Anal. Chem. 1983, 55, 787-790
is available a t no charge to academic laboratories. Inquiries may be addressed to the corresponding author.
ORACLE
ACKNOWLEDGMENT The authors thank Ani1 Kumar for technical assistance. LITERATURE CITED (1) (2) (3) (4)
(5) (6) (7) (8) (9) (10) (11)
Carman, C. J.; Wllkes, C. E. Macromolecules 1974, 7 . 40-43. O’Nelll, I. K.; M. A. Pringuer Org. Mag. Reson. 1974, 6 , 398-399. Sarneskl, J. E.; Reilley, C. N. Anal. Chem. 1978, 48, 1303-1308. Levy, G. C.; Hewitt, J. M. J . Assoc. Off. Anal. Chem. 1977, 60, 241-243. Abldl, S. L. Anal. Chem. 1982, 5 4 , 510-516. Forsyth, D. A.; Hediger, M.; Hahmoud, S. S.; Glessen, B. C. Anal. Chem. 1982, 5 4 , 1896-1898. Shoolery, J. N. Prog. NMR Spectros. 1977, 1 1 , 79-93. Marecl, T. H.; Scott, K. N. Anal. Chem. 1977, 49, 2130-2136. Thiault, B.; Mersseman, M. Org. Mag. Reson. 1978, 8 , 28-33. Blunt, J. W.; Munro, M. H. G. Aust. J . Chem. 1978, 2 9 , 975-988. LaMar, G. N. J . Am. Chem. SOC. 1971, 9 3 , 1040-1041.
787
(12) LaMar, G. N. Chem. Phys. Len. 1971, 10, 230-232. (13) Freeman, R.; Pachler, K. G. R.; LaMar, G. N. J . Chem. Phys. 1971, 5 5 , 4586-4593. (14) Natusch, D. F. S. J . Am. Chem. SOC. 1971, 9 3 , 2566-2567. (15) Levy, G. C.; Cargloll, J. D. J . Magn. Reson. 1973, IO, 231-233. (16) Levy, G. C.: Edlund, U. J . Am. Chem. SOC. 1975, 9 7 , 4482-4485. (17) Freeman, R.; HIII, H. D. W.; Kaptein, R. J . Magn. Reson. 1972, 7 , 327-329. (18) Canet, D.; Levy, G. C.; Peat, 1. R. J . Magn. Reson. 1975, 18, 199-204. (19) “Sadtler Standard I3C NMR Spectra”; Sadtler Research Laboratories, Inc.: Philadelphia, PA: 1976, Vol. 1; 1977, Vol. I O . Vol. 15. (20) Pearson, G. A. J . Magn. Reson. 1977, 27 (2), 285-272.
RECEIVED for review October 12,1982. Accepted January 3, 1983. Financial support from the National Science Foundation (Grant CHE 81-05109) and the Division of Research Resources, National Institutes of Health (Biotechnology Research Resource, RR 01317), is gratefully acknowledged.
