Energy & Fuels 1994,8, 258-265
258
Characterization of Coal Liquefaction Heavy Products Using 2s2CfPlasma Desorption Mass Spectrometry John W. Larsen,* Andrzej R. Lapucha, Patrick C. Wernett,? and William R. Anderson Department of Chemistry, Lehigh University, Bethlehem, Pennsylvania 18015 Received July 1,1993. Revised Manuscript Received September 2 9 , 1 9 9 P
Californium plasma desorption mass spectrometry (PDMS) has been used to analyze heavy distillation residues obtained from direct coal liquefaction processes. The characteristics of the 2a2Cf PDMS technique for the analysis of these nonpolar materials were determined, especiallythe efficiency with which molecules of different chemical type are ionized and detected. The molecular weight distributions of several THF-soluble portions of nondistillable residual materials (850 O F + “resids”) obtained from the Wilsonville pilot plant were determined. These data are compared to results obtained by field ionization mass spectrometry (FIMS) and gel permeation chromatography (GPC). In general, number-average molecular weights for all three techniques agreed well. The molecular weight distributions for these resids produced under a range of conditions are quite similar. The separation of the resids into chemical classes by medium-pressure column chromatography (MPLC) on silica gel is irreversible.
Introduction Determining accurate molecular weight distributions for coal derived materials is a long-standing experimental problem. The materials to be analyzed are usually complex mixtures of molecules covering a very broad molecular weight range and a broad range of chemical types. In this situation, it is very difficult to separate only by molecular size. Association and adsorption complicate all techniques. Detection is also a problem because many detectors are responding differently to different chemical classes. The difficulties in measuring molecular weight distributions are probably less with coal conversion products than with coal extracts or other materials whose structures have not been simplified by removal of heteroatoms during hydrogenation. The range of functional groups has been greatly reduced and the molecular weight range narrowed from that of starting coal. The most convenient approaches to measuring the molecular weight distributions of these materials are gel permeation chromatography and a variety of mass spectrometries. The use of plasma desorption mass spectrometry to analyze heavy distillation residues obtained from direct coal liquefaction processes is the subject of this work. Field ionization and field desorption mass spectrometries have been applied to the analysis of coal liquefaction producta.1-8
* To whom correspondence should be addressed.
address: Speciality Minerals Inc., Bethlehem, PA 18017. Abstract published in Advance ACS Abstracts, November 15,1993. (1)St. John, G. A.; Buttrill, S. E.; Anbar, M. Organic Chemistry of CoakLaraen, J. W.,Ed.;ACS SymposiumSer. No.71;AmericanChemical Society: Washington, DC; pp 223-239. (2) Solomon,P. R.; Hamblen, D. G.; Best, P. E. DOE Quarterly Reporta, Contract No. DE-AC21-81FE05122, 1981. (3) Solomon, P. R.; Hamblen, D. G. In Chemistry of Coal Conversion; Schlosberg, R. H., Ed.; Plenum Press: New York, 1985; pp 128-129. (4) Aczel, T.; Laramee, J.; Hansen, G.J. R o c . 30th Annu. Conf. Mass Spectrom. Allied Top. 1982, 808-809. (5) Aczel, T.; Dennis, L. W.; Reynolds, S. D. Proc. 35th Annu. Conf. Mass Spectrom. Allied Top. 1987, 1066-1067. (6) Laraen, B. S.; Fenselau, C. C.; Whitehurst, D. D.; Angelini, M. M. Anal. Chem. 1986,58, 1088-1091. t Present @
