Mass Spectrometric Tracer Pulse Chromatographic Investigations of

Anal. Chem. 2008, 80, 6708–6714

Mass Spectrometric Tracer Pulse Chromatographic Investigations of Eluent Sorption with Bonded RPLC Packings Mei Wang, Jennifer Mallette, and Jon F. Parcher* Chemistry Department, University of Mississippi, University, Mississippi 38677 The experimental technique of mass spectrometric tracer pulse chromatography was used to investigate the uptake of RPLC eluents by a C18-bonded packing. The experiments were carried out with eluents consisting of binary aqueous mixtures with acetonitrile, methanol, and tetrahydrofuran over the full range of eluent composition at 25 °C. The primary experimental data obtained were excess volumes of sorption for the eluent components. The excess volume data were then used to determine the absolute volume of each eluent component in the stationary phase as a function of composition. The absolute volumes were calculated by utilizing a series of strategies specific to limited eluent composition range. The linear inflection region of the excess volume isotherms was used to calculate the volume and composition of the eluent in the stationary phase for organic-rich eluents. Three different assumptions were used and evaluated for the waterrich eluent compositions. The assumptions were (i) constant volume of the sorbed layer, (ii) constant amount of water sorbed, and (iii) no water sorption. The latter assumption was adopted as the most satisfactory. The calculated void volume data were compared with the retention volume of thiourea and uracil, commonly used dead time markers. Neither thiourea nor uracil proved to be an accurate measure of the kinetic void volume. Tracer pulse chromatography with stable isotopes and refractive index detection has been used for decades to measure the adsorption or partition isotherms of both analytical solutes and eluent components in RPLC systems.1–6 Refractive index detection, however, suffers from a number of disadvantages including low sensitivity, flow rate, and temperature influence on response, and particularly the lack of specificity. The detector may produce positive or negative peaks for various solutes, and overlap of peaks may be difficult to resolve. This detection system will also respond to concentration changes in the eluent due to the injection of isotopes to produce extraneous system peaks that may be difficult to distinguish from solute peaks. * To whom correspondence should be addressed. E-mail: [email protected]. (1) Alvarez-Zepeda, A.; Martire, D. E. J. Chromatogr. 1991, 550, 285–300. (2) McCormick, R. M.; Karger, B. L. Anal. Chem. 1980, 52, 2249–2257. (3) Zhu, P. L. Chromatographia 1985, 20, 425–433. (4) Foti, G.; Reyff, C.; Kovats, E. Langmuir 1990, 6, 759–766. (5) Slaats, E. H.; Markovski, W.; Fekete, J.; Poppe, H. J. Chromatogr. 1981, 207, 299–323. (6) Poppe, H. J. Chromatogr., A 1993, 656, 19–36.

6708

Analytical Chemistry, Vol. 80, No. 17, September 1, 2008

Radioactive isotopes have also been used for tracer pulse experiments;7–10 however, the experimental, regulatory, and environmental issues associated with chromatography involving radioactive materials are prohibitive. Mass spectrometric detection of stable isotopes, such as 2H, 13 C, 18O, and 15N, has been used extensively to investigate retention mechanisms with gas and supercritical fluid systems.11,12 However, the application of tracer pulse methods to RPLC systems was inhibited by the lack of suitable LC/MS interface and ionization instrumentation. This situation has improved with the development of commercial, atmospheric pressure electrospray, and chemical ionization sources. Recently, Samuelsson et al.,13 applied mass spectrometric detection with tracer pulse chromatography to measure the adsorption isotherms of methyl and ethyl esters of mandelic acid with an eluent of fixed composition (0.30 volume fraction of acetonitrile in water) on a C8-bonded packing. These authors claimed that the tracer pulse techniques had the advantages of applicability to multicomponent systems and the production of direct isotherm data without the need for integration required for concentration pulse experiments.13 Previous studies of adsorption on or absorption in RPLC packings have concentrated on either the uptake of analytical solutes for modeling preparative chromatographic systems or the uptake of eluent components to elucidate the role of the stationary phase in the retention mechanisms of analytical solutes at low concentration. Detection systems based on the determination of bulk eluent properties, such as refractive index, are not suitable for the investigation of multicomponent systems. UV-visible detection is universally applied with spectroscopically transparent eluent components, which limits their application to the investigation of solutes that absorb UV or visible light. On the other hand, a quadrupole mass spectrometer operated in the selected ion monitor mode allows the simultaneous detection of both individual solutes and isotopic tracers of eluent components. This mode of operation eliminates the high background response normally associated with eluent components or solutes at prep-scale concentrations by eliminating the unwanted ions before they reach (7) Knox, J. H.; Kaliszan, R. J. Chromatogr. 1985, 349, 211–234. (8) Samuelsson, J.; Forssen, P.; Stefansson, M.; Fornstedt, T. Anal. Chem. 2004, 76, 953–958. (9) Gilmer, H. B.; Kobayashi, R. AIChE J. 1965, 11, 702–705. (10) Peterson, D. L.; Helfferich, F.; Carr, R. J. AIChE J. 1966, 12, 903–905. (11) Parcher, J. F. J. Chromatogr. 1982, 251, 281–288. (12) Wang, M.; Hou, S.; Parcher, J. F. Anal. Chem. 2006, 78, 1242–1248. (13) Samuelsson, D.; Arnell, R.; Diesen, J. S.; Tibbelin, J.; Paptchikhine, A.; Fornstedt, T.; Sjoberg, P. J. R. Anal. Chem. 2008, 80, 2105–2112. 10.1021/ac800776c CCC: $40.75  2008 American Chemical Society Published on Web 07/31/2008

