Anal. Chem. 1983, 55, 1370-1375
1370
recent work of Haber (14) may suggest alternative eluents that have these properties.
(6) Mould, D.L.; Synge, R. L. M. Biochem. J . 1954, 58, 571. (7) Myseis, K. J. "Introduction to Colloid Chemistry"; Interscience: New York, 1976; Chapters 15 and 16. (8) Cantwell, F. F.; Puon Anal. Chem. 1979, 5 1 , 623-632. (9) Tsuda, T.; Namura, K.; Nakagama, G. J . Chromatogr. 1982, 248, 241-247. (10) Whatman Liquid Chromatography Product Guide, Catalog No. 126, 1982, Whatman Chemical Separations Inc., 9 Bridewell Place, Clifton, NJ. (11) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (12) Rice, L. L.; Whitehead, R. J . Phys. Chem. 1965, 6 9 , 4017-4024. (13) Hunter, R. J. "Zeta Potential in Colloid Science"; Academic Press: London, 1961; Chapter 3. (14) Haber, N. Roc. Nafl. Acad. Sci. U . S . A . 1982, 79, 272-276 (Biochemistry).
ACKNOWLEDGMENT The valuable counsel of John Evans and Paul Himes, of our laboratory, during this work is gratefully acknowledged. Registry No. Methanol, 67-56-1; acetonitrile, 75-05-8.
LITERATURE CITED (1) Heftmann, E. "Chromatography", 2nd ed; Reinhoid New York, 1976; Chapter 10. (2) Pretorius, V.; Hopklns, B. J.; Schieke, J. D.J . Chromafogr. 1974, 9 9 , 23-30 _. (3) Jorgenson. J. W.; Lukacs, K. D. J . Chromatogr. 1981, 219, 209-216. (4) Prlvate communicatlon with John Evans, Dow Chemical USA, 574 Bullding, Midland, MI; 21, May 1979. (5) Business Week Magazine I979 (May 14), 218 8-H.
RECEIVED for review January 12, 1983. Accepted April 12, 1983.
Mobile-Phase Transport Properties of Liquefied Gases in Near-Crit ical and Superc rit icaI FIuid Chrornat ography Henk H. Lauer," Douglass McManlglll, and Robert D. Board Hewlett-Packard Laboratories, Applied Physics Laboratory, 165 1 Page Mill Road, Palo Alto, California 94304
The physlcal aspects of chromatographlc separatlon for several solute, moblle-phase, and statlonary-phase comblnatlons have been studled in the neartrltlcal reglon. I t Is observed that the logarlthms of the coefficient of blnary dlffuslon and relative retention factor, at constant density, exhlbit smooth llnear dependence on reclprocal temperature as the critical Isotherm is crossed, whlie moblie phase vlscoslty remains constant. Examples of chromatographlc separations of model compounds at sub- and supercritical temperatures are demonstrated for COP, N20, and for subcrlticai NH,. These Investlgatlons Indicate that the useful range of applications for supercrttlcal fluids can be extended to include thermally iablle compounds and to fluids which have inherently hlgh critlcal temperatures.
