638
Anal. Chem. I98qS56,638-642
(7) Chaplin, M. F. Anal. Blocbem. 1982, 723,336-341. (8) Bryn, K.; Jantren, E. J . Chromatogr. 1982, 240, 405-413. (9) Manius, 0. J.; Liu, T. M. Y.; Wen, L. F. I. Anal. Blochem. 1979, 99,
365-37I. (10) Mawhinney, T. P.; Feather, M. S.; Barbero, G. J.; Martinez, J. R. Anal. Blochem. W80. 101, 112-117. (11) Marier, R. L.; Milllgan, E.; Fan, Y. D. J . cm. Mlcrob/o/. 1982, 16, 123-128. (12) Dmltriev, 8. A.; Backlnowsky, L. V.; Chizhov, 0. S.;Zolotarev, E. M.; Kochetkov, N. K. Carbohyd. Res. 1971, 19, 432-435. (13) McGlnnls, G. D. Carbohyd. Res. 1982, 708, 284-292.
(14) Stegllch, W.; Hofie, 0. Angew. Chem., Inl. Ed. Engl. 1069, 8 , 981. (15) Wachowiak, R.; Connors, K. A. Anal. Chem. 1979, 51, 27-30. (16) Varma, R. S.;Varma, R.; Allen, W. S.; Wardl, A. H. J . Chromatogr. 1974, 93,221-228.
RECEIVED for review November 8, 1983. Accepted January 3,1984. Use of trade names is for identification only and does not imply endorsement by the Public Health Service or by the U S . Department of Health and Human Services.
Effect of Pore Diameter on Diffusive Sample Retention in Gas Chromatography Hikoyuki Kaizuma Department of Chemistry, The Konan University, Okamoto, Kobe 658, Japan
Conslderatlon was glven to dlffuslve sample retentlon In relation to the pore dlameter and the dlffuslblllties of sample and carrier molecules. Three klnds of mlcrobead slllca gel havlng dlfferent pore slze were used as packlng materlal. Inert gases, He, Ne, and Ar, were Injected Into carrier flow as the sample. Hydrogen and nltrogen gases were used as carrlers. A constant term, O , In the hyperbollc equatlon which was prevlously offered by the author can be related to the pore dlameter. I t Is also affected by the dlffuslblllty and/or the mean free path of the carrler gas. As the ratlo of the mean free path of the carrler gas agalnst the pore dlameter of the pcklng gets larger, the dlffuslve sample retentlon becomes more effectlve.
The practice of gas chromatographic separation usually adopts a rather fast flow rate above the optimum where the HETP becomes lowest. It is seldom operated below the optimum because of the following shortcomings: The lower flow rate takes longer time for the migration and, as the result, the peak becomes broader. It is considered that nonsorbed samples migrate with the carrier at the same velocity and retention times of nonsorbed samples should become the same (I). This is true for a nonporous column, while the result is quite different for packed columns of porous material. The retention time of nonsorbed sample in porous column differs from sample to sample and the difference becomes larger as the flow rate decreases (2). Since the samples are nonsorptive, the factors affecting the migration in the column would be the diffusibilities of the sample molecules and the micropores in the packing material. During the process of migration, sample molecules diffuse into the micropores on the packing and are trapped for a while and return back to the flow by diffusing out from the pores. Then, in a steadily flowing carrier, a sample having a smaller diffusibility prolongs the migration by such diffusive displacement and results in longer retention time. In the case of nonporous packing material, the relationship among the retention time (tRo), the interdiffusion coefficient of the sample into the carrier (Drfi),and the peak spreading (go, the standard deviation of the peak width) is given by eq 1, where y is an obstructive factor (3). tRO
=-
2yDm
(1)
0003-2700/84/0356-0838$01 SO/O
On the other hand, when a nonsorbed sample is applied on a column of porous packing, the length of the column is equal to that of the former, the diffusive displacement of the solute occurs between the flowing phase and the stagnant phase in the pores. Retention time of the latter (tR) would be longer than that of the former (tR"). A relation similar to eq 1 may be arrived at for tR
where Q is the standard deviation of the peak spreading and y' is the obstructive factor including intrapore diffusion. D is the apparent diffusion coefficient. The R value can be expressed by using these two retention times
(3) where td = t R - tR". The term td means the time consumed in intrapore diffusion and has following relation which is derived from eq 1and 2:
where 6 is defined as 6 = y ' ( ~ ' ) ~ / y uThe ~ . time ratio, h-J/tRo in eq 3 can be expressed by using above relation as
- Drii - 6D 6D
Again, a new term Dm0 is defined by eq 6 D m 0 = Dlii - 6D
(5) (6)
Taking the relation into eq 5, eq 7 is obtained
(7) Substituting eq 7 into eq 4, one will find that 0 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
639
Rearranging eq 8, the final relation, eq 9, will appear, which is the relation called the "hyperbolic equation" by the author.
