Anal. Chem. 1997, 69, 2541-2549
High-Efficiency Solvating Gas Chromatography Using Packed Capillaries Yufeng Shen and Milton L. Lee*
Department of Chemistry and Biochemistry, Brigham Young University, Provo, Utah 84602
In this study, column efficiency in packed capillary column solvating gas chromatography (SGC) was investigated. Long (>3 m) fused silica capillaries with an inner diameter of 250 µm were packed with 10 and 15 µm spherical porous (300 Å) octadecyl bonded silica particles using a CO2 slurry packing method. A 336 cm × 250 µm i.d. fused silica capillary containing 10 µm particles provided a total column efficiency of 264 000 plates (k ) 0.41), corresponding to a reduced plate height of 1.27, using CO2 as the mobile phase at a column inlet pressure of 260 atm. A minimum plate height of 12.7 µm and a maximum plate number per unit time of 813 plates/s were obtained using packed capillary SGC. Retention factors were dependent on the column inlet pressure but independent of the pressure gradient along the column. Gasoline and diesel samples were separated under SGC conditions, and the results were comparable to those obtained using typical open tubular column gas chromatography. Packed column gas chromatography (GC) was extensively studied in the past.1-10 In recent years, high-efficiency GC has been primarily associated with open tubular columns. To improve the total column efficiency when using packed columns, either long columns or small particles are necessary. However, these columns produce a significant column resistance to mobile phase flow, and a high column inlet pressure is needed to force the mobile phase through the column. Therefore, high performance packed columns are always associated with high column inlet pressure. Packed column efficiency can be described by total column efficiency (N), column efficiency per column length (n), plate height (H), reduced plate height (h), or column efficiency per unit time (Nt). Myers and Giddings prepared an extremely long column (4000 ft) packed with large particles (50-60 mesh) and obtained a total column efficiency on the order of 106 with a column inlet pressure of 2500 psi, corresponding to a reduced plate height of 2.2 This is the highest plate number and lowest reduced plate height (1) Giddings, J. C. Anal. Chem. 1964, 36, 741-744. (2) Myers, M. N.; Giddings, J. C. Anal. Chem. 1965, 37, 1453-1457. (3) Myers, M. N.; Giddings, J. C. Anal. Chem. 1966, 38, 294-297. (4) DiCorcia, A.; Liberti, A.; Samperi, R. J. Chromatogr. 1978, 167, 243-252. (5) Lu, P. C; Zhou, L. M.; Wang, C. H.; Wang, G. G.; Xia, A. Z.; Xu, F. B. J. Chromatogr. 1979, 186, 20-35. (6) Carter, H. V. Nature 1963, 197, 684. (7) Cramers, C. A.; Rijks, J. J. Chromatogr. 1972, 65, 29-37. (8) Huber, J. F. K.; Lauer, H. H.; Poppe, H. J. Chromatogr. 1975, 112, 377388. (9) Lauer, H. H.; Poppe, H.; Huber, J. F. K. J. Chromatogr. 1977, 132, 1-16. (10) Welsch, Th.; Engewald, W.; Poerschmann, J. J. Chromatogr. 1978, 148, 143-149. S0003-2700(97)00011-5 CCC: $14.00
© 1997 American Chemical Society
obtained in packed column GC, and these results confirmed that a high pressure drop and a large mobile phase linear velocity gradient had limited effects on column efficiency. However, because of the use of large particles, a low column efficiency per unit time (less than 80 plates/s) and a large plate height (larger than 500 µm) required an intolerably long analysis time to obtain this high efficiency. To improve column efficiency per unit time and per column length, microparticles must be used. Myers and Giddings also prepared a 2 m long column packed with 13 µm particles, and a minimum plate height of 82 µm was obtained, corresponding to 24 000 plates/column and a reduced plate height of 6.3.3 Corcia et al. packed a 21 cm long column containing 20-25 µm particles, and they obtained 17 000 plates/ m, corresponding to a plate height of 60 µm and a reduced plate height of 2.5-3.0.4 This is the lowest reduced plate height value obtained in packed column GC when microparticles were used as packing materials. Lu et al. prepared a 10 cm long column containing 7 ( 2 µm particles, which produced 45 000 plates/m, corresponding to a reduced plate height of 3.1.5 This is the largest plate number per meter obtained in packed column GC. However, these shorter columns had a limited total column efficiency. The selection of mobile phase in packed column GC is important in order to carry out high-performance separations. Carbon dioxide as a mobile phase in packed capillary GC has the following advantages: (1) higher column efficiency can be obtained using CO2 as the mobile phase in packed column GC than by using other lighter gases;5 (2) the solvating power of dense CO2 can offset the large retention