The use of supercritical fluids as mobile phases in chromatography was f i t reported by Klesper, Corwin, and Turner (I). Since that time a considerable volume of publications has appeared, and the technique has been demonstrated for a variety of mobile phases, stationary phases, and potential analytical applications. Reviews of the recent literature have been given by Schneider (2),Novotny (3), Randall ( 4 ) , and Peaden (5). Recently the technique has been demonstrated for capillary columns (6-9) and for packed columns using a straightforward modification of available liquid chromatography hardware (IO). Previous work has emphasized the desirable physical characteristics of supercritical fluids used as mobile phases. In particular, viscosity, solute diffusivity, and solubility are regarded in the literature as favorable for chromatography. Considerable experimental difficulty has been tolerated to maintain the mobile phase above the critical conditions which, for solvents like NH3, requires pressures and temperatures above 115 bar and 133 OC. We have studied the chromatographic usefulness of several inorganic gases (COz,N20, NH3) as mobile phases, which have
been compressed to liquid densities and maintained a t near critical temperatures. We find that the logarithms of the coefficients of binary diffusion, as well as the logarithm of relative retention factors, at constant mobile phase density, exhibit a smooth linear dependence on reciprocal temperature when the critical isotherms are crossed, while mobile phase viscosity remains constant. It appears that the chromatographic attributes of commonly used supercritical fluid mobile phases virtually remain intact for these liquids, greatly reducing the burden on the design of the hardware. For example, COz, with its moderate critical parameters of 31 "C and 74 bar, is well suited for use in equipment of present technology, but due to its nonpolar nature it is of limited usefulness as a chromatographic solvent. More polar mobile phases, such as NH3, tend to have higher critical temperatures, which can create special problems for columns, injection devices, and detectors. Thermally labile materials may decompose during analysis a t these elevated temperatures. Adequately sealing the connections to valves, injectors, and detectors is difficult with polar mobile phases at high temperature due to the rapid degradation of polymeric seals. Accordingly, the use of subcritical mobile phases allows the hardward requirements to be relieved considerably and the range of applications extended to include thermally labile materials.
EXPERIMENTAL SECTION Apparatus. A modified HP1084B liquid chromatograph, described in a previous paper (IO),was used for both diffusion and chromatographic experiments. The diffusion measurements were carried out in a Teflon-lined stainless steel (8s) tube with a coil diameter of 220 mm, length 1829 & 2 mm, and of 1.262 & 0.002 mm i.d. inserted in the column oven and connected in the usual way to an injector (Rheodyne 7125) and W detector by 0.25 mm i.d. ss tubing so that the system total dead volume did not exceed 50 mm3. Constant low flow rates through the tube could be assured by maintaining constant temperature and pressure with a pressure regulator (Tescom,
0003-2700/63/0355-1370$01.50/0 0 1983 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 55, NO.
8, JULY 1983 1371
Table I. Viscosities of CO,, NH,, and Common Liquidsa at Different Temperatures and Pressures 104q,P
T, "C -30 20
25 40 50 60
P, bar 400 100 200 400
CO,b 22.9 8.3 10.3 13.0
NH3b 31.4 16.6 17.0 17.8
100 200 400
4.9 8.0 10.8
13.3 13.6 14.5
100 200 400
2.4 6.1 9.1
9.8 10.3
-_ a Liquid viscosities at 1 atm.
-
CH,CNC
MeOHd
H,Od
35.3
59.7
100.2
34.5 29.2
54.1 45.6
89.0 65.3
24.5
35.1f
46.1
Hzol CH3CNe (60/40 (v/v))
H,O/ MeOHe (60/40 (v/v1)
C,HWd 32.6
96.1
142.4
29.4 27.1 24.8
11.5
From ref 12.
From ref 42.
26-1721-:24-043)before the injection device and a valve restrictor (Whitey SS2, RS2) at the detector outlet. Pressure drops over the diffusion tube were negligible. The injection valve, detector, and connectingtubing were kept at ambient temperature, whereas the pressure regulator and the restriction valve were heated to 45 "C. Subambient temperatures in the oven compartment were achieved by a built-in heat exchanger connected to a recirculating cold bath (Poly Science Corp.). Oven temperatures and pressures were maintained at kO.2 "C and f0.7 bar, respectively. Volumes of approximately 0.3 mm3 of a solution (80 gg/mm3) of naphthalene (Fisher Scientific Co.) in n-hexane (Baker) were injected at ambient temperature into a flow of COz (anaerobic grade, 99.99% (v/v), Liquid Carbonic). Measurement of the flow rate and the temperature of the exiting gas, along with compressibility data, allowed calculation of the internal flow velocity of the dense gas and the inner radius of the diffusion tube (see eq 6). In the chromatographicexperiments COz,N20 (Union Carbide), and NH:, (anhydrous,electronic grade, Union Carbide) were used as the mobile phases. PRP-1 (100 X 4.6 mm, 10 pm, Brownlee Laboratories) and Spheri-b, RP-18 (100 X 3.2 mm, 5 pm, Brownlee Laboratories) columns were used. Nicotine, caffeine, theophylline (highest grade available, Eastman), and the other samples (Chem. Service) were dissolveld in methanol (HPLC grade, Malinckrodt). Sample concentrations ranged from 0.1 to 2.0 pg/mm3 and injection volumes from 2.5 to 8.0 mm3. In pumping NH3, Vespel seals in the injector and pump were replaced by Teflon and stainless steel materials, respectively. Retention factors (k? were calculated in the usual way, assuming that the unretained peak was adequately reflected by a base line disturbance due to solvent elution.