the detector. Thus, there is no need to dilute the eluent or analytical sample to maintain ion concentrations within the allowable or linear range of the detector. Simultaneous detection of multiple ions without a high background due to eluent components is the major advantage of mass spectrometric tracer pulse chromatography. The major disadvantages are the complex and costly instrumentation required and the need for isotopic analogues of the eluents or samples that must be distinguishable from the natural (unlabeled) components of the eluent. A secondary problem encountered with LC/MS systems is the need for efficient ionization of the isotopic tracer components. Atmospheric pressure chemical ionization (APCI) and atmospheric pressure electrospray ionization (APESI) sources can be operated in either positive or negative mode, and it is usually relatively simple to choose a mode of operation and tuning method that will produce detectable ions for eluent components and solutes. The impetus for the measurement of sorption isotherms of eluent components in RPLC is elucidation of the effect of stationary-phase properties on the retention of analytical solutes. This area of research has been pursued extensively with different experimental approaches for many years.14–16 However, there is still considerable uncertainty concerning the retention mechanisms in RPLC systems. It has been shown repeatedly that most binary aqueous-organic eluents interact significantly with bonded stationary-phase packings.1–3,6,17 The extent of uptake of eluent components depends upon the bonded phase, the eluent components, the eluent composition, and to a lesser extent pressure and temperature. The question of whether the eluent components adsorb on or absorb (partition) into the bonded phase is unsettled, and there is abundant evidence that both types of interactions may operate in any given system. Computer simulations support this dual-mechanism model.18 It is also possible that the uptake mechanism can change with eluent type and composition. This then leaves the question of what experimental parameters control the relevance of partition and adsorption processes with a given solute, eluent and packing? In order to elucidate solute retention mechanisms in RPLC systems, accurate knowledge of the volumes of the mobile and stationary phases, which may vary with eluent composition, is vitally important. Experimental methods for the accurate determination of the void volume, dead volume, and mobile-phase volumes in RPLC columns have been reviewed.16 The exact definitions of these terms have been discussed by numerous authors.6,15,16 Various “unretained” solutes have been used as void volume markers.16 However, most experimental methods for the determination of the mobile- and stationary-phase volumes in RPLC columns involve indirect measurements based on direct experimental data in the form of excess sorption isotherms. In order to obtain actual phase volumes from excess sorption data, it is necessary to invoke a secondary assumption for the sorption mechanism of at least one component.15 This process is complex, (14) Gritti, F.; Guiochon, G. J. Chromatogr., A 2005, 1099, 1–42. (15) Wang, M.; Mallette, J.; Parcher, J. F. J. Chromatogr., A 2008, 1190, 1–7. (16) Rimmer, C. A.; Simmons, C. R.; Dorsey, J. G. J. Chromatogr., A 2002, 965, 219–232. (17) Kazakevich, Y. V.; LoBrutto, R.; Chan, F.; Patel, T. J. Chromatogr., A 2001, 913, 75–87. (18) Rafferty, J. L.; Zhang, L.; Siepmann, L.; Schure, M. R. Anal. Chem. 2007, 79, 6551–6558.