252Cf-PDMS,a different kind of mass spectrometry,”14 has been incompletely explored for the analysis of coal products. The 252Cfplasma desorption technique utilizes energetic fission fragments from the decay of 2a2Cfto volatilize and ionize a solid sample. The ionization source is typically a 5-15-pCi 262Cfsample sandwiched between two thin sheets of nickel foil. To obtain a mass spectrum, a sample is dissolved in an appropriate solvent and the solution is deposited (often electrosprayed) onto an 0.8 cm diameter circular foil disk made of aluminized poly(ethy1ene terephthalate) (Mylar disk). 252Cf-PDMSis capable of producing mass spectra consisting primarily of molecular ions from a wide variety of compounds, including amino acids,9J7 glycolipids,lE peptides,19 proteins? nucleotides,20 and other natural products.21y22G e ~ p o r p h y r i n sfatty , ~ ~ acids,24-26polynu(7) Recheteiner, C. E.; Attoe, T. H.; Boduezynski, M. M. Proc. 33rd Annu. Conf. Mass Spectrom. Allied Top. 1985, 937-938. (8)Malhotra, R.; McMillen, D. F. “Characterization of Coal Liquids by Field Ionization Maee Spectrometry”; DOE Final Report, Contract No. DE-AC22-89PC89883,November, 1991. Malhotra, R.; McMillen, D. F. Energy Fuels,submitted for publication. (9) Macfarlane, R. D.; Torgeson, D. F. Science 1976,191, 920-925. (10) Macfarlane, R. D. Anal. Chem. 1989,55, 1247A-1264A. (11)Cotter, R. J. Anal. Chem. 1988,60,781A-793A. (12) Campana, J. E. Anal. Instrum. 1987,16, 1-14. (13) Macfarlane, R. D.; Hill, J. C.; Jacobs, D. L. Anal. Instrum. 1987, 16, 51-69. (14) Kamensky, I.; Craig, A. G. Anal. Instrum. 1987,16, 71-91. (15)Lytle, J. M.; Tinaey, - . G. L.; Macfarlane, R. D. Anal. Chem. 1982, 54, 1881i1883. (16) Zingaro, R. A.; Macfarlane, R. D.; Garcia, J. M.; Vindiola, A. G.; Zoeller, J. H. Chemistry of Coal Liquefaction; ACS Symposium Ser. Vol. 29, No. 5; American Chemical Society: Washington, DC, 1984, pp 22-30. (17) Bondarenko, P. V.: Zubarev. R. A.:. Knvsh. - . A. N.:. Rozvnov. _ .B. V. Org. Mass Spectrom. 1991,26, 765-769. (18)Jardine, I.; Scanlan, G.; McNeil, M.; Brennan, P. J.; Lindhart, R. J. Proc. 36th Annu. Conf. Mass Spectrom. Allied Top. 1988, 72-73. (19) &ai, M.; Demirev, P.; Fenselau, C.; Cotter, R. J. Anal. Chem. 1986,58,1303-1307. (20) Scanlan, G.; Benson, L.; Tsarbopoulos, A.; Jardine, I. Proc. 36th Annu. Conf. Mass Spectrom. Allied Top. 1988, 1289-1290. (21) Chait, B. T. Int. J.Mass Spectrom. Ion Processes 1989,92,297329.
oaa~-os24194125oa-o258$04.5~10 0 1994 American Chemical Society
Coal Liquefaction Heavy Products
Energy & Fuels, Vol. 8, No. 1, 1994 259
clear aromatic hydrocarbon^,^^ polyethersZ8 and other synthetic polymers2s36 have been analyzed by PDMS. Recently, PDMS was applied to ionic coordination comp o u n d ~and ~ ~crown ether-metal complexes.37 Gel permeation chromatography (GPC), originally designed for polymer a n a l y ~ i s has , ~ ~been * ~ ~used to measure molecular weight distributions of coal and crude oil The limits of the method and lack of calibration standard for coal-derived materials are wellkn0~11.41~46When analyses are performed using the same columns, solvent, and experimental conditions, the relative average molecular weight values can be satisfactorily compared.
spectrometer (AppliedBiosystemsInc., Foster City,CA) equipped with a PDP-11 microcomputer data acquisition system (DAS). The instrument has a 13-pCiZ2Cf source and a short 15-cmflight tube. We used a positive 10kV acceleration voltage and the time resolution was set to 1 ns/channel. Data collection for each spectrum was terminated when lo6 primary ion events were acquired. Using H+ and Na+ as calibrants, the mass accuracy of the instrument is 0 . 1 4 2 % of the measured mass. The mass resolution ( M / d M ) is typically below 500. With the time-offlight technique, the theoretical mass range is not limited by the instrument. The practical limit is set by sample to be analyzed. The simplest sample preparation technique for PDMS is to apply a droplet of the sample solution directly onto a support disk. In some cases the electrospray technique produces better spectra.sl Resid samples were dissolved in THF (10 mg/mL) and 40 pL of each solution was electrosprayed onto aluminized Experimental Section Mylar or nitrocellulose coated aluminized Mylar disks using the Bio-Ion electrospray apparatus. Thin homogeneous coverages Materials. Twenty-five THF-soluble portions of solid nonwere obtained. Aluminized Mylar and nitrocellulose coated distillable residual (850 OF+ “resids”) obtained from the Wilaluminized Mylar disks were obtained from Applied Biosystems sonville pilot plant during direct liquefaction of different coals Inc. and used without modification. were the subject of this study. They are brittle, pitch-like We developed a software package that can be used on any materials, supplied as -60-mesh brown powders. Despite difIBM compatible PC to calculate average molecular weights using ferent coal feeds and processing conditions, all the samples show data imported from the Bio-Ion 20 PDP-11 computer. In similarities. Tables 4 and 5 summarize physicochemical paaddition, this software package allows us to subtract the rameters of five selected resids. A complete list is supplied as background spectrum which comes from the Mylar film. The supplementary material. software package supplied with the spectrometer (BIO-ION 20 Plasma Desorption Mass Spectrometry. Mass spectra were Data Acquisition System) adds counts from the surrounding obtained using a BIO-ION 20 commercial time-of-flight mass channels to enhance the signal-to-noiseratio. This is undesirable for the determination of molecular weight distributions and has (22) Hunt, J. E.; Michalski, T. J.; Katz, J. J. Int. J. Mass Spectrom. been removed from data analysis. Ion Phys. 1983,53,335-336. Gel Permeation Chromatography. The molecular weight (23) Wood, K. V.; Bonham, C. C.; Chou, M. M. Energy Fuels 1990,4, 747-748. distributions of five selected THF-soluble resids (no. 7, 13, 16, (24) Zingaro, R. A.; Vindiola, A. G.; Zoeller, J. H. Int. J.Mass Spectrom. 19,and 22 provided by CONSOL Inc.) dissolved in pyridine have Ion Phys. 1983,53,349-352. been measured by a computer-interfaced gel permeation chro(25) van Veelen, P. A.; Tjaden, U. R.; van der Greef, J. Int. J. Mass matography with a mass-sensitive detector. The components of Spectrom. Ion Processes 1991,110,93-101. (26) Bolbach, G.; Viari, A.; Galera, R.; Brunot, A.; Blais, J. C. Int. J. the instrument were all commercially available. It consists of a Mass Spectrom. Ion Processes 1992, 112, 93-100. Waters liquid chromatograph (Model ALL 201) equipped with (27) Zoeller, J. H.; Zingaro, R. A.; Macfarlane, R. D. Int. J. Mass a Rainen loop injector and four p-styragel columns (1@+ 109 + Spectrom. Ion Processes 1987, 77, 21-30. 500 + 100 A). The mass detector (Applied Chromatography (28) Chait, B. T.; Shpungin, J.; Field, F. H. Znt. J.Mass Spectrom. Ion Processes 1984,58, 121-137. Systems Ltd., Model 750/14) works by nebulizing the liquid (29) Jordan, E. A.; Macfarlane, R. D.; Martin, C. R.; McNeal, C. J. Int. stream emerging from the columna and then passing this uniform J. Mass Spectrom. Ion Phys. 1983,53, 345-348. distribution of droplets down a heated tube in a flow of dry (30) Macfarlane, R. D.; McNeal, C. J.; Martin, C. R. Anal. Chem. 1986, nitrogen, evaporating the solvent. The resulting solid passes a 58,1091-1097. (31) Gehrig,C.C.;Wood,K.V.Proc.38thASMSConf.MassSpectrom. light scattering detector whose output is directly proportional to Allied Top. 1990, 255-256. the mass of solid material passing in front of it. Between the (32) Feld, H.;Leute, A.; Zurmuhlen, R.; Benninghoven, A. Anal. Chem. mass detector and the columns is a digital thermal pulse flow 1991,63,903-910. meter. The outputs from the mass-sensitive detector and the (33) Quinones, L.; Schweikert, E. A. J. Vac. Sci. Technol. (A) 1988,6, 946-949. flow meter are directed to a Zenith Model 158 microcomputer, (34) Park, M. A.; Cox, B. D.; Schweikert, E. A. J. Vuc. Sci. Technol. the former through a 12-bit resolution A to D converter (Data (A) 1991,9,1300-1306. Translation Model DT 2805). The programs for data collection (35) Lacey, M. P.; Keough, T. Anal. Chem. 1991,63, 1482-1487. and manipulation were written in Asyst (McMillan Software Co.). (36) van Veelen. P. A,: Tiaden. U. R.: van der Greef.. J.: Haee. R. Ora. Mass kpectrom. 1991,26,f4-80.’ The instrument was calibrated as previously described with (37) Malhotra, N.; Roepstorff, P.; Hansen, T.; Becher, J. J.Am. Chem. commercially available polystyrene standards (Polymer LaboSOC. 