Anal. Chem. 1993, 65, 268-276
268
Origin of Indirect Detection in the Liquid Chromatography of a Neutral Sample with an Ionic Probe Using an ODs Bonded Phase and Aqueous Mobile Phase Laura L. M. Glavina and Frederick F. Cantwell’
Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
Indirect UV detectlon of the nonlonlc sample butanol Is achieved on a Partlsll 10 ODS-3 HPLC column by adding a constant concentratlon of anlonic probe naphthalene-2sulfonate (NS) to the mobile phase. Butanolelution Is detected as a posltlve peak and Isfollowed by a negative system peak. This pattern of peaks suggests that In the presenceof butanol, the sorptlon of NS- Is decreased. The reason why this Is so was Investigated by measuring the slmultaneous sorptlon of butanoland NS-on a small (-0.1 g) column of the O D s packing via the column equlllbration technlque. A plot of moles of NSsorbed versus moles of butanol sorbed from mobile phases contalnlng a constant concentratlon of NS- and Increasing concentrations of butanol was linear wlth a negative slope. I n addltion, a plot of moles of butanol sorbed versus moles of NS- sorbed from mobile phases contalnlng a constant concentratlonof butanol and increasingconcentrationsof NSwas also linear with a negatlve slope. These observatlons reveal that butanol and NS- compete with one another for space In the ODS stationary phase and that butanol, in the concentratlonrange 0-0.04 mol/l, does not dgnlflcantlyalter elther the solvent strength of the moblie phase or the sorbent strength of the bonded phase, for NS-. This Is the origin of indirect detection of butanol wlth NS-.
INTRODUCTION When indirect UV detection of a non-UV-absorbingsample species is performed on a reversed-phase-bonded-phase (RPBP) liquid chromatography column, a UV-absorbing species (the probe) is added as a component of the mobile phase. Before a sample is injected, the probe distributes itself between the mobile and stationary phases, establishing equilibrium. When a sample is injected, the probe distribution equilibrium is disturbed, with the result that a peak in the positive direction or in the negative direction appears a t the retention time of the sample species and another peak, of opposite direction, appears a t the probe retention The latter peak is normally referred to as the system peak. The sample and system peaks arise from local changes in probe concentration, since it is only the probe which absorbs UV light. The pattern of sample and system peaks, in terms of their direction and their elution sequence, is under~tood.l-~,*-8 When probe sorption is decreased by the presence of the sample, the sample peak will be positive and the system peak will be (1)Schill, G.; Crommen, J. Trends Anal. Chem. 1987,6, 111-116. (2)Schill, G.; Arvidsson, E. J. Chromatogr. 1989,492,299-318. (3)Bidlingmeyer, B.; Warren, F. V. Anal. Chem. 1982,54,2351-2356. (4)Barber, W. E.;Carr, P. W. J. Chromatogr. 1984,316, 211-225. (5)Levin, S.; Grushka, E. Anal. Chem. 1986,58,1602-1607. (6)Stranahan. J. J.: Demine. S. N. Anal. Chem. 1982.54.154&1546. (7)Vigh, G.; Leitold, A. J. bromatogr. 1984,312, 345-356. (8)Takeuchi, T.;Watanabe, S.; Murase, K.; Ishii, D. Chromatographza 1988,25,107-110. 0003-2700/93/0365-0266$04.00/0
negative if the sample elutes before the probe. However, if the sample elutes after the probe, then the system peak will be positive and the sample peak will be negative. On the other hand, when probe sorption is increased, rather than decreased, by the presence of the sample, then the above patterns will be reversed. Previous studies in indirect detection have used both ionic and nonionic species as probes. When the probe is ionic, it is found that sample species which have a sign of charge opposite to that of the probe increase the sorption of the probe, while sample specieswhich have the same sign of charge as the probe or have no charge (neutral sample species) decrease the sorption of the probe. A question which has often been asked is why does a particular sample type (i.e sign of charge) produce the observed increase or decrease in probe sorption. For systems where both the probe and the sample are ionic, the suggested explanations for the increase or decrease in probe sorption brought about by the presence of the sample have usually been formulated in terms of the particular physicochemicalsorption model (i.e. “retention mechanism”) that is favored by ita authors to explain so-called “ion-pair” chromatography. One model is based on the assumption that ion pairs formed between the probe, sample, and other mobile phase ions are sorbed onto the stationary phase.*lZ In this model, competition for space on the stationary phase is expressed in terms of mixed Langmuir isotherms for sorption onto a particular type of surface site. The stationary phase may have either one or two types of surface sites, and ion pairs may be present either only in the stationary phase or in both the stationary and mobile phases. A second model invokes an ”ion-interaction r n e ~ h a n i s m ” ~ Jwhich ~ J ~ may include the influence of local changes in ionic strength.14 A third model is an ”electrostatic model” in which the electrical potential of the stationary-phase surface is influenced by sorption of sample ions as well as probe ions and competition for space in the stationary phase is accommodated by employing mixed surface potential-modified Langmuir isotherm~.’~ The presently reported work deals with a neutral sample. Several studies have previously been reported where the sample is neutral and the probe is either neutral or i o n i ~ . Generally ~ ~ ~ ~speaking, ~ J ~ ~the~presence of the sample can alter the extent to which the probe is sorbed because of ~~
~~~
(9)Sokolowski, A. Chromatographia 1986,22,177-182. (10)Hackzell, L.; Rydberg, T.; Schill, G. J. Chromatogr. 1983,282, 179-191. (11)Denkert, M.;Hackzell, L.; Schill, G.; Sjogren, E. J. Chromatogr. 1981,218,31-43. (12)Crommen, J.; Schill, G.; Westerlund, D.; Hackzell, L. Chromatographia 1987,24,252-260. (13)Michaelis, R.; Cassidy, R. M. Adu. Ion Chromatogr. 1990,2,2143. (14)Rigas, P. G.; Pietrzyk, D. J. Anal. Chem. 1988,60,454-459. (15)Stahlberg, J.; Almgren, M. Anal. Chem. 1989,61, 1109-1112. (16)Freiser, H.; Gnanasambandan, T. Anal. Chem. 1982,54,12821285. 0 1993 American Chemlcal Society
ANALYTICAL CHEMISTRY, VOL. 85, NO. 3, FEBRUARY 1, 1993
269