The equation contains two constant terms, i.e., tRo and Dm". The term tRomay be called absolute retention time and has the following meaning: It is the time required for the samples permeating with the carrier through only interparticle pores in the column. There is no diffference among samples at a constant carrier's flow. Thus, the quotient of the column length L divided by t R o gives the average h e a r velocity of the carrier. This interpretation was already discussed in the previous papers (2,4). The term Dm0 has the same dimension as the diffusion coefficient, and it may be considered to have a certain relation to the pore diameter on the packing and to the diffusibility of the carrier gas. Rewriting eq 9, Dmo is expressed as being incorporated with the retention value R (=tRo/tR)as eq 10. In other words, the Dm0 = D m ( 1 - R) (10) intrinsic factors of the system, which affect diffusive sample retention, would be collected in the constant term DmO. The main purpose of this paper is to examine the factors affecting Dmo so that the key to understanding the phenomena can be provided. The columns packed with spherical material having different pore distribution were used, and any difference observed among columns must be the result of the difference in intrapore diffusion, since the packing structure of the bed of spherical material is regarded as analogous (5).
EXPERIMENTAL SECTION Apparatus. A commercially available apparatus, Yanagimoto HSG1, was used. A part of the flow path was modified as shown in Figure 1,such that two TC detectors (TCD) were mounted on both ends of the column so that a sample peak could be detected twice at the top and the end of the column as described in the previous paper (4). A precolumn was inserted between the sample injection valve (SIV) and the detector on the column top to set flow lines in good order, eliminating a shock arising from sample injection. Carrier and Sample Gases. Hydrogen and nitrogen were used as carrier gases. Flow rate of the carrier gas was controlled with a pressure control valve (PCV). Outlet pressure at the column end was constantly maintained at 0.1 kg/cm2 by means of a needle valve. Inert gases, He, Ne, and Ar, supplied as pure standard gases from a manufacturer (Gasukuro Kogyo Co., Ltd., Tokyo, Japan) were adopted as sample gases. Columns and Packing Materials. Three kinds of microbead silica gel, 3A, 4B, and 5D (Fuji-Davidson Chemical, Ltd.), were used as packing material. Particle size was selected within a narrow range of 30 to 40 mesh by sieving. Irregular shaped beads were removed by rolling the beads down on a slightly inclined glass plate. Three stainless steel tubes of identical size (length, 3 m; 3 mm i.d.) were prepared. Three kinds of selected microbead silica gel were packed in these tubes. Shimalite Q (80/100 mesh, Shimadzu Seisakusho), which consists of crushed quartz and is monoporous, was packed in the precolumn (length, 2 m; 3 mm i.d.1. Fine screens made of stainless steel wire were soldered on both ends of the column to prevent the packing from leaking out. Details of the columns are listed in Table I. Measurements were carried out under a controlled temperature of 50 OC. Treatment of the data and method of calculation were similar to those described in the previous paper ( 4 ) . RESULTS AND DISCUSSION Porosity. The appearance of the packing material is spherical granules, but the pore distribution differs. An acceptable value of the interparticle porosity of such packed
f-7
I
v.l.-F-q Integrator # 500A
Flow diagram: (PCV) pressure control valve; (SIV)sample injection valve; (TCD) thermal conducthrity detector; (Pg, Pi, Po) pressure gauges for applied, inlet, and outlet pressures, respectively. The outlet pressure was kept 0.1 kg/cm2 by means of a needle valve and the whole apparatus was thermostated at 50 OC. Figure 1.