1990,112, 3709-3710. ratories Ltd., UK).” HPLC grade pyridinewas used as the eluting (38) Collins, M. J.; Zeronian, S. H.; Marshall, M. L. J.Macromol. Sei.solvent with a flow rate of 0.96-1.00 mL/min. A 50-pL volume Chem. 1991,28A, 775-792. (39) Sanayei, R. A.; Odriscoll, K. F. J. Macromol. Sci.-Chem. 1991, of each sample with concentration of 10 mg/mL was loaded 28A, 987-1000. through the 100-pLloop of the injector valve. Data for each run (40) Wong, J. L. In Coal Science and Chemistry; Volborth, A., Ed.; were collected for 60 min. Elsevier: Amsterdam, 1987; pp 461-471. (41) Bartle, K. D.; Mills, D. G.; Mulligan, M. J.; Ameachina, J. C.; Medium-pressure liquid chromatography (MPLC) was Taylor, N. Anal. Chem. 1986,58, 2403-2408. used to separate the THF-soluble portions of resids into the (42) Bartle, K. D.; Mulligan, M. J.; Taylor, N.; Martin, T. G.; Snape, following compound classes: “oils”, “asphaltenes”, and “preasC. E. Fuel 1984,63, 1556-1560. phaltenes”. These chromatographically separated fractions will (43) Larsen, J. W.; Choudry, P. J. Org. Chem. 1979, 44, 2856-2859. differ from those isolaed by solubility differences so quotes are (44)Larsen, J. W.; Mohammadi, M.; Yiginsu, J.; Kovac, J. Geochim. Cosmochim. Acta 1984,48,135-141. used to avoid confusion. The chromatography system used for (45) Larsen, J. W.; Wei, Y. C. Energy Fuels 1988,2, 344-350. these separations consists of a Kontes Flex column (2.5 X 20 cm) (46) Buchanan. D. H.; Warfel, L. C.; Bailey, S.; Lucas, D.Energy -.Fuels equipped with a flow adapter, Teflon tubbing, an FMI (Fluid 1988,2,32-36. Metering Inc.) lab pump model RPSY with l/4-in. piston, and an (47) Himmel, M. E.; Tatsumoto, K.; Grohmann, K.; Johnson, D. K.; Chum, H. L. J. Chromatogr. 1990,498, 93-104. ISCO Model 1220fraction collector. Silica gel (Merck Grade 60, I
(48) Khan. 2.H.: Hussain, K. Fuel 1989. 68,1198-1202. (49) Reynolds, J..G.; Bigga, W. R. Acc. Chek. Res. 1988,21,319-326. (50) Evans, N.; Haley, T. M.; Mulligan. J.: Thomas. K. M. Fuel 1986. 65,694-703.
(51) McNeal, C. J.; Macfarlane, R. D.; Thurston, R. J. Anal. Chem. 1979,51, 2036.
Larsen et al.
260 Energy &Fuels, Vol. 8, No. 1, 1994 Table 1. Number-Average Molecular Weights Obtained from Field Ionization Mass Spectrometry and Plasma Desorption Mass Spectrometry
Mn
0
resid no.
FIMS
PDMS
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
600 600 600 620 580 600 620 600 620 610 600 580 570 570 610 590 610
590 580 610 610 610 590 600 590 590 600 610 570 570 570 600 590 610 610 580 590 620 600 620 590 590
-n
-
Not available.
230-400 mesh, 60 A) and HPLC grade solvents hexane, benzene, and pyridine (Aldrich) were the stationary phase and eluents. Thirty grams of silica gel was dispersed in 80 mL of hexane and this slurry was put into the column. To pack the stationary phase, 500 mL of hexane was pumped through the column at a flow rate of ca. 6-8 mL/min; 700-800 mg of the resid samples was dissolved in 2 mL of THF and deposited carefully on the top of the column. Elution was made with sequential 250-mL amounts of hexane, benzene, and pyridine (flow rate of 6-8 mL/min). Three different colored fractions, one from each solvent, were collected. Solvents were removed using a rotary evaporator and aspirator vacuum, and residues were vacuum oven dried. At least 98.6% of all samples was recovered. Gas Chromatography. A Hewlett Packard 5880 %, gas chromatograph equipped with FID detector, split injection port, and an intermediate phase polarity Supelco SPB-20 capillary column (15 m long,0.25 mm ID, 0.25 pm film thickness) was used for the qualitative comparison of the original resids and samples prepared by remixing chromatography separated fractions (recreated "originals"). Samples dissolved in THF were injected using the following conditions: (1) carrier gas and flow rate: He, 10 mL/min; (2) oven temperature: 100-250 "C programmed at 10 "C/min; (3) injection port temperature: 250 "C; (4) detector temperature; 250 "C; (5) split ratio: 1OO:l.
