Anal. Chem. 1981, 53, 197-201
197
Fourier Transform Infrared Spectrometric Detection in Size-Exclusion Chromatographic Separation of Polar Synfuel Material R. S. Brown, D. W. Hausler, and L. T. Taylor” Department of Chemistry, Virginia Polytechnic Institute & State Unlvers&, Blacksburg, Virginia 2406 1
R. C. Carter Department of Chemical Sciences, Old Dominion University, Norfolk, Virglnia 23508
The complexlty of solvent-refined coal (SRC) is such that no slngle means of analysis can hope to achieve a substantial characterization of the whole SRC material. It Is therefore necessary to use a varlety of analysis schemes. Liquld chromatographic (LC) separatlon with conventlonal LC detectors is widely used but is dmicult to apply broadly because of the heterogeneous nature of coal. Infrared spectrometric detection offers high selectivity for the analysis of many organic functionalltles and slmpllfles the separatlon required. The use of Fourler transform infrared spectrometry (FTIR) allows slmultaneous monitorlng of multiple functlonalltles. Appiicatlon of Infrared detectlon to the size separation of various SRC fractlons prevlously separated on a silica column has been achleved with FTIR detection. Several characterlstlc absorptlons were observed in these samples and functionality assignments have been made for several SRC fractlons.
The several synfuel processes being developed to better utilize our coal resources require methods of analysis which can yield highly specific information when applied to extremely complex mixtures. The chemical nature of coal-derived materials such as solvent-refined coal (SRC) is so bewildering that preseparation procedures on the total product do not significantly simplify the mixture for informative analysis (1, 2). At this point, complete chromatographic resolution of the synfuel product has not been achieved for one or more of the following reasons: (a) no suitable chromatographic solvents, (b) inadequate inexpensive column packings, and (c) diversity of chemical entities in the product. Consequently,the application of monitoringtechniques which respond to specific chemical species in crude synfuel fractions appears to be a viable option for addressing this problem. A fluorine “tagging” method employing trifluoroacetyl chloride and I?F NMR has been perfected for the quantitative analysis of primary, secondary, and tertiary alcohols and variously substituted phenols ( 3 , 4 ) . Chromatography coupled with a highly specific, highly sensitive detector “on-line” is potentially capable of providing speciation information. Speed of analysis and a smaller (hopefullyfiner) analytical sample are apparent advantages of the latter “flow” method. The recent demonstration of lH NMR detection in the separation of coal-derived jet fuels on silica gel is significant in this regard (5). Our current work makes use of previously developed HPLC techniques for the size exclusion separation of SRC material (6). Detectors employing differential refractive index and ultraviolet spectrometry provide little information other than perhaps relative molecular weight distributions. Infrared spectrometry as a more selective detector system especially
for heteroatom organic species (e.g., ethers, phenols, ketones, acids, esters, lactones, etc.) is currently under consideration. To date our work (7) has employed a low-cost single-beamIR detector (Miran IA) for “on-line” analysis. With this detector system we were able to observe selective detection and molecular weight distribution of several heteroatom-containing functionalities. Limitations of this system include low wavelength resolution due to the band-pass filters employed to select wavelengths, multiple chromatographyruns required to monitor various wavelengths, and relatively low sensitivity. Fourier transform infrared spectrometry (FTIR) offers many unique advantages over conventional IR as an LC detector. Most important for the application to LC are the rapid scan rates over large spectral regions which permit monitoring numerous wavelengths during a single chromatography run. This fact, coupled with signal averaging capability, computer assisted data collection, and manipulation and operator selected resolution, makes FTIR an unique LC detector with much potential. Its only drawback with respect to conventional IR is its high cost which might preclude its use in many laboratory applications. Recently, HPLC-FTIR has been demonstrated in the separation of industrial waste water effluent (81, silicone polymers (9),dye components (IO),and parafin oils (11).Its application for the separation and characterization of coal-derived products “on-line” has not been reported up to now. In fact, the FTIR technique has only recently been applied (12) to coal-derived products. This involved the determination of functional groups present in three SRC samples obtained from the same parent coal but under different processing conditions and selected isolated chromatographic fractions of SRC from a silica gel column. A considerable degree of heterogeneity was observed in the chromatographic fractions, and further separation of the SRC was deemed highly desirable. This paper reports not only additional separation efforts aimed at these fractions but also describes (a) the employment of FTIR as a highly specific “on-line” detector and (b) the information which FTIR can provide on extensively separated coal liquids.