RESULTS AND DISCUSSION Figure 1 shows the pressure-density isotherms for COz. According to Schneider (II), fluid extraction with COz is of interest within the boundaries P = 74-400 bar, p = 0.25-1 g/cm3, and T = Tc-62 "C, whereas with regard to chromatography an extension into the liquid state down to -30 "C seems feasible. For this extension to be chromatographically useful, the viscosity must not increase dramatically in the liquid state, and the diffusivity of solute in the mobile phase must remain high enough to achieve effective mass transfer with thle stationary phase. The diffusivity of naphthalene in COz was investigated by crossing the isotherms in the phase diagram at isodensities of 0.70, 0.80, and 0.90 g/cm3 and via an isobar a t 171 bar (broken lines in Figure 1). The chromatographic experiments were carried out by crossing the isotherms at 0.80 g/cm3. Viscosity. A plot of the dynamic viscosity of COz as a function of temperature and pressure is shown in Figure 2 (12). If, for chromatographic purposes, a temperature and pressure
From ref 43. e IAom ref 44.
f
From ref 45.
740
400 300
220
-
150
cc Y
VI
74 50
37
I -90 (
0.25
0.50
0.75
1.0
1.25
DENSITY ~ g i c r n ~ i
Flgure 1. Phase diagram of COP. Broken lines indicate isotherm crossing under isobaric (171 bar) and constant density (0.70, 0.80, 0.90 g/cm) conditions.
range of 10 "C, 400 bar and 100 "C, 75 bar is considered, the figure shows that the dynamic viscosity changes from :L6 X lo4 to 2 X lo4P. However, in following constant density loci, changes in viscosity can be avoided (13) as shown by the broken lines in Figure 2. The viscosities of COz and NH, together with those of some commonly used chromatographic solvents are shown in T'able I for a broad temperature range. At ambient temperature (20 to 25 "C) the viscosities of the liquefied gases are much lower than those of the common solvents (at STP) whereas the mixtures of water with acetonitrile and methanol (often used in HPLC) show high viscosity values. Even at low temperature (-30 "C) the rheologic behavior of the liquefied gases will be close to that of n-hexane at ambient temperature. Diffusivity. The rate of diffusion of a compound in different fluids under varying conditions (P,T,p ) is largely responsible for the exchange of mass between mobile and stationary phase in a chromatographic column. High diffusion rates (or diffusion coefficients) favor good separation char-
1372
ANALYTICAL CHEMISTRY, VOL. 55,
NO. 8, JULY 1983
Table 11. Comparison of Experimental Diffusivities with Data from the Literature (29-32) compound
fluid
benzene naphthalene caffeine aniline
concn, wg/mm3
T, "C
P, bar
co, CO,
40.1
COa CH,OH
40.1 40.8 15.0
160.3 160.3 160.3 157.5
40.1
solvent
pure 40 80
105D,, cm"s lit. calcd
105~,b cm*/s
a
12.6 11.2 8.7 2.40 1.53c
1.30
C6HM CH,Cl, CH,OH
80
SO I
1.30 1.19 1.16
ref
12.9 11.2 7.9
30 31 31 2.46d 1.57d
1.49
33
a s l o = Asymmetry factor calculated from 10% of the peak height ( 3 4 ) . Average value from four separate measurements with an overall precision of 2.9 * 0.6%. Conversion of experimental value at 40.8 "C to 15.0"C by means of Wilke's equation (DQ/T = constant) (33). Calculated from the Wilke-Chang equation (32) with an association parameter of 1.9 for methanol.