Results and Discussion
Twenty-five THF-soluble samples of solid 850 OF+ materials ("resids") obtained from the Wilsonville pilot plant during direct liquefaction of different coals were analyzed by zszCfplasma Desorption Mass Spectrometry. Molecular weight distribution curves were obtained and number-average molecular weights (M,) were calculated from these curves. These data are compared to results obtained on the same samples by Malhotra and McMillen using field ionization mass spectrometry (see Table 1).* The optimum PDMS conditions were established using resid 7. The amount of material deposited on the Mylar disk, the type of support, the addition of NaCl to enhance
the spectrum intensity, the accelerating voltage, and the number of counts per spectrum were varied. Disks were coated by electrospray using 40 p L of THF solutions of the analyte (resid 7) having concentrations of 10 and 20 WgIpL, respectively. The spectra show only very small differences in the intensity of the peaks. As is known from the literature, several matrix materials improve the molecular ion yield compared to direct desorption from a m e t a l f ~ i l The . ~ ~most ~ ~ ~commonly used matrix is nitrocellulose. Solutions of resid 7 were electrosprayed onto Mylar and onto nitrocellulose-coated Mylar disks. Mass spectra of the analyzed resid obtained using nitrocellulose coated Mylar did not show any significant improvement in ion intensities compared to uncoated Mylar. When polymers are applied directly to targets, they often give poor peak shapes and low responses. Addition of sodium chloride to the polymer solution prior to electrospraying increases the number of sodium attachments, enhancing the overall response.31@ This approach was tried with resid 7. Unfortunately, mass spectra with and without NaCl were almost identical. However, the presence of sodium chloride improves the instrument calibration because signal produced by trace amounts of Na+ is used to calibrate the mass spectrum. Increasing the acceleration voltage from +10 to +15 kV gave a peak intensity enhancement of about 25 % . Negative ion PDMS spectra of resid 7 recorded at -10 kV, showed poor peak shapes and a response about 80% lower than the corresponding positive ion spectrum. On the basis of the above observations, we chose the following routine operation conditions: (1)nitrocellulosecoated aluminized Mylar disks as sample holders; (2) 40p L (10 mg/mL in THF) samples with a few crystals of NaCl; (3) acceleration voltage +10 kV; (4)acquisition of lo6 counts (time 13 min). Figure 1shows the plasma desorption mass spectra of Mylar film (a) and of resid 7 (b) in the range between 1 and 5000 mlz. The Gaussian shape of the curve is characteristic of PDMS spectra of recycle resids and coal extracts. These mass spectra were sent via KERMIT communication software to a Zenith 386 20 MHz PC computer and converted from binary to a format readable by Statgraphics by our home-made software package (called CFINT). Statgraphics was used to calculate the number-average and weight-average molecular weights. In addition, CFINT allows the subtraction of data files making it possible to subtract the background which comes from Mylar from a resid spectrum. Statgraphics was also used to smooth by simple moving average (five points on either side of target) and to plot the molecular weight distribution curves. Figure 2 showsthe software converted spectrum of resid 7 which includes the Mylar background (a) and of resid 7 with the Mylar background subtracted (b). The following equation was used to calculate numberaverage molecular weights from the PDMS molecular weight distribution curves, where Ni = ith ion intensity
-
(52) Macfarlane, R. D. Acc. Chem. Res. 1982, 15, 268-275. (53)Jonsson, G. P.; Hedin, A. B.; Hakansson, P. L.; Sundqvist, B. U.; Save, G. B. S.; Nielsen, P. F.;Roepstorff,P.;Johansson, K. E.;Kamensky, L.; Lindberg, M. S. L. Anal. Chem. 1986,58, 1084-1087. (54) Blak,J. C.;Viari,A.;Cole,R.B.;Tabet,J.C.Znt.J.MassSpectrom. Ion Processes 1990, 98, 155-156.
Coal Liquefaction Heavy Products
Energy &Fuels, Vol. 8, No. 1, 1994 261
0
1.2
0.8
0.4
2 (X 1000)
1.6
mass
400 Lo k
3 0
300
1
200
-
100
-
u
O LL
i
0
. . .
l . 0.4
.
.
l . 0.8
l
.
l l 1.2
mass
.
.
l . 1.6
.
.
I 2
i
I
( X 10001
Figure 2. Software-converted PDMS spectrumof resid 7 which includes the Mylar background (a) and of resid 7 with the Mylar background subtracted (b).
Figure 1. PDMS spectra of the Mylar background (a) Ad of resid 7 (b).
and Mi = ith ion mass:55@
M, = ~ N i M i j ~ N i 1
i
In Table 1, PDMS results are compared with those obtained from FIMS.8 The PDMS M, values are essentially identical to those from FIMS. Careful analysis and comparison of the corresponding PDMS and FIMS mass spectra show similarities in the mass range of 100-1200 mlz. Below 100 mlz the PDMS spectra are disturbed by remnants of peaks from the aluminized Mylar disks (mainly aluminum oxide cluster ions from the film).57This is due to imperfect subtraction of these intense peaks. There is a significant difference between the results of these two techniques above 1200amu. The FIMS spectra do not show any peaks with molecular weights higher than 1200-1300 mlz (Figure 3). The PDMS molecular weight distribution does not reach the base line until 2000-2500 mlz. There are three possible explanations for that. First, PDMS is detecting a significant population of high (55) Yon, W. W.; Kirkland, J. J.; Bly, D. D. Modern Size-Exclusion Liquid Chromatography; John Wiley & Son: New York, 1979;pp 8-9. (56) Billmeyer, F. W. Textbook of Polymer Science, 2nd ed.; John Wiley & Son: New York, 1971;p 78. (57)Juvet, R.S.;Allmaier, G. D.; Schmidt, E. R. Anal. Chim. Acta
1990,241,241-248.