36 ' A
(3A)
(48) (5D) Figure 2. Comparison of volume fraction ratios of solid part and pores In three classes of microbead silica gel. ~~
Table I. Column Dimensions column tubes length (cm) internal diameter (cm) packing material shape and mesh size
stainless steel 300 0.30 microbead silica gel spherical, 30/40
class
3A
4B
5D
average pore diameter ( A ) density (g/cm3) packed density (g/cm3) total porosity
23 2.127 0.917 0.578
80 2.291 0.477 0.791
170 1.031 0.363 0.648 (0.839)a
a Calculated using the density of pure silica (2.25 g/cm3).
columns of spherical material would be 0.36 (5-7). The intraparticle porosities of these packing material can then be easily estimated as the difference in the interparticle porosities relative to the total porosities for respective packings. Considering that the specific gravity of 5D packing is about half that of others, relatively large amounts of blind pores must be included within the particle. This inference may be agreed upon due to the fact that it looks opaque and milky, while the other two are transparent. The porosity occupied by the blind pores may be estimated by refering to the densities of 5D packing and pure silica. The density of the latter lies between 2.2 and 2.3 (8). Volume fractions of the respective pores (interparticle, intraparticle, and blind pores) for these columns are compared in Figure 2. Figure 3 shows how far micropores and interdiffusion coefficients contribute to the sample retention. Although only a few data, obtained at inlet pressure of 0.16 kg/cm2, are shown as examples, the contribution is significant. Considering that samples are inert gases and they are not sorptive, the result was unexpected and rather surprising. The characteristics of the plots in Figure 3 are similar to those for the porous material such as C-22, activated alumina, and others as already reported (2, 4 ) .
640
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
Table 11. Estimated Values of tRo, 0 2 ,and [ h a " ] " in H, carrier
Pi,b packing kg/cm2 3A 0.20 0.16 0.14 0.12 4B 0.18 0.16 0.14 0.13 5D 0.22 0.20 0.18 0.16
tRn,
DZ,
S
cm2/s
60.0 100.0 154.1 352.8 99.4 133.9 186.7 286.1 81.4 99.6 120.8 166.5
0.470 0.484 0.491 0.532 0.312 0.321 0.346 0.322 0.139 0.134 0.141 0.149
in N , carrier
[Dm" 1, cm2/s
0.427
D 2 , cmz/s
tRo 7 S
0.017 (4.0%)
[ D m " ] cm"s ,
155.3 2 59.9 382.8 826.6 223.6 312.7 459.4 589.5 177.5 210.7 263.6 359.4
0.0918 0.0943 0.0963 0.0944 0.281 * 0.011 (3.7%) 0.0618 0.0618 0.0687 0.0714 0.123 t 0.004 (3.5%) 0.0197 0.0208 0.0220 0.0266 Values at the standard state ( 0 'C, 1 atm). Inlet pressure; pressure guage reading. i:
0.0823
i:
0.0016 (1.9%)
0.0560 * 0.0049 (8.8%)
0.0194
* 0.0023 (11.6%)
Table 111. Data Concerning the Permeability of Each Columnu
Pi column 3A
4B
5D
7
kg/cm' 0.20 0.16 0.14 0.12 0.18 0.16 0.14 0.13 0.22 0.20 0.18 0.16
p, kg/cm2 1.186 1.166 1.153 1.143 1.174 1.163 1.153 1.148 1.194 1.184 1.174 1.163
LA,", cm/s
ap*/L,
g/(cm' s2) 326 196 131 65 261 196 131 98 392 326 261 196
H, 5.00 3.00 1.95 0.85 3.02 2.24 1.61 1.05 3.69 3.01 2.48 1.80
10ZK,(cm3 s)/g
N2 1.96 1.15 0.78 0.36 1.34 0.96 0.65 0.51 1.69 1.42 1.14 0.84
Column length (cm); L = 300.0. Viscosity at 50 "C (g/(cm s));
H, 1.58
106Kq,cm2
N, 0.607
H, 1.51
N, 1.17
1.17
0.508
1.12
0.977
0.948
0.437
0.907
0.840
q ( H 2 )=
0.938 x
q ( N 2 )=
1.883 X
Estimation of D f i , DfiO, and tRo. The interdiffusion coefficient, Dm, is calculated according to the equation proposed by Fuller et al. (9).
(d+ &)
112
400
(1.00 X 10-3)P75
Drn =
L48
(11)
p[(cAvi)1'3 +(~BVL)~'~]~
where terms have the following meaning and dimension: Dm, interdiffusion coefficient (cm2/s) between molecule A and B under the average pressure P; MA, MB,molecular weight (g/mol); T,absolute temperature (K); CAVi, ZBVi, diffusion volume (9), respective atomic special diffusion parameters V, are summed over atoms in molecules A and B; P, average pressure in the column (atmosphere) defined as P = JP,where J is called James Martin's factor (10, 11)
J=
2(Pi- P0)3 3(P' - PJ2
and P, and Po are inlet and outlet pressure, respectively. Rewriting eq 9, a linear relation can be obtained between Dm and Dm/tR as shown by eq 12. D f i = tRo(Dfi//R) D f i o (12)
+
Values of tRo and D m o can be estimated as the gradient and intercept of the plots of eq 12, respectively, which are listed in Table 11. Propriety of t R o and Dm O Values. As mentioned above, L/tRo,where L is column length, gives the linear flow velocity of the carrier through the column, and so the velocity L/tRo should be proprotional to the pressure difference @*/L as predicted by Kozeny (12),where A P is the pressure difference
Hz C A R R I E R
0
; c
2
200-
-4
I
t "0
Ar
I O
"48
t
- 3A
He
NO
05
-
d .
