Superheated Water Eluent Capillary Liquid Chromatography

Talanta 56 (2002) 977–987 www.elsevier.com/locate/talanta

Superheated water eluent capillary liquid chromatography

T. Scott Kephart, Purnendu K. Dasgupta *

Department of Chemistry and Biochemistry, Texas Tech Uni6ersity, Lubbock, TX 79409-1061, USA Received 24 January 2001; accepted 12 February 2002

Abstract

A capillary scale reverse phase liquid chromatography (LC) system using a super hot water eluent is described. The system, constructed in-house from readily available components, has been shown to operate at temperatures as high as 370 °C and pressures in excess of 10 000 psi. The capability of the system is demonstrated with the separation of a mixture of polar and non-polar benzene derivatives on polybutadiene and elemental carbon modiﬁed zirconia packings with or without temperature gradients. Six benzene derivatives can be separated in 2 min. © 2002 Elsevier Science B.V. All rights reserved.

1. Introduction Throughout the temperature range above its nor- mal boiling point up to the supercritical tempera- The environmental impact of organic solvents ture, liquid water exhibits a much lower dielectric and the consequent economics of their disposal constant than room temperature water and is has provided much impetus to limit their con- likely less aggressive than supercritical water. The sumption or to eliminate their use altogether. In dielectric constant, viscosity and hydrogen bond- 1995, Hawthorne et al. demonstrated that super- ing of water changes continuously from room critical water at 400 °C and 350 bar can be used temperature up to supercritical conditions; these to extract non-polar analytes such as polycyclic are the properties that are likely to influence the aromatic hydrocarbons (PAHs) and polychlori- behavior of water as a liquid chromatographic nated biphenyls (PCBs) from contaminated soil eluent. samples [1,2]. Under these extreme conditions, the Foster and Synovec were the first to explore the dielectric constant of water is greatly reduced [3]. reduction of the stationary/mobile phase ratio to Supercritical water is a very aggressive solvent exploit the use of pure water as an LC eluent with that readily oxidizes or decomposes many sub- highly polar stationary phases. Separations could stances; supercritical water treatment has been be accomplished at room temperature but efficien- studied as a method to decompose toxic waste [4]. cies for non-polar analytes were modest [5]. In the same year, the spectrometric advantage of a pure superheated water eluent, in that it was transpar- * Corresponding author. Tel.: +1-806-742-3067; fax: +1- 806-742-1289. ent even down to 190 nm was pointed out by E-mail address: [email protected] (P.K. Dasgupta). Smith and Burgess [6]. Miller and Hawthorne first

0039-9140/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S0039-9140(02)00049-8 978 T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 used a traditional reversed phase (ina2mm to 400 °C and pressures up to 11,000 psi with column format) and utilized the temperature ef- water as an eluent. When no back pressure is fect on the solvent properties of water, with ther- applied at the column exit with an FID as a mal gradients to 175 °C to separate alcohols, detector, the effluent is obviously in the gas phase. polyhydroxybenzenes and amino acids [7]. About At what point does the liquid turn in to gas? the same time, Smith and Burgess separated phe- Early studies of gas chromatography with FID nols, barbiturates and parabens on traditional detectors that use steam as the mobile phase are PS-DVB and ODS-bonded silica columns with well known [16,17]. Do the present separations water as hot as 200 °C [8]. Yang et al. were the with FID detection represent gas, or liquid phase first to study the elution of non-polar aromatic separations, or both? hydrocarbons on traditional reversed phases using water up to 200 °C [9]. It has been shown that buffering components can also be put in the hot 2. Experimental section water eluent [10]. The absence of carbon containing compounds The pumping system shown in Fig. 1 is similar in a pure water eluent allows the use of detection to the home-built gradient pumping system de- methods that are difficult or impossible to use scribed in a previous paper [18], except that only with conventional eluents. The use of a flame a single pump is required in the present work. The ionization detector (FID) can be particularly at- output from the pumping system is connected tractive [7,11]. In addition to the traditional UV– with 0.25 mm i.d. stainless steel tubing to a high Vis detector, Smith et al. showed the facile pressure inline check valve CV (cartridge applicability of NMR and MS detectors to a CV3000; Upchurch scientific, Oak Harbor, WA). superhot D2O eluent LC system with tempera- The check valve housing was machined in-house tures to 190 °C [12]. They also authored a thor- out of poly(etheretherketone) (PEEK) to handle ough review documenting the manifold utility of a pressures over 10 000 psi. The high-pressure side water eluent as hot as 240 °C [13]. One particu- of the check valve is then connected to a pressure larly noteworthy aspect is that of many analytes sensor and gauge PG (0–10 000 psi with 50% investigated, very few actually decomposed under overranging capability, model SP70-A10000, the conditions of the separation. Senso-Metrics, Simi Valley, CA) to continuously A few questions remain: for thermally stable monitor system pressure. analytes, what is the practical upper limit of the separation temperature given the most thermally stable stationary phase currently available? How fast can such separations be carried out if one takes advantage of the decreased viscosity? We attempt to answer these questions using a capillary scale system. The low flow rates and the small thermal mass in the capillary scale allow rapid temperature ramps and greatly facilitate radial heat transfer, thus minimizing radial temperature gradients [14]. The complexity of interfacing to detectors that are intrinsically compatible with a low flow rate (e.g. mass spectrometers) is also reduced. Coupling of a capillary scale water eluent LC system to an FID was reported during the Fig. 1. Hot water chromatography system, schematically preparation of this paper [15]. We report here the shown. CV, check valve; PG, pressure sensor and gauge; SSC, capillary scale separation of hydrophilic and hy- silica saturation column; CH, column heater; BPC, back pres- drophobic benzene derivatives at temperatures up sure column. T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 979