Table 111. Investigation of Solvent Effects on the Diffusivity of Benzene in CO,
a
1.0 10
I
20
I 5w
I , I l l
, 40
60
80 1W
,
,
,
solvent
concn, wg/mm3
T, "C
P , bar
none CH,Cl, C6HM CH30H
80 80 80
40.1 40.1 40.1 40.1
160.3 160.3 160.3 160.3
105~,a cma/s 12.6 12.5 12.2 12.4
As in Table 11.
Under the conditions of eq 4 the first term in eq 1can be neglected and the binary diffusion coefficient calculated from
I
1wo
TEMPERATURE 1°C)
Figure 2. Viscosity of C02 as a function of temperature and pressure. Broken lines indicate constant density conditions.
acteristics in column chromatography. To study this effect in GO2,the binary diffusion coefficient of a model compound (naphthalene) was measured in a flow-through open tube. According to Taylor-Aris (14-1 7) the broadening of a concentration profile per unit length can be described as (18-21)
The tube radius, ro, was calculated from
in which F = volume flow rate through the tube. Plate height, was calculated according to
Hi,
2Dif
rtu0
UO
2403
Hi = - +-
Hi = L(ati/tR)'
in which Hiis the theoretical plate height of component, i, in the fluid (0,Di is the diffusivity of i in f, uo is the linear velocity = L/tR, L being the tube length and tR the mean residence time, and ro is the tube inner radius. For straight tubes a virtually Gaussian concentration profile is obtained if (22, 23) Deff/u&
< 0.01
(2)
in which the effective diffusivity (Deff)can be described as
(3) For coiled tubes eq 1 only applies if (24-29) De S C ~< /10 ~
(4)
where the Dean (De) and Schmidt (Sc) numbers are defined as follows
and Re stands for the well-known Reynolds number.
(7)
in which C T is ~ the half width of the peak at 0.607 of its height in time units as observed on the recorder plot. Effects of external band broadening on plate height due to injection and instrument dead volume were neglected. The reliability of the experimental equipment was investigated by comparing the observed binary diffusion coefficients with data obtained from the literature (30-33). The results and the quality of agreement are shown in Table 11. The solid samples, such as naphthalene and caffeine, had to be dissolved in a suitable solvent prior to injection. Possible solvent effects on experimentally obtained diffusivities appeared to be negligible (within experimental error) for benzene and are shown in Table 111. Therefore, it was assumed that solvent effects are negligible for naphthalene and caffeine as well (see also Table 11). Experimental diffusivities of naphthalene (dissolved in n-hexane to a concentration of 80 pg/mm3) are shown in Table IV. A plot of In D vs. 1/T is depicted in Figure 3 together with data of Feist and Schneider (31) a t a density of 0.60 g/cm3. All data show a smooth linear dependence of In D vs. 1/T with no anomalies in passing through the critical temperature of GO2 at the densities 0.80 and 0.90 g/cm3. At a density of 0.70 g/cm3 no reliable data below the critical temperature could be obtained due to possible phase separation (gas-liquid), as can be seen from Figure l. In addition, I , of the Table IV shows that the asymmetry factor (34),SO recorded concentration profiles at a density of 0.70 exceeds 1.40, which probably affects the absolute accuracy of the data
ANALYTICAL CHEMISTRY, VOL. 55, NO. 8, JULY 1983
Table IV. Experimental Diffusivities of Naphthalene in CO, at Different Densities P,a
105~,b
T , "C
.P, bar
g/cm3
s lo
15.2 25.2 40.5 54.9 20.11 25.0 40.4 54.9 15.1 25.1 40.4 49.8 40.7 47.0
172.0 l.70.6 171.3 170.6 67.2 90.7 163.8 238.9 111.4 179.6 351.3 351.3 116.2 l42.0
0.943 0.893 0.806 0.704 0.802 0.801 0.796 0.800 0.900 0.900 0.898 0.900 0.695 0.705
1.10 1.15 1.18 1.41 1.20 1.21 1.29 1.26 1.16 1.15 1.19 1.18 1.46 1.42
Densities calculated from ref 46.
cm2/s 8.43 9.68 11.6 13.4 10.9 11.0 11.4 12.0 9.16 9.46 9.80 10.4 12.8 13.2
Table V. Critical Parameters of Different Fluids ( 1 2 ) fluid
l',, "C
CO, N2O
31.06 36.41 132.4
NH,
TEMPERATURE
COLUMN
P,, d c m 3
P,, bar 73.825 72.45 114.80
0.464 0.452 0.235
loci
- PRP-1.4.6 x lOOmrn
As in Table 11.