of solutes in packed columns; (3) the high column inlet pressure can be easily controlled using a supercritical fluid chromatography (SFC) pump. It should be mentioned that CO2 is a gas at pressures below 72.9 atm. Beyond this point, it becomes a supercritical fluid. Therefore, when using CO2 as the mobile phase with long packed columns containing microparticles (which require high column inlet pressures), the mobile phase at the inlet end of the column is a supercritical fluid, and at the outlet end it is a gas. We have called this method “solvating gas chromatography (SGC)”.11 When using other lighter gases such as He and N2 as mobile phases in packed column GC, the term “high-pressure gas chromatography (HPGC)” is probably more suitable because the mobile phase behaves as a gas along the whole column length. A previous study showed that SGC was more suitable to carry out high speed separations than HPGC when packed columns were used.11 In the past, there have been two limitations in the performance of packed column GC. First, a reduced plate height as low as 2 was never achieved using microparticles as packing material in (11) Shen, Y.; Lee, M. L. J. Chromatogr., submitted.
Analytical Chemistry, Vol. 69, No. 13, July 1, 1997 2541
GC, which is typical of packed column liquid chromatography (LC). Second, the total column efficiency and plate number per unit time were never as high as the values obtained by using open tubular column GC (∼100 000 plates/column, 100-600 plates/ s). This has limited the practical use of packed columns for the separation of complex samples. The aim of this study was to resolve these two limitations by using packed capillary SGC. EXPERIMENTAL SECTION Materials and Instrumentation. Spherical porous (300 Å) octadecyl bonded silica (ODS) particles having diameters of 10 µm were purchased from Phenomenex (Torrance, CA). Spherical porous (300 Å) ODS particles having diameters of 15 µm were purchased from YMC (Wilmington, NC). Fused silica capillary tubing was purchased from Polymicro Technologies (Phoenix, AZ). Column connections were made using PEEK tubing and zero dead-volume unions (Valco Instruments, Houston, TX). The column packing and SGC experiments were carried out using a Lee Scientific Model 600 SFC instrument (Dionex, Salt Lake Division, Salt Lake City, UT). SFC grade CO2 (Scott Specialty Gases, Plumsteadville, PA) was used for the preparation of packed capillary columns and as the mobile phase. Helium (Scott Specialty Gases) was used as a carrier gas in open tubular column GC and HPGC. Open tubular column GC was performed using a 22 m × 250 µm i.d. fused silica capillary coated with 0.25 µm SE-54 stationary phase (Supelco, Bellefonte, PA). An HP-5890A series gas chromatograph (Hewlett-Packard, Avondale, PA) was used for open tabular column GC experiments. HPGC was performed using the same system as used for SGC; however, He was used as carrier gas instead of CO2. Other chemicals used were purchased from Sigma (St. Louis, MO) and Aldrich (Milwaukee, WI). Preparation of Packed Capillary Columns. A previously reported CO2 slurry packing method12 was modified and used to prepare long packed capillary columns containing 10 and 15 µm particles. One end of the fused silica capillary column was connected to a zero dead-volume union using PEEK tubing to position a steel screen (2 µm pores, Valco) to support the particles, and the other end of the union was opened to the atmosphere. The column was then connected to a steel vessel, in which microparticles were introduced. Using CO2 at room temperature and 40 atm, the particles were packed into the column to a length of approximately 40 cm, and then the column was placed in an ultrasonic bath to help settle the particles at the end of the column. Particles were then introduced into another 40 cm length, and the process was repeated while the pressure was gradually increased. At approximately 120 atm, the column was fully packed, and it was conditioned at room temperature and 180 atm in an ultrasonic bath for 10 min. Finally, the column was left overnight to allow for slow depressurization. SGC Experiments. SGC experiments were carried out using a Lee Scientific Model 600 SFC instrument. A manual liquid injector (Valco Instruments) with a rotor volume of 0.2 µL was used for introduction of samples. A tee was connected to the injector valve by using a 10 cm × 125 µm i.d. steel tube. A 4 cm × 15 µm i.d. fused silica capillary was used for the split line. Separation columns were connected to the tee using PEEK tubing. A 10 cm × 50 µm i.d. fused silica capillary was used to connect the separation column to the FID. CO2 was used as the mobile (12) Malik, A.; Li, W.; Lee, M. L. J. Microcolumn Sep. 1993, 5, 361-367.
