Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A
jou rn al h om epage: www.elsev ier .com/ locat e/chroma
xtended zones of operations in supercritical fluid chromatography
bhijit Tarafder, Georges Guiochon ∗
niversity of Tennessee, Knoxville, TN 37996, USA
r t i c l e i n f o
rticle history:eceived 7 July 2012eceived in revised form8 September 2012ccepted 20 September 2012vailable online 1 October 2012
a b s t r a c t
The pressure and temperature ranges within which supercritical fluid chromatography is operated aregenerally decided based upon limitations imposed by the instrument or by the stationary phase. Becausethe maximum pump outlet pressure of most commercial instruments is near 400 bar and the maximumtemperature at which most chiral stationary phases are stable is usually below 318 K, the possibility ofperforming analyses at sub-ambient temperatures (e.g., below 293 K) and under sub-critical pressures(i.e. below 73.8 bar) should be explored. This work investigates the performance of separations made in
eywords:sopycnic plotFCFC under sub-ambient conditionsFC under subcritical conditions
One of the main advantages of which supercritical fluid chro-atography (SFC) is often credited with is the “tunability” of the
olvent properties, which originates from the use of carbon dioxides the primary component of the mobile phase. Either in the sub-r in the supercritical state, carbon dioxide is significantly moreompressible than the organic solvents typically used in HPLC. Anmportant outcome of this high compressibility is the possibilityf altering the mobile phase density considerably by changing itsressure and/or temperature within the limits in which SFC instru-ents can be used. Because changes in its density strongly affects
ll the physical properties of CO2, its usage as the main eluent com-onent provides practitioners with an opportunity to “tune” theobile phase properties by manipulating the operating pressure
nd/or the temperature.During the design of SFC operations, the possibility to adjust the
obile phase properties by varying the temperature and the pres-ure of the solvent is rarely employed and often not even evaluated.he main emphasis of those who design SFC separation methodseems to be on selecting the best composition of an isocratic mobilehase or the best characteristics of a solvent gradient, using someffective organic modifier. This shifts the focus of design method-
logy from developing operations in which the elution and theeparation of the sample components are based on tuning the tem-erature and the pressure to developing separations in which these
properties are tuned by adjusting the concentration of a strong sol-vent in the mobile phase, as is done in HPLC. This shift has takenplace progressively over time, influenced by multiple reasons. Nei-ther the scope nor the objectives of this report were to analyzethese reasons; it merely lists a few of the events that might haveinfluenced this shift.
1. There has long been confusion regarding the selection of themost suitable operating conditions in SFC. Working in the super-critical region of neat CO2, where diffusion coefficients are low,hence the column efficiencies are expected to be high, is attrac-tive. However, this is often considered as dangerous, due to theefficiency loss of columns operated under conditions close to thecritical point of CO2, especially under pressures below 100 barand at temperatures less than 373 K. This issue and the origins ofthe efficiency loss were recently discussed in some detail [1–3].
2. Analysts feel compelled to add relatively polar organic modifiersto CO2 since it was realized that the applications of SFC couldbe significantly expanded by this addition to the non-polar CO2mobile phase. The inclusion of modifiers introduces a new setof tuning options to the optimization of SFC operations, whichcould be more effectively utilized to manipulate solute reten-tions in SFC. Additionally, the similarity between this approachand the one used to optimize gradient HPLC operations increasedits appeal among practitioners, especially for those who hadacquired experiences in reversed-phase LC operations.
3. There is reluctance at selecting operating temperatures higherthan 313–318 K, due to the possible degradation of certainorganic analytes, of most biomolecules, and also of stationaryphases, especially of some chiral stationary phases [4]. This
Fig. 1. Pressure and temperature constraints on the operations of some SFC sys-tems. The higher pressure constraint (400 bar) is mainly instrumental, whereas the
66 A. Tarafder, G. Guiochon / J. C
constraint practically dissuades practitioners from trying to usethe temperature as an operational tool because in many situ-ations it would not be effective enough to bring the desiredretention change by manipulating the temperature over ashorter range, given the current design of SFC instruments.
. Changing the column average pressure affects the mobile phasedensity, hence the equilibrium constants of the compoundsbetween the mobile and the stationary phases in the col-umn. However, the pressure can only be manipulated within arange that is too narrow to be effective in optimization studies.Most practitioners perceive 150 bar to be the lowest reasonableback pressure while the pump highest pressure rarely exceeds400 bar. Quite often, especially with high modifier concentra-tions, the pressure drop is around 200 bar, which leaves a narrowpressure range to optimize operations and does not provideattractive dividends to consider pressure optimization.
. There seems to be a general experience-based perception thatmore robust operating conditions can be achieved by stayingin the CO2 sub-critical region than by conducting operations inthe higher temperature region. Sub-critical temperatures, e.g.300 K [5], are selected for operating the column, to stay awayfrom regions where compressibility is high. Such selection ofsub-critical operating temperatures is often made on the basisof the critical temperature of neat CO2, even when modifiers areused in the mobile phase.