A silica pre-saturator is used to minimize the For thermal gradient separations using an FID dissolution of silica from the capillary wall (not detector, the column was placed in a gas chro- the column, the packing is not silica based) by the matography oven (Model 4300SX, Varian Inc., superhot water, a silica saturator (4.6×40 mm controlled by Varian Star 4.5 software) with the stainless steel guard column packed with 200 column outlet being directly connected to the FID mesh silica gel) was placed ahead of the injector. inlet (held at 275 °C). Others have previously A separate siliconized band heater was used to looked into optimizing the hydrogen and air flows heat the pre-saturator. This column is heated to a to the FID for maximizing the signal to noise temperature approximately 50 °C lower than the ratio. We made only crude adjustments to obtain column temperature: overheating the pre-satura- reasonable signals, no detailed sensitivity opti- tor can result in dissolved silica subsequently de- mization was carried out since improving detec- positing in the injector which is cooler (vide tion sensitivity was not an objective of this work. infra). An electrically actuated injection valve No back pressure regulator was used after the equipped with a 20 nl internal sample loop (model column and no modifications were made to the C2XL, Valco Instruments, Houston, TX, rated at FID. 15 000 psi) was connected to the guard column Water used in the experiments was distilled and and used for sample introduction. To prevent the then further purified in a Barnstead Nanopure sample from boiling before injection, the injector system. Fused silica capillaries (Polymicro Tech- loop was cooled externally by closed-loop pump- nologies, Phoenix, AZ), hydrogen (ultrapure grade, Airgas, Lubbock, TX), benzene derivatives ing of cold ethanol from a 100 ml reservoir kept (reagent grade, Aldrich) were obtained as in an ice bath through 2.5 mm i.d. Tygon tubing, indicated. wrapped around the sample loop casing with a Fused silica capillary columns (360 and 180 mm variable-speed pump drive (Model 75225, Cole i.d., Polymicro Technologies, Phoenix, AZ) were Parmer). No deposition of silica in the injector packed in-house with frits made after Kennedy was noted, either the eluent does not stay long and Jorgenson as previously described [18]. For enough in the injector or any deposited material is use with the UV detector, the frit was placed 10 washed out during sample loading. m cm from the exit end of the capillary. With the The analytical column was inserted in a 450 m FID, the frit was placed at the very end. The bed i.d. stainless steel tube around which heating tape length was 13 cm for all reported results. Two was wrapped. A layer of aluminium foil, a 1-inch different 3 mm diameter zirconia based packing thick layer of glass wool and another layer of materials were used: (a) ZrO2 modified with aluminium foil, each tightly wrapped, completed polybutadiene; and (b) ZrO2 modified with ele- the column heating enclosure CH. The tempera- mental carbon (ZirChrom–PBD and ZirChrom– ture was monitored by a platinum RTD in con- Carb, ZirChrom Separations, Anoka, MN). Only tact with the steel tube and was controlled by a for prolonged high temperatures studies with the PID temperature controller (Micromega CN770, FID, stainless steel capillaries (0.014¦ o.d., 0.007¦ Omega Engineering, Stamford, CT). i.d. tubing, ×20 cm, Small Parts Inc., Miami An absorbance detector (Linear UVIS 200, Lakes, FL) were packed with 3 mm diameter Spectra-Physics/Thermoseparation systems), de- ZirChrom–Carb. A frit was made by packing signed for on-column detection with capillaries, glass wool (0.25mm thick bed) in a 0.016 inch was used for detection. A detection wavelength of i.d., 0.020 inch o.d. stainless steel tube which 195 nm was used in all work. In order to keep the served as a butt-joint connector between the superheated liquid eluent from boiling while pass- column and an exit tube with the same dimen- ing through the detector, 10 mm i.d. capillaries of sions as the column. variable lengths were attached to the exit of the To pack columns, a 40% w/v slurry of the column to maintain a back pressure of 1000–3000 zirconia based particles in a 10% aqueous Triton psi. X-100 solution was agitated and a portion was 980 T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 transferred to the slurry reservoir. A Shandon HPLC packing pump (Shandon Southern Instru- ments Inc., Model 628 x51, Sewickley, PA) was used with water as the packing liquid. During the packing process, the slurry reservoir was con- stantly agitated and the pressure was increased from 0 to 7000 psi in 30 s, with complete column packing taking less than 3 min. After the desired column length was achieved, the packed column was then placed in an ultrasonic bath and soni- cated for 3 h with the column under 7000 psi. The pressure was then maintained at 7000 psi for 24 h. We did not use frits at the head of the column.

