Comparison of three-and five-micrometer column packings for

Apr 9, 1982 - (1) Cretier, G.; Rocca, J. L. Presented at Journées de Chromatographle en phase liquide, Paris, Déc 1981. (2) Glueckauf, E. Trans. Far...
0 downloads 0 Views 985KB Size
Anal. Chem. 1982, 54, 2277-2283

Table V. Purity p and Recovery Ratio r of the Collected DBP Fractions example %P %r n o 0 ,nn n, ' 3 0 . 1 (3'3.U)

N

B

--

98.9 (99.0)

89.8 (90.0) 90.8 (90.0)

device can deliver a perfect sample plug onto the preparative column.

ACKNOWLEDGMENT The authors thank the Jobin-Yvon Co. for the loan of the Miniprep LC chromatograph.

2277

(2) Giueckauf, E. Trans. Faraday SOC. 1955, 57, 34. (3) Snyder, L. R. J . Chromatogr. Sci. 1972, 10, 200. (4) Grenier-Loustalot, M. F.; Metras, F.; Campiiio, J. P.; Bonastre, J.; Grenier, P. J . Chromatogr. 1976, 178, 1. (51 . . Haarhoff. P. C.; Van Berge, P. C.; Pretorius. V. Trans. Faraday SOC. 1961, 57, 1838. (6) Rellley, C. N.; Hlldebrand, G. P.; Ashley, J. W., Jr. Anal. Chem. 1962,

.

i d i i,--. aa "_, 171 Coo. B.: Gonnet. C.: Rocca. J. L. J . Chromatow. 1975, 106, 249. i 8 j Cod; B.1 Cretier, G.; Rocca, J. L. J . Chromato&-. 1979, 786, 457. (9) Coq, B.; Cretier, G.; Rocca, J. L. J . Chromatogr. Sci. 1961, 79, 1. (10) Conder. J. R.: Shinaari, M. K. J . Chromatogr. Sci. 1973, 1 1 , 525. i l l j Conder, J. R. Chro&tographia 1975, 8 , 60: (12) Roz, 8.; Bonmati, R.; Hagenbach, G.; Valentin, P.; Gulochon, G. J . Chromatogr. Sci. 1976, 14, 367. (13) Hupe, K. P.; Lauer, H. H. J . Chromatogr. 1981, 203, 41.

LITERATURE CITEID (1) Cretier, G.; Rocca, J. L Presented at Journges de Chromatographle en phase liquids, Paris, DBc 1981.

RECEIVED for review April 9, 1982. Accepted July 28, 1982.

Comparison of Three!- and Five-Micrometer Column Packings for Reversed-Phase Liquid Chromatography Nelson H. C. Cooke," Benedlct G. Archer, Krlstlne Olsen, and Alan Berick Beckman Instruments, Inc., 1780 Fourth Street, Berkeley, California 947 10

Columns packed wlth 3-pm and 5-pm ODS bonded slllcas had similar reduced plate heights and separation impedances. Comparlsons of 3-pm and 5-pm partlcles, demonstrated the practicality of the smaller partlcles and confirmed thelr advantage In reduclng analysts times. Because peak volumes wlth 3-pm columns are vsry small, reduced efficiency can result when used wlth conventlonal absorbance detectors. Efflclency losses can hie minimized by reduclng flow cell volume and other extracolumn volumes, but a shortened pathlength wlll decrease detectabillty. Detectablllty wlth mass sensitive detectors may however be Increased. A potentially serlous llmltatlon to the use of 3-pm particles at hlgh flow rates is frictlonal heating which can lead to large efficiency losses. It Is concluded that 3-pm particles offer a unique advantage when separation speed is important but 5-pm partlcles may be better matched to conventional equipment and are recommended ifor general use.

Over the past several ,years, much effort has been directed toward increasing both speed and efficiency in liquid chromatography (LC). During this time there have been major advances in the understanding of band broadening phenomena in LC and in the dependence of speed, efficiency, pressure, detectability, etc. on particle size and column dimensions. At present, the compromises (e.g., between time and efficiency) which must be made in optimizing an LC separation are well understood (1-6)< Moreover, the art of packing LC columns has reached a level such that for a given particle size the efficiency of well-packed columns is quite predictable. At present, most analyses are carried out on columns 15-25 cm in length and 4-5 mm internal bore packed with 10-pm or 5-pm particles with a definite trend toward greater use of the latter. In particular, the versatility and high efficiency