2542
Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
phase in SGC, and its pressure was controlled using the SFC pump. The outlet of the pump was connected to a valve injector through a 50 cm × 1 mm i.d. steel tube, which was preheated in the chromatographic oven. HPGC Experiments. The same instrument and procedures as those used for SGC were used to carry out HPGC. Helium was used as the carrier gas. Open Tubular Column GC Experiments. Split injection with a split flow of 150 mL/min was used in open tubular column GC experiments. RESULTS AND DISCUSSION SGC and HPGC. In the introduction, we briefly described the difference between SGC and HPGC. In SGC, both SFC and GC operate in the column when CO2 is used as the mobile phase. This method is appropriately named as a form of GC because the elution of solutes depends on temperature. Solutes must be vaporized before they can be eluted from the column. The solvating power of the mobile phase, especially at the column inlet end, assists in the elution process. On the other hand, in HPGC, the mobile phase serves only as a carrier, even though high column inlet pressure is imposed. When using lighter gases such as H2, He, or N2 as mobile phases at high pressures, HPGC results. In SGC, the solvating power of the mobile phase can decrease the retention of solutes in the column. Interaction between the mobile phase and the stationary phase usually reduces interactions between the solutes and the packing material. However, slower diffusion of solutes in supercritical fluids and higher supercritical fluid viscosities can be disadvantages of SGC. In this study, HPGC and SGC using packed capillary columns were experimentally compared. Figure 1 shows HPGC and SGC chromatograms of normal and aromatic hydrocarbons. Great differences were observed. At 130 °C and 150 atm inlet pressure, naphthalene and 1-methylnaphthalene were not eluted in HPGC when He was used as the carrier gas. Although other lighter components were eluted, large retention and poor peak shapes were observed (Figure 1A). Upon increasing the temperature to 180 °C, all components were eluted; however, low column efficiency and poor peak shapes were obtained (Figure 1B). When the mobile phase was changed from He to CO2, excellent results were obtained (Figure 1C). At 130 °C and 150 atm inlet pressure, all components were eluted with excellent peak shapes. The shortened retention times and improvement in peak shapes could both be explained by the solvating power of the mobile phase and the interaction between the mobile phase and the stationary phase, which reduces the interactions between solutes and the stationary phase. It was found that when the same column, split tube, concentration of solutes (∼2% total concentration), temperature, and inlet pressure were used, SGC produced a much higher detector response than HPGC. Figure 1C was obtained by increasing the attenuation 10 times compared to that in Figure 1A, while all other experimental conditions remained the same. In our experiments, high-efficiency HPGC could not be obtained when using long packed capillaries containing microparticles. Generation of Mobile Phase Flow. At a specific column inlet pressure, the resultant mobile phase linear velocity is dependent on the viscosity of the mobile phase. Differing from HPGC, the average viscosity of CO2 in the column is higher than highpressure gases because the mobile phase in SGC is a combination of gas and supercritical fluid. The higher viscosity of the
Figure 2. Relationship between column inlet pressure and mobile phase linear velocity in SGC. Conditions: 130 °C, methane used as unretained marker, (O) 228 cm × 250 µm i.d. capillary packed with 15 µm porous ODS bonded particles, (b) 336 cm × 250 µm i.d. (9) 250 cm × 250 µm i.d., and ([) 118 cm × 250 µm i.d. capillaries packed with 10 µm porous ODS bonded particles; other conditions are the same as in Figure 1.