I
I
t-
3.1
3.3
3.2
-'
O
' 0 ,
-8.8
I
3.4
l l T x lo3 IK-']
Flgure 4. van't Hoff plots of model compounds in denslty. -8.7
1973
CO,
at constant
TEMPERATURE I'CI
i
20
F;kO
40
aim3
30 I
I
20
j
I
C O L U M N - P R P - 1 , 4 . 6 ~lOOrnm
4
-9.2 -gl[
-
i 05
3.0
-9.3
-9.4
t c L
29
30
3'
32
33
1 I T x lo3 I K
.\
I
i A 34
35
36
'I
Flgure 3. DiffustviYy of naphthalene in COPas a functlon of temperature at different densities. Broken llne with squares lndlcates isobaric conditions. Open circles (0)are data from Felst and Schnelder (30). at this density. Profiles with asymmetry factors in excess of 1.30 should therefore be ,rejected. Moreover, we were not able to reproduce Feist's (31) data at a density of 0.60 due to severe tailing (SIo > 1.80) which suggests adsorption of naphthalene on the Teflon inner wall of the tube. Feist and Schneider @I), however, used plain stainless steel tubing and reported severe tailing of caffeine a t densities below 0.70. The presence of tailing was confirmed also by our experiments with stainless steel at a higher density (0.80) but appeared nonexistent with the Teflon inner tube. In passing the critical temperature over an isobar ( 171 bar, see Table IV and broken lines in Figures 1 and 3) the consistency in behavior of the previous data was confirmed. This isobaric approach appears well suited to estimate diffusivities at lower temperatures via simple extrapolation. The diffusivity of naphthalene in COzat -30 "C (171 bar; p zz 1.07), obtained in this way, appears to be -4.4 X cm2/s. This is comparable to the calculated (32)diffusivity of naphthalene
-
1
8
32
I
33
34
liTxlO3IK1I
4
\
1
:I 1
Ftgure 5. van't Hoff plots of model compounds in N,O at constant
denslty. in n-hexane at 25 "IC (-3.5 X cm2/s) and, together with the before mentioned favorable rheologic behavior of C 0 2 at low temperatures, suggests that a powerful method for analysis of extremely thermolabile compounds could emerge. Chromatography. In a dense gas the retention behavior of a particular compound is commonly described by its relative retention or capacity factor, k', which is a strong function of pressure, density, and temperature as observed and descrilbed by many authors. The effect of temperature on retention in supercritical COz a t constant density has been reported by van Wasen (35,36). To investigate whether retention behavior changes drastically in going from the dense gas to liquid state, the van't Hoff plots (In k'vs. 1/T) were investigated for COz and NzO with a number of model compounds at a density of 0.80 g/cm3 on a PIIP-1 column. Lacking a suitable density equation for NzO, the law of corresponding states (equal reduced parameters) was used to calculate the experimental pressures and temperatures of this fluid from its respective critical values (see Table V). 'The resulting plots are (displayedin Figures 4 ((30,)and 5 (M20) and reveal that none of the eluted compounds in either fluid shows an anomalous behavior. Moreover, the average enthalpy of eluite interaction with the stationary phase (evaluated from the slope -AH/R) in approximately -6 kcal/mol which is very
1374
ANALYTICAL CHEMISTRY, VOL. 55, NO. 8, JULY 1983
3 "
a. W
0
z
4
m a 0 cn m
W
4
II:
0
z a
m
0 (0
m
--
a
,,
.