3. Results and discussion Fig. 2. Sixty times magnification of capillary column after structural failure. 3.1. Silica pre-saturator especially since it is impossible to provide quanti- The solubility of silica is greatly increased in tative surface coverage of whatever type of func- superhot water. Water was continuously pumped tionality is put on. Polymeric phases, especially through the column at a flow rate of 8.5 mlmin−1 poly(styrenedivinylbenzene) (PSDVB), do not and the temperature was alternately maintained at have solubility problems but these phases gener- 250 and 25 °C for 12 h periods. Without the silica ally exhibit too great a retention for non-polar pre-saturator, the column burst due to dissolution aromatic analytes for use with a water eluent. of the capillary wall in 3 days. With the silica Further, at temperatures above 250 °C, the con- pre-saturator, the column maintained its struc- tinually increasing absorbance of the column tural integrity and no voids were seen in 30 days effluent suggests that the packing is beginning to of continuous operation. At temperatures closer depolymerize. Any chemical modification that is to supercritical conditions, e.g. at 370 °C, a fused made to the surface functional group must also be silica column fails catastrophically in 1 h, gener- compatible with both the temperature and an ally developing picturesque helical cracking pat- aqueous environment. Zirconia based stationary terns that separate in sawtooth patterned phases for liquid chromatography (LC) was intro- hemicylindrical pieces (Fig. 2). The presence of duced by Carr and coworkers [19]. Pure water has the silica pre-saturator increased the operational been used as an eluent on zirconia based station- period under these conditions to 10 h; of course, ary phases [20], but high temperature purely this is still not long enough to be practical and aqueous use have not been much studied. As a some other column material is warranted. core material, zirconia is excellent due to its ex- ceptional thermal stability. Its aqueous solubility, 3.2. Thermally stable stationary phases even at elevated temperatures, is negligible. Both elemental carbon and polybutadiene modified zir- The selection of a stationary phase compatible conia based stationary phases are commercially with superheated water is obviously an issue. If available and both functional groups have good dissolution of silica from the very limited surface thermal stability and are compatible with aqueous area of a fused silica column is of concern at eluents. The elemental carbon phase is fully stable temperatures above 250 °C, the dissolution of at the maximum temperature used in this work; it silica from the core of any silica based stationary is likely that other commercially available carbon phase will obviously be a much greater problem, based phases will also be stable [13]. T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 981