CORRESPONDENCE Study of Chromatographic Retention with Spectroscopic Bandwidths Sir: An improved understanding of chromatographic retention behavior has been provided from statistical thermodynamic descriptions of solute distribution coefficients (1-3). Also, spectroscopic measurements have been developed to obtain specific structural information about bonded stationary phases ( 4 , 5 ) . The link between these two areas of study is the determination of the relationship between distribution coefficients and the structures of the chromatographic phases. The purpose of this work is to establish a first step toward developing an understanding of the intermolecular interactions giving rise to the molecular shape dependence of solute retention. The fundamental mechanisms that determine the structure dependence of chromatographic retention are presently not well understood. It is widely accepted that the polar mobile phase gives rise to the retention selectivity for polycyclic aromatic hydrocarbrons using n-alkylsilane stationary phases (6). The selectivity is thought to arise from hydrophobic exclusion in the mobile phase, with the stationary phase being inert. This idea is consistent with the observations that the selectivity decreases markedly upon changing mobile phase composition and less upon changing the nonpolar stationary phase. A mobile phase origin of PAH selectivity is also consistent with the inverse correlation between retention time and molecular size (7). On the other hand, there are recent chromatographic data that are more consistent with the stationary phase as the origin of selectivity. These data show that the retention indexes of polycyclic aromatic hydrocarbons (PAH)correlate with length-to-breadth ratio in LC separations employing a C18stationary phase with an acetonitrile-water mobile phase (8). The fact that solutes having higher length-to-breadth ratios are retained longer by the n-octadecyl stationary phase suggests that the shape selectivity originates from the stationary phase rather than the mobile phase. Chromatographic retention measurements alone are insufficient to identify unambiguously the contributions to selectivity from each phase because retention inherently senses the difference between two phases. Spectroscopy allows study of the phases individually. It can potentially provide information pertinent to chromatographic retention mechanisms
because the intermolecular interactions that determine retention also control the positions and widths of spectroscopic bands. The relation between spectroscopic and chromatographic measurements is thus the attractive and repulsive intermolecular interactions that influence both measurements. The observed molecular shape selectivity of chromatographic retention suggests that steric interactions between the solute and its environment are controlling selectivity among isomers. Steric interactions for nonpolar molecules occur through short-range repulsive forces between molecules. Spectroscopic bandwidths are known to be affected by repulsive and attractive forces that cause modulations of the quantum states of the molecule. The relation of these interactions to vibrational line shapes is the subject of vibrational dephasing theory (9-11). Vibrational spectra are not studied in this application because a large, unknown contribution from lifetime broadening diminishes the reliability of such measurements. Since the same interactions operate on electronic spectra, the ideas developed in vibrational dephasing theory are adapted to this discussion. Both the temporal properties and the strengths of the attractive and repulsive interactions contribute to spectroscopic bandwidths. For nonpolar molecules, where the interaction strengths are weak, it is likely that the repulsive, steric interactions largely control the bandwidths; for polar phases, both attractive and repulsive interactions contribute. The fact that the steric interactions are evident in the spectroscopic bandwidths is the principle behind this work. The purpose of this work is 2-fold: first, the existence of a relationship between spectroscopy and chromatography is explored, and second, the spectroscopic measurements are applied to the study of the origin of retention in the reverse-phase separation of polycyclic aromatic hydrocarbons. To study the shape dependence of the solute-solvent steric interactions, the retention times and the electronic spectral bandwidths of solutes having varying shapes were measured. The bandwidths for the solute in each solvent should thus reveal the extent to which the corresponding phase contributes to shape selectivity in chromatographic retention. A series of dimethylnaphthalene derivatives was chosen to achieve a
0003-2700/83/0355-0787$01.50/00 1983 American Chemical Soclety
788
ANALYTICAL CHEMISTRY, VOL. 55, NO. 4, APRIL 1983
Table I. Correlation of Spectroscopic and Chromatographic Data naphthalene derivative
retention timea
bandwidth difference,b cm-'
n-dodecane bandwidth, cm-'
85% AN-W bandwidth, cm-'
LIBd
1,4-dimethyl 2,3-dimethyl 1,3-dimethyl 2,6-dimethyl
0.917 0.932 0.961
6.8 f 1.7c 10.2 f 1.0 27.2 f 2.5 50.0 t 2.0
230 21 3 190 170
237 223 217 220
1.10 1.38 1.30 1.53
1.000
Difference between the bandwidth in 85% acetonitrile-water and that in a Relative to 2,6-dimethylnaphthalene. n-dodecane. 7 5% confidence limits. Length-to-breadth ratio. variety of molecular shapes without significantly perturbing the polarizabilities of the ground and excited states. The electronic 0-0 bands of the first excited singlet states were studied in order to minimize lifetime relaxation effects on the bandwidths. The difference between the bandwidth in the polar phase and that in the nonpolar phase was used to represent the additional repulsive contribution from the polar phase over that from the nonpolar phase. Polar solvents acetonitrile-water and methanol, and nonpolar solvents ndodecane and benzene were used in the spectroscopic experiments. The results are compared with chromatographic retention measurements on comparable phases. The same octadecylsilane column used to determine the shape correlation of ref 7 was used in this work for the chromatographic measurements. It is assumed that the steric interactions between the solutes and the bulk nonpolar phase approximate those between the solute and the bonded stationary phase consisting of the same type of functional group.