E1
-_-)I
2
0
4
2
0
4
TIME (min.1
Flgure 6. Separation of PAH's in C02 at sub- and supercrltical conditions: elution order, 2-methylnaphthalene, phenanthrene, fluoranthene, pyrene; column, Spherid, RP-18; flow rate, 1.20 cm3/mln; wavelength, 254 nm; sample, 2.5 mm3; at 23.8 OC, inlet pressure 98 bar. outlet pressure 82 bar; at 40.0 OC, 177 and 163 bar, respectively.
0
2
0
4
2
T I M E (min.1
Figure 8. (a) Separation of a test mlxture in liquid NH: elution order, dimethyl phthalate, diethyl phthalate, biphenyl, o-terphenyl; column, PRP-1; flow rate, 2.50cm3/mln; wavelength, 265 nm; sample, 6 mm3; temperature, 40.0 OC; inlet pressure, 177 bar; outlet pressure, 172 bar. (b) Separation of alkaloids in liquid NH,: elution order, caffeine, theophylline, nlcotine; temperature, 30.0 OC; inlet pressure, 180 bar; outlet pressure, 174 bar; other conditions, as in Figure 8a.
W
0
z 4
m a: 0 u)
m 4
- -
0
1
2
0
1
2
TIME [min.)
Flgure 7. Separation of PAH's in N20 at sub- and supercrltlcal conditions. Elution order and other conditions are given In Figure 6: at 23.8 OC, inlet pressure 87 bar, outlet pressure 69 bar; at 40.3 OC, 163 and 149 bar, respectively.
close to those values reported (37-41)in reversed-phase liquid chromatography. Figures 4 and 5 also show interesting selectivity differences on the PRP-1 column with the investigated fluids. Caffeine did not elute in C 0 2 and the other model compounds, also, were more strongly retained when COz wm used as the mobile phase. The model compounds were observed to be less strongly retained on a hydrocarbonaceous bonded phase column. For this reason chromatograms of a mixture of PAH's separated on a RP-18 column in both fluids at sub- and supercritical conditions are shown in Figures 6 and 7. The promising combination of liquid NH3 and a PRP-1 column is shown in Figure 8. The easy and fast separation (Figure 8b) of compounds with polar functional groups such
as caffeine, theophylline, and nicotine merits special attention. In practically useful separations, however, the capacity factor ( k ? of the first peaks in Figures 7 and 8 has to be increased in order to allow separation from possible interferences. The unusually rapid increase in peak width, as shown in Figure 8a, may arise from relatively slow mass transfer in this stationary phase. ACKNOWLEDGMENT The authors appreciated the skillful experimental assistance of Peter Daetwyler (Ciba-Geigy, Basle, Switzerland) during his visit at our laboratories and acknowledge the review of the manuscript by Harry E. Weaver, Jr., of our laboratories. Registry No. NHB,7664-41-7;C02, 124-38-9;N20, 10024-97-2; naphthalene, 91-20-3. LITERATURE CITED (1) Klesper, E.; Corwin, A. H.; Turner, D. A. J. Org. Chem. 1982, 2 7 , 700 .
(2) Schneider, G. M.; van Wasen, U.: Swald, I. Angew. Chem., Int. Ed. Engl. 1980, 19, 575. (3) Novotny, M.; Springston, S.R.; Peaden, P. A.; Fjeldsted, J. C.; Lee, M. L. Anal. Chem. 1981. 5 3 . 407A. (4) Randall, L. G. Sep. Scl. Techno/. 1982, 17, 1-116. (5) Peaden, P. A.; Lee, M. L. J. Ll9. Chromatogr. 1982, 5 , 179. (6) Peaden, P. A.; Fjeldsted, J. C.; Lee, M. L.; Sprlngston, S. R.; Novotny, M. Anal. Chem. 1982, 5 4 , 1090. (7) Peaden. P. A.; Lee, M. L. J . Chromatogr. 1983, 259, 1-16. (6) Springston, S.R.; Novotny, M. Chromatographia 1981, 4 , 83. (9) Fieldsted, J. C.: Jackson, W. P.: Peaden, P. A.: Lee. M. L. J . Chromatogr. SCI., In press. (10) Gere, D. R.; Board, R.; McManlgill, D. Anal. Chem. 1982, 54, 736. (11) Schneider, G. M. Angew. Chem., Int. Ed. Engl. 1978, 17, 716. (12) "Gas Encvclooaedia. L'Alr Iloulde": Elsevier: Amsterdam. 1976. (13j Giddings, 2. Cr; Bowman, L. M., Jr:; Myers, M. N. Anal. Chem. 1977, 49, 243. (14) Taylor, G. Proc. R . SOC.London, Ser. A 1953, 219, 186. (15) Taylor, G. Proc. R . SOC.London, Ser. A 1954, 223, 446. (16) Taylor, G. Proc. R. SOC.London, Ser. A 1954, 225, 473. (17) Arls, R. Proc. R. SOC. London, Ser. A 1958, 235, 67. (16) Giddings, J. C.; Seager, S. L. J. Chem. Phys. 1960, 33, 1579. (19) Giddlngs, J. C.; Seager, S.L. J. Chem. Phys. 1981, 35, 2242. (20) Seager, S. L.; Geertson, L. R.; Giddings, J. C. J. Chem. Eng. Data 1983, 8, 166. (21) Huber, J. F. K.; van Vught, G. Ber. Bunsenges. Phys. Chem. 1985, 6 9 , 612.