t
t
15
20
INTERDIFFUSION COEFFlClENTDrii (crn2/s)
Flgure 3. Typical plots showing the hyperbolic relation. Considering that the packings are spherical and similar in mesh size, the differences in retention times among the packings arise from the differences in volumes and distributions of pores in the respective packings.
across the column of L length corrected for the compressibility of gas as eq 15 (13). The proportional constant, K ((cm3s)/g)
P,Z - P,2 (15) 2P in the plots of L/tRo and P / L , is called permeability. The hp*
=
~
permeability of hydrogen is almost twice as much as that of nitrogen, while the product with viscosity, Kq,yields similar magnitude in the two carriers for the respective columns. The product Kq (cm2)is called specific permeability, and it should be constant and independent of carrier. If the bed consists of nonporous packing and has simple structure, the fact may be true. However, the results obtained here do not show this. The reason may arise from the variation in the effective porosity. Detailed discussions can be found elsewhere (2),since
ANALYTICAL CHEMISTRY, VOL. 56, NO. 4, APRIL 1984
641
0.61
PFj
D 0.1,
O0 -
0 2 04 0 6 0 8 10 12 X I O - ~ MEAN FREE PATH OF CARRIER (cm)
Figure 4. Plots of the constant In diffuslve term, Om", and the values converted to the standard state (0 "C, 1 atm), [Om"],against the mean free path of the carrier. Dots express Dm " under the operated condition and circles correspond to [Om"]at the standard state. The extrapolation of respective plots gathers on the abscissa at around 300 A.
it is not the object of this paper. The term Dm" should have the same dimension as Dm. It may be considered that Dfi' is inversely proportional, like Dm, to the average pressure (eq 11). The bar mark above the letter in D m and Dm" indicates that it is a value under the average pressure in the column. Defining [ D n " ]as the value of Dm0 at the standard state (0 "C, 1 atm), which is converted from Dm" assuming that the contribution of temperature and the pressure to Dfi" is similar as in eq 11,we get
[ D m " ] = D R 0 P273( ~1.76 ) where T and P are column temperature (K) and average pressure in the column, respectively. [Dm"]would be constant to respective columns and carriers. This is shown in Table
11. Applicability of t h e Hyperbolic Equation. The hyperbolic equation (eq 9) has a form of xy = 1. Thus, it can be separated into two functions, one of which consists of the terms for retention time and another related to the diffusion terms as follows:
Using the values of tR" and Dm" in Table 11, f ( t ) and g(D) were calculated to respective data. The mean values of the product, Le., f(t).g(D)were 1.012 f 0.012 in H2carrier and 1.012 f. 0.010 in Nz carrier, respectively. These become sufficient confirmation of the hyperbolic equation. This fact shows that only diffusive retention is related to sample migration in these systems. Physical Meaning of DB The extrapolated two lines which connect three points of [Dm"]respective carriers are coincident on the pore diameter axis at around 250 8,as seen in Figure 5. This means that the pores exceeding this limit are not effective to the diffusive sample retention. Such large pores may be treated as a part of the interparticle pore. As pointed out in the previous paper ( 4 ) ,the diffusibility of the carrier is closely related to Dm" and/or [Dm"]. The mean free path, A, would be a proper measure expressing the mobility of gas molecules as well as diffusion coefficient. Plots of [Dm"] against mean free path (14) of carrier become a slightly concave line for respective packings as shown in Figure 4. The extrapolation of each line shows that it crosses the abscissa (mean free path) at about 300 A. Even if the pores
".