A direct effect of this shift in the operational procedures, pla-ing lesser emphasis on the selection of appropriate temperaturesnd pressures and more on that of the eluent composition, can bebserved in many reports published within the last 5 years. Thenalysis of these reports show that, albeit employing quite differ-nt temperatures and back pressures in their investigations, theractitioners rarely provide any explanation regarding the ratio-ale behind their selection of these P–T points. For example, Westnd Lesellier [6] selected an operating temperature of 25 ◦C to keephe column under sub-critical conditions at all the mobile phaseompositions that were used in their studies and selected a backressure of 150 bar. The mobile phase was a CO2/methanol 90:10v/v, %) mixture. Cazenave-Gassiot et al. [7] undertook SFC analy-es at 35 ◦C with a 100 bar back pressure to evaluate the effects ofncreasing the concentration of ammonium acetate as an additiven SFC, using methanol as the modifier (10%, v/v).When develop-ng a systematic approach to evaluate new chiral stationary phasessing methanol as the modifier, Pirzada et al. [8] selected 35 ◦C and00 bar. Pell et al. [9] investigated weak anion-exchange type of chi-al stationary phases at 40 ◦C, under 150 bar, using 25% methanols the modifier. Patel et al. [10] used 60 ◦C as the operating temper-ture and 140 or 120 bar as the operating pressures for differentodifier and their additive combinations. This operating temper-
ture had been optimized for an earlier publication, but no reasonas given for the selection of this back pressure. Hamman et al. [11]
elected a temperature of 40 ◦C, a back pressure of 120 bar, and aethanol gradient, without further explanations. Berger worked
t a temperature of 50 ◦C, under pressures of 150 and 170 bar andifferent concentrations of methanol as the modifier [12] for theharacterization of a 2.6 �m Kinetex HILIC column in SFC; the pro-edure used to select the P–T points was never discussed.
In general, these reports suggest that, although the modifier andts compositions were carefully selected, according to the chal-enges of the separation, the selections of the temperatures andhe pressures were relatively random. The general motivations forhese choices seem to have been of keeping the back pressure
arkedly higher than the CO2 critical pressure, i.e. 73.77 bar andf staying clear from the critical temperature, 304.13 K.
The broad motivation of this work was to study the extentf the influences of the temperature and the pressure on the
higher temperature constraint (318 K) is due to physical limitation of some station-ary phases to maintain the functionality. The lower constraints are mainly due toincorrect perceptions, which vary among practitioners and are arbitrary.
operating parameters in the design of SFC separations, whetheranalytical or preparative. The ability of tuning the mobile phaseproperties by adjusting the concentration of a modifier is certainlya major positive aspect of the modern possibilities of SFC. However,the possibility of using pressure and temperature as additional vari-ables during method development involves negligible additionalcost but seems worth evaluating. The more specific objective of thiswork was to explore the possibility of operating SFC in temperatureand pressure zones that are not generally considered for SFC oper-ations. To further explain the motivation of this work, we brieflydiscuss the operational limitations of SFC for some separations.
During the separations of certain pairs of enantiomers or thepurifications of some biomolecules, the temperature should notexceed 318 K to avoid damaging the stationary phase [4]. Althoughpressure variations do not seem to adversely affect either solutesor stationary phases, most commercially available SFC instrumentscome generally with a pump that has an upper pressure limitaround 400 bar. These constraints on the higher values of the pres-sure and the temperature are due to either instrumental or physicalreasons, as schematically shown in Fig. 1. They are justified. Theconstraints on the lower temperature and pressure values (Fig. 1),on the other hand, are mainly due to perceptions that widelyvary among SFC practitioners and are often arbitrary. To symbol-ize these perceived constraints, the 150 bar isobar and the 300 Kisotherm are shown in Fig. 1 as the conventionally perceived lowestpressure and lowest temperature for SFC operations, respectively(actually, temperatures lower than 300 K are rarely used for prac-tical reasons). When designing an SFC procedure and selecting theoperating points, these two constraints restrict the choices to anarrow zone in the pressure–temperature plane and discouragethe exploration of possibly better separations through manipula-tions of the pressure and the temperature. Understandably, undersuch circumstances, it often seems more pragmatic to select a “safe”pressure–temperature condition, as is generally done and to focusmore on designing the modifier gradients.
Our objective in this work was to investigate the perceived con-
straints of SFC operations, to explore the pressure–temperatureregion beyond the lower conventional boundaries of pressure andtemperature, and to clarify the physical limits of operations in thesezones. We first elaborate on the physical limits of operations of neat
A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175 167
perati
CcCpv
2l
potwahpG
rIcitbpbnascifa
papbmd
m
Fig. 2. Schematic diagrams showing plausible areas of o
O2 and of CO2/methanol mixtures with methanol volume per-entages of 5, 10 and 20. Then, similar limitations for a few otherO2/modifier mixtures are qualitatively discussed and, finally, thehysical boundaries of CO2/methanol mixtures are experimentallyerified.
. The physical limits of operations at low pressures andow temperatures
The physical reason for which the choice of low operational backressures may be limited is the location of the two-phase regionf the mobile phase on the pressure–temperature plane. A phaseransition of the mobile phase during a chromatographic separationould ruin it and it should be avoided. Some additional factors may
lso restrict the low-pressure choices, like (1) the location of theigh compressibility zone and (2) that of the low temperature gashase zone of the mobile phase, since this zone is more suitable forC than SFC operations.
Schematic diagrams of these zones are shown in Fig. 2 that rep-esents the phase diagrams of a pure fluid and of that of a mixed one.n Fig. 2a the areas shown in light gray offers favorable operatingonditions for SFC operation with a pure fluid. The so-called subcrit-cal region is a part of the liquid zone, where the back pressure canheoretically be set as low as the boundary of the VLE curve and cane much lower than the critical pressure of the pure fluid. So anyerception that an SFC separation cannot be performed at pressureselow the critical pressure is incorrect. There is no physical reasonot to venture into the liquid CO2 zone for operations and, actu-lly, this is frequently done. The lowest pressure under which SFCeparations can be performed with liquid CO2 is not limited by theritical pressure but rather by the VLE line, which is the actual phys-cal boundary. In other words, subcritical pressures are applicableor chromatographic operations at subcritical temperatures, as longs the local pressures are above the VLE line.