molecular weight species missed by FIMS because it is not volatile. We do not believe this is occurring to a significant extent. Second, and probably the dominant process, the instrument is detecting so-called uncorrelated events and the base line actually never reaches zero. Up to 3% of the events measured with the BIO-ION 20 spectrometer are uncorrelated. These uncorrelated events occur because of the random probability that a second 252Cffission will occur during the time-of-flight measurement and the products from this second fission will always be recorded a t a higher mass.% If a second 252Cfdisintegration occurs while the ions are in the flight tube (an uncorrelated event), an artificially high mass will be recorded by the instrument. Because of the artificially elevated baseline caused by this, calculated weight-average molecular weights (M,)are unreliably high and are not included in our discussion. Finally, the PDMS ionization process may be producing some dimers.24927 Next, the PDMS results were compared with those obtained using gel permeation chromatography. Several analyses of commercially available polystyrene samples of different molecular weight were performed to confirm reliability of our gel permeation chromatography system. The data in Table 2 show generally fair to good agreement with the reported polystyrene molecular weights. Exceptional is the case of polystyrene of MW 90 000. The GPC results are much lower than expected. The data demonstrate that the system can be safely used up to 15 OOO amu. All the recycle oil samples contained material of molecular weight lower than 3500. Polystyrene has not been considered to be an adequate standard for coal(58)Wolf, B.; Macfarlane, R. D. J. Am. SOC.Mass Spectrom. 1992,3, 706-715.
262 Energy & Fuels, Vol. 8, No. 1,1994 I. 2
Larsen et al.
.
-
--
"95:
Figure 3. FIMS spectrum of resid '7 (from ref 8). Table 2. Number-Average and Weight-Average Molecular Weights Obtained from Gel Permeation Chromatography and Plasma Desorption Mass Spectrometry sample polystyrenea MW 580 MW 1240 MW 2350 MW 3600 MW 15000 MW 90000
resid 7 resid 13 resid 16 resid 19 resid 22
Mn 470 900
1900 3020 6950 26600
GPC Mw
MwfMn
M,
Mw Mw/Mn
570 1140 2240 3420 15050 28280
1.2 1.3 1.2 1.1 2.2 1.1
500
590
PDMS
600
600
570 610 500
570
700
1.2
590
580 600
a Polystyrenestandards. Their given molecularweightsare weight averages.
derivedmaterials because of differences in molecular shape and functionality between coal products and polystyrene. It has not proved to be an adequate standard for coal extracts. The problems with polystyrene standards increase with the molecular weight of the materials in question. The resids which are being studied here are of relatively low molecularweight and have had most of their heteroatom content removed. Both of these factors should improve their fit to the polystyrene standard. The improvement is such that polystyrene seems to be an adequate standard. The following formula was used to calculate molecular weights from the GPC curves, where hi is the GPC curve height at the ith volume increment and Mi the molecular weight of the species eluted at the ith retention volume: 55,s
In plasma desorption mass spectrometry, nonpolar polymers, especially polystyrene, give very low responses and/or poor peak shapes. We were able to obtain satisfactory PDMS data only for polystyrene of MW 580. Polystyrenes with higher molecular weights gave no peaks in the expected range, even in the presence of alkali metal
cations,59 but did give peaks in the "fingerprint" region between 0 and 200 m/z.60*61 Mn values of the five resids (no. 7, 13, 16,19, and 22), selected for chromatographic separation are also collected in Table 2. Results obtained from GPC are compared with those from PDMS. GPC Mn values are similar to those from PDMS and also from FIMS (Table 1). The good agreement between two mass spectrometric techniques and GPC in measuring of Mn is gratifying. If number-average molecular weights are desired, any of the three techniques is satisfactory. A central question for the use of PDMS is the possible variable response of the technique to different chemical classes. Will a Clo aliphatic compound behave similarly to a Clo aromatic component? Specifically, will the volatilization and ionization efficiencies of different compound classes (e.g., aliphatic vs aromatic) be the same? If they are not, accurate molecular weight distributions for their mixtures will not be obtained. To address this issue, we separated several resids into chemical classes and studied them individually. Our initial plan was to sum the spectra of the individual fractions weighted by their relative amountsand compare the resulting spectrum with that of the unseparated resid. This will reveal any differences in ionization or desorption efficiencies. Medium-pressure liquid chromatography was used to separate five carefully selected recycle oils wmples into the following categories: "oils", "asphaltenes", and "preasphaltenes". The followingfractions were obtained: yellow/ orange oils (hexane), brownish/red waxes (benzene), and black solids (pyridine). Yields of the corresponding fractions are presented in Table 3. Studies comparing the original resid with the resid reconstituted by mixing the separated fractions show differences between them (see Table 6). The chromatographic separations made permanent changes in the structure of these recycle oils. PDMS results are generally similar to GPC data, especially for "oils" and "asphaltenes". "Preasphaltenes" give higher average molecular weights in GPC than in PDMS. (59) Demirev, P.;Fenyo, D.; Sundqvist,B. U. R. Org. Mass Spectrom. 1991,26,471-475.