Figure 5. Three-dimensional expression among [Om "1, pore diameter of the packing, and mean free path of the carrier. The effective region for diffusive sample retention becomes a plane twisted and slightly concave and includes points ABCD. Hatched part is not effective because of the low mobility of carrier, even the pores are small enough. Crosshatched region is also noneffective because the pores are too large to retain the samples diffusively.
are small enough, the diffusive sample retention does not occur when the carrier's mean free path is smaller than this limit. Actually such molecules are significantly large, or it is a case under relatively high pressure. Three-dimensional expression is tried in Figure 5 , among [Dm"],pore diameter, and mean free path of the carrier. The relationship becomes a twisted and slightly concave plane. The diffusive sample retention does not occur when [ D n " ] is zero. That region is outside of two cross-lines, AB and BC in Figure 5. In such a case, tR becomes equal to t ~ " . It is important to point out that tR = t~' does not always mean that all the pores in the column can be available for the carrier's flow. The region where the pores are too large to retain the samples is on the right side of the line BC (crosshatched), while in the region (hatched), on this side of the line AB, the pores would be left behind from the carrier's flow, trapping the carrier molecules inside them. In such a case, the effective porosity which is the volume fraction of pores available to carrier's flow is smaller than the total porosity (2). The contribution of pore size to the diffusive sample retention is rather easily understood. An unexpected result, however, is that the carrier's mobility is affected in such a manner as obtained here. It is necessary to mention the difference between the sieving effect and the diffusive sample retention. The former is usually observed for molecular sieves such as zeolite, where samples are trapped irreversibly in the pores whose diameters correspond with those of molecules to be trapped. While in the latter case, the diffusive sample retention, the behavior of the sample molecules is diffusively reversible and the pore diameter is ten or a hundred times larger than that of the molecule. ACKNOWLEDGMENT The author greatly appreciated the kind offering of FujiDavidson Chemical, Ltd., of packing material and is grateful to Hiroyuki Hatano for his perpetual encouragement and useful discussions. LITERATURE CITED (1) Giddings, J. C. "Dynamics of Chromatography, Part I"; Marcel Dekker:
New York, 1965; p 200. (2) Kaizuma, H.; Nakamura, J.; Sugano, T. J . Chem. SOC.Jpn. 1980, 1415. (3) Reference 1, p 29, 35.
842
Anal. Chem. 1984, 56, 642-646
(4) Kaizuma, H. Anal. Chem. 1982, 54, 732.
(11) Purnell, H. "Gas chromatography"; Wiley: New York, 1962; p 68. (12) Reference 11, p 63. (13) Kaizuma, H.; Ogawa, K.; Yamakita, I. J . Chem. SOC.Jpn. 1975, 935. Translated into English and kept at Los Alamos Science Laboratories, University of Callfornla, Code 51-7339-2. (14) Perry, J. H., Ed. "Chemical Engineering Handbook"; McGraw-Hill: Tokyo, 1963; pp 17-30.
(5) Kubo, T.; Suito, E.; Nakagawa, U.;Hayakawa, S. "Funtai"; Maruzen: Tokyo, 1962; p 208. (6) Kaizuma, H.; Yamakiata, I. Abstracts of the 20th Annual Meeting of the Chemical Soclety of Japan, 1972; 2103. (7) Reference 1, p 198. (8) Lange, N. D., Ed. "Handbook of Chemistry"; McGraw-Hili; New York, 1967: D 318. (9) Fuller; 'N.D.; Schettler, P. D.; Giddings, J. C. I n d . Eng. Chem. 1966, 58, 21. (10) James, A. T.; Martin, A. J. P. Biochem. J. 1950, 50, 679.
RECEIVED for review October 14,1982. Resubmitted October 24, 1983. Accepted December 6, 1983.
Determination of Manufacturing Impurities in Heroin by Capillary Gas Chromatography with Electron Capture Detection after Derivat.ization with Heptafluorobutyric Anhydride James M. Moore,* Andrew C. Allen, and Donald A. Cooper
Special Testing and Research Laboratory, Drug Enforcement Administration, 7704 Old Springhouse Road, McLean, Virginia 22102
A unique derlvatlzatlon reaction used In the gas chromatographlc-electron capture detection of selected manufacturlng lmpurltles In llllclt heroln Is described. The reaction involves the perfluoroacylatlon of A'6~1'-dehydroherolnlumchloride, A15*16-dldehydroheroIn, noscaplne, and related Impurities wlth heptafluorobutyrlc anhydride (HFBA) In the presence of 4(dlmethylamino)pyrldlne (4-DMAP). Thls reaction Is novel In that It Introduces a heptafluorobutyryl (HFB) group, In high yleld, at carbon sites, namely, at C-15 In A15*16-dldehydroheroln and C-4 In noscaplne. These electrophlles, along wlth the expected 0 - and Ill-HFB derlvatlves of other heroln Impurltles, are easily detected on-column at plcogram levels by using a fused slllca column In the splltless mode. The method Is also appllcable for the low-level detection of the Ill-oxides of alkaloids such as morphlne.