Anal. &em. (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38)
1983, 5 5 , 1375-1379
Levenspiel, 0.; Smith, K. Chem. Eng. Scl. 1957, 8 ,227. Levenspiel, 0.; Bischoff, K. B. Adv. Chem. Eng. 1963, 4 , 95. Adler, M. 2.Angew. Math. Mech. 1934, 14, 257. Koutsky, J. A.; Adier, A. J. Can. J. Chem. Eng. 1964, 43, 239. Horvath, C. S.;Preiss, B. A.; Lipsky, S.R. Anal. Chem. 1967, 39, 1422. Tyssen, R. Chromatographla 1970, 3 , 525. Tyssen, R.; Wlttebrood, R. T. Chromatographia 1972, 5 , 288. Moulyn, J. A.; Spyker, H.; Kolk, J. F. M. J . Chromatogr. 1977, 142, 155. Swaicl, I.; Schnelder, G. M. Ber. Bunsenges. f h y s . Chem, 1979, 83, 969. Feist, R.; Schneider, G. M. Sep. Sci. 1982, 17, 261. Wilke, C. R.; Chang, P. AIChE J . 1955, 1 , 264. Wiike, C. R. Chem. €ng. frog. 1949, 45, 218. Kirkiand, J. J.; Yau, W. W.; Stoklosa, H. J.; Dllks, C. H., Jr. J . Chromatogr. Sci. 1977, 15, 303. van VVasen, U. Thesls, Bochum, 1978. van VVasen, U.; Schneider, G. M. Chromatographla 1975, 8, 274. Knox, J. H.; Vasvari, G. J . Chromatogr. 1973, 83, 181. Horvath, C. S.;Melander, W.; Molnar, I . J . Chromatogr. 1976, 125, 129.
1375
(39) Melander, W.; Campbell, D. E.; Horvath, C. S. J . Chromatogr. 1976, 158, 215. (40) Molnar, I.; Horvath, C. S.Ciin. Chem. (Wlnston-Salem, N.C.) 1976, 22, 1497. (41) Horvath, C. S.,Ed. "Hlgh-Performance Llquid Chromatography"; Academic Press: New York, 1980; Vol. 2. (42) Perry, J. H., Ed. "Chemlcal Englneers Handbook", 4th ed.; McGmwHili: New York, 1983. (43) "Handbook of Chemistry and Physics", 60th ed.; CRC Press: Boca Ratan, FL, 1979-80. (44) Abbott, S.R.; Berg, J. R.; Achener, P.; Stevenson, R. L. J . Chromatogr. 1975, 128, 421. (45) Bretsznajder, S."Predlctlon of Transport and Other Physical Propetilies of Fluids"; Pergamon Press: New York, 1971. (46) Reynolds, W. C. "Yhermodynamlc Properties in SI";Stanford Universlty Mechanlcal Englneering Department: Stanford, CA, 1979; p 122, eq P-3.