of reversed-phase columns packed with alkyl bonded silica has become widely recognized (7,8). Separations requiring 10 000-20 000 theoretical plates can be performed on such columns in 5-15 min. Although, in theory, for specified efficiency and pressure, column optimization for minimum analysis time could be done by varying particle size, until very recently particles smaller than 5 pm were not generally available. It is well recognized that for many applications, the use of well-packed columns containing 2-3-pm particles would offer a substantial advantage in decreasing separation time. The potential of these benefits has provided stimulus for a number of workers to explore the use of small (3 pm) particle columns. Thus, some time ago, Halasz ( 4 ) , Unger (9),and Kirkland (10) demonstrated the use of 3-pm particles in liquid-solid adsorption chromatography. The first use of 3-pm particles in the reversed-phase mode was reported by Cooke et al. in 1980 (8, 11). More recently, reports of applications on commercially available 3-pm reversed-phase columns have appeared (12-15). The lengths of columns with similar efficiencies packed equally well with different sizes of particles will be in proportion to their particle sizes. Thus substitution of a short 3-pm column for a longer one packed with larger particles will result in smaller peak volumes which are susceptible to serious extracolumn dilution unless extracolumn volumes (e.g., capillary tubing connections, flow cell, etc.) are stringently minimized. If, for example, flow cell volume is not sufficiently reduced, or the detector time constant is too large, the shorter analysis time on the 3-pm column may be achieved at the expense of decreased resolution. If, on the other hand, the flow cell volume and path length are reduced in order to avoid excessive extracolumn losses in efficiency and resolution, without an equivalent reduction in base line noise sample detectability may be decreased. Beyond these considerations is the higher operating pressure which always results when particle size is reduced. Thus, the practicing analyst must

0003-2700/82/0354-2277$01.25/0 0 1982 American Chemical Society

2278

ANALYTICAL CHEMISTRY, VOL. 54, NO. 13, NOVEMBER 1982

be aware of the compromises necessary during column selection in order to assess the potential benefits of short 3-pm columns for a particular application problem. In light of the recent interest in 3-pm columns, it seems appropriate at this time to critically assess the use of these columns relative to 5-pm particle columns. Moreover, although column optimization for 5-pm and larger particles is well understood, additional considerations such as extracolumn viscous heating (17, 18),etc. may become espeeffects (16), cially significant for particles less than 5 pm. It is the specific purpose of this report to examine the pratical merits of these columns for RPLC relative to (1)column performance and extracolumn band broadening effects, (2) chromatographic speed, and (3) sample detectability. We restrict, here, our discussion to isocratic chromatography and will explore gradient elution in a subsequent paper. EXPERIMENTAL SECTION Apparatus. All measurements were made with a Beckman Series 341 chromatograph consisting of a Model 112 pump, Model 210 injector with 5 - p L sample loop, and Model 160 detector operated at 214 nm with standard 20-yL (10 mm path) or 5 - p L (4.5 mm path) flow cells. Approximately 16 cm of 0.18 mm capillary tubing connected the 3-pm columns to the injector (4 cm) and flow cell (12 cm including 4 cm heat exchanger). A slightly longer (6 cm) piece of 0.18 mm capillary connected the 5-pm columns to the injector. Approximately 20 cm of 0.25" capillary of which 6 cm served as heat exchanger connected the 5-pm columns to the 20-pL flow cell. A Trilab I11 (Trivector Systems, Ltd., U.K.) laboratory data system was used to collect, store, and analyze chromatograms. The integrated detector signal was sampled at 20-ms to 100-ms intervals by a voltage/frequency converter, and digitized chromatograms were stored on magnetic tape. A BASIC language program was used to analyze the raw data. Retention times were estimated either by the first normalized moments or by the maximum of each peak. Plate numbers were calculated from the first and second peak moments (19), and also from width at half-height correcting for asymmetry at 10% height, Ao.l (10). Nonchromatographic peak dispersion, Le., system variance, was estimated directly as the second moment of a 5-pL injection made with the injector connected directly to the detector. Columns and Chemicals. Ultrasphere ODs, maximum coverage, fully endcapped packings (Beckman Inc., Berkeley, CA) were used for all experiments. Column dimensions were 4.6 mm internal diameter by 75 mm or 150 mm in length for the 3-bm and 5-pm particles, respectively. All columns were thermostated by circulating water at 30 "C through column jackets. The mobile phase, except where stated otherwise, was 80% acetonitrile/20% water prepared by mixing the required volumes at ambient temperature. Test solutes for column evaluation were a homologous series (C,C,) of n-alkyl bromides (Aldrich),made up and injected in mobile phase. Other samples and reagents used are described in figure legends. Procedures. Mean packing particle sizes were estimated both by light microscopy and by dynamic means from the specific permeability, B. B was estimated from the slope of a plot of pressure against linear flow velocity according to the formula AP = uqL/B where u is the superficial linear velocity, 9 is the viscosity, L is the column length, and AP is the preasure drop (20). B is related to particle diameter, dp, by the Kozeny-Carman relation dp2c,3 B= (1) 180@(1 - E,)* where $ is a parameter equal to 1.3 for porous spherical particles (21)and E", the interstitial porosity, was taken to be 0.4 (9). The dynamically measured particle diameters were used to calculate all reduced parameter values. Dead volumes, VO,were estimated as the retention volume of modifier (injectedas modifier-enriched mobile phase) at the mobile phase composition giving the least retention of modifier, ca. 65% acetonitrile,assuming dead volume remains constant to 80% acetonitrile where most measurements were made (22). Column porosities of approximately 0.55 can be estimated from these dead volumes indicating very tightly

Table I. Performance Parameters for 3-ym and 5 y m Columnsa dp/,umb L/cm 3(3.4)

5(4.7)