(60)Park,M. A.;Gibson,K. A.;Quinones,L.;Schweikert,E. A.Science 1990,248,988-990. (61) Park,M. A.; Schweikert,E. A.; da Silveira, E. F. J. Phys. Chem. 1992, !%, 3206-3210.
Coal Liquefaction Heavy Products
Energy & Fuels, Vol. 8, No. 1, 1994 263
Table 3. Composition of Five Recycle Oils (mg) Based on MPLC Separations hexanebenzenepyridinetotal sample no. soluble fracn soluble fracn soluble fracn yield (%) 7 (730) 13 (700) 16 (800) 19 (780) 22 (820)
270 210 310 260 260
230 250 250 290 250
220 230 230 220 300
98.6 98.6 98.8 98.7 98.8
Table 4. Origin of the Analyzed Samples resid no.
coal feed
source/conditions
7 13 16 19 22
Illinois No. 6 Texas Lignite Pittsburgh No. 8 Wyodak Wyodak
catalytic/catalytic thermal/catalytic catalytic/catalytic thermal/catalytic; H/La catalytic/thermal, L/Ha
a H/L= high/low; L/H= low/high (first stage/second stage reactor temperature).
Table 5. Analysis of the Soluble Portion of the Distillation Resids. wt%
C H
N S 0 condensed aromatics uncondensed aromatics cyclicalpha alkyl alpha cyclic beta alkyl beta gamma phenolic-OH conc. a
resid 7
resid 13
resid 16
88.75 89.39 90.54 8.15 7.12 6.74 0.91 0.66 1.23 0.14 0.09 0.39 1.42 2.30 2.17 Proton Distribution (%) 22.4 26.1 16.7
resid 19
resid 22
90.90 6.32 1.11 0.03 1.64
89.46 6.69 1.13 0.05 2.67
32.2
22.5
4.4
4.3
5.4
5.1
5.1
17.9 9.4 18.6 20.4 12.5 0.50
19.2 9.5 15.9 18.5 10.3 0.71
18.9 9.4 14.8 14.6 10.8 0.69
18.1 9.2 13.6 13.9 1.8 0.67
16.8 9.3 14.4 21.2 10.8 0.89
Data provided by Dr. Sue Brandes-ee
supplementary material.
Gas chromatographic analyses of the five selected original resids showed one broad peak with a retention time of 28.7 min. “Oils” and ”asphaltenes” gave one sharp peak at 24.9 min. “Preasphaltenes” did not elute within 60 min at 250 “C. The fractions were remixed to recreate the “original” material. This remixed “original” gave one peak at 24.9 min, significantly different from the virgin original material. Gas chromatography does not detect the entire sample. It does demonstrate that a significant portion of the sample is changed by the separation, and therefore the whole resid must be altered in some way by the chromatographic separation. This rough test indicates that the liquid chromatographic separations made some permanent changes in the structure of original resids. Structural data on chromatographically separated resids should be used cautiously. 252Cf PDMS and GPC results for the five fractionated resids are collected in Table 6, Figures 4(a-d) and 5(a,b). Analysis of remixed fractions after 4 weeks standing did not show any significant changes, despite the fact that a little precipitation was observed. In all cases, the remixed fractions have a lower molecular weight distribution ( M , values) than the original. There are three possible explanations: (1)disruption of intermolecular non-covalent association; (2) covalent bond cleavage; if noncovalent associations were being disrupted, they should re-form with time; there is no evidence for this; (3) removal of a minor polar component through irreversible adsorption
Table 6. Number-Average Weights Obtained from Gel Permeation Chromatography and Plasma Desorption Mass Spectrometry
Mn resid no.