EXPERIMENTAL SECTION one or more of the following reasons. ( 1 ) The sample alters the solvent strength of the mobile Chemicals. Sodium naphthalene-2-sulfonate(NaNS)was phase. The presence of a neutral sample in the mobile phase prepared from 2-naphthalenesulfonicacid (Eastman Kodak) has been suggested to alter its strength, including the by converting the acid to the sodium salt followed by possibility of complex formation, when either an ionic recrystallization from water. n-Butanol (A&C American probel6J7 or a neutral probe8J9!23,25 is used. Chemicals) and 1-pentanol (Aldrich) were reagent grade and (2) The sample competes with the probe for space in the were used as received. Phosphoric acid (BDH Chemicals) stationary phase. Sorption of a neutral sample may displace and sodium hydroxide (BDH Chemicals) were reagent grade. sorbed probe from the stationary phase25i26or it may reduce Water was distilled and deionized (Barnstead NANOpure the effective surface area available for probe system, Boston, MA). Methanol (Anachemia) and ethanol Mixed Langmuir isotherms12J9and interfacial tension ef(Commercial Alcohol Ltd.) were reagent grade and were fects7*24,29 have been suggested to describe this competition. distilled before use. In a contrary sense, the results of one study were interpreted Mobile Phases and Eluent Solutions. In the elution as demonstrating that no competition occurs when a neutral chromatography experiments, the mobile phase contained sample is sorbed.3O 2.00 X mol/L NS- probe in 0.04 mol/L H3P04/0.03 moUL (3)Thesample alters thesorbent strength of thestationary NaH2PO4 buffer at pH 2. The sample solution to be injected phase. It is possible that the chemical properties of the was prepared by including 1.09 X moVL butanol in the stationary phase may be modified by the presence of a same mobile phase. ample.^^^^^ Alcohols, in particular, can solvate the stationary In the column equilibration experiments,the mobile phases phase and alter its polarity and structure when they are contained various concentrations of NS- and butanol in 0.04 present at significant concentrations in the mobile ~ h a s e . ~ l - ~ ~ moVL H3P04/0.03 moVL NaH2P04 buffer at pH 2. For the When an ionic probe is used, the surface potential created by various experiments,the following concentrationswere used: the probe sorbing onto the stationary phase may also be altered (a) For measurement of the butanol isotherm, the butanol by the ample.^^,^^ concentration was varied from 0 to 0.654 moVL. (b) In the The purpose of the present study was to consider the case study of the effect of NS- on butanol sorption, the butanol in which the neutral sample, butanol, elutes before the ionic concentration was kept constant at 1.09 X mol/L and the probe, naphthalene-2-sulfonate,and to determine the proNS- concentration was varied from 0 to 0.157 mol/L. (c) For cesses which cause the probe to be less sorbed in the presence measurement of the probe isotherms in the absence and in of sample than in ita absence. Naphthalene-2-sulfonate(NS-) the presence of butanol, the butanol concentration was kept was chosen as the probe because it has been used previously constant at 0 and at 2.18 X mol/L, respectively. (d) In in the literature in 0.05 moVL phosphoric acid to perform the study of the effect of butanol on NS- sorption, the NSindirect UV detection.lP2J1J2 Similarly, alcohols, including concentration was kept constant at 2.00 x moVL while n-butanol, have been used as neutral samples in inthe butanol concentration was varied from 0 to 4.36 X direct detection studies using both ionic and nonionic mol/L. After achievementof column equilibration with each probes.7,12.1"17,20,zl mobile phase, the sorbed NS- and butanol were eluted with In the present study the mutual effects of the sample on methanoVwater (1:l v/v). probe sorption and of the probe on sample sorption were In both the elution and the column equilibration experstudied by the column equilibration technique in which iments, all solutions were filtered before use through a 0.45solutions containing probe and sample are pumped through pm pore size Nylon 66 filter (Alltech Associates Inc., Guelph, a short column packed with an octadecylsilyl (ODS) bonded ON). Mobile phases and eluent solutions were degassed by phase until equilibrium is achieved. After equilibration, both sparging with helium (Linde), after which a blanket of helium sample and probe are eluted from the column and determined was maintained over the solution. quantitatively. This technique allows measurement of the ODS Sorbent. The octadecylsilyl packing used in the amounts of both probe and sample that are simultaneously column equilibration experiments was Partisil 10 ODs-3 sorbed, at equilibrium. From the results of these studies, it (Batch No. 101409, Whatman Inc., Clifton, NJ) which has a was possible to identify the reason why the sorption of the 10-pm particle diameter. This packing is prepared from a NS- probe is decreased when the sample is present, during trifunctional silane and is reported to be a polymeric bonded indirect detection LC. phase.35 It is described by the manufacturer to be 'highly end capped" with 953' 5 surface coverage.36 (17) Berthod, A.; Glick, M.; Winefordner, J. D. J. Chromatogr. 1990, Apparatus and Procedure. The apparatus for elution 502, 305-315. (18) Hackzell, L.; Schill, G. Chromatographia 1982,15, 437-444. chromatography consisted of a Waters Model 590 pump, a (19) Crommen, J.; Schill, G.; Heme, P. Chromatographia 1988, 25, Rheodyne 7120 injection valve fitted with a 20-pL loop, a 397-403. Waters Model 481 lambda max UV-visspectrophotometer, (20) Parkin, J. E.; Lau, H. T. J. Chromatogr. 1984, 314, 488-494. (21) Parkin, J. E. J. Chromatogr. 1984,287, 457-461. and a Recordall Series 5000 recorder (Fisher Scientific Co.). (22) Herne, P.; Renson, M.; Crommen, J. Chromatographia 1984,19, The analytical column (25-cm X 0.40-cm i.dJ was acommercial 274-279. column of Whatman Partisil 10 ODs-3 (Catalog No. 4228(23) Banerjee, S. Anal. Chem. 1985,57, 2590-2592. 001, Whatman Inc., Clifton, NJ). The column was thermo(24) Stranahan, J. J.; Deming, S. N. Anal. Chem. 1982,54,2251-2256. (25) McCormick, R. M.; Karger, B. L. J. Chromatogr. 1980,199,259stated precisely to fO.O1 OC by circulating water from a water 273. bath (Model R20, Haake, Berlin, Germany) whose temper(26) Geng, X.; Regnier, F. E. J. Chromatogr. 1986, 332, 147-168. ature was maintained less precisely at 25.00 f 0.04 OC,through (27) Scott, R. P. W.; Simpson, C. F. Faraday Symp. Chem. SOC.1980, 15, 69-82. the use of a specially designed column jacket which will be (28) Bartha, A.; Vigh, Gy.; Billiet, H.; De Galan, L. Chromatographia describedbelow. Other important parameters were flow rate, 1985,20,587-590. 1.0 mL/min; detection wavelength, 276 nm; and detector (29) Tang, M.; Deming, S. N. Anal. Chem. 1983, 55, 425-428. (30) Bidlingmeyer, B.; Deming, S.;Price, W.; Sachok, B.; Petrusek, M. range, 0.05 AUFS. J. Chromatogr. 1979, 186, 419-434. (31) Stahlberg, J.; Almgren, M. A d . Chem. 1985, 57, 817-821. (32) Carr, J. W.; Harris, J. M. Anal. Chem. 1986, 58, 626-631. (33) Michels, J. J.; Dorsey, J. G. J. Chromatogr. 1988, 457, 85-98. (34) Knox, J. H.; Hartwick, R. A. J. Chromatogr. 1981, 204,3-21.
