.,
Markidcs, K. E., Eds.; Chromatography Conferences: Provo, UT, 1990. (2) Yocklovich, S. G.; Sarner, S. F.; Levy, J. M. Am. Lob. 1989, 21, 26.
882
Amslyticel Ctwmistty, Vol. 66, No. 6, Merch 15, 1994
EXPER IMENTAL SECTION The MPS 225 SFE/SFC system was purchased from the Suprex Corp. (Pittsburgh, PA). For any part of the study requiring chromatography, a 10 cm X 1 mm Deltabond Cyanopropyl packed column (dp= 5 pm) (KeystoneScientific, Bellefonte, PA) was inserted into the system. Otherwise, a 1 m X 0.010 in. i.d. stainless steel transfer line was employed in lieu of a column to provide ample time to begin FT-IR background data collection prior to peak elution. Stainless steel extraction vessels (0.18 mL) used in the study were purchased from Keystone Scientific. The Nicolet 710 SX FT-IR (Madison, WI) used for the study was equipped with a Nicolet SFC/FT-IR flow cell interface. The flow cell was thermostatically controlled at 35 OC, while the transfer lines were maintained at oven temperature (75-130 "C). Data were collected at 8-cm-l resolution. SFC-grade carbon dioxide was purchased from Scott Specialty Gases (Plumsteadville, PA). Other necessary gases, such as air and hydrogen for the FID and syphon-grade carbon dioxide for the cryotrap, were purchased from Airco (Roanoke, VA). Finished and unfinished fiber samples as well as the individual neat finishes were provided by the DuPont Co. Any necessary solvents were purchased from Fisher Scientific (Raleigh, NC). RESULTS AND DISCUSSION Three basic fiber types were studied, each employing a different finishfor textile processing. The basic fiber structures (3) Kirschner, C. H.; Taylor, L. T. Presented at the 4th International Symposium on Supercritical Fluid Chromatographyand Extraction, Cincinnati, OH,May 19-22, 1992. (4) Kirschner,C. H.;Taylor,L.T.;Seemuth.P. D. Prescntedatthe 1992Pittsburgh Conference and Expositionon Analytical Chemistryand Applied Spectroscopy, New Orleans, LA, March 9-12, 1992; Paper 1064. (5) Kirschner, C. H.; Taylor, L. T. Anal. Chem. 1993, 65, 7 8 .
0003-2700/94/03680882$04.50/0
0 1994 Amrican Chemlcal Soclety
-
3 0 - C - N9H - R - N H - C - 9 0-R
0 0
+C-R-C-NH-R~-NH+
Bramid:
Figuro 1. Chemlcal structures of three test fibers.
1. 1. 2. 3. 4.
5. 6.
Polyurethane poly(dimethylsiloxane)oil Polyamide glycerol triesters alcohol ethoxylates alcohol PO-EO blocked surfactants phosphites poly(ethy1eneglycol) derivatives fatty acid soaps
T a b 2. On-Unr Exirrctkn R l c o v o r h tor Thrn Ftkr F1nbh.s’
% recoverede fiberb
A
B
C
% SD/RSD
polyurethane
98.3 89.9
0.0 2.0
1.3/1.3 1.2/1.3
83.7
6.4
93.8 91.9 96.9 74.1 99.6 91.9 99.6 80.4 89.0 99.8 98.9 26.7 34.4
polyamided 1 2 3 4 5 6 aramid 1 2 3 4
0 SFE conditione: 76 O C , 360 atm COI, = 0.811 g / d , J . 2 mL/+ li uid COZflow rate for 10min, 8 mL of methylene chlonde tra p sjvent. b The numbers indicate the individual components of aa ordered in Table 1. A, f i t 10 min of SFE B, second 10 min of SFE C, total recovered from A + B. n * 3;60-76-mg sample size. Heated solvent collection vial required.