Figure 1. HPGC and SGC chromatograms of test solutes. Conditions: 228 cm × 250 µm i.d. fused silica capillary column packed with 15 µm spherical porous (300 Å) ODS bonded particles, 150 atm column inlet pressure, (A) 130 °C, He carrier gas, (B) 180 °C, He carrier gas, and (C) 130 °C, CO2 mobile phase, flame ionization detector (FID). Peak identification: (1) benzene, (2) toluene, (3) n-octane, (4) p-xylene, (5) n-nonane, (6) n-decane, (7) butylbenzene, (8) n-undecane, (9) naphthalene, (10) n-dodecane, and (11) 1-methylnaphthalene.
supercritical fluid imposes an increased column inlet pressure in SGC to obtain a specific mobile phase linear velocity. In our experiments, no separation was obtained between methane and ethane when using columns packed with porous ODS bonded particles under SGC conditions. Therefore, it was assumed that methane had no retention in these columns, and it was used as a marker to measure the dead time. Figure 2 shows the relationship between column inlet pressure and mobile phase linear velocity for various length columns packed with 10 and 15 µm porous (300 Å) ODS bonded particles under SGC conditions. Longer columns and smaller particles led to less change in mobile phase linear velocity with pressure. When the 10 µm particle packed column length was reduced 1.35 and 1.87 times from 3.36 to 2.50 to 1.80 m, the linear velocity was increased 1.44 and 2.05 times, respectively, at any specific column inlet pressure. Retention Factor and Mobile Phase Linear Velocity. Since the solvating power of the mobile phase in SGC is affected by the pressure or density of the mobile phase, the mobile phase linear velocity, generated by imposing a certain column inlet pressure, is always associated with the retention factor. Figure 3 shows the experimental relationships between retention factor and mobile phase linear velocity for different columns. For each
Figure 3. Relationship between retention factor and mobile phase linear velocity. Conditions: n-octane as test solute; other conditions and identification of data points are the same as in Figure 2.
column, the retention factor rapidly decreased when the mobile phase linear velocity was increased. At a specific linear velocity, the longer column produced a much smaller retention factor, and the column containing 15 µm particles produced a much larger retention factor than the column containing 10 µm particles. This can be expected because the higher mobile phase linear velocities, longer columns, or smaller particles require higher column inlet pressures and result in a decrease in the retention factors of solutes. The decrease in retention factor of solutes with increasing mobile phase linear velocity produces both advantageous and disadvantageous effects on the chromatographic performance. A decrease in retention factor can reduce the resolution, but it can also speed up the elution of solutes and shorten the analysis time. This effect is somewhat similar to that obtained by increasing the temperature in GC. In our experiments, it was found that the retention factor was a direct result of the column inlet pressure and not the pressure or density gradient of the mobile phase along the column. Table Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
2543
Table 1. Relationship between k and Pi in SGCa 180 cm b
c
250 cm d
e
b
c
336 cm d
e
b
c
Pi (atm)
k1
k2
k3
k4
k1
k2
k3
k4
k1
k2
k3d
k4e
120 140 160 180 200 220 240 260 280
1.88 1.56 1.31 1.09 0.90 0.78 0.68 0.59 0.54
3.63 2.86 2.26 1.82 1.45 1.24 1.04 0.88 0.77
6.54 5.04 3.94 3.13 2.49 2.07 1.74 1.47 1.37
8.95 6.60 5.02 3.90 3.05 2.51 2.08 1.73 1.53
1.57 1.29 1.06 0.88 0.77 0.68 0.57 0.53
2.88 2.27 1.81 1.44 1.23 1.04 0.89 0.78
5.11 3.97 3.15 2.50 2.10 1.76 1.49 1.31
6.67 5.10 3.93 3.08 2.53 2.10 1.78 1.53
1.30 1.08 0.90 0.77 0.66 0.57 0.52
2.25 1.79 1.45 1.21 1.00 0.86 0.76
3.94 3.11 2.49 2.04 1.70 1.44 1.26
4.96 3.85 3.02 2.44 2.01 1.71 1.45
a Conditions: various length columns having an inner diameter of 250 µm packed with 10 µm porous (300 Å pores) ODS particles, 130 °C, CO 2 mobile phase, methane used as unretained marker, FID. b Retention factor of toluene. c Retention factor of n-nonane. d Retention factor of e n-butylbenzene. Retention factor of n-dodecane.
Figure 4. Effect of column inlet pressure on selectivity in SGC. Conditions: 250 cm × 250 µm i.d. fused silica capillary column packed with 10 µm porous ODS bonded particles; other conditions are the same as in Figure 1. Peak identification: (1) naphthalene and (2) n-dodecane.