3.3. Properties of superheated water m=85.31−0.2901 t+2.572×10−4 t 2, r 2 =0.9965, (1) An oft-cited paper by Hawthorne and Yang [1] contains an informative figure on the variation of p (@7200 psi)=(t−0.916977) 18 906, dielectric constant of water as a function of tem- water perature and how these values compare with com- r 2 =0.9947, (2) mon hydrorganic eluent compositions used in reversed phase LC. Similar information has been where t is the temperature in °C. The majority of subsequently published [21]. In Fig. 3, we show the change in the viscosity occurs between room both the viscosity and the dielectric constant of temperature and 200 °C. The viscosity of water water over a more extended rage of temperature. changes by 400% within this range [24], permitting Both viscosity and the dielectric constant decrease a four times longer column or a fourfold greater with increasing temperature, approximately in an flow rate at the higher temperature. The Stokes– exponential fashion [22,23]. For liquid water, the Einstein relationship [25] invokes that the diffu- viscosity is only weakly pressure dependent; we sion coefficient varies in proportion to the ratio of have chosen the viscosity data at a pressure of the absolute temperature to the viscosity. Further 7200 psi, typical of our column operations in this gains in chromatographic performance may there- study. Over the entire range of temperature of fore be attainable at high flow rates at tempera- interest, the relationships below satisfactorily ex- tures beyond 200 °C, albeit no further significant press the dielectric constant m and the viscosity p gains are likely to be realized in terms of de- of water: creased column pressure beyond this temperature.

Fig. 3. Dielectric constant and viscosity of water at 7200 psi along with the viscosity and dielectric constant of both pure ACN and a 50% ACN. 982 T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987

to compete with organic modifiers in RPLC, the system has to be able to elute polar as well as non-polar analytes. A series of benzene derivatives were separated using both the elemental carbon and polybutadiene modified 3 mm zirconia phases. The injected sample contained a mixture of seven analytes, 0.1% (v/v) phenol and 0.5% (v/v) benzene, nitrobenzene, toluene, ethylbenzene, n-propylbenzene and n-butylbenzene. A complete mixture was injected in every run; however, all the analytes did not elute from the column at lower temperatures. Fig. 5 shows the nature of these separations. At 370 °C, n-butylbenzene elutes in under 7 min (not shown in the figure). The temperature stability of ZirChrom– PBD is lower and 300 °C is the maximum practical operating temperature. At all temperatures, the PBD-phase was observed to be more reten- tive for the alkylbenzene analytes than the car- Fig. 4. Dielectric constant of water as a function of pressure. bon phase. This behavior is opposite to that observed for ambient temperature hydroorganic The dielectric constant of water is pressure-de- eluent separations on these phases [26,27]. pendent; this is more pronounced at higher tem- The retention times over the studied range of peratures. Fig. 4 shows the pressure dependence temperatures vary over such a large range that of the dielectric constant at 200, 300, 400 and the chromatograms are plotted with a logarith- 550 °C [22,23]. Even when an exit back pressure mic time axis for best viewing. The elution of is applied to keep the effluent in the liquid state the sample solvent (acetonitrile) occurs in the with UV absorption detectors (it is doubtful that this practice is essential, vide infra), the void volume and detector response is primarily pressure at the column exit is far smaller than due to its different refractive index. Apparently the column head pressure. As such, there is a the refractive index difference between water continuous and significant pressure gradient and acetonitrile increases with increasing tem- across the column. If the column is at the same perature. As such, the solvent peak increases in temperature along its length, then the dielectric magnitude as the operating temperature in- constant/polarity also varies continuously along creases. It is interesting to note that nitroben- the length of the column. A solvent polarity zene and toluene changes retention orders with gradient along the column has the same effect temperature on the carbon phase but not on the as decreasing the stationary phase capacity from PBD-phase. In a traditional hydroorganic eluent the start to the end. The solvent is more polar system, similar behavior has been observed by at the head of the column than towards the others as the polarity of the eluent is increased end; the difference can be especially pronounced [26]. The optimum temperature for the carbon at low pressures, just above the critical pressure. phase for this particular separation appears to However, for high-speed separations driven by be 170 °C while the separation continues to an increased flow rate, most of the column will improve with increasing temperature on the not experience such a low pressure. PBD phase. Most previous studies using superheated liq- Fig. 6 shows high speed separations on: (a) uid water have been limited to separating polar the ZirChrom–Carb phase at 300 °C and a flow organic analytes. For superheated liquid water rate of 20 mlmin−1 (equivalent to 13 ml min−1 T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 983 on a 4.6 mm column); and (b) the ZirChrom– 3.4. Does separation mechanism change with PBD column at 240 °Cataflow rate of 27 ml temperature? Thermodynamics of solute transfer min−1 (equivalent to 17.5 ml min−1 on a 4.6 mm column). To our knowledge, these separations are A sensitive probe for any changes in the separa- faster than any other isothermal isocratic liquid tion mechanism as a function of temperature is a chromatographic separation performed on com- Van’t Hoff plot (log k% vs. 1/T). A rectilinear plot parable compounds, not just with water as eluent. indicates that the same separation mechanism pre- Faster separations are indeed possible with hy- vails across the entire temperature range of inter- droorganic eluent gradients; this also helps main- est. Recently, Yang et al. reported that Van’t Hoff tain chromatographic efficiency for late eluting plots for the separation of substituted benzene peaks. The equivalent to this in the present system derivatives on a Nucleosil C-18 phase using a is a thermal gradient and is discussed in a later superhot water eluent were non-linear, thus sug- section. gesting a change in retention mechanism [21].