F a
n
Aramid
1. 2. 3.
substituted phenol C41.g triglycerides
sorbitol derivatives (2)
are given in Figure 1, and Table 1 shows the basic ingredients of the finishes used for each fiber. The polyurethane fiber was primarily coated with a poly(dimethylsi1oxane) oil finish, present on the fiber at about 3% by weight. The remaining two fibers were treated with much less finish than the polyurethane (Le., 0.5-1%). As shown in Table 1, the nylon polyamide fiber required more complex finishing agents. Kevlar aramid was treated with a multicomponent finish that consisted predominantly of triglycerides very similar to those found in natural oil. In order to determine the extractability of each fiber finish with pure supercritical carbon dioxide, off-line dynamic extractions of each finish employing a liquid solvent trap of methylene chloride were conducted. A sample of 50-75 mg of neat finish was measured into a 0.18-mL vessel and mixed with just enough (25 mg) Celite (diatomaceous earth) matrix to prevent pressurizing the finish out of the vessel rather than extracting it. The vessel was sealed, inserted into the system, and dynamically extracted for a set period of time (i.e., 10 min). During extraction, the effluent was passed from the heated vessel into a half-filled 15-mL collection vial outside the oven that contained methylene chloride. A small section (20 cm) of 50-pm4.d. fused silica with a slight end taper was used as a transfer line, which sufficiently restricted the flow to maintain the necessary back pressure. After extraction, the collection vial was removed. The recovered analytes were then determined gravimetrically by evaporating the solvent and subtracting the previous weight of the empty vial from its current weight (Table 2). These data cannot be directly related to on-line data due to changes in sample quantity, extraction time, and extraction method (dynamic vs static/ dynamic) but were used strictly to demonstrate the feasibility of using 100%C02 as the supercritical fluid. Integrity of the
6.6/6.3
r~
lma 2-
I
m a Pl(.
1
4mlr )?J
Flgure 2. SFC of polyurethane fiber flnish: condltbns In text.
extracted sample was determined by resolvating the extract in the vial and chromatographing the analyte solution. For all three fiber finishes, the extract was determined to be no different from the neat finish (Le., no visible decomposition) as judged by a comparison with the chromatograms of the original neat finishes. The chromatography of each finish is given in Figures 2-4. Each finish was analyzed at 75 OC, with an initial pressure of 100 atm C02 (2 min) and a final pressure of 400 atm C02 (ramped at 10 atm/min). In observing the chromatographic results, it is interesting to note that, for the multicomponent finishes, not all of the ingredients were seen in the chromatogram. This is due to the low percentages in which they exist in the finish, the number of peaks they produce when chromatographed (Le., a single ingredient vs a homologous series), and the relative solubility of each component in supercritical C02. An ingredient which produces a 20-peak homologousserieschromatogram will need to be 20 times more concentrated in a given solution than will a component which produces a single peak, assuming equal detector response. As can be seen in Table 2, the polyurethane finish proved to be the easiest finish for extraction (Le., most CO2-soluble). The remaining two finishes were more problematic for various AnelytlcalChemlstty, Vol. 66, No. 6, March 15, 1094
883
A
Flgure 3. SFC of polyamide fiber finish: condltlons in text.
n
B
I
C
Flgure 4. SFC of aramid fiber finish: condltlons In text.
reasons. Restrictor plugging was a constant problem for the polyamide finish. This dilemma was best solved by controlling the temperature of the solvent trap more effectively (Le,, positioning the solvent vial within a warm water bath). When this was done, the restrictor no longer plugged, and recoveries of 92 & 2% were attained. The heating bath technique did not improve the recovery results of the other finishes, however. The recoveries of the polyamide and aramid finishes were determined to be lower due to the low solubility of certain finish components in COz. Off-line dynamic SFE (350atm COz, 75 "C,1.2 mL/min liquid flow rate, 20 min) of the individual sorbitol-based components, for example, resulted in recoveries of only 27-30% (Table 2). This t y p of finish component is present in the aramid finish and in fact comprises 20-30% of finish. Other components (Le., fatty acid soap, certain copolymers) may be difficult to extract, but these are present at very low percentages in the finishes. Even given these limitations, the recoveries of all three finishes were still relatively high and it was decided that all were sufficiently soluble in C02 to pursue SFE/FT-IR analyses of each finish on fiber. For on-line analysis it is desirable to eliminate any trapping of the finish to reduce analysis time and maintain the simplicty of the instrumentation. The system was therefore modified to yield the flow paths shown in Figure 5 , but in lieu of a column, a 1 m X 0.010in. stainless steel line was installed for transfer of analytedirectly to the FT-IR flow cell. This system initially performs static extraction of the analyte, followed by 664
Ane&tIC%l CheMtty, Vol. 66,No. 6,Merch 15, 1994
Flgure 1. Statlc/dynamic direct SFE/FT-IR sctw"ec: A, equlilbration mode; B, statlc extraction mode; C, dynamic extractlon mode.