indirect, and laborious. However, currently it is the only reliable method for the measurement of phase volumes in RPLC. The objectives of the present investigation are to (i) develop a mass spectrometric tracer pulse method for the determination of absolute isotherms of eluent components into or onto the stationary phase in RPLC systems, (ii) use the measured excess sorption data for several eluents to produce reliable volumes for the stationary and mobile phases, and (iii) evaluate various “unretained” solutes used by previous investigators as dead volume or dead time markers to determine if any solute consistently gives correct mobile-phase volumes. MATERIALS AND METHODS Apparatus. HPLC: The liquid chromatograph was a HewlettPackard model HP 1050. This system consisted of a single pump, an autosampler, and a multiwavelength UV detector. MS: The mass spectrometer was an Agilent 6120 Single Quad LC/MS. The system included a multimode ionization source and a single quadrupole mass analyzer. The ionization source was capable of either APESI or APCI mode. Chromatographic Column. The RPLC column was obtained from SGE, Inc. The column dimensions were 250 × 4.6 mm. The column was packed with 2.0 g of Exsil ODS. The packings had a particle diameter of 5 µm, pore size of 80 Å, and surface area of the base silica was 300 m2/g. The reported bonding density was 3.4 µmol/m.2 Chemicals. The HPLC grade eluents methanol, acetonitrile, water, and the Optima grade tetrahydrofuran were purchased from Fisher Scientific. Isotope samples of methanol (MeOH-d4; 99.8 atom % D), acetonitrile (ACN-d3, 99.8 atom % D), tetrahydrofuran (THF-d8, 99.5 atom % D), and water (D2O, 99.8 atom % D) were obtained from Aldrich. Uracil (>99%) and thiourea (g99.0%) were purchased from Sigma Aldrich. Experimental Procedures. HPLC procedure: The experiments were carried out over the full range of eluent compositions from 0 to 100%. The binary eluent mixtures were prepared by measuring volumes separately and then mixing together. All the eluents were degassed prior to use. The column temperature was controlled at 25 °C (±0.1 °C). The tracer sample was made by diluting either methanol-d4, acetonitrile-d3, or tetrahydrofuran-d8 with D2O, and then thiourea was added to each solution. The concentration of the D2O is much higher than the rest of the compounds in the mixture because the MS response of D2O was very low compared to other compounds. A five-microliter sample was injected, and triple injections were made for each composition. The retention volume was corrected for the extracolumn contribution by subtracting the retention volume measured without a column from the retention volume measured with a column in the system. MS procedure: The mixture of tracers and dead time probe samples were detected in positive APCI mode. The instrument was operated in selected ion monitor (SIM) mode at m/z values characteristic of the labeled isotopes of each eluent component or dead time probe. The APCI source was particularly sensitive to tuning parameters at low mass (m/z < 50). In order to detect the isotopic tracers of water (m/z ) 19), methanol (m/z ) 36), acetonitrile (m/z ) 45), and tetrahydrofuran (m/z ) 81), a manual tune was performed to establish the optimum voltages for both the octopole Analytical Chemistry, Vol. 80, No. 17, September 1, 2008

6709

and fragmentor for each mass. Thus, in the SIM mode, the voltages were varied with mass to optimize the response for each monitored mass. The optimum voltages varied with eluent type and composition, so the manual tuning process was repeated when sensitivity loss was observed for each experiment. Only the retention time, not the peak area, was measured for the tracer pulse experiments so a linear detector response was not a prerequisite for the experiments. The effect of isotopic labeling on the retention times of the probe solutes was determined by measuring the relative retention times of labeled and natural probe solutes (ACN, MeOH, THF) in a column with a mobile phase that did not contain the probe component. The labeling effect was negligible (3 µmol/m2), the order of uptake was THF > ACN > MeOH. The values for the volumes of

ViXS ) ViS - θiMV S

(9)

(22) Schay, G. In Surface and Colloid Science; Wiley-Interscience: New York, 1969; Vol. II

organic eluent

ViS (mL)

VS (mL)

θiS

linear θiM range

methanol acetonitrile tetrahydrofuran

0.10 0.44 0.55

0.12 0.52 0.62

0.84 0.85 0.89

0.5-0.8 0.5-0.8 0.5-0.8

(23) Everett, D. H. Pure Appl. Chem. 1986, 58, 967–984. (24) Gritti, F.; Kazakevich, Y. V.; Guiochon, G. J. Chromatogr., A 2007, 1169, 111–124.