EXPERIMENTAL SECTION Samples. SRC samples derived from Indiana no. 5 feed stock were obtained from a Southern Services Inc. pilot plant (Wilsonville, AL) funded by Electric Power Research Institute and operated by Catalytic, Inc. A chloroform soluble fraction was generated by stirring in chloroform (40:lw/w) for 3 h followed by filtration. This was followed by a preliminary separation into rough chemical classes by a modification of the sequential elution selected solvent chromatography (SESC) procedure developed by Farcasiu et al. (13)in which the SRC material is sequentially eluted from silica gel by use of various mobile-phasecompositions. Of the nine fractions generated (14),SESC no. 5 , 7 , and 9 (designated basic nitrogen heterocycles, polyphenols, and increasing oxygen-nitrogen content and basicity, respectively) were chosen
0003-2700/81/0353-0197$01.00/00 1981 American Chemical Society
198
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981
r
CHCI, SRC
soluble
L
CHCI,
insoluble
SESC
105
Scheme I SESC no. 5 SESC no. 7 SESC no. 9
5
SEC-FTIR
90
f
75
1
I
SRC
because of their suspected high heteroatom content for analysis by LC-WIR. Appropriate amounts of each fraction were weighed and dissolved in chloroform in an attempt to prepare 1% solutions. Limited solubility of the samples in chloroform only permitted the examination of the resulting saturated solutions. The solubility of the separated material differs from the solubility of the total SRC. We have observed that lower SESC fractions have greater solubility in chloroform; whereas, the higher fractions are increasingly less soluble. Chromatography. Size exclusion chromatographyemployin a Waters 6000A reciprocating piston pump and a Waters 100p-Styragel column with 14000 plates/m was performed on all samples. A flow rate of 0.5 mL/min with chloroform as the mobile phase and a Nicolet flow cell (KBr, 1 mm pathlength, 7.5 pL volume) designed for use with the Nicolet FTIR system were empolyed. A differential refractometer (LDC Model 1107)was used as an auxiliary detector during the separation. A Nicolet 7199 FTIR spectrometer was empolyed as an LC detector. The system used a KBr beam splitter and TGS detector with a usable range of 5000-400 cm-l. An MCT detector system would have been desirable but was not available. A total of 4096 data points were transformed and a Happ-Genzel apodization function was applied giving a resultant resolution of -4 cm-l for stored spectra. Each stored spectrum (representing -28 s elution time or -0.25 mL of elution volume) was the result of averaging 32 individual spectra. Up to five IR window regions with 16-cm-’ resolution could be selected over which the total integrated absorbance can be plotted vs. time to produce a pseudo-real-time chromatogram. Concurrentlyindividual averaged spectra are filed whenever an operator predetermined threshold value is exceeded in any IR window region. The software employed in this study was that developed by the manufacturer and provided with the instrument.
-
1
RESULTS AND DISCUSSION Fourier transform infrared spectrometry (ffIR) has been employed as a liquid chromatographic detector in the size exclusion separation of selected CHCl,-soluble solvenbrefined coal (SRC) fractions. Indiana no. 5 feed coal served as the source of SRC. Crude separation of the sample into nine fractions (SESC no. 1-9) was achieved via separation on a silica gel column utilizing a constant elution volume (twice the bed volume) for each eluting solvent. SESC no. 5,7, and 9 dissolved in CHC1, provided the specific samples for SECFTIR. Scheme I outlines the history of these samples. Prior to size separation, FTIR spectra were measured on approximately 1% (w/w) CHC1, solutions of SESC fractions 5,7, and 9. The 4000-3200-~m-~ region (Figure 1) is similar for all samples. Bands at -3690 and 3600 cm-l dominate this part of the spectrum and are assignable to an 0-H stretching vibrational mode. The lower wavenumber value is no doubt due to nonbonded phenolic and/or alcoholic functionality. The band at 3690 cm-l is tentatively assigned to traces of moisture in the sample. Bonded 0-H absorbances which dominate the Nujol mull spectra (12) are of little consequence here since our spectra are of dilute solutions. Weak absorbance in the 3600-3400-~m-~ region is observed which may be due to bonded OH and/or N-H. While solvent interference obscures the aromatic C-H region, there appears to be a progression to less aliphatic C-H in the higher SESC fractions. Estimates of this type are difficult at this stage since the solubility of each fraction in CHC&is different. Each FTIR spectrum exhibits a sharp absorption around 1600 cm-l at-
4000
3600
3200
2800
2400
2000
1600
WAVENUMBER
Flgure 1. IR spectra of whole SESC fractions 5 7 , and 9 in CHCI:, employing Nicolet flow cell.