The phase diagram of a mixture on the pressure–temperaturelane is comparatively complex (Fig. 2b). For mixtures, there is
two-phase region spanning a wider range of temperatures and
ressures in place of the VLE curve. This two-phase region isounded by the bubble-point and the dew-point curves whicheet at the critical point of the mixture, the coordinate of which
epend on the mixture composition.1 These curves separate the
1 This means that in Fig. 2b, Pc and Tc depend on the actual composition of theixture.
on for neat CO2 and for mixtures of CO2 with modifiers.
two-phase region from the liquid and the vapor states respectively.To avoid phase separation during an SFC operation with mixedmobile phases, the accurate locations of the bubble-point and thedew-point curves corresponding to all the relevant compositionsof the mobile phase should be identified. Although, theoretically,the back pressure can be set close to the bubble-point curve of themobile phase, the main challenge is the lack of reliable data on thesecurves, which complicates the design of the operating conditionsin this region.
In the rest of this section we discuss the actual physical bound-aries of the two-phase region, i.e. the coordinates of the bubblepoint and the dew point curves of neat CO2 and of CO2/methanolmixtures. Our selection of methanol as the modifier is due to factthat (1) methanol is arguably the most popular modifier in SFC and(2) among the CO2/modifier mobile phase combinations that arenormally employed in SFC, the CO2/methanol mixture is the onlyone for which almost all the physico-chemical data relevant to SFCoperations are available from NIST (National Institute of Standardsand Technology, Denver, CO) over a very wide range of condi-tions, through the REFPROP software. In the following, we focuson exploring the operational viability of the low pressure subcriti-cal zones because this is where a clear understanding on this issueis missing.
2.1. Boundaries of the neat carbon dioxide regions
Although neat CO2 is rarely used now as a mobile phase inSFC, understanding its physical properties and operational limita-tions is important for a better understanding of the properties ofCO2/modifier mixtures. The identification of plausible operationalareas in SFC using neat CO2 can be done with high confidence due tothe availability of high accuracy data from the REFPROP software[14]. Fig. 3 shows the isopycnic plots of neat CO2, over pressureand temperature ranges between 0 and 200 bar, and 273 and 373 K,respectively. The area within the dotted and the dashed lines on theplot represents the possible extension of operating points beyondthe perceived constraints or boundaries, which were explained inSection 1 (Fig. 1).
The dotted line on the 150 bar isobar (Fig. 3) represents theperceived lowest pressure line, whereas the dashed lines repre-sent limitations for selecting operating conditions due to physical
reasons. The dashed line on the 318 K isotherm shows the uppertemperature limit which is found for some chiral stationary phases.The low temperature constraint was set at 273 K. Theoretically,the temperature of operation could be as low as ca. 220 K, since
168 A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175
Fig. 3. Isopycnic plots of neat CO2 on the pressure–temperature plane. The areacircled by the dashed and the dotted lines indicates where SFC operations can beci
tadctio
petbwell(hauceaico
ibodoaiaoptllt
Fig. 4. Isopycnic plots of CO2 with 5% methanol on the pressure–temperature plane.The area circled by the dashed and the dotted lines indicates the possible operatingzone which would be neglected if the current operational conventions are accepted.
of this curve are provided later (Section 2.5). Although the data of
arried out without phase separation. (For interpretation of the references to colorn the text, the reader is referred to the web version of this article.)
he freezing point of CO2 is 217.01 K at 25 bar, and 219.64 Kt 150 bar. But the main physical limitation could probably beue to the presence of water vapor in industrial CO2, whichould lead to ice formation and column clogging for tempera-ures below 273 K. The new constraint on the lowest pressures shown by the dashed line passing through the VLE curvef CO2.
It is probable that SFC operations at any pressure–temperatureoint inside the bounded area should not be drastically differ-nt from those around operating points within the areas outside,hose where SFC operations are currently carried out. This cane understood by considering the positions of the isopycnic lineshich transcend the perceived boundaries. However, there are
xceptions within the bounded area, roughly within the triangu-ar zone (Fig. 3) bounded by the high-temperature limit and theow pressure limit (dashed lines) and the 0.75 g/mL isopycnic lineshort-dashed line). In this area, the compressibility of CO2 is quiteigh, which may cause unexpected deformations of band profilesnd separation results [1]. Fig. 3 shows that any operation madender less than −120 bar, at temperatures higher than the criti-al temperature (red dot on Fig. 3) will cause the mobile phase toxperience high density drop along the column, which should bevoided. At subcritical temperatures, however, the fluid compress-bility is significantly lower (Fig. 3) and the operating back pressuresan be set close to the VLE region without any consequences in theperation.
These results clearly show how the region of reliable operat-ng points can be extended and why the use of liquid CO2 shoulde explored. The positions of the isopycnic lines show that theperations carried out in the supercritical zone do not differ fun-amentally from those carried out in the liquid CO2 area. Thenly physical difference is due to a slightly higher density and
lower temperature, which may bring only moderate changesn the viscosity and the diffusion coefficient. The lower temper-tures would result into higher retention factors. The benefits,n the other hand, would be the opportunity to employ backressures that would be much lower than the perceived limit,o allow operations at much higher flow rates, with markedly
onger columns, providing higher efficiencies and better reso-utions for no or for only a moderate increase in the analysisime.