1 clearly shows this phenomenon. Two normal and two aromatic hydrocarbons were used to investigate this relationship, using different length columns packed with 10 µm particles. At any specific column inlet pressure, solute retention factors were the same, regardless of how long the column was or what the pressure gradient was. In GC, the separation selectivity is determined only by the temperature and the stationary phase. However, in SGC, it was found that the separation selectivity and even the elution order of solutes were affected by the column inlet pressure or mobile phase linear velocity. Figure 4 shows this phenomenon. At a constant temperature of 130 °C and increasing the column inlet pressure from 140 to 220 atm, the elution order of naphthalene and dodecane was reversed. This is because the retention factors 2544
Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
of the solutes were influenced by both solvating power and temperature of the mobile phase. Therefore, in SGC, both retention mechanisms of GC and SFC are in effect. Column Efficiency in SGC. Packed columns containing microparticles create a significant pressure drop along the column. In SGC, this pressure drop results in a linear velocity gradient of the compressible mobile phase (both gas and supercritical fluid) and a retention factor gradient of solutes along the column because the solubility of the mobile phase is dependent on its pressure or density. Giddings1 theoretically treated the effect of the mobile phase linear velocity gradient on column efficiency and pointed out that the loss of column efficiency could be corrected using a compressibility factor (f2). A maximum correction of f2 (Pi . Po) was 1.125, which means that the maximum loss of column efficiency due to the linear velocity gradient was 12.5%. Using a long packed column (4000 ft) and a large column inlet pressure (2500 psi), Myers and Giddings experimentally showed that the minimum reduced plate height could be as low as 2.2 This suggests that the mobile phase linear velocity should not be a major factor affecting the column efficiency when compressible mobile phases are used. There have been few publications which theoretically analyzed the effect of retention factor gradient on column efficiency. It is difficult to treat this theoretically because this effect is always associated with the mobile phase linear velocity gradient. During a study of column efficiency in packed column SFC, this was experimentally investigated. Gere et al. showed that, when 3 µm particles were used in packed column SFC, a reduced plate height of less than 2 could be obtained.13 Berger et al. connected several LC columns together and found that the column efficiency was proportional to the column length.14,15 However, these studies were based on conventional packed columns. In our present study, packed capillaries were used, and their performances were investigated by determining total plate number (N), plate number per second (Nt), plate height (H), and reduced plate height (h) under SGC conditions. Table 2 shows the experimental relationship between N and average mobile phase linear velocity (u) for various columns under SGC conditions. Increasing the column inlet pressure and, consequently, the mobile phase linear velocity, the total column (13) Gere, D. R.; Board, R.; McManigill, D. Anal. Chem. 1982, 54, 736-740. (14) Berger, T. A.; Wilson, W. H. Anal. Chem. 1993, 65, 1451-1455. (15) Berger, T. A.; Blumberg, L. M. Chromatographia 1994, 38, 5-11.
Table 2. Relationship between Total Plate Number (N) and Mobile Phase Linear Velocity (u, cm s-1) in SGCa column 1
column 2
column 3
column 4
u
N
u
N
u
N
u
N
1.05 1.20 1.36 1.45 1.55 1.67 1.76 1.82 1.92
68 594 86 562 106 014 112 500 112 503 121 702 118 341 73 426 70 738
0.86 0.95 1.04 1.11 1.19 1.26 1.33 1.39
104 197 143 602 124 895 151 515 155 819 184 817 154 968 173 843
0.66 0.72 0.77 0.82 0.86 0.90 0.96
123 860 125 585 190 146 213 368 227 966 264 088 225 008
3.93 4.45 4.85 5.26 5.67 5.94 6.11 6.38
88 485 76 487 55 400 43 350 39 397 38 752 34 245 35 456
a Conditions: n-octane used as test solute, methane used as unretained marker, CO used as mobile phase, 130 °C, FID. Column 1, 180 cm × 2 250 µm i.d. capillary column packed with 10 µm (300 Å pores) ODS particles; column 2, 250 cm × 250 µm i.d. capillary column packed with 10 µm (300 Å pores) ODS particles; column 3, 336 cm × 250 µm i.d. capillary column packed with 10 µm (300 Å pores) ODS particles; and column 4, 225 cm × 250 µm i.d. capillary column packed with 15 µm (300 Å pores) ODS particles.