(35) Products Catalog; Chromatographic Specialities Inc.: Brockville, ON, 1992. (36) TechnicalBulletin, LC-lll-6/83; Whatman Inc.: Clifton,NJ, 1985.
270
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993
The column equilibration technique has been discussed previo~sly.~ In~the , ~ ~present study, only the upper half of the apparatus was used, consisting of a pump P1 (Model 501, Waters Associates)for the mobile-phasesolution, an injection valve V1 (Part No. 7010, Rheodyne Inc.), a pump P2 (Model 590, Waters Associates) for the eluent, and a precolumn C1 which contained the ODS packing. The precolumn was either a 2.0- x 0.30-cm-i.d. or a 2.0- x 0.40-cm-i.d. stainless steel guard column (Part No. 84550, Waters Associates). The smaller precolumn was dry-packed with 0.0871 g of packing and was used for the probe isotherm experiments. The larger precolumn was packed with 0.1540 g of packing when used for the butanol isotherm experiment and with 0.1510 g of packing when used for the other column equilibration experiments. The precolumn was inserted in place of the injection loop of valve V1. The precolumn and injection valve were placed in a water bath whose temperature was maintained at 25.00 f 0.04 "C. The holdup volume, which includes the void volume of the packed bed and frits and the volume of the connectingtubing, was measured, as previously d e ~ c r i b e dby ,~~ pumping water through the precolumn, elutingwith ethanol, adding methanol as an internal standard to the volumeteric flask in which the eluate was collected, and determining the amount of water by gas chromatography. The chromatography was performed at 110 "C on a 2.9-m X 1.6-mm-i.d. column of 50/80 mesh Porapak Q-S(Waters Associates) using a Model 3700 gas chromatograph (Varian Associates) with a thermal conductivity detector. The holdup volumes of the three precolumns described above,with their 95 % confidence limits, were 0.127 f 0.007,0.219 f 0.003, and 0.218 f 0.006 mL, respectively. In the column equilibriumexperiment,a solutioncontaining NS-,and/or butanol, was pumped through the precolumn until equilibrium was achieved between the ODS packing and solution. This was the loading step. Valve VI (see refs 37 and 38) was then switched to allow the eluent to elute NSand/or butanol into a volumetric flask, up to the calibration mark. This was the elution step. The total amounts of each of the species eluted from the precolumn, including that in the holdup volume, were then measured by determining the concentration of each of these species in the solution in the volumetric flask. The concentration of NS-was determined spectrophotometrically at 276 nm on a Model 8451A Hewlett Packard diode array spectrophotometer. The concentration of butanol was determined by gas chromatography on a 3.0-m X 2.1mm-i.d. column packed with 7% w/w Carbowax 20 M (F&M ScientificCorp.) on Chromosorb WAW-DMCSmesh size l00/ 120(Manville)using a Model 3700gas chromatograph (Varian Associates) with flame ionization detection. The flow rates used were helium (carrier) = 30 mL/min, hydrogen = 30 mL/ min, and air = 300 mL/min. The temperature was programmed to be held initially at 65.0 "C for 6 min, followed by an increase at a rate of 4 "C/min for 2 min, and then at a rate of 3 "C/min for the remainder of the chromatogram. 1-Pentanol was used as the internal standard. The peak areas were measured with a Spectra-Physics Autolab minigrator. The concentration of NS- or of butanol sorbed onto the ODS packing, C, (mol/kg), was calculated by the following
where nT is the total number of moles eluted including that (37)May, S.;Hux, R.; Cantwell, F. F. Anal. Chem. 1982,54, 12791282. (38)Hux, R.;Cantwell, F. F. Anal. Chem. 1984,56, 1258-1263. Cantwell, F. F. Anal. Chem. 1991,63,993-1OOO. (39)Liu, H.;
in the holdup volume, C, (moVL) is the concentration in solution, V M (L) is the holdup volume, as measured above, and W,(kg) is the weight of packing in the column. The sorption isotherm is a plot of C, versus Cm and is obtained by repeating the experiment at several different C,. The distribution coefficient for the probe was calculated as
It was demonstrated by varying the loading times that the initial rise toward equilibriumin the early stages of the loading curve was rapid and depended upon the volume of solution passed through the column, while the final approach to equilibrium in the later stage of the loading curve was slower and depended upon the length of time that the solutionflowed through the column. As a result, a flow rate of 3.00 mL/min was used for the first 20 min (i.e. 60 mL), followed by a flow rate of 1.00 mL/min for 100min (i.e. 100mL),for atotal time of 2 h and a totalvolume of 160mL. It was also demonstrated, by varying the elution times, that an elution time of 10 min at a flow rate of 1.00 mL/min (i.e. 10 mL) was sufficient to completely elute both NS- and butanol. Calibration. Passage of mobile phase through the analytical column and precolumn for long periods of time was found to slowly deteriorate the ODS stationary phase such that the effective number of sorption sites was reduced. Although no evidence was found for dissolution of the silica substrate itself and the volume of packing in the columns was not visibly reduced, this decrease in the number of sorption sites can formally be taken into account as a decrease in the weight of stationary phase ( W,) in eq 1. To account for this, a solutionwhich contained 2.00 X moVL NS- in phosphate buffer, but no butanol, was used as a standard and was run between experiments. Using the amount of NS-sorbed from this standard at any time, t, the corrected weight of packing to be used in place of W,in eq 1was calculated as ( nNS,t we,= -nNS,o ws,o
(3)
where nNS,t is the amount of NS- sorbed from the standard solution at time t,nNs,ois the amount of NS-sorbed from the standard solution by the freshly packed precolumn, and W,,, is the weight of packing in the precolumn.
