6 August / September 2019
There were further tests where a 15 cm length of 65, 120, 170, and 250 µm tubes were inserted between the column outlet and the detector inlet, with no significant effect.
The plumbing modifications made were similar to those made to create the early UHPLC’s, and should have decreased extra-column variance by more than an order of magnitude. Surprisingly, these plumbing changes had minimal effect on total dispersion using either a 2.1x150 mm, HILIC, or a 2.1x100 mm HILIC 1.8 µm column. With a little retained solute, k = 2.3, the reduced plate height was h ≈ 4, and a long retained solute, k = 10.8, h ≈ 2.87. The latter is similar to Guillarmes [6-7] results with an unmodified SFC, with a extra-column variance of ≈ 80 µL2
using a 3 mm ID column.
Aspects of the injection volume and the sample solvent composition were found to be most important in determining efficiency. The fixed injection loop was replaced with a length of 50 µm ID tubing yielding a fixed injection volume of 0.4 µL. The sample solvent consisted of a 1:1 mixture of hexane/ IPA with various concentrations of ethanol added. Above 10% ethanol there was noticeable loss of efficiency. This superficially appears to be a ‘strong sample solvent’ effect although the sample isn’t really strong. The mobile phase was 8% methanol in CO2 which is probably not a much different solvent compared to 10% ethanol in hexane:IPA. The results all suggest that the columns were probably relatively poorly packed, but the characteristics of the injection problems were difficult to explain.
Armstrong
In 2015 while developing ultra-fast chiral SFC separations, Armstrong [11] stated: “
...column technology is ahead of existing chromatographic instruments”. He meant, of course, SFC instruments, as delivered by the manufacturers, were inadequate to see the full efficiency of columns packed with sub-2 µm particles. A Jasco semi-prep SFC was used with 4.6x50 mm column packed with 1.9 µm teicoplanin particles. The optimum flow on such a 4.6 mm ID column is ≈ 3.5-5 mL-min-1
, depending on modifier
concentration. In an effort to minimise extra- column effects, near optimum linear velocity, various ID tubes; 50, 75, 127, 254, and 508 µm, of PEEK or silica lined PEEK were used, with 11.5 cm in front of and 20 cm after the column. The column oven was bypassed (no thermal control of column temperature). Rather surprisingly, it was found that
254 µm tubing marginally produced the highest plate counts. These results are also non-intuitive. However, the differences in efficiency between 75, 120, and 254 µm tubing were minimal (≈ 2%), and all were poor, with hmin
≈ 5.85, k < 2. The smallest
tube ID ought to yield the lowest dispersion, but did not. The most reasonable conclusion was that this column was also not very efficient.
Berger
Many of the plumbing modifications made by Broeckhoven [8] were similar to those previously made by Berger [9] in 2010 when the first use of sub-2 µm particles in SFC was demonstrated. The chromatograph consisted of an Agilent 1260SL HPLC, with an Aurora SFC conversion module. A 10cm length of 125 µm tubing was used as the injection loop (≈ 1.25 µL). The sample solvent was methanol. All the tubing from the injection valve to the detector was 125 µm ID stainless steel tubing of the shortest possible length (≈70 cm total). However, the 2 heat exchangers with 175 µL tubing were retained. The photodiode array detector flow cell was the same 1.7 µL, 6 mm cell used by Broeckhoven [8]. A larger ID 3x100 mm column packed with 1.8 µm particles delivered a reduced plate height of ≈ 2.48 at k = 3.63, which was somewhat better than [8]. The same instrument was used to perform the first use of superficially porous 2.6 µm particles [10] in SFC. The Kinetex 2.6 µm column was 4.6x150 mm, and produced reduced plate heights as low as 1.62.
Later, Berger [12] modified a commercial Agilent 1260 Series I instrument (a more developed instrument compared to [9,10]) to achieve a reduced plate height of 2 with k = 2, using 3x100 mm columns packed with 1.8 µm particles. Many of the modifications were the same as used previously [8,9]. The injection loop was replaced with a 10 cm length of 125 µm tubing (≈ 1.25 µL). The sample solvent was methanol. The electronic filter was set to 80 Hz. All the 175 µm connector tubing was replaced with 125 µm versions, as before, but, in this case, the built-in heat exchangers (25 cm and 12.5 cm of 175 µm ID) were replaced with low dispersion HX’s (125 µm ID). The column was connected to the detector with 510 mm of 125 µm tubing (including the HX). The 13 µL detector flow cell was replaced with a 2 µL, 3 mm path length version. The measured variance (using w1/2) was ≈ 7 µL2
.
Achieving h ≈ 2 required very well packed columns. In fact, a fairly large number (8)
of 3x100 mm columns with 1.8 µm particles were tested before several columns were found to produce reduced plate heights of ≈ 2 at k =2. Without such highly efficient columns, there is always the question of what is real. Since 6 of 8 columns failed to produce reasonable efficiency, while the calculated plumbing dispersion showed it was feasible, the rest were poorly packed. Attempts to use 2.1x100 mm columns with 1.8 µm particles yielded poor efficiencies (hmin
= 2.85) even with k > 8 [12]
The same instrument and modifications were used to produce a hmin
= 1.93 with a 4.6x50
mm, 1.8 µm R,R Whelk-O chiral column [13]. This column had a theoretical column variance of ≈ 220 µL2
was required. , at k = 2. The plumbing
was adequate for use with 1.8 µm particles in such a column, since an extra-column dispersion of only < 44 µL2
On the other hand, h = 2.78 was obtained using a 3x50 mm, 1.6 µm column packed with an immobilised polysaccharide (IA-U) chiral column [14], using the same plumbing (column dispersion ≈ 36.8 µL2
). Since the
extra-column dispersion was ≈ 1/5th the column dispersion it probably indicates the column was relatively poorly packed or the chiral stationary phase was heterogeneous.
Gasparrini Gasparrini [15] modified a Waters UPC2 SFC
for use with 1.8 µm particles with another chiral selector (Whelko-O1). He replaced the 600 cm long, 175 µm ID tubes in front of and after the column with 250 mm, and 350 mm tubes respectively. The ID of these tubes was progressively decreased from 180, to 130, to 100 µm. This shortening of the tubing apparently involved the oven design which was replaced with an in-house built unit. Similarly, the autosampler with a 10 µL loop, was replaced with an external valve with a fixed 200 nL loop. This external, very small loop injector seemed to eliminate (or avoid) the injection problems of Broeckhoven [8]. He also replaced the 8 µL detector flow cell with a 3 µL cell. With all the modifications in place he obtained σ2
e-c ≈ 2.2 µL2 , producing
h = 1.88 on a 4.6x50 mm column with 1.8 µm R,R-Whelk-O1 particles. This is probably the lowest reduced plate height ever using a chiral stationary phase, let alone one with 1.8 µm particles.
Gasparrini paid particular attention to the onset of turbulent flow [16,17] and its beneficial effects on reduced extra-column dispersion. However, he did not recommend the use of 100 µm tubing since it performed no better than the 130 µm tubes, but
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68