Figure 5. SGC chromatogram of test solutes. Conditions: 336 cm × 250 µm i.d. fused silica capillary column packed with 10 µm porous ODS bonded particles, 130 °C, 260 atm column inlet pressure; other conditions are the same as in Figure 2. Peak identification: (1) benzene, (2) toluene, (3) n-octane, (4) p-xylene, (5) n-nonane, (6) n-decane, (7) n-butylbenzene, (8) n-undecane, (9) n-dodecane, (10) naphthalene, (11) 1-methylnaphthalene.
efficiency was increased for all columns containing 10 µm porous ODS bonded particles. All columns produced more than 100 000 total plates. A maximum plate number of 264 000 was obtained by using a 336 cm × 250 µm i.d. column containing 10 µm porous ODS particles at a mobile phase linear velocity of 0.90 cm/s, corresponding to a column inlet pressure of 260 atm. This is the highest plate number reported in packed column chromatography using microparticles as packing materials. The column efficiency decreases with an increase in retention factor (k) when k is less than 5, and Figure 5 shows a chromatogram to illustrate the changes in N with k for the separation of normal and aromatic hydrocarbons. Even with a k value greater than 4, more than 118 000 plates were obtained using packed capillary SGC. These column efficiencies are comparable with those of typical open tubular column GC. Furthermore, excellent peak shapes were obtained. A 228 cm × 250 µm i.d. column packed with 15 µm porous (300 Å) ODS bonded particles produced a total plate number of nearly 90 000 in the experimental range of mobile phase
Figure 6. SGC chromatogram of test solutes. Conditions: 226 cm × 250 µm i.d. fused silica capillary column packed with 15 µm porous ODS bonded particles, 130 °C, 260 atm column inlet pressure; other conditions are the same as in Figure 2. Peak identification: (1) benzene, (2) toluene, (3) n-octane, (4) p-xylene, (5) n-nonane, (6) n-decane, (7) n-butylbenzene, (8) n-undecane, (9) naphthalene, (10) n-dodecane, and (11) 1-methylnaphthalene.
linear velocities studied. However, better column permeability allowed this column to carry out fast separations at lower column inlet pressures. Figure 6 shows a fast SGC chromatogram of the same solutes as shown in Figure 5. At the same temperature (130 °C) and column inlet pressure (260 atm), the separation was finished within 3 min. Of course, the resolution was reduced compared with that in Figure 5. Table 3 shows the relationship between Nt and u for various columns under SGC conditions. Comparable plates per second of 100-800 as typical of open tubular column GC were obtained in SGC. For a 180 cm column containing 10 µm particles, the plate number per second increased with increasing the mobile phase linear velocity until the linear velocity reached 1.8 cm/s. A maximum value of 780 plates/s was obtained for this column. For 250 and 336 cm long columns containing 10 µm particles, the plate number per second increased with increasing mobile phase linear velocity throughout the experimental range of the linear velocities studied. Comparing the plate numbers per second obtained using 10 µm particle packed columns, it can be seen that, at a specific mobile phase linear velocity, increasing the column length Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
2545
Table 3. Relationship between Plate Number per Second (Nt, plates s-1) and Mobile Phase Linear Velocity (u, cm s-1) in SGCa column 1
a
column 2
column 3
column 4
u
Nt
u
Nt
u
Nt
u
Nt
1.05 1.20 1.36 1.45 1.55 1.67 1.76 1.82 1.92
187 289 426 525 598 732 780 525 563
0.86 0.95 1.04 1.11 1.19 1.26 1.33 1.39
180 297 305 421 483 633 579 669
0.66 0.72 0.77 0.82 0.86 0.90 0.96
131 156 271 340 401 504 464
3.93 4.45 4.85 5.26 5.67 5.94 6.11 6.38
736 813 693 628 631 702 664 738
Conditions and identification of columns are the same as in Table 2.