Analytical Chemistry, Vol. 80, No. 17, September 1, 2008

6711

the stationary phase calculated from regression of molar excess data in the inflection region reported in the figures24 were 0.07, 0.34, and 0.45 mL for MeOH, ACN, and THF, respectively. These data were reported for 150-mm columns. The data reported in Table 1 were measured on a 250-mm column. The data from Table 1 for VS corrected to a 150-mm column would be 0.07, 0.31, and 0.37 mL. Thus, the volumes of stationary phase measured with tracer pulse chromatography are in reasonable agreement with data previously determined with concentration pulse experiments. In order to generate a full sorption isotherm, the volumes of the organic eluent in the stationary phase must be determined at compositions less than 0.5 and greater than 0.8 volume fraction. If it is assumed that the total volume of eluent in the stationary phase is constant for 0.5 e θiM e 1.0, then ViS f VS as θiM f 1. This allows the calculation of the organic-rich region of the isotherm. Calculation of the isotherm data for the water-rich domain requires an additional, secondary assumption. Two basic strategies have been reported in the literature. These assumptions are either constancy of the dimensions (volume or thickness) of the stationary phase or constancy of the quantity of components in the stationary phase over the full eluent composition range. The experimentally measured excess volumes of the organic eluent decreases to zero as the concentration of organic component in the eluent approaches zero. Thus, an assumption of constant stationary-phase volume implies that the stationary-phase composition must vary, whereas an assumption of constant uptake implies that the volume of the stationary phase must vary. Several authors24–27 have used the assumption of constant adsorbed-phase volume or thickness with the volume or number of adsorbed film layers determined from linear regression. If it is assumed that VS measured in the linear range of the excess volume isotherm persists at eluent compositions lower than 50%, then the absolute volume of any eluent component in the stationary phase can be determined from the measured excess volume from eq 9 in the form ViS ) ViXS + θiMV S

(10)

The absolute isotherm for acetonitrile calculated from eq 10 and the values from Table 1 are shown in Figure 3. Other investigators have assumed that the amount of uptake of one eluent component (usually water) was constant over the full eluent composition range. Zhu28 developed a unique method for measuring VS in the linear region of the excess isotherms without resorting to linear regression. He reasoned that in this region where ViS was constant, the retention volume equation for concentration pulse chromatography, i.e., VR,i ) V M + ∂ViS ⁄ ∂θiM

(11)

where VR,i is the retention volume of a concentration pulse. Equation 11 would reduce to VR,i ) VM because the derivative term would be zero. Thus, the value of VM and VS were (25) Kazakevich, Y. V. J. Chromatogr., A 2006, 1126, 232–243. (26) Chan, F.; Yeung, L. S.; LoBrutto, R.; Kazakevich, Y. V. J. Chromatogr., A 2005, 1082, 158–165. (27) Gritti, F.; Guiochon, G. J. Chromatogr., A 2007, 1155, 85–99. (28) Zhu, C.; Yun, K. S.; Parcher, J. F. Anal. Chem. 1995, 67, 1596–1602.

6712

Analytical Chemistry, Vol. 80, No. 17, September 1, 2008

Figure 3. Absolute volumes of acetonitrile in the stationary phase as a function of eluent composition. Line calculated with three assumptions S S for θiM < 0.5: ---, VS is constant; - · -, Vwater is constant; s, Vwater ) 0.

determined at a specific eluent composition where VR,i was a minimum for a concentration pulse. The absolute volume of a component in the stationary phase could then be determined from eq 10 with experimental data for ViXS. In order to determine the absolute isotherm at other eluent compositions, Zhu assumed that the volume of water in the stationary phase was constant for all eluent compositions. The volume of eluent in the mobile phase could then be determined from the retention time of an isotopic tracer for water (D2O) from the relation over the full range of eluent composition / M V M ) VR,D - VDS2O ⁄ θD 2O 2O

(12)

Elimination of VM from the retention volume equation for component i, yields an equation for the absolute isotherm of the organic eluent component

[

/ / ViS ) θiM VR.i - VR,D + 2O

VDS2O M θD 2O

]

(13)

In the present investigation, the value VM was not determined from concentration pulse data as suggested by Zhu. However, the data in Table 1 provides a value for the volume of water in the stationary phase from the relation VSD2O ) VS - ViS. If it is assumed that this value is constant, the absolute isotherms for the three eluents could be calculated from the tracer pulse chromatographic data. The results for acetonitrile are also shown in Figure 3. The final method for the calculation of absolute phase volumes in RPLC systems was developed by Alvarez-Zepeda and Martire in 1991.1 These authors assumed that no water was present in the stationary phase at eluent compositions of less than 50% water. The rationale for this assumption was the observation that C18bonded phases were not wetted by water at these compositions. * M If this hypothesis is valid, then from eq 2, VR,D and the 2O ) V absolute isotherm for the other component can be determined from the relation / / ViS ) {VR,i - VR,D }θiM 2O

(14)

Figure 4. Absolute volumes of water in the stationary phase as a function of eluent composition. Line calculated with three assumptions S S for θiM < 0.5: ---, VS is constant; - · -, Vwater is constant; s, Vwater ) 0.