RESULTS AND DISCUSSION Temperature Effect. When indirect detection is performed by liquid chromatography, the absolute value of the mobile-phase optical absorbance is what constitutes the baseline signal of the chromatogram. This baseline absorbance is relatively high, while the change in absorbance which is observed when a sample or system peak elutes is relatively small. This is because analytical chromatography is performed under conditions where the sample isotherm is linear, which requires a low sample concentration. Consequently, the additional number of moles of probe transferred from the stationary phase to the mobile phase, or vice versa, in the sample zone usually represents only a small fraction of the moles of probe that are present in the mobile phase in the absence of sample. Another variable, in addition to the presence of sample, which can produce changes in the amount of probe sorbed from a given mobile phase, is temperature. Since changes in probe sorption induced by the presence of the sample are small, it is possible that small changes in temperature, such as the normal fluctuations in a constant temperature bath, can produce baseline absorbance shifts that are comparable in magnitude to sample and system peaks. Previous reports of studies in which NS- was used as a probe have mentioned the need for thermostatting,lOJ1but
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, lQQ3 271 0.920
0.900
0.880
--
0.860
-
0.840
-
L
1
0.820
100
150
200
250
300
Volume (mL) Flgure 1. Plot of absorbance versus volume of mobile phase passed through the analytical column with changlng temperature. The dashed line is the baseline absorbance. The Initla1 temperature was 25.1 "C. At point a, the temperature of the column was decreased to 23.4 O C . At point b, the temperature was increased to 23.8 "C. Conditions: Column, 2 5 c m X 0.40cm-i.d. Partisll 10 OD53 analytical column; mobile phase, 2.00 X lo4 mol/L NS- in 0.04 mol/L H3PO4/0.03mol/L NaH2P04buffer at pH 2; flow rate, 1.00 mL/mln.
none has quantified the effect of temperature or demonstrated how critical the temperature control can be. Under the conditions employed in the present study, these temperature effects were profound. Shown in Figure 1are the absorbance changes that are brought about by small temperature changes. The baseline (dashed line) in Figure 1 represents the absorbance of the mobile phase at 25.1 "C, after the probe has achieved completebreakthrough on the column. At point a, the column temperature is lowered by 1.7 O C and held at the new temperature (23.4 "C). This increases the distribution coefficient of NS- along the whole column and causes the absorbance to decrease. The difference in volume between the descending front, at 130 mL, and the ascending rear, at 190 mL, of the broad negative plateau following point a is equal to the retention volume of the probe. After the plateau, the column has been completely reequilibrated at 23.4 "C. This is indicated by a return of the signal to the original baseline since this represents the absorbance of the constant composition mobile phase entering, and now exiting, the column. A t point b, the temperature is raised by 0.4 OC to 23.8 "C which decreases the probe distribution coefficient from ita previous value at 23.4 "C and causes the probe to desorb into the mobile phase. After the resulting positive plateau, the signal again returns to ita original baseline value. The baseline absorbance (0.880 AU) would normally be assigned a value of zero (offset) when indirect detection is performed. To show how sensitive the system is to changes in temperature, a plot of the plateau absorbance at various temperatures is shown in Figure 2. The plot is linear over this small temperature range, with a slope of 0.0220 & 0.0002 AU/"C. This means that the baseline absorbance will shift by 0.022 AU for a 1-deg temperature change. This is seen to be a significant temperature dependence when it is realized that sample and system peaks typically produce absorbance changes of 0.02 AU and that the detector sensitivity is often set at 0.05 AUFS. At this detector sensitivity, short-term temperature fluctuations of the water bath, as small as f0.04 OC,will produce baseline excursions of 2% of the full scale deflection which appears as baseline noise in the chromato-
21
22
23
24
25
26
Temperature ("C) Figure 2. Plateau absorbance versus column temperature. Column and mobile phase are as In Flgure 1. gram. Thus, it is desirable to thermostat the system with a precision closer to f0.01 "C. To achieve the necessary high-precision temperature control while using a water bath that is rated at f0.04 "C precision, a special double-walled water jacket was used. Stagnant water filled the inner jacket and bathed the column contained therein, while thermostated water (25.00 f 0.04 "C) from the constant-temperature bath circulated through the outer jacket. By having the column immersed in stagnant water rather than in direct contact with the circulating water, it was less sensitive to the short-term (i.e. several seconds) fluctuations in water bath temperature. Elution Chromatography. Indirect detection of a butanol sample using NS- as probe in 0.05 moVL phosphoric acid has been reportad on a p Bondapak phenyl reversedphase c01umn.'~ In order to verify that the same behavior would be obtained on the PartisillO ODs-3 column using a similar mobile phase, elution chromatographywas performed in the present study. In the elution chromatogram (not shown), a positive sample peak with k'BuoH = 8.55 f 0.05 was followed by a negative system peak with k' = 26.4 f 0.1. This is the usual response pattern observed with a neutral sample that is less strongly retained than the pr~be.'J.~The positive sample peak suggests that some of the probe is desorbed from the stationary phase in the presence of butanol. The assignmenta of the system peak (contains no butanol) and the sample peak (contains butanol) were verified as follows. To verify that the second peak was the system peak, solutions were injected which had the same composition as the mobile phase except that they contained slightly more or slightly less NS-. In each case a single peak was obtained with a k' that agreed with that for NS-.To verify that the first peak was the sample peak, its k' (=8.65 f 0.05) was compared with the k' (=8.4 f 0.6) calculated for butanol from the linear region of its sorption isotherm measured in the presence of 2.00 X mol/L NS- (not shown). Column Equilibration Studies. In the column equilibration experiments, the amounts of NS- and butanol simultaneouslysorbed onto the PartisillO ODs-3 stationary
272
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993
0.0
0.2
0.4 Cm,BuOH
0.6
0.8
(mouL)
b
E
1
3
I
b
O.OOe+O
v
. u“ 0
1.00e-4
2.00e-4
(mol) Figure 4. Plot of moles of butanol sorbed versus moles of NS- sorbed in the column equlllbratlon experiment. The butanol concentration mol/L. The NS- concentration was was kept constant at 1.09 X varied from 0 to 0.157 mol/L. The weight of ODs packing In the precolumn was 0.1510 g. ns,NS
1000 2000 3000 4000 5000