Figure 7. Open tubular column GC, packed capillary SGC, and packed capillary HPGC chromatograms of a gasoline sample. Conditions: temperature programming from 40 °C (4 min) to 300 °C at 2.5 °C min-1. (A) 22 m × 250 µm i.d. fused silica capillary column coated with 0.25 µm SE-54 stationary phase, He carrier gas. (B) 180 cm × 250 µm i.d. fused silica capillary column packed with 10 µm porous ODS bonded particles, CO2 mobile phase, pressure program from 110 to 150 atm at 0.5 atm min-1. (C) He carrier gas, 150 atm column inlet pressure. Other conditions are the same as in Figure 1.
increased the plate number per second. For example, at a linear velocity of 1.2 cm/s, the 180 cm long column produced approximately 350 plates/s, while the 250 cm column produced approximately 500 plates/s. However, as shown in Figure 2, the shorter column produced a larger mobile phase linear velocity 2546 Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
when using a specific column inlet pressure, which favors higher values of plates per second. Therefore, at a specific column inlet pressure, a higher plate number per second can be obtained by using shorter columns. For example, at a column inlet pressure of 220 atm, 732 plates/s were obtained using a 180 cm length
Table 4. Relationship between Plate Height (H, µm) and Mobile Phase Linear Velocity (u, cm s-1) in SGCa column 1
a
column 2
column 3
column 4
u
H
u
H
u
H
u
H
1.05 1.20 1.36 1.45 1.55 1.67 1.76 1.82 1.92
26.2 20.8 17.0 16.0 16.0 14.8 15.2 22.0 25.4
0.86 0.95 1.04 1.11 1.19 1.26 1.33 1.39
24.0 17.1 20.0 16.5 14.0 13.5 16.1 14.4
0.66 0.72 0.77 0.82 0.86 0.90 0.96
27.0 22.8 17.7 15.7 14.8 12.7 14.9
3.93 4.45 4.85 5.26 5.67 5.94 6.11 6.38
25.7 29.8 41.1 52.6 57.8 58.8 66.5 64.0
Conditions and identification of columns are the same as in Table 2.
Figure 8. Open tubular column GC and packed capillary SGC chromatograms of a diesel sample. Conditions: Temperature program from 60 °C to 300 °C at 2.5 °C min-1. (A) 22 m × 250 µm i.d. fused silica capillary column coated with 0.25 µm SE-54 stationary phase, He carrier gas. (B) 226 cm × 250 µm i.d. fused silica capillary column packed with 15 µm porous ODS bonded particles, CO2 mobile phase, column inlet pressure program from 160 to 200 atm at 0.5 atm min-1. Other conditions are the same as in Figure 1.
column, while only 340 plates/s were obtained using 336 cm column.
It was found that, by increasing the particle size from 10 to 15 µm, the plate number per second was improved under SGC Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
2547
Table 5. Relationship between Reduced Plate Height (h) and Mobile Phase Linear Velocity (u, cm s-1) in SGCa column 1
a
column 2
column 3
column 4
u
h
u
h
u
h
u
h
1.05 1.20 1.36 1.45 1.55 1.67 1.76 1.82 1.92
2.62 2.08 1.70 1.60 1.60 1.48 1.52 2.20 2.54
0.86 0.95 1.04 1.11 1.19 1.26 1.33 1.39
2.40 1.71 2.00 1.65 1.40 1.35 1.61 1.44
0.66 0.72 0.77 0.82 0.86 0.90 0.96
2.70 2.28 1.77 1.57 1.48 1.27 1.49
3.93 4.45 4.85 5.26 5.67 5.94 6.11 6.38
1.70 1.97 2.74 3.51 3.86 3.92 4.44 4.29
Conditions and identification of columns are the same as in Table 2.