Fig. 5. Separations on (left) ZirChrom–Carb and (right) ZirChrom–PBD columns; detection at 195 nm, flow rate of 8.6 mlmin−1. s, solvent (acetonitrile): 1, phenol; 2, benzene; 3, toluene; 4, nitrobenzene; 5, ethylbenzene; and 6, n-propylbenzene. Eluent, pure water. All subsequent figures has the same numerical identification for analytes. Note that abscissa scaling in this chromatogram is logarithmic to clearly visualize the initial portion of this isothermal isocratic separation. 984 T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987

the intercept of the plots in Fig. 7, respectively [28]. The highly parallel nature of the alkyl benzene behavior in the plots are readily observed. Detailed data are presented in Table 1. The entropy terms should be considered merely relative because we can only estimate the gross (as op- posed to the active) volume of the stationary phase. Nevertheless, the data readily show that while the retention of the alkylbenzenes, especially with increasing carbon number, is promoted due to entropic reasons, the retention of the more polar phenol and especially nitrobenzene are largely governed by enthalpy.

3.5. Thermal gradients

One clear advantage of the capillary scale in high temperature LC is its ability to facilitate rapid and uniform heating throughout the

Fig. 6. High speed separations: (a) ZirChrom–Carb; and (b) ZirChrom–PBD columns at 300 and 240 °C, respectively. Eluent, pure water.

Fig. 7 illustrates that rectilinear Van’t Hoff plots are obtained for all solutes on both the ZirChrom phases over an even greater range of temperature than studied by Yang et al. The difference in the observations is therefore phase related. The behavior observed by Yang et al. could indeed be due to a shift from an adsorption to a partition like process if the C-18 chains unfold as the dielectric constant of the water decreases with increasing temperature. Unlike the C-18 packing used by Yang et al., the polybutadiene modiﬁer always lies on the surface due to extensive cross linking and thus can not collapse in an aqueous environment or open up in a non-polar environment. The nature of the carbon phase also likely remains unaltered throughout the temperature range studied. The standard enthalpy and entropy of the so- Fig. 7. Van’t Hoff plots for the retention of benzene deriva- lute transfer can be calculated from the slope and tives. T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 985

Table 1 Thermodynamic properties for solute transfer

AnalyteZirChrom–Carb ZirChrom–PBD

Enthalpy (kJ mol−1)Entropy (J (mol K)−1) Enthalpy (kJ mol−1) Entropy (J (mol K)−1)

Benzene −8.56 −12.29 −8.00396 −7.36 Toluene −9.49 −4.41 −8.19004 −5.02 Ethylbenzene−9.10 3.12 −7.93 13.01 Propylbenzene−9.02 12.74 NM NM Phenol −12.48 −25.69 −8.72 −21.41 Nitrobenzene −16.89 −23.26 −7.55 5.31 column. Fig. 8 shows a thermal gradient separa- how does it affect the resulting separation? Is it tion with flame ionization detection. The average merely all a continuum between LC with super- retention times (n=4) and the % RSD values for heated water and gas chromatography with hot the individual analytes in a six-component mix- steam [16,17]? In chemical engineering practice, ture were as follows: phenol (1.20 min, 0.80%), the back-pressure created by fluid flow through a benzene (1.35 min, 0.96%), nitrobenzene (1.65 packed bed is traditionally computed by the Er- min, 0.86%), toluene (2.02 min, 1.10%), ethylben- gun equation [29]: zene (2.44 min, 1.55%) and propylbenzene (3.16 DP/DL= min, 1.56%). The average RSD of retention times (1−b)zu 2[(150p(1−b))/(zud )+1.75]/(b 3d ) under gradient conditions was thus 1.20%. It p p (3) should be noted that this performance is obtained with a homemade pumping system that costs un- where DP is the pressure drop across a portion of der $1500 to fabricate. a packed column of length DL, z, u and p are,