This behavior can be explained in terms of a competition between NS-and butanol for available space in the stationary Figure 3. Butanolsorptionisotherm. (a)Experimentalisothermobtained by Stineman interpolation (see ref 66) through the experimental points phase. The distribution coefficient for butanol, K D , B ~ O H ,is (circles) using Cricket Graph 1.3. The dashed line is the Langmuir given by isotherm generated from the Langmuir constants obtained from the straight line in panels b and c. See text for details. (b) Plot of 1/ C s . ~ ~ ~ (4) K D , B ~ O H = Cs,BuOH/Cm,BuOH = nS,BuOH/AsCm,BuOH versus l/Cm,BuoH for the experimental isotherm (solid line) in panel a. where ~,,B,,oH is the moles of butanol sorbed onto the stationary (c) Expansion of the lower left hand corner of the plot in panel b. phase are measured. This simulates the situation in the butanol sample zone during an elution chromatogram. Simultaneous sorption data for NS- and butanol are evaluated to determine the reason why butanol sample causes decreased sorption of the probe. The evaluation is made in terms of the three categoriesgiven in the Introduction, i.e. change in mobile phase strength, competition for space in the stationaryphase, and change in sorbent strength. In addition to looking at the effect of butanol on probe sorption, the reverse case is also considered, i.e. effect of probe on butanol sorption. Effect of Probe on Butanol Sorption. The butanol sorption isotherm, measured on Partisil 10 ODs-3 in the present study, is shown in Figure 3a. At the highest butanol concentrations employed (0.654 mol/L) the isotherm has not achieved the plateau that would indicate complete surface coverage. Since high butanol concentrations were of no special interest in the present work, the isotherm was not measured above 0.654 mol/L. Sorption of organic modifiers, including butanol, onto RPBPs has previously been studied.27,4041 To determine how the probe affected butanol sorption, column equilibration experiments were performed in which butanol was kept a t the low and constant concentration of 1.09 x 10-3 mol/L while the NS- concentration was varied from 0 to 0.157 mol/L. Higher concentrations of NS- could not be used because the solution was nearly saturated a t 0.157 mol/L. The amount of butanol sorbed is plotted versus the amount of NS- sorbed in Figure 4, where it is seen that the amount of butanol sorbed decreases linearly with the amount of NS- sorbed. The slope is -0.0148 f 0.0010 mol of butanol sorbed/mol of NS- sorbed and the z intercept is (1.50 f 0.12) X loW4 mol of NS- sorbed. (40)Scott, R.P.W.; Kucera, P. J. Chromatogr. 1979,175, 51-63. (41)McCormick, R. M.; Karger, B. L. Anal. Chem. 1980,52,22492251.
phase and A, is the space available in the stationary phase for butanol sorption. When NS- is also present in the mobile phase, then the available space for butanol sorption is decreased as a result of ‘blockage” of some of the surface by sorbed NS-. This is expressed as As
=
- Aa,NS
(5)
Here, A,,t is the total space available and Asps is the space occupied by NS- which can be expressed as
A~,Ns= A N S ~ ~ , N S
(6)
where ANS is the space occupied per mol of NS- and ~ , , N s is the moles of NS- sorbed onto the stationary phase. By combination of eqs 4-6, the amount of butanol sorbed onto the stationary phase in the presence of NS- is %,BuOH
= (KD,BuOHCm,BuOHAs,t)
- (KD,BuOHCm,BuOHANS)ns,NS
(7) Assuming that the space occupied per NS- molecule is independent of the fractional coverage of the stationary phase by NS- and considering a constant value of Cm,BuOH, the linearity of the plot of ns,BuOH versus n s , N s in Figure 4 suggests, via eq 7, that a competition for space is occuring. The linearity of the plot also shows that K D , B u 0 H is constant and is therefore independent of the concentration of NS- in both the mobile and stationary phases. This means that neither the solvent strength of the mobile phase nor the sorbent strength of the stationary phase changes significantly in the presence of NS-. The space occupied per mole of NS-can be calculated from the slope and intercept of the plot in Figure 4 slope (8) AN, = interceptwas where W, is the weight of Partisil 10 ODs-3 packing in the precolumn and S is the total space available per unit weight of packing. If “space” is considered in terms of surface area,
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, lgQ3 278
then S is approximately 340 m2/gfor Partisill0 ODs-3. This value was calculated starting with the manufacturer's values of 400 m2/g for the specific surface area of the parent silica and 10% carbon loading.35 This estimated decrease of 15% in specificsurface area in going from silica gel to 10% carbonbonded CISphase is consistent with the approximately 12% decrease measured for bonded phases.42 Also, the estimated area of 340 m2/g is reasonably close to the value of 309 m2/g measured by the BET method on a different batch of Partisil 10 ODS-3.39 Using S = 340 m2/g in eq 8, ANS is calculated to be (34 f 2) X lo4m2/mol. This corresponds to a molecular area of (56 f 3) X m*/molecule,which is in agreement with the area of 65 X m2/moleculemeasured on a spacefilling model (CPK Models) of NS-, by taking the average of two possible areas that NS-could present to the surface (flat or side-on). The standard deviations given for ANSand for the molecular area reflect the scatter in the data in Figure 4 but do not include the uncertainty in the value of S which probably has a relative standard deviation in the vicinity of 10%. Effect of Butanol on NS-Sorption. Using the column equilibration technique, NS- isotherms were measured without and with butanol present (Figure 5) in order to demonstrate the effect of butanol on NS- sorption. In both curves a and b, the NS- concentration was varied from 0 to 3.62 X 10-3mol/L. In curve a, no butanol was present, and in curve b, the butanol concentration was kept constant at 2.18 X mol/L. Fromvisual examinationof the isotherms,it is obvious that the presence of butanol reduces the amount of NSsorbed. This is expected, on the basis of the peak pattern in the elution chromatogram described above. However, the reason for this effect (i.e categories 1-3 in the Introduction) is not clear from the data in Figure 5 alone. If the decrease in NS- sorption in the presence of butanol resulted from a decrease in the distribution coefficient (KD,Ns) alone, then it would imply that the presence of butanol strengthens the mobile phase as a solvent for NS- and/or it weakens the stationary phase as a sorbent for NS-. On the other hand, if the decrease in NS- sorption in the presence of butanol results from only a decreasein available space in the stationary phase, then it implies that butanol is exerting its effect by competing for space. From the results discussed above in connection with Figure 4, it is certain that at least part of the reduction in the amount of NS- sorbed must be due to a competition between butanol
Flgure 5. Plot of NS- sorption isotherms wlthout and wlth butanol present. In curvea, no butanol was present. I n curve b, 2.18 X mol/L butanol was present. Solid lines are emplrlcal; points are experimental.