1200
-
800
1% (w/w)
tributable to aromatic ring vibrations. Weaker unresolved absorptions at slightly higher wavenumber are apparent in most all fractions and are likely due to low concentrations of a variety of carbonylic material. This becomes most pronounced in fraction no. 9 where now the major peak in the spectrum appears at 1800 cm-l. The 1200-900-cm-’ window demonstrates the most dramatic spectral changes in going to higher SESC fractions. Aliphatic ethers, aryl-alkyl mixed ethers, and alcoholic functionality exhibit absorption in this region. Considering the complexity of each fraction the assignment of individual bands to specific moieties is conjecture. Detailed interpretation of these and other spectral bands has been deferred to smaller cuts of further size-separated SESC fractions. As might have been expected, this procedure has provided considerably more detail concerning the speciation of each SESC fraction. Real-time size-exclusion chromatograms employing “aliphatic C-H” (C-Hdph) detection, “aromatic ring” (cm0) vibration detection, and “alkyl etheral” (C-0-C) detection for SESC fractions 5, 7, and 9 are reproduced in Figure 2. “Alkyl ethers” are most pronounced in no. 5 and occur in predominantly large size molecules. The distribution of C0 4 is similar in no. 5 and 7 whereas no. 9 contains practically no large size molecules possessing C-0-C moieties. The same is true of c-Hdph with no. 9 having few species of any size with C-Hdph. The c,, chromatograms show an interesting pattern with major elution in the K D = 0-0.5 region followed by a significant group of materials near the totally permeated region. Again the amount of detectable material in no. 9 is less. The relatively low solubility of no. 9 in CHC13 probably accounts for this observation rather than the absence of organic functionality. While this type of data provides an overall picture of the distribution of specific functionalities with respect to (a) individual samples and (b) molecular size for a specific sample, an examination of file spectra gives a more complete description of the eluting species.
-
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981 No. 5
INDIANA SESC
199
SOLUBLE
CHCL3
5
SESC
7
SESC
9
040 2980-2850m-'
005
-'Oo2 -010
,020
2980 - 2 8 5 0 ~ 6 ' 010
Y \lh
- 005 Oo2
.1
,017 OZ5
-
1630 157Ocm-' 005
1630
-
1570crf'
isL_
2970
1
- 289Ocm-'
025
-.005 .OIO
;;;j
.055 ,040
1630 -1580~16'
- 005 ,055
,040 II 5 5
-
97 5 06',005
1155
-.OlOC
0
,
,
I
600
1200
I800
975cm-I
- .0°2 005
L 0
TIME SECONDS 00
-
d-.-
600
1060 - 980cm-1 ,025
,005 010
le00
1200
1 c--.l0
TIME SECONDS 00
KD
10
IO
600
SECONDS 10
0.0
KD
I800
I200
TIME
KD
I
22
40
60
FILE
m
FILE No
No
+
O
FILE
No
Flgure 2. Selected window regions monitored during the chromatographic separation of SESC 5, 7, and 9. Conditions: flow rate 0.5 mL/min CHCI,, 100-pL injection, Nicoiet flow cell p-Styragel column (30cm X 7.8 mm i.d.).