3.6. When does liquid water turn in to gas?

With a detection technique like ﬂame ionization or mass spectrometry, that actually handles the sample in the gas phase, the issue of using a back pressure capillary after the detector is moot. A small diameter exit restrictor between the column and the detector can be used to keep the mobile phase in the liquid state in the column but coupling such a restrictor between the column and the detector generates unavoidable and undesir- able post-column band broadening. If the column terminus is directly coupled to the detector, some of the end portion of the column is effectively the restrictor and the mobile phase will go from liquid to the gas phase at some point in the column. In terms of experimental performance and simplicity, we found that the best results are indeed obtained Fig. 8. Thermal gradient performed in GC oven using a FID detector. One hundred and eighty micrometer i.d., 13 cm silica with the column directly coupled to the detector. capillary with ZirChrom–Carb packing, eluent, pure water, The interesting question that then arises is at ﬂow rate, 8.6 mlmin−1. Temperature gradient started at what point does the eluent boil in the column and 100 °C and was ramped to 250 °Cat50°Cmin−1. 986 T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987

respectively the density, nominal flow velocity and on the column is greater than Pb. Otherwise, e.g. the viscosity of the fluid, dp is the particle diame- if we are using a 20 cm column at 370 °Cand b −1 ter and is the void fraction in the column. The putting 8.6 mg min H2O through it the entire fluid velocity u is most conveniently calculated column will contain water in the gaseous state. In from the known mass flow rate G pumped by the actual experiments with a 20 cm stainless steel pump, the density of the fluid at the location of capillary column and connected directly to the the column under consideration (z) and the inner FID, separation of injected analytes were ob- radius of the column r: served up to temperatures 300 °C; however, as the temperature was raised to 350 °C, only a u=G/(yr 2z). (4) single undifferentiated response was observed. Calculator versions of the Ergun equation are These results therefore suggest that in this particu- available on the web where upon entering the lar packed capillary mode, without a back pres- requisite parameters, the desired parameter is au- sure on the column, operations should be limited tomatically computed [30]. We used the Ergun to temperatures at which the majority of the equation in an iterative fashion as follows. Con- column contains the eluent in the liquid phase. It sider that a column is being operated at t °C. The is interesting to note that these calculations also vapor pressure of water at this temperature is suggest that as long as mixtures of gases and known to be Pb atm, at the point in the column liquids are not formed within the detector cell the pressure drops below this value, liquid water causing consequent problems, even with an opti- is converted into the vapor phase. A constant cal detector it is not essential to have a back given temperature is assumed in any calculation. pressure capillary, absorbance can be measured

Starting then at t °CandPb atm we use an just as well in the gas phase. However, due to a iterative BASIC routine that utilizes a software greater refractive index mismatch; light through- addressable version of the steam tables [31] to put through the cell may be poorer. calculate z and p. The Ergun equation is then In summary, we have described an affordable, used to calculate the pressure drop DP over a environmentally friendly capillary scale reverse small length DL (1 mm was used in our calcula- phase LC system using superheated water eluent tions). The calculation is then repeated for a new that can operate in isothermal and temperature pressure P−DP. This procedure is repeated until gradient modes and is capable of separating both the new pressure becomes equal to the ambient polar and non-polar compounds. There is, of pressure. If the total number of iterations to reach course, no barrier to using a solvent gradient at this point is n, the total length of the column from the same time if a binary pumping system is used. the end in which the liquid water has turned into vapor is then nDL.Ataflow rate of 8.6 mg min−1 of water (same conditions as in Fig. 8), we Acknowledgements calculate for our columns (void fraction in these columns is typically 0.56) that the transition to We would like to thank ZirChrom Separations vapor phase begins respectively 0, 0.044, 0.35, for some of the phases used in this work. 2.38, 8.15, 28.3 and 48.0 cm at temperatures of 100, 150, 200, 250, 300, 350 and 370 °C. The relationship of the length with temperature is approximately exponential (r 2 =0.9858). If the References packing contains small pores, the resulting capillary effect will reduce vapor pressure. The length [1] S.B. Hawthorne, Y. Yang, D.J. Miller, Anal. Chem. 66 of the column filled with gaseous eluent as com- (1994) 2912. [2] Y. Yang, S. Bowadt, S.B. Hawthorne, D.J. Miller, Anal. puted above will be a small overestimation. Also, Chem. 67 (1995) 4571. these computations are meaningful only if a given [3] G.C. Arkelof, H.I. Oshry, J. Am. Chem. Soc. 72 (1950) column is sufficiently long and the head pressure 3844. T. Scott Kephart, P.K. Dasgupta / Talanta 56 (2002) 977–987 987