EXPERIMENTAL SECTION Absorption spectra were obtained on a Cary 118C UV/visible absorption spectrometer operating with an approximate spectral bandwidth of 20 cm-l in the spectral region of interest. The spectra consist of several bands which are overlapped by different amounts for the different compounds. The 0-0 bands were used in all cases, with the assignments of the 0-0 bands for the various compounds obtained from published work (12). The bandwidths for the 0-0bands were quantified as the half-width at half-height, taken from the band center to the low frequency side. This procedure minimizes the effect of overlap with higher frequency bands on the measured bandwidths. The spectra for the four dimethylnaphthalene compounds used are given in Figure 1to illustrate the degree of band overlap. Hot bands do not appear to contribute significantly to the widths at half-height. All bandwidths reported are estimated to be precise to A5 cm-l, and the bandwidth differences to A7 cm-l, except as otherwise noted. The fwhm of the bands are typically 400 cm-'. To estimate the sytematic error on the bandwidths from the 20 cm-l instrument band-pass, the convoluted bandwidth can be approximated to be a Gaussian of half-width [A2+ B2]1/2,where A is 20 cm-l and B is 400 cm-', and both the bandwidth and band-pass are approximated to be Gaussians (13). The systematic error due to band-pass is therefore less than 1 cm-l, which is negligible. Liquid chromatographic separations were performed with a Separations Group Vydac 201TP polymeric octadecylsilane column. Two different mobile phases were used 85% (v/v) acetonitrile-water and methanol. A Waters Associates Model M-45 solvent delivery system was used with a Waters Model U6J loop injector and a Waters Model 400 absorption detector operating at 254 nm. 2,6-Dimethylnaphthalene was used as an internal standard for the naphthalene derivatives, which were not well resolved. The accuracy of the relative retention times is estimated to be A0.2%. RESULTS AND DISCUSSION The data plotted in Figure 2 show that there is a strong correlation between reversed-phase chromatographic retention and the differences in spectroscopic bandwidth between 85% acetonitrile-water and n-dodecane phases. The four dimethylnaphthalene derivatives occur in the same order in
Wavelength
--+
Flgure 1. Electronic spectra of the four dimethylnaphthalene com-
pounds illustrating the extent of band overlap. The 0-0 band is the lowest energy band in each case.
I
I
.92
1
I
0.94 096 0 98 R E L A T I V E RETENTION TIME
t
1.00
Flgure 2. Relationship between spectroscopic bandwidth difference and chromatographic retention time. The error bars represent 75% confidence levels.
spectroscopy as in chromatography. The shape factor is also shown to correlate reasonably well with retention time; however, the order for the close-lying 1,3- and 2,3-derivatives is predicted correctly by spectroscopy, whereas the shape factor predicts the wrong order in this case. The correlation is consistent with the prediction that the spectroscopic bandwidths describe the repulsive interactions controlling retention selectivity. A quantitative evaluation of the relation between bandwidth difference and retention, based upon these data, would not be realistic due to the small range of possible shapes obtainable by rearranging methyl groups. Other systems are to be studied in future work. The data of Table I show that the retention correlates with the spectroscopic bandwidths themselves in n-dodecane but not with the bandwidths in 85% acetonitrile-water. The bandwidths in the latter solvent are nearly equal for all solutes except 1,4-dimethylnaphthalene, which is possibly due to the
ANALYTICAL CHEMISTRY, VOL. 55, NO. 4, APRIL 1983
Table 11. Effect of Polar and Nonpolar Phase on the Bandwidth Differences (cm-' ) bandwidth differences naphthalene derivative benzene 100% AN-W 85%AN-W 60% AN-W 1,4-dimethyl 2,3-dimethyl 1,3-dimethyl 2,6-dimethyl
methanol
-1 5
-1 1
7
8
0
-9 3
-6
10
12
11 40
27 50
19 34 44
22
17