Each file spectrum represented approximately 0.25-mL elution volume and was the result of averaging 32 individual scans. As Figure 2 attests major chromatographic peak maxima from the SESC no. 5 separation coincide with files 28,44, and 54 for the IR windows monitored. An examination of each file confirms the observation that no new information pertaining to functionality is supplied beyond what is revealed in the files noted above. Figure 3 illustrates the C-Hdph, C, and C-0-C regions for these files. The larger size fraction (file 28) contains appreciable methyl (2965 cm-l) and methylene (2930 cm-') along with aromatic (1588 cm-l) moieties which either may contain heteroatoms or have extended conjugation. The assignmentof the band at 1588 cm-l is based on the relatively high peak intensity. Dominant bands at 1153 and 1108 cm-' are assigned to ether functionalities. Aryl ethers are obscured by CHCls absorption;however, aliphatic, acyclic, and alkyl aryl ethers give rise to IR activity in the 11001000-cm-l region. The lack of IR bands above 3200 cm-l suggests that there is no contribution from large size alcoholic or phenolic materials which would otherwise also absorb in this region. Less intense bands at 1168 and 1015 cm-l may therefore be attributed to smaller concentrations of alkyl containing ethers and/or the symmetric C-0-C stretching vibration for aryl ethers, respectively. In going to intermediate size molecules (files 43 and 441, etheral has reappeared but it has somewhat different character. The band at 1108 cm-I (fide 43) msigned to aliphatic ethers is extremely weak whereas the C-H,, is surprisingly intense. This could be due to the presence of high molecular weight aliphatic or alkyl aryl ethers. The 1108-cm-l band intensifies and broadens on going to file 44 but the C-Hdph has essentially disappeared. This suggests a variety of mixed ethers with minimal aliphatic content. Bands at 1042 and 1033 cm-l may again arise from aryl ethers (symmetric stretch). This assignment is supported by IR activity around 1600 cm-l. The intense 1168-cm-l peak possibly arises hom an alicyclic ether such as benzofuran. The doublet at 1606 and 1588 cm-l may alternatively be due to pyridinic material. This would be consistent with the designation of SESC no. 5 material as basic nitrogen heterocycles. The only other significant absorbances in the regions that we are able to monitor elute in the near totally permeated region (file 54). Intense absorption is noted at 1606 cm-l with si-
il
INDIANA
No 5
SRC
CHCLa SOLUBLE SESC SEC
5
FILE
w
4
A
-
2960
2840
272C
No
54
44
43
lk
28
u '651
1573
1495
1183
1105
1027
949
WAVENUMBER
Figure 3. Portions of selected files collected during the separation of
SESC no. 5 showing various functionality absorption regions. Conditions are as in Figure 2. Resolution is 4 cm-'. multaneous onset of IR activity at 3600 cm-'. With the evidence presented in Figure 4 coupled with the absence of etheral and alcoholic bands in the C-0-C and C-Hdph region, we believe the eluting species to be phenols with traces of moisture present (3691 cm-I). File spectra 153 and 154 obtained during the size exclusion separation, Figure 5, of SESC no. 7 reveal carbonylic (1771 cm-l) material eluting with the larger size molecules. Its concentration is low since it is not readily observed in the static spectrum of SESC no. 7 and it only appears in a few file spectra. The high frequency band precludes simple aldehydes, ketones, amides, and esters. Aroyl halides absorb in the region of interest but they should not survive the SESC chromato-
200
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981 I N D I A N A No. 5 SESC 7
INDIANA No. 5 S R C CHC13 S O L U B L E SESC 5
3691
SEC
S EC FILE
54
r. 3800
I
,
,
3680
3560
3440
FILE
No
,
I
3320
3200
WAVENUMBER
Figure 4. OH/" region of file 54 collected during SESC no. 5 separation corresponding to small size material. Condltions are as in Figure 2. INDIANA No. 5 CHC13 SOLUBLE SESC 7 SEC
4
FILE
1
NO.
160
SRC
FILE
No.
1
FILE
No.