The absolute isotherms for all three eluents were calculated from eq 14. The results for acetonitrile are shown in Figure 3. Comparison of the three calculation strategies illustrated in Figure 3 shows that the derived isotherms are similar in magnitude, so it is problematic to specify which is more realistic. On the other hand, the calculated isotherms for water based on the three strategies vary significantly as shown in Figure 4. Assumption of constant stationary-phase volume implies that significant amounts of water must be sorbed from water-rich eluents in order to maintain the Gibbs dividing surface at a fixed distance for the surface. Assumption of constant amount (volume) of water sorbed also results in the predicted uptake of water from water-rich eluents. Both of these assumptions seem to run counter to the observed hydrophobicity of C18-bonded phases. The assumption of no water uptake from water-rich eluents coupled with the assumption of greater uptake in organic-rich eluents produces a reasonable isotherm. This assumption also agrees with the experimentally observed nonwettability of C18-bonded phases. Figure 5 illustrates the results for the volume of eluent in the stationary phase for all three eluents based on the assumption of no water uptake at water-rich compositions. The measured retention volumes of both thiourea and uracil were compared with the calculated volume of the mobile phase using the assumption of no water uptake from water-rich eluents. The retention volumes for thiourea and D2O along with the thermodynamic void volume are shown in Figure 6 for three eluents. The results show that the retention volume thiourea was consistently higher by 5-10% than the calculated kinetic void volume, VM. The measured retention volumes for uracil were even higher than those observed for thiourea. The relatively small differences between the retention volumes of the dead time probe solutes and the calculated kinetic void volume (mobile-phase volume) illustrated in Figure 6 have a profound effect upon the calculated volume of eluent in the stationary phase. The volume of acetonitrile in the stationary phase, ViS, was calculated from eq 2 using the retention volumes of thiourea and D2O in place of VM. The results of these calculations are shown in Figure 7 along with the isotherm calculated from regression analysis with the

Figure 5. Absolute volumes of three components in the stationary phase as a function of eluent composition. Assuming that no water is sorbed at θiM < 0.5: O, methanol; 4, acetonitrile; 0, tetrahydrofuran.

Figure 6. Evaluation of thiourea as a measure of the kinetic void S volume, VM: - · -, VM, assuming Vwater ) 0 for θiM < 0.5; s, V0; `, thiourea; ], water isotope.

assumption of no water uptake from water-rich eluent (Figure 3). The results clearly indicate the magnitude of error introduced by the use of inappropriate dead time markers. CONCLUSIONS Tracer pulse chromatography is an accurate experimental method for the determination of the excess volume uptake of RPLC eluents by bonded stationary-phase packings. Atmospheric pressure chemical ionization with selected ion monitoring using a single quadrupole mass analyzer is a convenient and timely method for the detection of stable isotopic tracers of eluent components in a background of natural eluent. Analytical Chemistry, Vol. 80, No. 17, September 1, 2008

6713

secondary strategies are required for the determination of the absolute isotherms of the eluent components. Direct comparison of the retention volume of “unretained” dead time markers and the calculated volume of the mobile phase bring into question the suitability of thiourea and uracil for the accurate determination of the kinetic void volume of RPLC columns. The fact that in RPLC systems the stationary and mobile phases are usually similar and may only differ in composition means that it is improbable that a given solute would be soluble in the mobile phase but not in the stationary phase over the full range of eluent composition. Water would appear to be the most promising candidate for an unretained solute; however, this solute is only unretained with eluents that are low in organic component. This difficulty of identifying a truly unretained solute has plagued liquid chromatography for decades. Figure 7. Absolute volumes of acetonitrile in the stationary phase as a function of eluent composition. Calculated with three assumptions: s, isotherm from Figure 3; ---, using thiourea as a dead time probe; · · · , using D2O as a dead time probe.

The directly measured excess sorption data for three eluents can be used to indirectly determine the absolute volume of eluent components taken up by the bonded-phase packings. Various

6714

Analytical Chemistry, Vol. 80, No. 17, September 1, 2008

ACKNOWLEDGMENT This research was supported by grant CHE-0715094 from the National Science Foundation. Received for review April 18, 2008. Accepted June 14, 2008. AC800776C