conditions. For a 228 cm column packed with 15 µm ODS bonded particles, values of 700-800 plates/s were obtained at high mobile phase linear velocities, and a maximum value of 813 plates/s was obtained using this column. Although relatively high values of plates per second were obtained using columns packed with porous (300 Å) particles, such columns have not produced more than 1000 plates/s. Using nonporous particles (3 µm nonporous ODS bonded particles) as packing materials, the value of Nt was increased to 1200 plates/ s.11 Table 4 shows the relationship between H and u for various columns under SGC conditions. All three 10 µm particle packed columns produced plate heights of less than 15 µm. A minimum value of 12.7 µm was obtained by using the 336 cm column. This is the lowest value reported in packed column GC. Higher plate heights (e.g, 25.7 µm) were obtained using a 15 µm particle packed column than from a column packed with 10 µm particles. Therefore, reducing the particle size favors smaller plate heights in SGC. Table 5 shows the relationship between h and u for various columns under SGC conditions. Minimum reduced plate heights of 1.48, 1.35, and 1.27 were obtained for 180, 250, and 336 cm columns containing 10 µm particles. Since all columns containing 10 µm particles produced minimum reduced plate heights of less than 1.5, this suggests that the CO2 slurry packing method is suitable to prepare repeatable and highly efficient capillaries for SGC. The minimum reduced plate height of 1.27 obtained using a 336 cm × 250 µm i.d. column containing 10 µm porous ODS particles with a mobile phase linear velocity of 0.90 cm/s is the lowest reduced plate height reported in packed column chromatography for a column inner diameter to particle diameter ratio >10. Theoretical considerations pointed out that a well-packed column can produce a minimum reduced plate height of 2, and this value decreases when the ratio of column inner diameter to particle size is less than 8.16 Very narrow bore (25-50 µm i.d) packed capillaries have been prepared, and a reduced plate height of ca. ∼0.8 has been obtained in LC when this ratio approached 8.17,18 However, in our experiments, a minimum reduced plate height of 1.27 was obtained, even though the ratio of column inner diameter to particle diameter was 25. (16) Knox, J. H.; Parcher, J. F. Anal. Chem. 1969, 41, 1599-1606. (17) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 61, 1128-1135. (18) Cole, L. J.; Schultz, N. M.; Kennedy, R. T. J. Microcolumn Sep. 1993, 5, 433-439.
2548
Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
It is should be mentioned that the data in Tables 4 and 5 cannot be used to quantitatively analyze the sources of peak broadening using van Deemter or Knox equations. The reason for this is that every data point has a different retention factor under SGC conditions. Separations of Complex Samples Using Packed Capillary SGC. The separation of authentic, complex samples can be a challenge to any chromatographic technique. In GC, long open tubular columns are primarily used for such problems. However, the low sample capacities of open tubular columns are a disadvantage. Packed columns, including packed capillary columns, can provide much larger sample capacities than open tubular columns and would be desirable for many applications if they could provide similar resolution as open tubular columns. In chromatography, the separation can be performed using temperature, pressure (or density), and mobile phase composition programming. Temperature programming is primarily used in GC, pressure (or density) programming is widely used in SFC, and composition programming is mostly used in LC. In SGC, all of these methods can be used. A composition gradient can be used to adjust the solute retention factors. However, gas composition gradients are experimentally difficult to generate using readily available chromatographic instrumentation. Currently, temperature and pressure (or density) programming can be easily accomplished. Packed capillaries are more effective than larger, conventional packed columns for temperature programming because of their better heat transfer characteristics. For typical open tubular column GC (∼20 m columns), the column efficiency is usually about 100 000 plates/column. This column efficiency is similar to efficiencies produced using 2 m packed capillary columns containing 10 and 15 µm particles. In this study, a commercially available 22 m × 250 µm i.d. fused silica capillary column coated with 0.25 µm SE-54 stationary phase (Supelco) was used to carry out separations of gasoline and diesel samples, and the results were compared with those obtained by using packed capillary SGC. In these experiments, it was found that, when using the same temperature program, a longer analysis time was needed for SGC at a specific column inlet pressure. By increasing the column inlet pressure, the analysis time was decreased; however, the separation of lighter components in the samples became worse. In this study, pressure programming was used together with temperature programming. Figure 7 shows separations of a gasoline sample using open tubular column GC (22 m column) and packed capillary SGC (1.8 m column). In approximately the
same analysis time, packed capillary SGC provided better resolution of sample components, especially for the lighter components in the sample. This resulted from either the higher retention on the packed capillary or the different selectivities of the stationary phases. Figure 8 shows separations of a diesel fuel sample using open tubular column GC and packed capillary SGC. Larger particles (15 µm) resulted in lower packed column efficiency compared to the open tubular column; however, smaller particles resulted in increased retention of the higher molecular weight components in the sample and longer analysis time. Further work is under way to optimize the conditions for both speed and efficiency. Since separations similar to those possible with open tubular column GC can be obtained using SGC, this technology has
potential in practical use for the separation of complex samples. A previous study showed that SGC could provide high column efficiency per unit time (1200 plates/s) when using small particles, and was suitable for high speed separations.11 The study of special stationary phases, including inert and selective packing materials, is important for further developments in SGC.
Received for review January 2, 1997. Accepted March 31, 1997.X AC970011J
X
Abstract published in Advance ACS Abstracts, May 15, 1997.
Analytical Chemistry, Vol. 69, No. 13, July 1, 1997
2549