unusually strong band overlap for the l,4-derivative. That the correlation holds for the n-dodecane bandwidths alone indicates that the nonpolar phase provides most of the retention selectivity. This assumes that any differences in bandwidths among the derivatives arises only from repulsive interactions rather than from (1)attractive interactions, (2) inhomogeneous broadening due to a distribution of solvation environments, (3) inherent differences in polarizability, or (4) lifetime contributions to the bandwidths. The last two contributions have been minimized by our selection of compounds. The importance of the first two contributions may be determined from a band-shape analysis of the spectroscopic bands because the repulsive contribution is the only one that imparts a Lorentzian shape to the bands. The attractive contribution is expected to be Gaussian; the distribution of environments is not known but it would be fortuitous if it were a Lorentzian. While a quantitative band-shape analysis is currently under way, the profiles all appear to be Voigt. The bandwidths are of comparable magnitude for both solvents. The profiles for the solutes in n-dodecane appear to be nearly Lorentzian, thus supporting the conclusion that the bandwidth differences, and hence the chromatographic selectivity, arise from steric repulsive interactions. The profiles for the solutes in acetonitrile-water are nearly Gaussian, suggesting that steric repulsions are much smaller than for n-dodecane and that, of course, attractive contributions are much larger. These considerations of the data in Table I strongly support the conclusion that the steric interactions in the stationary phase give rise to selective retention of PAH. One limitation to the applicability of electronic bandwidth measurements in probing solute-solvent steric interactions is that the spectra only sense the repulsions at the aromatic part of the molecule, while much of the repulsion related to selectivity may occur at the methyl groups. It is likely that the methyl groups, which are only a few angstroms from the center of the ring, sufficiently disrupt the structure of the solvation environment that the increased repulsive interaction is sensed by the ring. Chromatographic retention behavior is strongly dependent upon the choice of phases, and the band broadening differences should exhibit the same strong dependences. Studies of bandwidths were undertaken for these same solutes in polar and nonpolar phases where chromatographic data were available. The effect of varying the nonpolar phase composition on the spectroscopic bandwidth differences is shown in the data of Table 11. For solvent benzene, the bandwidth differences for the solute molecules are less distinguishable than those for n-dodecane. This is consistent with the report that benzyl bonded phases provide poorer chromatographic resolution for substituted naphthalenes than do octadecyl bonded phases (6). Measurements for two types of polar phases were made: acetonitrile-water mixtures of different compositions and methanol. The data of Table I1 show the effecta of these polar phases on the spectroscopic behavior of the dimethylnaphthalenes. Changes in the composition of water in acetonitrile-water does not affect the range of bandwidth differences but does shift the values. This is consistent with the report that the water composition has little effect upon the
5
789
retention time (MeOH) 0.979 0.975 0.972 1.000
chromatographic selectivity of PAH with the C18stationary phase (14). The data show that the use of methanol as a polar phase results in bandwidth differences that are closer together and probably arranged in a slightly different order. The 2,6-derivative has a distinctly different bandwidth from the others. The low spectroscopicresolution predicts low chromatographic resolution, according to the arguments made earlier. In chromatography, methanol does give lower resolution, as shown in Table 11. In addition, the 2,6-derivative is distinguished from the other three, as it was in the spectroscopic data. The retention order among the three other compounds does not agree with the bandwidth order. The small quantitative disageement can be attributed to the failure of one of the