[4] M. Onobu, A. Suzuki, J. Surf. Finish Soc. Jpn. 51 (2000) [18] T.S. Kephart, P.K. Dasgupta, J. Microcol. Sep. 11 (1999) 11. 299. [5] M.D. Foster, R.E. Synovec, Anal. Chem. 68 (1996) 2838. [19] L. Sun, M.J. Annen, F. Lorrenzano-Porras, P.W. Carr, [6] R.M. Smith, R.J. Burgess, Anal. Commun. 33 (1996) 327. A.V. McCormick, J. Colloid Interface Sci. 163 (1994) 464. [7] D.J. Miller, S.B. Hawthorne, Anal. Chem. 69 (1997) 623. [20] W.W.C. Quigley, C.G. Fraga, R.E. Synovec, J. Microcol. [8] R.M. Smith, R.J. Burgess, J. Chromatogr. A 785 (1997) Sep. 12 (2000) 160. 49. [21] Y. Yang, A.D. Jones, C.D. Eaton, Anal. Chem. 71 (1999) [9] Y. Yang, M. Belghazi, A. Lagadec, S.B. Hawthorne, D.J. 3808. Miller, J. Chromatogr. A 810 (1998) 149. [22] K. Heger, M. Uematsu, E.U. Franck, Ber Bunseges. [10] R.M. Smith, O. Chienthavorn, Chromatographia 50 Phys. Chem. 84 (1980) 758. (1999) 485–489. [23] M. Uematsu, E.U. Franck, J. Phys. Chem. Ref. Data 9 [11] B.A. Ingelse, H.G. Janssen, C.A. Cramers, J. High Reso- (1980) 1291. lut. Chromatogr. 21 (1998) 613. [24] L. Haar, J.S. Gallagher, G.S. Kell, National Bureau of [12] R.S. Smith, O. Chienthovorn, I.D. Wilson, B. Wright, Standards/National Research Council Steam Tables, S.D. Taylor, Anal. Chem. 71 (1999) 4493. Hemisphere Publishing Corp, Bristol, 1984. [13] R.M. Smith, R.J. Burgess, O. Chienthavorn, J.R. Bone, [25] F. Daniels, R.A. Alberty, Physical Chemistry, John Wiley LC GC Mag. 17 (1999) 938. & Sons, New York, 1955, pp. 503–506. [14] H. Poppe, J.C. Kraak, H.H.M. Vandenberg, Chro- [26] T.P. Weber, P.T. Jackson, P.W. Carr, Anal. Chem. 67 matographia 14 (1998) 515. (1995) 3042. [15] E.W.J. Hoijschuur, C.E. Kientz, U.A.Th. Brinkmann, J. [27] J. Li, P.W. Carr, Anal. Chem. 68 (1996) 2857. High Res. Chromatogr. Chromatogr. Commun. 23 (2000) [28] S.O. Akapo, C.F. Simpson, Chromatographia 44 (1997) 309. 135. [16] A. Nonaka, Anal. Chem. 44 (1972) 271. [29] S. Ergun, Chem. Eng. Prog. 48 (1952) 89. [17] B.A. Rudenko, A. Baydarovtseva, V.A. Kuzovkin, V.F. [30] http://www.processassociates.com/process/ﬂuid/rx–1.htm. Kucherov, J. Chromatogr. 104 (1975) 271. [31] www.Taftan.com.