1166
1098
1032
1064
IO33
996
WAVENUMBER
Figure 6. C-0-C region of selected flies collected during the separation of SESC no. 7. Conditions are as in Figure 2.
u SESC
- -
166
153
1780
-
1760 cm-'
025
I
1790
1480
1790
1480
3800
3510
3220
WAVENUMBER
Figure 5. Portions of selected files collected during the separation of SESC no. 7 showing various functionality absorption regions. Conditions are as in Figure 2.
i:: 010
- 005 0
9
600
1200
TIME
SECONDS
00
KD
IO
LT1--7--$1 104
graphic separation. Anhydrides also appear around 1770 cm-l but they have symmetric and asymmetric IR-active modes. The band at 1771 cm-l is a singlet. Nonbonded acids would appear to be candidates. Even more likely would be cyclic carbonyl materials (Le., ketones, lactones, and lactams) of relatively small ring size. It should be noted that no carbonylic or potential carbonylic (e.g., THF) material has been employed in either the SESC or SEC procedures at this point. Little additional absorption is detected in the IR windows that can be monitored. Very weak, poorly resolved C-Hdph appears in files 153 and 154. (Most of the C-Hdph is seen in files 140-150.) The aromatic carbon region (CaJ exhibits little absorption as does the 0-H, N-H stretching region. Bands in the ether region which are unique only to files 153 and 154 are 1159 and 1038 cm-I (Figure 6). The assignment of one or both to C-0-C stretching modes in an a-lactone is tempting. Phenyl acetate appears to be an excellent prototype which exhibits absorbance at 1770 cm-l and around 1100 cm-'. Material eluting immediately after (files 155-159) the carbonylic species shows the familiar doublet around 1600 cm-l observed in the SEC of SESC no. 5. Again no 0-H, N-H activity and weak C-Hdph absorbances were found. The appearance of the doublet, however, is accompanied by a con-
is00
120
140 FILE No
Figure 7. Carbonyl window region monitored durlng the separatlon of SESC no. 7. Conditions are as In Figure 2. siderable number of new peaks in the C-O-C region (Le., 1151, 1067,1032, and 993 cm-l), Figure 6. Mixed aryl alkyl ethers are known to absorb in the 1075-1020-~m-~ region. SESC no. 7 has been previously designated polyphenols (13). No files except 166-168 show 0-H, absorbances. This feature, as was the case with SESC no. 5, is accompanied by the appearance of a 1607-cm-' peak suggestive of phenols. Larger size phenols appear to be absent. However, evidence from an independent model study has shown a non-size-exclusionmechanism for phenols (6). The original designation of SESC no. 7 as polyphenols was based on limited model studies with low molecular weight polyphenolic material (e.g., 2,3-dihydroxynaphthalene) (17). Several of the phenolic vibrational modes are obscured by CHC13absorbances (e.g., C-Oak). Furthermore, the possibility of intramolecular association may cause the O-Hat, region to be weak and broadened. Alternatively, the eluting material may be so broadly distributed that many species are undetected. Although we have been concerned with CHC13-solubleSRC, an appreciable amount of material still appears in SESC no.
ANALYTICAL CHEMISTRY, VOL. 53, INDiANA
No. 5 9
SESC SEC
117
118 I200
1132
1064
962
996
WAVENUMBER
Figure 8. C-0-C region of selected files collected during the separation of SESC no. 9. Conditions are as in Figure 2. INDIANA .
NO.
s
CHC13
SOLUBLE
SESC
9
SRC
SEC
FILE
No
NO. 2,
FEBRUARY 1981
201
Relatively sharp bands at 2922, 1464, and 1380 cm-l can be assigned to the asymmetric methyl C-H stretching band and the asymmetric and symmetric methyl bending vibrational modes, respectively. For files 115-118 the C-0-C region is also uniquely active with bands at 1178,1044, and 993 cm-’, Figure 8. The assignment of these bands to cyclic esters and/or mixed ethers seems appropriate as discussed in SESC no. 7. An examination of files 130-133 reveals phenolic material again of apparent small size, Figure 9. Surprisingly little C,, vibrations are observed in any of the files before the elution of phenols, Several points should be made regarding this investigation. Relative to more conventional detectors, FTIR coupled with chromatography yields a much clearer picture of the composition of a typical synfuel. Highly colored material elutes continuously from the column; yet, all files are not rich in IR bands. This observation could mean that considerable material which exhibits infrared absorption in several of the nontransparent regions obscured by CHC&is eluting from the column, or an insignificant part of the material which possesses extraordinarily high molar absorptivities is eluting. Clearly, quantitative elution data are required in this regard. The observation of the intermediate size carbonylic material which appears to be concentrated in SESC no. 9 (yet which appears in SESC no. 7) is a novel finding which must be extended to other synfuels, Significant “spill-over” from one SESC fraction to another apparently occurs and the chemical class designation of specific SESC fractions should be broadly interpreted. The feasibility of FTIR-LC for synfuel studies has been demonstrated. Clear differences in chemical functionality have emerged between different size fractions of SRC. The modeling of our FTIR-LC study with mixtures of known compounds and the application of synfuels to existing separation technology with FTIR detection is mandatory.