The short-dashed line indicates the bubble point curve derived from experimentalresults [17] and the solid circles represent the P–T points used as the experimentaltemperatures and back pressures in this study.
2.2. Boundaries of carbon dioxide with 5% methanol
The identification of the plausible operational areas when usinga modifier becomes quite challenging due to the absence of reli-able physico-chemical data in a wide range of operating conditionsfor most possible CO2/modifier combinations. A most importantcontribution in this field is the data on CO2/methanol mixtures pro-vided by the REFPROP software (NIST), which are available now andallow fair estimates of the properties of CO2/methanol mixturesover a wide range of concentrations. The errors made in the pre-diction of these data can, however, exceed markedly those madein predicting the properties of neat CO2. A discussion on this issuewas recently provided [15].
The critical pressure and temperature of the 95:5 (v/v, %)CO2/methanol mixture is 103.83 bar and 325.75 K respectively [15],which was calculated based on published experimental data [16].To identify the two-phase region of this CO2/methanol mixture (seeFig. 4) the scheme was to calculate the dew and the bubble pointtemperatures corresponding to pressures 1–103.83 bar, at an inter-val of 1 bar. Although the temperatures corresponding to pressures1 bar to 85 bar could be calculated with the REFPROP software, therest of the temperature points had to be interpolated because REF-PROP could not estimate these points, i.e. the dew and bubble pointtemperatures corresponding to 86 bar and above. These remainingtemperatures were estimated via a linear interpolation betweenthe critical temperature and the last calculated dew and bubblepoint temperatures (at 85 bar) by the REFPROP software, respec-tively. Any other interpolation scheme was not employed as noclear information is available regarding the nature of the dew pointand bubble point curves of this mobile phase composition at thoseconditions. Details on this procedure was recently published [15].
The low-pressure boundary of the extended region of operation,based on the bubble point calculation scheme described above, ismarked by a dashed line. Another short-dashed line is also providedjust above this first curve. This curve was drawn based on experi-mental results found in the literature [17]. Details on the calculation
the properties of this mixture are less reliable than those for neatCO2, this plot could be useful in the selection of the operation con-ditions in the absence of other reliable sources. The importance of
A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175 169
Fig. 5. Isopycnic plots of CO2 with 10% methanol on the pressure–temperatureplane. The area circled by the dashed and the dotted lines indicates the possibleoperating zone which would be neglected if the current operational conventionsaee
tb
pdoopvopto
ztnos2
2
(fcwbtgidhtcltnp
Fig. 6. Isopycnic plots of CO2 with 20% methanol on the pressure–temperatureplane. The area circled by the dashed and the dotted lines indicates the possibleoperating zone which would be neglected if the current operational conventions
re accepted. The short-dashed line indicates the bubble point curve derived fromxperimental results [17] and the solid circles represent the P–T points used as thexperimental temperatures and back pressures in this study.
he error does not seem to be significant and the approach coulde practically useful.
The extended operational area on the 5% methanolressure–temperature diagram includes isopycnic plots forensities between 1 and about 0.65 g/mL. The properties of theperational points on these lines should be similar to thosef the points on the isopycnic lines in the other part of theressure–temperature diagram; the retention factors shouldary similarly. Under isothermal conditions, the retention factorf any compound should decrease with increasing operationalressure; the retention factor should also decrease with increasingemperature along the isopycnic lines. It should increase, on thether hand, with decreasing temperature along the isobars.
The uniform extension of the isopycnic lines into the boundedone in Fig. 4 indicates that operations in the subcritical tempera-ures and pressures of 95:5 (v/v, %) CO2/methanol mixtures shouldot be fundamentally different from those in the other areas ofperation. So, if necessary, we should be able to set the back pres-ure at 45 bar if the operating temperature is sufficiently low, e.g.80 K.
.3. Boundaries of carbon dioxide with 10% methanol
The two-phase region of 90:10 (v/v, %) CO2/methanol mixturesFig. 5) was calculated using the same procedure as describedor the 95:05 (v/v, %) mixtures. The two low-pressure boundaryurves (dashed and short-dashed lines) were obtained in the sameay as the same lines in the previous case. The main difference
etween the 95:05 and 90:10 (v/v, %) mixtures, can be noted fromhe isopycnic plots inside the bounded regions of the phase dia-rams of these mixtures. The compressibility of the mobile phases lower for the 90:10 than for the 95:05 (v/v, %) mixtures. Anotherifference is the marginal shift of the two-phase region towardigher temperatures. The density of the mixtures varies from 0.75o 1 g/mL within the extendable boundary, which is not insignifi-ant. This also demonstrates another advantage of working in the
ow pressure region, since this allows a considerable adjustment ofhe solvent density by varying only the temperature over a ratherarrow range, which increasingly broadens when the operatingressure is raised.
are accepted. The short-dashed line indicates the bubble point curve derived fromexperimental results [17] and the solid circles represent the P–T points used as theexperimental temperatures and back pressures in this study.
2.4. Boundaries of carbon dioxide with 20% methanol
The two-phase region of the 80:20 (v/v, %) CO2/methanol mix-ture was calculated in the same way as it was done for the othercompositions (see Section 2.2). Fig. 6 shows that the two phaseregion is shifted further toward the higher temperatures and higherpressures, and that the mobile phase density inside the boundedarea is less compressible than to the mixtures of lesser methanolconcentrations discussed earlier in this section. The shift of thetwo-phase region provides opportunity of working with lower backpressures at increasingly higher temperatures, as illustrated bycomparing Figs. 4–6.