(42)Sander, L. C.; Glinka, C. J.; Wise, S. A. Chemically Modified Surfaces: Silanes, Surfaces and Interfaces; Leyden, D. E., Ed.; Gordon and Breach Science Publishers: New York, 1986; p 431. (43)Schoenmakers, P. J.;Billiet, H. A. H.; Tijssen, R.; De Galan, L. J. Chromatogr. 1978,149,519-537. (44)Schoenmakers, P. J.; Billiet, H. A. H.; De Galan, L. Chromatographia 1982,15,205-214. (45)Schoenmakers, P. J.;Billiet, H. A. H.;De Galan, L.J. Chromatogr. 1983,282,107-121. (46)Dorsey, 3. G.; Dill, K. A. Chem. Rev. 1989,89,331-346. (47) Dill, K.A. J. Phys. Chem. 1987,91,1980-1988. (48)Cheong, W. J.; Carr, P. W. Anal. Chem. 1989,61,1524-1529. (49)Barton, A. F. M. CRC Handbook of Solubility Parameters and Other Cohesion Parameters; CRC Press: Boca Raton Florida, 1983;pp 142-149. (50)Scott, R.P.W.; Simpson, C. F. J. Chromatogr. 1980,197,ll-20. (51)Sander, L. C.; Wise, S. A. Crit. Reu. Anal. Chem. 1987,18,299415. (52)Gilpin, R.K.;Gangoda, M. E. Anal. Chem. 1984,56,1470-1473. (53)Locke, D. C. J. Chromatogr. Sci. 1974,12,433-437. (54)Bedard, P. R.;Purdy, W. C. J. Liq. Chromatogr. 1986,8,24172443. (55)Yonker, C. R.;Zwier, T. A.; Burke, M. F. J . Chromatogr. 1982, 241,257-268. (56)Colin, H.; Guiochon, G. J. Chromatogr. 1978,158, 183-205. (57)Hammers, W. E.; Meurs, G. J.; Deligny, C. L. J. Chromatogr. 1982,246,169-189. (58)Hammers, W. E.; Verschoor, P. B. A. J. Chromatogr. 1983,282, 41-58.
The symbols have meanings analogous to those in eq 7, as implied by their subscripts. The linearity of the plot in Figure 6 indicates that KD,NSis constant and, consequently, that neither the mobile-phase eluent strength nor the stationaryphase sorbent strength is altered to a detectable extent by the presence of these small amounts of butanol. That is to say, while the presence of even these low concentrations of butanol undoubtedly produces some changes in the mobilephase eluent strength (and perhaps also in the stationaryphase sorbent strength), the magnitudes of these changes are too small to be detected in the presence of the large competitive effect. The linearity of the plot in Figure 6 also implies that the space occupied per mole of butanol is independent of the fractional coverage of the stationary phase by butanol. Also by analogy with the previous case, the area occupied per molecule of butanol can be calculated. It was found to be (54 f 2) X m2/molecule,which is close to the value of 35 X 10-2Om2/moleculedetermined from a molecular model by taking the average of three possible areas that butanol could present to the surface.27 Langmuir Isotherm Behavior. It is noteworthy that the linearity implied by eqs 7 and 9 and observed in Figures 4
8.k-2 I
O.Oe+O
1.0e-3
2.0e-3
3.0e-3
4.1 e-3
Cm,NS(mol/L)
and NS- for space because a competition for space must be mutually reciprocal. In order to determine how much of the reduction in NS- sorption is due to the competitive effect, a column equilibration experiment analogous to the one represented in Figure 4 was performed in which the NSconcentration in the mobile phase was kept constant at 2.00 X mol/L while the butanol concentration was varied from 0 to 4.36 X mol/L. (Indirect detection liquid chromatography of butanol as a sample would generally not be performed outside this range because the butanol isotherm becomes distinctly nonlinear at higher concentrations). The amount of NS- sorbed is plotted versus the amount of butanol sorbed in Figure 6. A linear relationship is observed. The slope is 4.0133 f 0.0005 mol of NS- sorbed/mol of butanol sorbed and thex intercept is (1.59 f 0.07) X 10-4mol of butanol sorbed. A similar treatment can be done as in the previous case. When a constant concentration of NS- is considered (i.e. constant C m , ~ s an ) , equation similar to eq 7 is obtained
274
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993 3.OOe-6
I
i
I
0.0
0.2
0.1
o.OOe+o O.OOe+O
1.00e-4 ~ S , B ~ O H
2.00e-4
300
(mol)
Flgure 6. Plot of moles of NS- sorbed versus moles of butanol sorbed in the column equlilbratlon experiment. The NS- concentratlon was kept constant at 2.00 X lo4 mol/L. The butanol concentration was varied from 0 to 4.36 X lo-*mol/L. The weight of ODS packing in the precolumn was 0.1510 g.