immediate dynamic extraction and transfer directly into the flow cell. By the incorporation of a static extraction step, the finish may first concentrate within a small quantity of fluid. The subsequent dynamic extraction will sweep the concentrated plug of fluid into the flow cell while extracting any further finish from the fiber. The only place any unsolvated finishshould therefore remain after the hyphenated experiment would be inside the vessel, which can be removed and cleaned after every extraction. For the initial testing of the system, only the polyurethane finish was used. Of all three finishes, it is the only single-ingredient finish, and it had the highest finish recoveries in the off-line studies. It was determined during the early stages of this study that flow rate plays a highly critical role in the optimization of direct static/dynamic SFE/FT-IR. The flow rate needs to be fast enough to extract analytes from the matrix quickly, yet not so fast that the IR signal becomes excessively noisy (approximately 0.18422mL/min liquid). In order to more effectively control the flow rate through the IR flow cell, the tapered fused silica restrictor exiting the cell was replaced with a 50-rm4.d. linear fused silica transfer line. This line was passed up through what was formerly an FID port. The FID assembly was removed, leaving only the heating block behind. This exit line was then connected to a high-pressure needle valve manufactured by Sno-Trik (Dibert Valve &
A
Own Flgm 6. Position of beck-pressure regulator for SFE/FT-IR system.
Fitting Co., Roanoke, VA.). The valve functioned as a backpressure regulator for the system and enabled any necessary changes in the flow rate through the vessel and cell to be made much more readily than with a fixed restrictor. Extending from the other side of the valve was a 12 cm X 50 pm i.d. piece of fused silica. The tip of this line was inserted back into the FID port so that the tip resided well inside the heated region. This prevented any line plugging as C02 exited the valve. The position of the valve and transfer lines is shown in Figure 6. A series of polyurethane finish standards in methylene chloride were made, and 0.1 8-mL extraction vessels were filled with unfinished polyurethane fiber matrix (about 2-3 mg). A known standard was then injected (80 pL) onto the fiber matrix bed, and the solvent was allowed to evaporate (about 2-4 h, ambient conditions). The vessel was then sealed and inserted into the system, and its contents were extracted using the static/dynamic configuration. Quantitation was performed via the FT-IR data. Usually, quantification would be done directly from the GramSchmidt reconstruction. This was not possible for the polyurethaneanalysis, however, due to the presence of several coextractives. This was demonstrated by reinserting the analytical column into the system and analyzing an extract from 1-2 mg of supposedly “unfinished” polyurethane fiber. The resulting chromatogram (Figure 7) showed that more than residual finish (seen as low concentrationpeaks between 4-10 min) had been extracted from the fiber. The identities of each unknown peak were determined from the IR data files (Figure 8). Peak A is a glycol derivative. The large peak (B) was determined to be low molecular weight polyurethane. Thus, the presence of these coextractives is a problem, since the quantities of these coextractives will vary greatly with the quantity of fiber used as a matrix. The coextractives seemed to be most problematic for the unfinished fiber. Extraction of finished fibers surprisingly showed only a small amount of glycol and a definite decrease in polyurethane. For this reason, finished fibers were preextracted for use as calibration matrices, rather than naked fibers which had not ever received finish.
Flgurr 7. SFE/SFC chromatogram of flnlsh-spiked “unfinidred” polyurethane flber. SFE conditions: 350 atm C02; 100 OC; 2 mL/mln llquU flow rate, 10 min; trap at -10 OC. SFC conditions: 150 atm Cor (0.5 min); ramp to 400 atm (10 atm/min); 100 OC; trap at 180 OC.