GPC
PDMS
7 7Ha 7Bb
600 210 360 1330 560 570 170 330 960 400 610 150 350 990 510 500 120 260 730 310 700 170 310 1120 570
600 250 330 790 520 570 160 320 710 420 590 150 350 760 520 580 140 260 720 340 600 210 320 750 620
1P‘ 7HBPd 13 13H 13B 13P l3HBP 16 16H 16B 16P 16HBP 19 19H 19B 19P 19HBP 22 22H 22B
22P 22HBP
a H = “oils”. b B = “asphaltenes”. e P = “preasphaltenes”. d HBP = remixed fractions.
on the column, a component which associates strongly (probably bridges) with nonpolar components increasing the measured molecular weight. We do not have data permitting an unequivocal choice between these three, although the last seems most probable. As can be seen in Table 6, molecular weights of resid 19 are much lower than the corresponding fractions of the other samples. This sample has the highest percentage of condensed aromatics and is low in sulfur and beta and gamma protons (Table 5). For this sample PDMS and GPC M , values are in overall good agreement. Resids 7 and 22 contain relatively low amounts of condensed aromatics, large amounts of cyclic and alkyl beta and gamma protons, and the highest percentage of oxygen. These resids have much higher molecular weights, especially for the pyridine-soluble and remixed fractions. Comparison of PDMS molecular weight distributions to the GPC and FIMS shows that number-average molecular weights generally agree well. For hexane and benzene solubles and remixed fractions, both techniques agree well. Both the “preasphaltene” and the original resid samples give smooth PDMS molecular weight distributions with no individual peaks standing out. The “oils” and “asphaltenes” fractions show sharp, intense peaks at mlz 208, 224,254,266,282,392, etc. These peaks are also present in the spectra of remixed fractions as seen in Figure 6(a,b). A series of prominent peaks between 240 and 340 Da were also observed in FIMS data and assigned to benzoperylene derivatives.8 We cannot distinguish between two possible explanations: (1) Chromatographic separations change the recycle oils in a way which makes their individual components more accessible for PDMS analysis. This could be irreversible dissociation or bond cleavages. (2) New molecules are chemically generated by reaction on the silica gel.
Larsen et al.
264 Energy &Fuels, Vol. 8, No. 1, 1994
i
500
’
‘
-
-
I
7
tm
tLo
5
5
u 0
V 0
0
0.4
0.8
1.2
1.6
mass
0
2 (X ioooi
0.8
0.4
1.2
1 6
2
mass
(X 100Ol
Ln c z
3 u 0
0
0.4
0.8
1.6
1.2
mass
2 ( X 100Ol
0
0.8
0.4
1.2
1.6
2
mass
(X 1000)
Figure 4. Software-convertedPDMS spectra of “oils”(a),“asphall;enes” (b), “preasphaltenes” (c),and remixed fractions (d) compared to the original resid 7. 2000
1 I
0
b,
4
i
I5O0 h
400
1000 -
1
500
t
200
-I 240
260
320
360
400
mass ( X 100) I
,
-T
1500 -
L
h loo0
1
500 -
,-0-
7 -
0
I
3
I
5
G
L
log M Oiiginal Samplc
Rcmixcrl
rr n c l i o n s
Figure 5. GPC molecular weight distribution of “oils”, “asphaltenes”, and “preasphaltene” fractions (a), and remixed fractions (b) compared to the original resid 7.
Conclusions 1. 2 W f PDMSis easy to operate, fast, and less expensive than most other mass spectrometric techniques and is
0
- -L-L-
200
240
280
320
360
400
mass
Figure 6. PDMS spectra of original resid 7 (a) and the remixed fractions (b).
suitable for quantitative measurements of the molecular weight distributions of some coal conversion products.
Coal Liquefaction Heavy Products
2. Comparison of PDMS molecular weight distributions to the GPC and FIMS shows that, in general, numberaverage molecular weights agree well. 3. In order to realize the full potential of PDMS for analytical determinations of low molecular weight species and/or fragment ion studies, the CFINT conversion/ subtraction utility appears to be an alternative to eliminate the majority of background interfering peaks.
Energy 6 Fuels, Vol. 8, No. 1, 1994 266
Energy Contract No. DE-AC22-89PC89883. We are most grateful to Dr. Osamu Yamada for writing the GPC software and to Dr. Ripudaman Malhotra and Dr. Don McMillen of SRI International for a critical reading of an earlier version of this paper. Registry No. Hexane, 100-54-3; benzene, 71-43-2; pyridine, 110-86-1; polystyrene 9003-53-6; silica gel, 11292600-8.
Acknowledgment. We thank Dr. Susan D. Brandes at the CONSOL Inc. for providing samples for this study. This research was financed by the subcontract from CONSOL Inc. under the United States Department of
Supplementary Material Available: Listing of physicochemical parameters of all the resids (26 pages). Ordering information is given on any current masthead page.