ACKNOWLEDGMENT The assistance of Edwige Denyszyn in the preparation of this manuscript is acknowledged. We thank Harry Dorn for helpful discussions regarding this investigation. The assistance of John Hellgeth in providing appropriate SESC fractions of SRC is appreciated.
+-AN----
-
LITERATURE CITED (1) Welsh, D. J.; Hellgeth, J. W.; Glass, T. E.; Dorn, H. C.; Taylor, L. T. ACS Symp. Ser. 1978, No. 71, 274. (2) Wooton, D.L.; Coleman, W. M.; Glass, T. E.; Dorn, H. C.; Taylor, L. T. Adv. Chem. Ser. 1978, No. 170, 37. (3) Sleevi, P. S.;Glass, T. E,; Dorn, H. C. Anal. Chem. 1979, 51, 1931. (4) Glass, T. E.; Dorn, H. C.; Taylor, L. T.; Manheim, A.; Sleevi, P. S., submitted for publication.
P
-
3900
A
A-
3600
3300
I800
115
1500
WAVENUMBER
Figure 9. Portions of selected files collected during the separation of SESC no. 9 showing various functionality absorption regions. Conditions are as in Figure 2.
9; even though, very few files exhibit IR activity. One reason for this no doubt is due to the fact that all of SESC no. 9 did not redissolve in CHC13for injection onto the SEC column. The 1770-cm-l band observed in select SESC no. 7 file spectra is much more prominent in the file and static spectra of SESC no. 9. A chromatogram employing 1780-1760-cm-’ detection is shown in Figure 7. The carbonylic material again elutes in a very narrow range of KD values corresponding to ‘‘C20alkane-size” molecules. This high-frequency carbonyl peak, as previously suggested, probably originates from a cyclic carbonyl. Several other IR bands track the 1770-cm-l band.
(5) Haw, J. F.; Glass, T. E.; Hausler, D. W.; Motell, E.; Dorn, H. C. Anal. Chem. 1980, 52, 1135. (6) Hausler, D. W.; McNair, H. M.; Taylor, L. T. J. Chromatogr. Scl. 1979, 17, 617. (7) Brown, R. S.;Hausler, D. W.; Taylor, L. T. Anal. Chem. 1980, 52, 1511. (8) Shafer, K. H.; Lucas, S. V.; Jakobsen, R. J. J. Chromaiogr. Sci. 1979, 17, 464. (9) Vklrine, D. W. J. Chrornafogr. Scl. 1979, 17, 477. (IO! Kuehl, D.; Grlfflths, P. R. J. Chromatogr. Sci. 1979, 17, 471. (11) Vldrlne, D. W,; Mattson, D. R. Appl. Spectrosc. 1978, 32, 502. (12) Painter, P. C.; Coleman, W. M. Fuel1979, 58,301. (13) Farcaslu, M. Fuel 1977, 58, 9. (14) Hausler, D. W.; Hellgeth, J. W.; Borst, J.; Cooley, B.; Taylor, L. T., accepted for publication In Fuel.
(15) Conley, R. T. “Infrared Spectroscopy”, 2nd ed.; Allyn & Bacon: Boston, MA, 1972. (16) Lambert. J. B.; Shurvell. H. F.; Verblt, L.; Cooks, R. G.; Stout, 0. H. “Organic Structural Analysls”; Macmillan: New York, 1976. (17) Whltehurst, D. D.; Farcaslu, M.;Mitchell, T. 0. EPRI AF-252, Research Project 410-1,“The Nature and Orlgin of Asphaltenes In Processed Coals”, 1976.
RECEIVED for review July 15,1980. Accepted October 30,1980. The financial assistance provided by the Commonwealth of Virginia is greatly appreciated.