assumptions that has been made, although the particular assumption is not identifiable without further experiments. The similar behavior of the solute molecules with respect to their bandwidths in methanol and n-dodecane suggests that similar steric interactions take place in each phase. It is thought from theoretical (15)and X-ray diffraction studies (16)that the bulk liquid structure of methanol consists of short, straight chains. The fact that both stationary and mobile phases are comprised of chains is consistent with the low steric selectivity for mobile phase methanol with CIS stationary phase. The agreement between the spectroscopic and chromatographic selectivity supports the predicted relation of these two measurements; however, caution must be used in quantitatively comparing different solvents. Attractive forces also contribute to spectroscopic band broadening and this contribution can be large for polar solvents due to large dipoleinduced dipole interactions. Hypothetically, the solutes in acetonitrilewater could have a larger bandwidth contribution from attractive forces than they do in methanol, thus minimizing the fraction of the acetonitrile bandwidths due to repulsive forces. While this situation could diminish the apparent shape selectivity of acetonitrile-water to hamper quantitative comparison, it is not likely to be so large as to change the interpretation of the results because the attractive interactions are approximately independent of the position of the methyl group. To accomplish quantitative comparisons among chromatographic phases, a more detailed analysis of the band shapes providing separation of the attractive and repulsive contributions is currently under way. The experimental results show that spectroscopic bandwidth measurements are sensitive to the subtle solvation interactions that control selective chromatographic retention and can thus be used to probe the fundamental retention processes. The approach in this work of using behavior in bulk solvents to infer bonded phase behavior is an approximation because, first, the bonded phase structures are not known but are unlikely to be the same as the bulk liquid structures, and second, the bonded phase structures may be altered by the mobile phase. Since the bulk and bonded n-alkanes are comprised of the same molecular structures, which determine the overall phase structure, the approximation is thought to be adequate for defining the relationship between the spectroscopic and chromatographic measurements. With further
790
Anal. Chem. 1983, 55,790-792
characterization, spectroscopy can provide valuable information aboutthe processes to aid in enhancing selectivity of chromatographic separations. - -
LITERATURE CITED Snyder, L. R. Anal. Chem. 1974, 46, 1384. Jaroniec, M.; Rozylo, J. K.; Golklewlcz. W. J . C h r m t o g r . 1979, 178,
(12) Petruska, J. A. In “Systematlcs of Electronic Spectra of Conjugated Molecules”; Platt, J. R., Ed.; Wlley: New York, 1964. (13) ~ ~ a c ~R.~ N.~ “The l l , ~~~~l~~ Transform and itS Appllcatlons”; McGraw-HIII: New York. 1978. (14) Ogan, K.; Katz, E.; Siavin, W.-Anal. Chem. 1979, 51, 1315. (15) Jorgensen, W. L. J . Am. Chem. SOC.1980, 102, 543. (16) Maglni, M.; Paschlna, G.; Piccaluga, G. J . Chem. Phys. 1982, 77, 2051.
Mary J. Wirth* David A. Hahn Ronald A. Holland
27.
Boehm, R. E.; Martire, D. E. J . fhys. Chem. 1980, 84, 3620. Lochmuller, C. H.; Marshall, D. 6.; Wlider, D. R. Anal. Chlm. Acfa 1980, 130, 31. Lochmuller, C. H.; Marshall, D. 6.; Harrls, J. M. Anal. Chlm. Acfa 1981, 731, 263. Colin, H.: Guiochon. G. J . Chromatoor. 1977. 141. 289. Sleight, R. B. J . Chromatogr. 1973,-83, 31. Wise, S. A.; Bonnett, W . J.; Guenther, F. R.; May, W. E. J . Chromatogr. Scl. 1981, 19, 457. Schweizer, K. S.; Chandler, D. J . Chem. Phys. 1982, 76, 2296. Oxtoby, D. W. J . Chem. Phys. 1981, 74, 5371. Qeorge, S. M.; Auweter, H.; Harrls, C. B. J . Chem. Phys. 1980, 73,
5573.
Department of Chemistry University of Wisconsin-Madison Madison, Wisconsin 53706
RECEIVED for review September 3,1982. Accepted December 22, 1982* We gratefully the research Of the National Science Foundation.