and 6 does not require that either the NS- or the butanol isotherms follow a particular equation (e.g. Langmuir). This is because the solution concentrations C m , ~ ufor~ eq ~ ,7, and C m , ~ for s , eq 9, are held constant in the experiments. That is, the experiments represented in Figures 4 and 6 are conducted at a fixed point on the butanol or NS- isotherm, respectively, at which K D , B ~ O Hor KQNSis expected to be a constant provided that solvent strength and/or stationaryphase sorbent strength are not changing. This simply means that ns,BuOH in eq 4,for example, has decreased in proportion to A,, the amount of space not "blocked" by NS-, so that C s , ~remains u ~ ~ constant. Thus, although the experiments represented in Figures 4 and 6 were carried out at concentrations falling in the linear regions of the butanol and NSisotherms, respectively (as can be seen from Figures 3a and 7a), this was not a necessary condition to obtain constant values of K D , B ~ O Hand KD,NS. Equations 7 and 9 reveal that the intercepts on the horizontal axes in Figures 4and 6 correspond to the maximum stationary-phase occupancy of NS- (ns,max,Ns) and butanol (&,-,B,,OH), respectively. Ifthe butanol and/or NS-isotherms could be described by an established theory, such as the Langmuir equation, then the interpretation of the experimental results could be carried further than is allowed by eqs 7 and 9. In particular, the limiting values predicted by curve fitting the Langmuir equation to the experimental data could be compared with the values from the horizontal intercepts in Figures 4 and 6. This possibility is now examined. A previous study showed that butanol sorption onto a different polymeric bonded phase (WhatmanPartisill0 ODS2) follows the Langmuir equation27which is given below:
b
h
200
E
B
u3 . -
100
0
0
10000
20000
3' 00
l/C,,,Ns (L/mol) Flgwa 7. NS- sorptknisotherm wlth 1.09X 1 O5 md/L butandp m n t . Data are taken from Flgure 4. (a) Experlmentai isotherm obtained by Stlneman interpolation through the experimentalpolnts (circles)using Cricket Graph 1.3. (b)Plot of 1/Cs,N~ versus 1/C,,,,NSfor the NS- sorption isotherm In panel a.
is made (Figure3b),the points for which 1/Cm,BuOHZ 12L/mol (Le. C m , ~5 u8 X ~ ~ mol/L) describe a straight line, from = the slope and intercept of which may be calculated 13 f 7 L/mol and C s , m r u , ~=u1.1 ~ ~f 0.6 mol&. However, as shown in Figure 3c, the points for which 1/Cm,BuOH 5 12 L/mol deviate from this line, indicating an apparent deviation from Langmuir behavior at high butanol concentrations. When the above values of KB~OH and Cs,max,BuOH are used in eq 10 to generate a plot of C,,B~OH versus Cm,BuOH, the dashed l i e curve in Figure 3a is obtained. Thus, the butanol isotherm follows the Langmuir equation in the concentration range (G8)X mol/L but deviates from it at higher concentrations. The experimental isotherm in Figure 3a (solid line) is not apparently approaching a limiting value of C s , - , ~ u ~ ~ at concentrations of C m , ~asu high ~ ~ as 0.654 mol/L. In contrast, Scott and S i m p ~ o nobserved ~~ Langmuir behavior over the concentration range 0-0.22 mol/L on Partieil 10 KBuOHCs,max,BuOHCm,BuOH ODs-2 and their isotherm is clearly approaching a limiting (10) Cs*BuoH = 1+ KBuOHCm,BuOH value at the latter c~ncentration.~' From the Langmuir constant, Cs,mru,BuOH, the Value of In the equation, K B ~ O(L/mol) H is the adsorption equlibrium constant for butanol and C s m ~ u (mol/ ~ ~kg) is the monolayer &,ma,BuOH may be calculated. It is found to be (1.7 f 0.9) X mol, which agrees within experimental error with the concentration of butanol. value obtained from the horizontal intercept in Figure 6, i.e. If the experimental data from Figure 3a conform to the mol. The vertical arrow shown in Figure Langmuir equation then a double-reciprocalplot of l / C e , ~ u ~ (1.59 ~ f 0.07) X 3a indicates the maximum butanol concentration that was versus 1/Cm,~,,0~ should be straight line.59 When such a plot used in studying the effect of butanol on probe sorption. Thus, the agreement of the horizontal intercept in Figure 6 with the (59) Tilly-Melin, A.; Askemark, Y.;Wahlund, K.-G.; Schill, G. Anal. Chem. 1979,51, 976-983. limiting value of the dashed isotherm in Figure 3a is not
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993 275
4.2
4.1
In KD on modifier concentration. Thus, the quadratic dependence of In KD,NS on 4 m S u O H t when considered in the
A 1
I
3.7 0.000
0.001
0.002
0.003
0.004
~ , B ~ O H
Flgure 8. Plot of the natural logarithm of NS- distribution coefficient versus the volume fraction of butanol in the mobile phase. Data are taken from Figure 6. The solid line is from nonlinear curve fltting to a quadratic equation. The circles are experlmentai points.