The characterization of manufacturing impurities in illicit drugs is important for forensic purposes, especially in sample comparison cases (1-9). We report here methodology suitable for the analyses of complex matrices of heroin manufacturing impurities at levels below 0.01 % by using capillary GC-ECD. The detection of manufacturing impurities in heroin, or other drugs, by GC-ECD usually requires prior derivatization with an electrophilic anhydride or acid chloride. The introduction of an electrophilic group in these impurities inevitably occurs a t 0 and N sites substituted with labile protons (10, 11). Although most heroin impurities in our study were amenable to these common acylation reactions, a select few reacted unexpectedly, in that in the presence of 4-DMAP, HFB groups were introduced in high yield a t carbon sites. Further investigations revealed that these reactions were also applicable for the sensitive detection of N-oxides of impurities such as morphine and codeine as C-HFB derivatives. We believe that this is the f i s t report of such reactions applied in the GC-ECD analyses of heroin and other drug manufacturing byproducts. Our derivatization procedure has been applied to unadulterated, illicit heroin samples as well as selected standards. The 0-,N - , and C-HFB derivatives of the heroin impurities were chromatographed in the splitless mode on a bonded, nonpolar, fused silica capillary column interfaced with a 63Ni ECD. Minimum on-column detectable quantities at low pi-
cogram (pg) or high femtogram (fg) levels for most HFB derivatives were easily achieved. The methodology was also shown to be reproducible, an important consideration in sample comparison cases.
EXPERIMENTAL SECTION Instrumentation. Nuclear magnetic resonance ('H NMR) spectra were obtained on a Nicolet (Fremont, CA) 200-MHz spectrometer interfaced with an 1180 data system and 293A pulser. Tetramethylsilane was used as an internal standard. All mass spectra (MS) were acquired on a Finnigan 4600 quadrupole mass spectrometer (Sunnyvale,CA). Electron impact (EI) mass spectra were collected at an ionizing potential of 50 eV and an ionizing current of 30 PA. Chemical ionization (CI) mass spectra and electron capture negative ion chemical ionization mass spectra (NICIMS)were obtained by using methane (0.5 torr uncorrected) as the reagent gas. The GC-MS was fitted with a 20 m X 0.20 mm i.d. fused silica capillary column coated with SE-54 (Hewlett-Packard, Avondale, PA). Hydrogen was used as the carrier gas. Infrared (IR) spectra were recorded in KBr on a Beckman 4240 spectrometer (Irvine, CA). Gas Chromatography-Electron Capture Detection. All standard and sample chromatograms were generated in the splitless mode on a Hewlett-Packard 5880A gas chromatograph fitted with a 15 m X 0.25 mm i.d. fused silica capillary column coated with DB-1 (J and W Scientific,Inc., Rancho Cordova, CA) at a film thickness of 0.25 wm. The GC was equipped with a 63Ni detector (15 mCi) and interfaced with a Hewlett-Packard Level IV data processor. Injector and detector temperatures were maintained at 300 "C and 275 "C, respectively. The oven temperature was multilevel programmed as follows: (level 1)initial temperature, 90 "C; initial hold, 1.8 min; temperature program rate, 25 OC/min; final temperature, 160 OC; final hold, 1.0 min; (level 2) temperature program rate, 4 OC/min; final temperature 275 "C. Helium ("Zero Grade") was used as the carrier gas at a flow rate of about 40 cm/s and measured at an oven temperature of 90 "C. An argon/methane (95/5) mixture was used as the detector makeup gas at a flow rate of 30 mL/min. The septa used were Thermogreen LB-1 (Supelco, Inc., Bellefonte, PA). All chromatogramswere recorded at an attenuation of 26 and a chart speed of 0.5 cm/min. During the splitless injection the solvent was vented after a 1.0 min hold. Reagents. The 4-DMAP was obtained from Alfa Products (Danvers, MA). Isooctane, acetonitrile, and diethyl ether were "Distilled in Glass" products of Burdick and Jackson Laboratories (Muskegon, MI). HFBA, supplied in 1-mL sealed ampules, was obtained from Pierce Chemical Co. (Rockford, IL). The pH 4 phthalate buffer was prepared according to the United States
This article not subJect to U.S. Copyright. Published 1984 by the American Chemical Society