2.5. Bubble point curves of CO2/methanol mixtures fromexperimental results
Bubble points measured by Reighard et al. [17] for CO2/methanolmixtures of different compositions are plotted in Fig. 7 (left).When used in conjunction with the values provided by REFPROP,these results can provide better estimates of the boundaries of thetwo-phase region. The methanol compositions considered in theseexperiments are quite different from those considered for this studybut we may interpolate them to extend these results into the rangeof compositions considered here. Fig. 7 (left) shows that the bub-ble point curves corresponding to CO2 mole fractions of 0.912 and0.821, or methanol mole fractions of 0.088 (8.8%, v/v) and 0.179(17.9%, v/v) respectively, almost overlap at 323 K and below, mean-ing that the bubble point temperatures corresponding to pressuresof 62, 67, 81 and 96 bar are nearly invariant for methanol concentra-tions between 17.9 and 8.8%. This may also indicate that when themethanol mole fraction is reduced to 5%, the location of the bub-ble point curve might not vary much. Based on this observation, weassumed that the bubble point curve for all three methanol compo-sitions considered here (5, 10 and 20%), for temperatures of 323 Kand below, follow the experimental bubble point curve for 8.8%methanol, which represents the highest extent of the two-phase
region in this temperature range. For temperatures above 323 K, itmay be safe to consider the bubble point curve for 0.71 CO2 (29%methanol), which represents the highest extent of the two-phaseregion. At temperatures 298 K and below, where we do not have any
170 A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175
F ethanp .918
et6ot
T
3tup
2
[ptpt
3
pppiippbskrtappbr4m4
ig. 7. Boundary of the two-phase regions with different mole fractions of CO2/mublished by Reighard et al. [17]. The mole fraction of CO2 increases from 0.105 to 0
xperimental data, we may extrapolate the bubble point tempera-ures of 298.8, 303.7, 313.5 and 322.6 K, corresponding to pressures1.91, 67.38, 80.76, and 96.32 bar, for 8.8% methanol. The extrap-lation leads to an exponential relation between the bubble pointemperatures and pressures:
bp = 0.2679 exp(0.0182pbp) (1)
The short-dashed lines on Figs. 4–6, follow this equation up to23 K and the bubble point curve of 29% methanol (Fig. 7) for higheremperatures. Admittedly, this is a simplistic scheme, but it can beseful to determine low back pressures near but out of the two-hase region.
.6. Bubble point curves of acetonitrile
The bubble points experimentally determined by Reighard et al.17] for CO2/acetonitrile mixtures of different compositions arerovided in Fig. 7 (right). In the absence of any other reference,hese results may be useful to determine the lower limits of backressures at different temperatures and mobile phase composi-ions, when using acetonitrile as the modifier.
. Operational advantages of using low back pressures
The main advantage of operating SFC separations at lower backressures is the possibility of increasing the limit of allowableressure drops of the system. Since the pressure drop is directlyroportional to the flow rate and the column length, this increase
n the allowable pressure drop provides the opportunity of increas-ng both flow rate and column length at constant inlet or outletressures. For example, if the maximum allowable pump outletressure of a system is 400 bar and the lowest limit at which theack pressure may be set is 150 bar, the maximum allowable pres-ure drop is 250 bar. Depending on the column permeability and theinematic viscosity of the mobile phase, this pressure drop will beeached for a certain flow rate, Q, and a column length, L. However, ifhe lowest limit for the back pressure is 50 bar, the maximum allow-ble pressure drop becomes 350 bar, which is 40% larger. Since theressure drop is directly proportional to the product of the mobilehase flow rate and the column length, this also means a possi-le 40% increase of the allowable value of this product, at a flow
ate of Q. During analysis this means the possibility of achieving a0% shorter analysis time or of using a somewhat longer henceore efficient column. During preparative separation the same
0% increase of the flow rate could provide a 40% increase of the
ol and CO2/acetonitrile mixture. The plots are derived from experimental resultsin the CO2/methanol mixture and 0.117–0.904 in the CO2/acetonitrile mixture.
productivity of the unit, a most significant improvement. Alter-nately, this possibility of increasing the pressure drop could betraded for a longer column.
There could be, however, some operational limitations inachieving such a large improvement. At higher flow rates, the col-umn efficiency may decrease due to an increase of the C term withincreasing flow rates. Fortunately, the C term in SFC is generallysmall, the column efficiency may not decrease rapidly with increas-ing flow rate. The situation may be more favorable for preparativeseparations, the band broadening due to column overloading beingmore important than that due to mass transfer. Thus, it might bepossible to increase the flow rate significantly, even accepting someloss of column efficiency. This is consistent with observations thatpreparative SFC tends to be carried out at high flow rates, becauseefficiency losses at these flow rates do not cause band overlappingto a degree that can greatly compromise production yields.
Other practical limitations in employing high flow rates mayoriginate from the designs of the SFC columns currently available.Some columns come with limits set by their manufacturers to theallowable pressure drops and to the column head pressures. Theselimits would restrict the implementation of this scheme if they areimposed for physical reasons because exceeding them would causeproblems, e.g. with the column bed, depending on the method ofpacking. They might as well come from the reluctance of the man-ufacturers to exceed conventional values. Practitioners should becareful about these possible consequences before implementingthis scheme of operation into the system.