Investigation of Adsorption of Benzene on Nickel(001) by Secondary Ion Mass Spectrometry Sir: It is now possible, using secondary ion mass spectrometry (SIMS) or fast atom bombardment (FAB), to form gas-phase molecular ions directly from organic and inorganic solids (1-3). This ionization scheme, where the sample is bombarded by a particle with 1-5 keV of kinetic energy, has proven particularly useful for the analysis of thermally labile organic solids such as amino acids (I,4 ) and for the characterization of biomolecules with molecular weights io the 1000-10000 dalton range (5). The mechanism of ejection of these large molecular species has been the subject of numerous speculations since it appears improbable at first thought that a large organic molecule held together by tens of electronvolts of energy could withstand impact by a 1000-eV primary particle. Although no rigorous approach has been developed to fully explain the observations, a number of conceptual advances in understanding the ejection mechanism have been made utilizing classical dynamics calculations to model the dissipation of energy of the primary particle (6). For benzene adsorbed on Ni(001) for example, molecular ejection is favored since (i) the primary ion energy is rapidly dissipated in the solid such that atomic collisions with the benzene molecule are often only a few electronvolts, (ii) the organic molecule possesses many internal vibrational modes which can absorb excess energy from a violent collision, and (iii) the metal substrate atom is larger than a carbon atom which allows it to strike two or three carbon atoms simultaneously, pushing the entire molecule in one direction (6). It remains to be shown how many of these concepts are extendable to molecules of much higher molecular weight. In this paper, we present preliminary SIMS experimental resulta for the adsorption of benzene on Ni(001) with the goal of comparing our observation to the previous benzene/Ni(001) classical dynamics calculations (6, 7). The results show that, in agreement with studies by other techniques, that the adsorption occurs molecularly at room temperature reaching monolayer coverage after approximately 1langmuir benzene exposure. In this environment, the only molecular ion that is observed with significant intensity is NiCsHe+,in contrast to the very complex spectra found from Ni surfaces exposed to much higher quantities of benzene. We also present evidence for the mechanism of cationization of benzene by Ni, which is predicted to occur via recombination of benzene with 0003-2700/83/0355-0790$01 S O / O
Ni+ ions over the surface of the solid. In general, this model system is utilized to better understand the SIMS spectra of organic solids and to evaluate the importance of various experimental parameters on the quality of the mass spectra.
EXPERIMENTAL SECTION Experiments were performed in an ion-pumped stainless steel torr after vacuum chamber with base pressure of 3.0 X bakeout. The primary ion was generated by a Physical Electronics sputter gun that had been modified to be differentially pumped by a 6-in. trapped diffusion pump. The gun could produce an ion beam with energies ranging from 75 to 1500 eV. With this arrangement we could perform SIMS measurements at an optorr with the ion gun normally erating pressure of 1.0 X operated at 3 X lo4 torr of Ar. An Extranuclear 162-8quadrupole mass spectrometer was employed to perform the secondary ion detection. The spectrometer was fitted with a Bessel box type energy analyzer with energy resolution typically between 1and 2 eV. In addition, the filter could be electrically scanned for use in observing secondary ion energy distributions. The primary ion was incident on the sample at 45O and the analyzer was positioned to detect ejected ions at 45” with a half-angle acceptance of 16”. The signal was amplified by a Channeltron electron multiplier coupled to a PAR 1120 amplifier-discriminator. The nickel(001) crystal was cut and finally polished with 0.05-pm alumina to i0.5” orientation. Before the crystal was mounted, it was etched in a 31:1:5 mixture of concentrated HN03, H2S04,H3P04,and CH,COOH and rinsed in deionized water and ethanol. The crystal was then mounted on a sample holder that could be resistively heated to 1300 K as monitored by a chromel-alumel thermocouple. The sample gas consisted of spectra grade benzene (Fischer) which had been degassed through a series of freeze-pump-thaw cycles. The benzene was admitted through a variable leak valve, with the pressure measured by a nude Bayard-Alpert ionization gauge calibrated for air. Crystal cleaning was restricted to cycles of heating in oxygen (5 X lo-‘ torr, 1200 K,10 min) and hydrogen (1 X lo4 torr, 1200 K, 10 min) followed by sputtering and annealing at 1200 K. The surface was considered clean when no Of and NiO+ peaks could be observed in the SIMS spectra and when the Niz+/Ni+ratio approached 0.25 (8). RESULTS AND DISCUSSION The SIMS spectrum of the clean Ni surface is given in Figure la. Note that the only peaks other than Ni+ and Niz+ are the ubiquitous Na+ and K+ contamination peaks. This 0 I983 Amerlcan Chemical Soclety