B
flgurr 8. FT-IR spectra of peaks A and B from Flgure 7.
In SFE/FT-IR there is nocolumn; therefore every extracted analytewill elute simultaneously if above its threshold density. The small amounts of polyurethane that could possibly be extracted with every run even from the preextracted fiber would therefore coelute with the finish, conceivably rendering the peak areas invalid. Thus, instead of using the GramSchmidt reconstruction (which denotes the total IR signal produced over the course of the run), a plot of the IR signal for a finish-selective frequency region over the course of the run was made. For the polyurethane finish, the Si-CH3 absorbance between 820 and 790 cm-1 was chosen, since the other dominant regions of the poly(dimethy1siloxane)spectrum are shared by the coextractant’s absorbancies. Examples of a GramSchmidt plot and the subsequent integrated transmittance plot for a 2-pg sample of polyurethane finish spiked onto preextracted polyurethane fiber are given in Figures 9 and 10. The presence of water in the analyte spectra seemed to be problematic for the direct static/dynamic extraction system. If intermediate trapping was used, the air present in the vessel is passed through the trap and vented to the atmosphere. In the direct static/dynamic extraction system, however, the air within the vessel is mixed slightly with C02 AnaWcal mmktty, Vd. 66, No. 6, March 15, 1994
885
A
ZOO
GSR PLUS EV
'
r - - . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 50 00 50 I 00 RETENTIONlTIYE IYIN) ibo 112 its 167 ifo 169 19s 168 z b i
:
Flgwr 0. &am-Schmidt reconstruction of SFE/Fr-IR analysis of polyurethane fiber finish: y-axis in units of vob/scan. 821- 791 C M - 1 RErONSTRuC'iOh
k Flgurr 10. Integrated transmittance plot (820-790 cm-l) from SFE/ FT-IR analysis of polyurethane fiber finish: y-axis in units of volts/ scan.
10
20
90
40
-0 r*Am W (n-4)
and then passed through the IR flow cell with the analyte. For this reason also it would be necessary to integrate the IR data over a specific wavenumber region in order to eliminate this water interference rather than quantitate directly from the GramSchmidt reconstruction. The peak area was determined from the integrated transmittance plot using a y-threshold of 20 V/scan. From these data, a calibration curve was successfully established for polyurethane finish. Using the propagation of errors technique,6the limit of detection (LOD) was calculated to be 3.7 pg for polyurethane finish, and the limit of quantitation was 12 pg. The RSDs for each of the four points were reasonable (5-98), with the highest error obtained at the low and high limits of the calibration curve. At high concentrations, the system may have been overloaded and restrictor plugging could have been an occasional problem. Error at the lowest concentrations was probably due to the smaller extracted plug sizegenerated, as noted for the previous Cz4 SFE/FT-IR ana lyse^.^ Since the extractability of the other polymer finishes with pure COZ was demonstrated in the off-line studies, direct extraction of their finish into the FT-IR flow cell was viewed as viable. For these experiments, solutions of the polyamide and aramid finishes (0.05-5.0 mg/mL) were made in methylene chloride, and 10 pL of a given standard was injected onto the appropriate preextracted fiber matrix. Four runs per calibration data point were made, with four points per calibration plot. The coextractive problem observed with the ( 6 ) Long,
888
G.L,: Wincfordncr, J. D.Anal. Chrm. 1983, 55, 712A.
AneWCal Ct"irv.