Another important issue related to the use of low back pressuresis the decrease of the average density of the mobile phase in thecolumn, which may increase the average retention times of ana-lytes. Thus, the possibility of reducing retention times by increasingthe flow rate at lower back pressures is in part compensated byan increase in the retention times due to the decreased averagedensity. Fortunately, the compressibility of the mobile phases inthe subcritical region is rather low (Figs. 4–6) and the increase inretention time expected should be small.
4. Operational advantages of using low temperatures
The main advantage of considering different temperatures ofoperation is that it confers the ability to manipulate the selectivity
of the component molecules. Retention in SFC depends consider-ably on the temperature and the density [14]. The variation of theretention factors with the temperature depends on the free energyof adsorption [18], which is different for different molecules. So a
A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175 171
Table 1Mobile phase flow rates (mL/min) used under the different experimental conditions selected for this study. Set back pressures of 50 bar were tentative; the actual backpressures are given inside the brackets beside the flow rates, in the respective columns. The unit of the numbers inside the brackets is bar. The first column gives themethanol concentrations in the CO2/methanol mixture. Note that all the flow rates represent the maximum allowable flow rates under those conditions when using theinstrument described in Section 5.
software tripped the pump for high pressure limitation. The flowrates employed for these nine operating points and for the differentmobile phase compositions are listed in Table 1. The retention fac-tors of only 12 of the components were measured, under the sets of
10 5 6.5 8(54) 5.5 720 3.2 4.25 5.5 (49) 4 5
a Operating points where no experiments were carried out.
hange in the column temperature will differently affect all reten-ion factors, hence the separation factors between the bands of theample components, as reported by Pell et al. [9], who found thatemperature changes may affect enantioselectivity, which tendso increase with decreasing temperature. An increase in the col-mn temperature, on the other hand, mostly leads to enhancedeak efficiencies, potentially causing a better resolution, and to
higher optimum flow rate, causing shorter retention times [9].ith current constraints, analysts operate columns at tempera-
ures higher than 318 K [4]. In a number of cases, they might benefitrom operating at lower temperatures. An examination of the isopy-nic plots of neat CO2 and of the CO2/methanol mixtures suggestshat decreasing the temperature to 273 K would not lead to any sig-ificant change of operation if the back pressure is also decreased.nder such conditions, the increases of the density and the viscosityre insignificant. So one could carry out SFC operations conve-iently near 273 K, with all the benefits of operating with a lowiscosity mobile phase at high flow rates.
. Experimental
An analytical TharSFC method station from Waters (Milford, MA,SA) was used for all the experiments. The system is comprisedf (1) a fluid delivery module with three parallel reciprocatingumps having their head temperatures maintained at 277 K; (2) anutosampler; (3) an oven; (4) a photo diode array detector (Waters998); and (5) an automated back pressure regulator (ABPR). Theystem was slightly modified to fit our purpose of operating atub-ambient temperatures.
A water-cooled jacket from Alltech (Deerfield, IL, USA), using aaake model A81 (Karlsruhe, Germany) water chiller, was substi-
uted to the column oven of the TharSFC system. The temperaturef the water chiller was continuously monitored using a digitalhermometer (Omega HH501BT, Stamford, CT, USA). This allowedetting the temperature below ambient. The tube connecting theolumn outlet to the detector inlet was insulated with layers ofTFE tape and foam.
The column used was a 0.46 cm × 15 cm Prototype C18 columnrom Waters, packed with 2.5�m particles. Industrial grade carbonioxide cylinders were bought from Airgas (Knoxville, TN, USA) andethanol was procured from Acros.The sample used was QTM PAH Mix, bought from
igma–Aldrich. It is a mixture of 16 PAHs in methylene chloride.he compounds present in the mixture, arranged by alpha-etical order are (1) acenaphthene, (2) acenaphthylene, (3)nthracene, (4) benz[a]anthracene, (5) benzo[b]fluoranthene, (6)enzo[ghi]perylene, (7) benzo[a]pyrene, (8) 2-bromonaphthalene,9) chrysene, (10) dibenz[a,h]anthracene, (11) fluoranthene, (12)uorene, (13) indeno[1,2,3-cd]pyrene, (14) naphthalene, (15)
henanthrene, and (16) pyrene. The concentration of the sampleeceived was 2000 �g/mL and was further diluted 50:50 (v/v, %)ith methylene chloride for all the experiments. SFC analysis
llowed the detection of all these components. Twelve of them
provided large peaks that were chosen for further analysis. Achromatogram is shown in Fig. 8.
6. Results and discussion
The main objective of this study was to check whether it is pos-sible to carry out SFC separations in the subcritical, sub-ambientregion of CO2 and CO2/methanol mixtures, as discussed in theprevious sections. An additional objective was to understand indetail the importance of the operational advantages of workingin the subcritical region, as discussed in Sections 3 and 4. Ninepressure–temperature points were selected (Table 1) to explorethe extended zone of operations. These temperatures were 280,290, and 300 K, the back pressures were 50, 100, and 150 bar, andthe methanol concentrations 0, 5, 10, and 20%. However, at 300 K,the lowest pressure had to be set at around 60 and not 50 bar,because the eluent would be in a two-phase state at 50 bar. It wasalways tried to set the back pressure at 50 bar under other con-ditions. However, this could never be exactly achieved due to thenon-linear nature of the VLE and the bubble-point curves, whichprevented from reaching very close to 50 bar at some tempera-tures but allowed to reach lesser pressures at other temperatures.The actual back pressures achieved are given between bracketsalong the flow rates in the respective columns of Table 1. For allexperimental points, we increased the flow rate until the system
Fig. 8. Chromatogram of the sample mixture. During the analysis, the retention fac-tors and separation factors of the components vary along the column, depending onthe local pressure and temperature, so the values measured and shown are averagesalong the column length.