7.5 15.0

voptC h,hc

9.5 9.7

1.8 1.9

cpd

Ee

840

2700 2600

710

N, 12700 16700

Determined in 80/20 (v/v) acetonitrile/water at 30 "C with 1-bromononane ( k ' = 9.2). Nominal values with dynamically measured values in parentheses. Values corresponding to minimum h in a plot of h vs. u . Optimal flow rates are 2.0 and 1.4 mL/min for the 3-pm and 5-rm columns, respectively, Optimal linear velocities are 0.40 and 0.29 cm/s, respectively. Column resistance Parameter ( 3 1 ) q5 = APt,(dp)Z/qL2. e Separation impedance ( 3 1 ) E = APt,/qN2 = hZ@. a

packed columns. Dead volumes consistently larger by 10-12% were estimated from extrapolations of the retention times of successive homologues (23). The estimates of dead volumes obtained by injecting modifier enriched mobile phase were used for all calculations. Reduced plate heights, h, were calculated from h = L/(N(dp)) where L and N are column lengths and plate numbers, respectively. Reduced velocities, v, were calculated in the usual way as u = u(dp)/(Dm). Diffusion coefficients in the mobile phase, Dm, were estimated from the Wilke-Chang equation (24) using a weighted mean of the product of solvent association parameters and molecular weights for the component solvents (3).

Column resistance parameters, 4, and separation impedances, E, were calculated as described in Table I. The values of 4 and E, both measures of overall column performance, are useful in comparisons between columns. The column resistance factor, 4, represents the pressure drop due to column factors, AP/q, multiplied by column dead time, to, both AP/9 and to being normalized with respect to particle diameter by the coefficient dp/L. The separation impedance, E, is a similar composite except that the normalization is with respect to one plate. RESULTS AND DISCUSSION Column Performance and Extracolumn Effects. The basic performance data for the columns compared in this study are compiled in Table I. As all the packings used in this study were spherical and have essentially the same internal porosities, specific surface areas, and selectivities, comparisons between columns are simplified and can be made considering only particle size, column length, and operating conditions (25). The values of vopt, &in, 4, and E, which are typical of measurements on several columns, easily meet the criteria established by Knox for well-packed, efficient columns (vOpt is the reduced velocity corresponding to /zmin) (5). A comparison of h,h and vOpt for the two particle sizes demonstrates that the previously encountered difficulties in packing particles smaller than 5 pm to obtain the same reduced performance parameter values have been overcome, i.e., similar values of hminclose to 2.0 are obtained for both particle sizes (Table I). In Table 11, the performance of 3-pm and 5-pm columns at optimum flow is compared for a range of retention. It is worth noting the discrepancies between the column plate numbers, N,for the early peaks computed from peak widths at half-height-especially as these values were determined from quite symmetrical peaks. Earlier eluting peaks are typically less symmetrical, and, consequently, have much more disparate N values when computed by the alternate methods (e.g., a 50% overestimate of N is expected if estimated by N1p for a peak with Ao,l = 1.4) (IO). It should be emphasized that meaningful comparisons of system or column performance are only possible if reliable methods of computing peak variances have been used. Except for the N l l z values in Table 11, all plate numbers, plate heights, and variances presented here are true values for the complete system computed from second central statistical moments.

ANALYTICAL CHEMISTRY, VOL. 54, NO. 13, NOVEMBER 1982

2279

Table 11. Typical. Performance of 5-pm and 3-pm Columns at Optimal Flow Ratesa

N

k’

N(u2)

Ao.1

dp = 3 fim, L = 7.5 cm 0.7 1.0

1.4 2.1 3.0 4.3 6.3 9.2

9 700 1 0 500 11 300 11900 1 2 200 1 2 300 1 2 100 1 2 000

11 200 11 800

1.08

12 300 12 400 12 600 12 600 1 2 300 12 000

1.00 1.00

1.13 1.04 1.04 0.95 0.97

dp= 5pm,L=15cm 0.7 1.0 1.5 2.1

3.1 4.5 6.3 9.5

17 000 17 900 18 300 18 500 18 500

1 2 800 13 900 1 5 900 1 6 900 1 6 400 16 900 17 000 17 000

18 400 18 100 17 800

1.11 1.06

1.04 0.99 0.99 0.95 0.96 0.93

a 2.0 and 1.4 mL/min for the 3-pm and 5-pm columns, resoectivelv.

Table 111. Peak Widths on Standard Columns at Optimal Flow Ratesa dp

=

3 Mm,

L = 7.5 cm --___ k‘

0 1

2 5

10

4av/pL 22 45

4ot/s

0.7 1.3 2.0 67 134 4.0 24 5 7.3 N = 12700

dp = 5pm,

L = 15cm -

4oV/pL

4ot/s

38 1.6 75 3.2 4.8 113 225 9.7 41 3 17.7 N = 16700

a 2.0 and 1.4 mL/min for the 3-pm and 5-pm columns, respectively.

Grushka et al. (19) have shown that computer-calculated second moments can carry large errors if base line determinations are inaccurate. A base line too high or too low by 1.5% of peak height caused measured peak variances to be in error by -12% or +35%, respectively. We estimate the uncertainty in base line to be