Vd. 66, No. 6, Much 15, 1994
I
Flgurr 11. Calibratbn plot of aramid flber finish vla SFE/FT-IR: top, straight line fit; bottom, curve flt.
unfinished polyurethane fibers was not a problem for any of the remaining two fibers. These fibers also required preextraction so that a known amount of finish could be applied in order to construct a calibration plot. Integrated transmittance plots were still used for quantitation due to the presence of water in the system. Each calibration plot was centered around the expected quantity of finish present on a 1-2-mg sample of the given finished fiber. For the polyamide finish, quantitation was performed with the v ( C - 0 ) stretch between 1120 and 1100 cm-l. The resulting plot was very linear (r = 0.999 93), with a calculated detection limit of 0.25 pg and a limit of quantitation of 0.84 pg. RSDs for each of the four points ranged from 6.5 to 21%, with the lowest error near the center of the testing range (i.e., 5-15 pg). The aramid finish did not produce as linear a calibration plot, however. As shown in Figure 11, the point placement resembled a curve more than a line, and the correlation coefficient (r = 0.992) further indicated this. RSDs for the aramidcalibration points ranged from 5.7 to 14%. For aramid fiber finish, integrated transmittancewas performed with the v(C-H) stretch at 292920 cm-I. Using the observed calibration curve data, the LOD for the aramid finish was calculated to be 4.6 pg with a LOQ of 15 pg. As demonstrated in the off-line SFE experiments, the aramid finish is not as readily extracted in pure COz as is the polyamide finish. Thus, the extract plug
Table 3. Comparkon ot SFEIFT-IR Results wlth Solvent ExtracLkn/IR Resutls for T h r n F l k r F1nkh.r Ullng the Statk/Dynamk Whod
fiber polyurethane polyamide aramid
avFOY*SD(%) SFE/FT-IRO solv extrac/IR* 1.46 f 0.102 0.243 0.022 0.770 f 0.080
2.09 f 0.19 0.422 f 0.014 0.888 0.083
n = 6. n = 10. FOY, percent fiiish on yarn.
tended to tail substantially. This forced the use of higher y-threshold values (Le., y = 30), which no doubt elevated the detection limits as a result. As a final means of method analysis, 2-10 mg of polyurethane, polyamide, and aramid finished fiber samples were analyzedvia SFE/FT-IR using the same conditions noted for generating each calibration curve. These results were then compared to the results obtained for the same batch of fiber analyzed by current DuPont Co. laboratory methods. The results of this comparison aregiven in Table 3. The resulting means and standard deviations were then compared. In all cases, a lower percent finish on yarn (FOY) is observed for the SFE data as compared to solvent techniques. This is not too surprising as organic solvents many times tend to extract components from a matrix more vigorously than supercritical COz and thus remove more of the oligomer and organic components present in the fiber. Analysis by the DuPont Co. of the polyurethane, polyamid, and aramid fibers extracted in our laboratory revealed little or no remaining finish on the fiber during subsequent liquid extractions and confirmed that SFE is an exhaustive means of removing a finish with little or no detriment to the character or makeup of the fiber.
In addition to carrying out the SFE/FT-IR experiments on the above-mentioned fibers it was desirable to determine possible matrix effects of fiber denier and composition. To accomplish this, a bulkcontinuous filament (BCF) polyamide yarn was used. The finish contained only two components, a dipotassium diacid salt and random EO-PO ester of a fatty acid. The experiment was carried out as previously discussed and quantitation performed using the v(C-H) stretch at 29452910 cm-l. The % FOY as calculated from this method was 1.5% as opposed to the 1.3% calculated from liquid solvent extractions. The RSDs for both methods remained slightly under 10%. Given these data, it was determined that the SFE/FT-IR method could be used for various fiber and yarn types and would be restricted by finish components rather than the matrix itself.
SUMMARY Solvent use is being rapidly curtailed in the United States; the development of methods excluding the use of such solvents bears both immediate and long-term rewards. The application of on-line SFE/FT-IR to fiber finish analysis is potentially an excellent means of achieving such rewards. All three finishes were readily extracted from an inert matrix with pure C02,as well as from their respective fiber matrices. RSDs for the calibration points of the three most extractable finishes ranged from 5.7% to 21%. SFE is a potentially "softer" extraction technique since it should remove less of the polymer from the fiber matrix than liquid solvent extraction. Research is underway to extend this technique to textile fabric and other textile matrix analytes. Received for revlew July 8, 1993. Accepted December 3, 1993.@ Abstract published in Advance ACS Abszracts, February 1, 1994.
Ana&t/mlChemisfry, Vol. 66, No. 6, Mrch 15, 1994
887