172 A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175
F er at 21 r. Thea
eoAcfColm
fwspamiAcrtehw
Ftao
ig. 9. Left: chromatogram of the EPA mix sample using 5% methanol as the modifi2) at different back pressures and temperatures with 5% methanol as the modifienalysis times significantly.
xperimental conditions indicated in Table 1. The separation factorsf each successive pairs of eluted components were also measured.lthough the measurements were made under the 36 operatingonditions listed above (three temperatures, three pressures, andour mobile phase compositions), only the results obtained withO2/methanol mixtures are reported here in detail because (1) theverall conclusions of all the results obtained with CO2 are simi-ar and (2) the data for the mobile phases containing methanol are
ore relevant for practical purposes.The practical limits at which the back pressure could be set dif-
ered from the theoretical limits, due to difficulties encounteredhen trying to maintain a sub-ambient temperature for the entire
ystem. When the chosen operating point is very close to the two-hase region and the temperature is sub-ambient, maintaining
low column temperature is insufficient. For accurate measure-ents, the temperature of the whole system must be kept low,
ncluding those of the detector and the back pressure regulator. post-column separation of the mobile phase into two phasesan disrupt the entire operation. Note that in all the experimentseported here, the average column pressures are much higher thanhe set back pressure, due to the significant pressure drop experi-
nced in these operations. However, the main point demonstratedere is that the back pressure of the column can be set at low valuesithout causing a loss of separation performance.
ig. 10. Left: retention factors of the 12 components identified in Fig. 8 as a function of theemperature on the retention factors is more important than that of the pressure. Right: snd the temperature. The separation factors at 300 K and under 150 bar are shown with thf many components.
80 K and under 50 bar as the back pressure. Right: elution time of the last peak (no. results show that using a lower back pressure and a higher flow rate reduces the
6.1. Experimental results with 5% methanol
The lowest pressure which could be reached with 5% methanolwas 58 bar at 280 K. The low pressure limit increased with increas-ing temperature. At 290 K the lowest pressure was 60 bar, andat 300 K it was 63 bar. The chromatogram obtained at the lowestpressure and temperature (58 bar and 280 K) with 5% methanol isshown in the left hand side of Fig. 9. Irrespective of the subcriticaltemperature and pressure that were selected for this experiment,the chromatogram is excellent, providing retention, resolution andthe elution of sharp peaks, with no unexpected effect. The righthand side of Fig. 9 shows the retention times of the last componentcorresponding to all nine chromatograms generated by the analy-ses carried out with 5% methanol. The results in Fig. 9 show thatthe analysis time can be significantly reduced, at all operationaltemperatures by employing a low back pressure and increasing theflow rate (Table 1) to the maximum possible pressure drop.
The retention factors of the 12 components and the separationfactors between successive pairs are shown in Fig. 10. It can beobserved that changing the temperature by 10 K has a more promi-nent effect on the retention factors data than changing the back
pressure by 50 bar. In fact, the influence of the pressure on theretention factors is generally moderate. The plot of the separa-tion factors shows that these factors could be widely affected by
back pressure and the temperature. It can be noted that the effect of a change in theeparation factors of the 2nd to the twelfth component as a function of the pressuree thick line. Note that a temperature variation affects strongly the separation factors
A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175 173
Fig. 11. Same as Fig. 9 but with 10% methanol as the modifier.
ith 10
crrUobiao
Fig. 12. Same as Fig. 10 but w
hanging the temperature to any value of the back pressure. Theesults of this section confirm our expectations. Excellent sepa-ations could be obtained with back pressures as low as 58 bar.sing such low back pressures permits significant increases of theperating flow rates (Table 1), which reduces analysis times. Also,
y manipulating the temperature, it was possible to considerably
nfluence the selectivity of the sample components. This provides most useful optimization approach during method developmentf difficult separations.
Fig. 13. Same as Fig. 9 but with 20
% methanol as the modifier.
6.2. Experiments with 10% methanol
The lowest pressure employed with 10% methanol was 54 barat 280 K. Increasing the temperature increases the low pressurelimit, which becomes 56 bar at 290 K and 65 bar at 300 K. The chro-
matogram obtained for the separation achieved at 54 bar and 280 Kis shown in the left hand side of Fig. 11. The right hand side of Fig. 11shows the retention times of the last component in the nine chro-matograms recorded in this series. The results in Fig. 11 show with
% methanol as the modifier.
174 A. Tarafder, G. Guiochon / J. Chromatogr. A 1265 (2012) 165– 175
ith 20
1ra
fnwsm
eo(i
6
wwe3opfi
7
brdcttpi
lflcbda
Fig. 14. Same as Fig. 10 but w
0% methanol like with in 5%, the analysis time can be significantlyeduced at all the temperatures, by decreasing the back pressurend increasing the flow rate (Table 1).