surprising since the butanol concentration range studied in Figure 6 lies entirely within the Langmuirian region of the butanol isotherm. Though the evidence from Figures 4 and 6 alone is conclusive, it is pleasing to find the above agreement which represents further support for the view that the reduction in the amount of NS-sorbed when butanol is present is due strictly to a competitive effect, i.e. category 2 in the Introduction. Different behavior is observed in the case of the NSisotherm, as shown in Figure 7. The points in Figure 7a correspond to the points in Figure 4, which were measured at a constant butanol concentration of 1.09 X mol/L. The presence of such a small concentration of butanol has very little effect on the sorption of NS-, as seen from the very early points in Figure 6. Thus, the NS- isotherm which would be measured in the absence of butanol would be virtually identical to the isotherm presented in Figure 7a. A plot of l/Cs,Ns versus 1/Cm,Ns is shown in Figure 7b. It is nonlinear over the entire concentration range, demonstrating that the NS- isothermdoes not exhibit Langmuirbehavior. As a result, a value for C s , m a r cannot , ~ ~ be estimated from the data in Figure 7 for comparison to the horizontal intercept in Figure 4. Changing Eluent and Sorbent Strength. Figure 8 is a plot of the natural logarithm of the distribution coefficient for NS- versus the volume fraction of butanol in the mobile phase, &,BuOH. The points in Figure 8 correspond to the points in Figure 6, which were measured at a constant NSconcentration of 2.00 x mol/L. The line in Figure 8 was obtained by nonlinear least squares curve fitting of the following equation to the experimental points:
(11) In KD,NS = A + 2 m , B ~ O H + % n , B u O H + The least squares values obtained for the constants are A = (1.55 f 0.27) X lo4,E = -180 10, and C = 4.256 f 0.005, and the correlation coefficient is r = 0.996. Current theories on the role of the mobile-phase composition in RPLC, such as the extended solubility parameter t h e 0 r y , ~ 3the ~ ~l 3 p solvatochromic the0ry,~3and the linear solvation energy relationship (LSER) theory,@when applied at low concentrations of “organic modifier” (i.e. butanol, in the present context), predict such a quadratic dependence of
*
absence of the correlations presented in Figures 4 and 6,might seem to suggest that the decrease in KD,NSwith increasing concentration of butanol results from an increase in the mobile-phase eluent strength accompanied, perhaps, by a consequent change in stationary-phase sorbent strength (i.e. categories 1 and 3 in the Introduction). However, comparison of the experimentally measured value of parameter A in eq 11 with the value predicted by theory does not support this view. Theory predicts that the value of A is given approximately as43*
where VNS is the molar volume of NS- (approximately 78 cm3/mol), 6 ~ and ~ 60 ~ are~solubility 0 ~ parameters (23.4 and 11.4 ~ a l W c m ~respectively, /~, from ref 49))R is the ideal gas constant (1.9865cal/(mol K)), and Tis 298 K. If one accepts the solubility parameter theory as a rough estimate of solution behavior, then the fact that the experimentally measured value of A (1.55 x 104) is over 3 orders of magnitude larger than that predicted by theory indicates that the quadratic relationship observed between ln KD,NS and dm,BuOH does not arise from a change in eluent strength but, rather, arises fortuitously from a combination of the linear dependence of &,NS on nsS,,oH (Figure6) and the curved dependenceof n s s u O H on C m , ~ (Figure u ~ ~ 3a) and, therefore, on at low C m , ~ uIf~ the ~ . effect of butanol on NS- sorption were due to the influence of butanol on solvent strength only, then the effect would have been very much smaller than the observed effect. The conclusion that competition, rather than a change in eluent or stationary-phase strength, is responsible for the decreased sorption of probe in the presence of sample can be strictly applied only to the NS-/BuOH system studied herein. However, it is likely that competition for space also occurs for other combinations of probe and sample and the p h s nomenon should be considered in studies of other systems used for indirect detection and also in studies designed to evaluate eluent strength. Adsorption versus Partitioning. Monomeric RPBPs have been studied more extensively than polymeric RPBPs, and the dependence of their alkyl chain conformation upon eluent composition and temperature suggests that the structure of monomeric stationary phases is more variable than that of polymeric phase^.^,^^^ Only a polymeric bonded phase has been employed in the present study. However, it is to be expected that in a highly aqueous mobile phase the same kind of competitive sorption behavior would be observed for BuOH and NS- on a monomeric bonded phase as is observed on the polymeric phase. A persistent theme in the literature on RPBP sorbents,46both monomeric and polymeric, is the dispute over whether the solute is sorbed by the stationary phase via adsorption onto its surface (i.e. onto the interface between the stationary and mobile phase^)^^^^^^ or via partition into the stationary phase.47-56.62p63 In the present study, it has been demonstrated that the two solutes, NS- (probe) and butanol (sample), are in direct competition with one another for space in the stationary phase. However, this should not be interpreted as conclusive proof that these solutes are sorbed via adsorption rather than via partitioning, for the following reasons. The thickness of a (60)Scott, R. P. W.; Kucera, P. J. Chromatogr. 1977,142,213-232. (61) Ketz, E. D.; Ogan, K.; Scott, R. P. W. J. Chromotogr. 1986,352, 67-90. (62) Lochmuller, C. H.; Wilder, D. R. J. Chromatogr. Sci. 1979, 17, 574-579. (63) Martire, D. E.; Boehm, R.E.J.Phys. Chem. 1983,87,1045-1062.
276
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993
bonded layer of CIS on the silica gel surface, whether monomeric or polymeric, is not more than about 20 A, when methanol is the mobile With a highly aqueous mobile phase this thickness would be, if anything, even somewhat less because of more extensive collapsing of the chains.63Even for a solute as small as butanolz7the bonded layer is only a few times thicker than the molecular diameter of the solute so that the fraction of the total stationary-phase volume occupied by the solute molecule, if partitioned, would be highly correlated with the fraction of the total stationary-phase area (64) Sander, L. C.; Glinka, C. J.; Wise, S. A. Anal. Chem. 1990, 62, 1099-1101. (65)Berendsen, G. E.; DeGalan, L. J. Chromatogr. 1980,196, 21-37. (66) Stineman, R.W. Creatiue Comput. 1980,654-57.
occupied by the solute molecule, if adsorbed. Even if the NS-and butanolwere completely ‘dissolved” in a ‘bulk liquid” bonded phase, there could be a competition between these sorbed species with one another for volume in such a thin layer of bonded phase. If only a limited part of the bonded phase interacts with the solute, as has been then this competition for “volume” would be even more pronounced. Thus, we have chosen to w e the more general terms “sorption” and “competitionfor space” rather than ‘adsorption” and ‘competition for area” and to adopt neither side in the adsorption-partition debate.
RECEIVED for review June 16, 1992. Accepted October 26, 1992.