RECEIVED for review January 13, 1983. Accepted March 24, 1983.
Separation of Aromatic and Polar Compounds in Fossil Fuel Liquids by Liquid Chromatography Atsushl Matsunaga Lubricants and Petroleum Products Laboratory, Nippon Mlning Co., Ltd., 3- 17-35, Niizo-Minami, l'oda, Saitama, 335 Japan
A study was made on the application of normal-phase llquld chromatography for the separatlon of aromatlc and polar compounds In fossll fuel Ilqulds. Several comrnerclally avallable packlngs lncludlng chemically bonded phases (NO,, NH,, CN, and sulfonlc acld), silica, alumina, and porous polystyrene gel were compared for the separation of aromatic compounds by rlng number, separation of polar compounds (nltrogem and oxygen-contalnlng compounds) from polynuclear aromatic hydrocarbons and the resolution between polar compounds by functional group. Model compounds expected to be present In these llqulds were used for the evaluation of the packhgs. I n the case of fuel llquld samples, the chromatographlc behavior of each compound class varled according to thelr source and bolllng range.
method by compound classes by several authors (5-13). E h t there are few reports concerning the separation of a wide variety of compounds including nonpolar hydrocarbons to very polar compounds. With the variety of column packing materials, HPLC can achieve various types of separation. But it seems apparent that normal-phase chromatography is preferred for the c1,sss separation of heavy fractions, since the contribution of lairge hydrocarbon groups governs the retention characteristics more than polar groups in the reversed-phase chromatography (IO, 14). This paper presents normal-phase HPLC separations of aromatic and polar compounds. It includes an evaluation of normal-phase HPLC packing materials with a variety of surface functionality, as they were expected to exert a unique selectivity. Some examples of the separation of practical samples are also presented.
Fossil fuel liquids are complex mixtures of a broad variety of chemi'cal classes from hydrocarbons to very polar compounds. Heavy petroleum fractions, shale oils, coal liquids, and other nonconventional fuel liquids contain a relatively large amount of polynuclear aromatic hydrocarbons and mitrogen and oxygen compounds. These compounds can cause serious problems both in processing and in the quality of the products. With the increasing use of alternative fuel sources, new methods for separation and characterization are required. Many works on compositional analyses of each liquid have been reported, most of which involve the combination of classical chromatographic separation and spectrometric characterization (1-4).Although this approach is undoubtedly most useful, it is time-consuming to apply to routine analysis for the purpose of process control or product evaluation. Also, poor separation due to the low efficiency or tailing problems of classical adsorbents causes cross-contamination of fractions in many cases. HPLC (high-performance liquid chromatography) has been applied for the purpose of fuel liquid fingerprinting, for hydrocarbon type analysis, and as a rapid preparative separation
Apparatus. The instrument used for retention measurements was a Varian Model 5020 liquid chromatograph equipped with a Valco 10 fiL loop injector, a UV 254-nm absorbance detector, and a ternary solvent programmer. A Varian Model 8000 liquid chromatograph with a variable wavelength detector (Variscan) was used for the stopped-flow measurements of ultraviolet spectrum of chromatographic peaks. Columns. The columns listed in Table I were examinedl in this work. All of theaie were packed by the slurry method. Silica was packed by the balanced-density method (15). Alumina was slurried with white mineral oil and packed with hexane. Nuclensil packings were slurriecl and packed with carbon tetrachloride, while hexane was used for TSK 111 gel. Reagents. Hexane, methylene chloride, and methanol of LC grade were obtained from Wako Chemicals, Tokyo. Their water content was 20,140, and 170 ppm, respectively, by Karl Fischer titration when received. Drying of solvents was performed by percolating them through silica gel (Wakogel Q12 from Wako Chemicals, Tokyo). Model compounds were obtained from Tokyo Chemical Industry Co., Tokyo, and used as received. Procedure. Most model compounds were dissolved in hexane, but some hexane-insoluble polynuclear aromatics and polar compounds were discsolved in methylene chloride. The concen-
EXPERIMENTAL SECTION
0003-2700/83/0355- 1375$0 1.50/0
0 1983 American Chemical Society