The retention factors of the 12 components and the separationactors of successive pairs (Fig. 12) show trends similar to thoseoted for 5% methanol, but the retention factors vary more stronglyhen the temperature is varied by 10 K than when the back pres-
ure changes by 50 bar. The separation factors could be widelyanipulated by changing the temperature.The results of this section confirm our expectations described
arlier. A back pressure of 58 bar provides a significant increasef the operating flow rates (Table 1), reducing the analysis timesFig. 11). The selectivity of the separation could be considerablynfluenced by manipulating the temperature.
.3. Experiments with 20% methanol
The general observations of the results of the experiments madeith 20% methanol (Figs. 13 and 14) are similar to those obtainedith lower concentrations. The lowest pressures which could be
mployed with 20% methanol are 49, 53 and 62 bar for 280, 290 and00 K respectively. These pressures (as well as the lowest pressuresbtained with other methanol percentages) do not represent thehysical limits of the pressure at which separations could be per-ormed but rather the limits of the instrument used and of thenstrumental set-up selected for this study.
. Conclusion
The results of this work clearly demonstrate that the physicaloundaries for SFC operations in the low pressure, low temperatureegions, for all methanol concentrations studied are profoundlyifferent from the perceived boundaries of operation. We couldonveniently conduct operations close to the boundaries of thewo-phase regions of the mobile phase used, which provides impor-ant advantages for SFC operations, while benefiting from solventroperties that are comparable with those achieved when working
n more traditional chromatographic areas.We showed that the main operational advantage of selecting
ower back pressure is the possibility of working with much higherow rates. By lowering the operating back pressure into the sub-
ritical zone, the retention and separation factors are barely alteredut the expanded window offered by allowing a higher pressurerop along the column and operating at higher flow rates provides
marked reduction of the analysis time, which can theoretically be
% methanol as the modifier.
up to 40%. Application of this new approach may provide significantadvantages in preparative chromatography since it offers the pos-sibility of increasing the productivity by using higher flow rates, ifa possible decrease in the column efficiency due to the increasedflow rate does not affect too much the yield of the pure product.
We also showed that by extending the possible operating tem-perature down to 273 K from the current value of 300 K, we canconsiderably expand the retention and selectivity windows andmanipulate to a great extent the selectivity by changing the tem-perature. This approach may be advantageously used to improvedifficult separations.
Finally, this study, which highlights the operational advantagesof employing sub-critical pressures and sub-ambient temperatures,indicates also possible areas in which improvements in instrumentand column design are desirable. It should be possible to set thecolumn temperature in commercial chromatographs for SFC belowambient temperature; back pressure regulators should work effec-tively in a pressure range going as low as 50 bar. Manufacturersshould offer SFC columns able to withstand high pressure dropsand high head pressure, in order to permit analysts to take fulladvantage of the wide range of possible SFC operations.
Acknowledgements
This work was supported in part by grant CHE-1108681 of theNational Science Foundation, by financial and technical support byWaters Technologies Corporation, and by the cooperative agree-ment between the University of Tennessee and the Oak RidgeNational Laboratory. We thank Martin Gilar (Waters TechnologyCorporation) for fruitful discussions and for his support. We alsothank Jeff Kiplinger (Averica Discovery Services) for highlightingpractical constraints of SFC operations and for many insightful sug-gestions.
References
[1] A. Tarafder, G. Guiochon, J. Chromatogr. A 1218 (2011) 7189.[2] K. Kaczmarski, D.P. Poe, G. Guiochon, J. Chromatogr. A 1217 (2010) 6578.[3] K. Kaczmarski, D.P. Poe, G. Guiochon, J. Chromatogr. A 1218 (2011) 6531.[4] L. Toribio, M. del Nozal, J. Bernal, C. Cristofol, C. Alonso, J. Chromatogr. A 1121
(2006) 268.[5] E. Lesellier, J. Chromatogr. A 1216 (2009) 1881.[6] C. West, E. Lesellier, J. Chromatogr. A 1110 (2006) 181.
[7] A. Cazenave-Gassiot, R. Boughtflower, J. Caldwell, L. Hitzel, C. Holyoak, S. Lane,
P. Oakley, F. Pullen, S. Richardson, G.J. Langley, J. Chromatogr. A 1216 (2009)6441.
[8] Z. Pirzada, M. Personick, M. Biba, X. Gong, L. Zhou, W. Schafer, C. Roussel, C.J.Welch, J. Chromatogr. A 1217 (2010) 1134.
hroma
[
[[[
[
A. Tarafder, G. Guiochon / J. C
[9] R. Pell, W. Lindner, J. Chromatogr. A 1245 (2012) 175.
10] M. Patel, F. Riley, M. Ashraf-Khorassani, L. Taylor, J. Chromatogr. A 1233 (2012)
85.11] C.D.E.S. Hamman Jr., M. Wong, M. Hayes, J. Chromatogr. A 1218 (2011) 7886.12] T.A. Berger, J. Chromatogr. A 1218 (2011) 4559.14] A. Tarafder, G. Guiochon, J. Chromatogr. A 1218 (2011) 4569.
[[[
togr. A 1265 (2012) 165– 175 175
15] A. Tarafder, K. Kaczmarski, D.P. Poe, G. Guiochon, J. Chromatogr. A 1258 (2012)
136–151.
16] J. Liu, Z. Qin, G. Wang, X. Hou, J. Wang, J. Chem. Eng. Data 48 (2003) 1610.17] T.S. Reighard, S.T. Lee, S.V. Olesik, Fluid Phase Equilib. 123 (1996) 215.18] C.R. Yonker, B.W. Wright, R.C. Petersen, R.D. Smith, J. Phys. Chem. 89 (1985)
