504
JOURNAL
PREPARATIVE-SCALE V. EXPERIMENTS
M. VERZELE
WITH
February
CHROMATOGRAPHY
VERY
LONG
COLUMNS
M. VERSTAPPE
of Organic Cltcmisiry, State
Laboratory (Reccivcd
AND
GAS
OF CHROMATOGRAPHY
gth,
Univ.wsity
of Ghcnt
(Belgizcm)
1965)
In earlier communications 1-4 of this series we investigated the factors that lead the optimum conditions for preparative-scale gas chromatography. The most important point was that for increased sample size, in narrow-bore columns, very coarse support materials gave nearly the same separating power as the more normal support mesh sizes. Coarse supports, however, permit the use of very long narrow-bore columns. Better results still can be expected with loosely ,packed ‘supports with a relatively low percent liquid phase loading and with the technique of “programmed In this paper we intend to discuss further developments of chromathermography”. the technique of preparative-scale gas chromatography. to
DEVIATIONS
FROM
NORMAL
CNROMATOGRAPRIC
BEWAVIOUR
Recently, we3 found that the second peak of binary mixtures showed a considerable increase in peak front appearance time when the sample size was increased. This is most probably caused by temperature effects. Evaporation at the rear portion of the first band cools the column locally and this retards the second substance. High partition factors should enhance these temperature effects and this is indeed so. This effect is general and also occurs on analytical size columns with larger samples. We tried to exploit this phenomenon to increase the capacity (sample size) in preparativescale gas chromatography by using thick-walled glass columns. The idea was that heat transfer would be slowed down by the thick glass wall and that the increase of appearance time for second peaks would be increased by this. This effect does occur in these thick-walled columns but is not much greater than in normal metal or glass columns. With regard to peak front retention changes with increase in sample size, we have already shown the influence of gas rate and of column resistance to gas flow (column length and grain size of support)3. Changes in function of the partition factor ‘$7.. .!*;, are now given in Table I. For low partition factors peak front retention decreaies with increasing sample size and for high partition factors it increases. Intermediate values showing no change are possible (toluene in Table I). Summarising, it can be said that peak front retention with increasing sample size is influenced by several counteracting factors: (I) It decreases because of exponential sample introduction. This is most obvious for substances with a small partition factor (higher temperatures), low gas flow rates on short columns with small pressure drop (coarse support). J.
Chvomatog.,
19
(x965)
504-51
I
PREPARATIVE-SCALE
GAS CHROMATOGRAPHY.
V.
505
I
TABLE CXANGES
IN THE
FUNCTLON
OF THE
PARTITION
FACTOR
WITH
SAMPLE
SYZE
m x g mm coiled glass column. Self made support3 of x5-zo mesh coated for 1/aof thti column length with 25 yO SE.30 and for 3/*with ~0% SE.30. Gas flow rate 200 ml H,/min; temp. 165’. 20
Samjde
size
(I4 50
100 200
5oo 1000 2000
Peak ,front retention in mm for
Benzene
Xoluene
Ethylbenzene
Cumene
62 62 61
89’ 89 9=
z; 53
:; 88
=27 129 135 =35 136 140
165 168 176 179 181 190
It imreases because of temperature effects in the column and because of increased resistance to gas flow rate with large samples. This is most strongly felt for substances with a high partition factor (lower temperatures), high gas flow rates and in longer columns with high pressure drop (small grained support). (2)
LIFE
EXPECTANCY
OF LONG
NARROW-BORE
COLUMNS !
Using longer narrow-bore columns with increased sample size, we come to’ a point where several ml of .liquid are injected in a column having only an 8 to, g mm opening. It could be that under these conditions the stationary phase is stripped rather quickly from the tist part of the column. This is the more likely when. “on column” injection is ‘used. In fact, on the longest column ovens discussed below we have not even movided an iniector heater. Several designs for heated injectors;,large and small, filled with steel turnings etc. were tried, but these did not -improve the results, rather the contrary compared to “on column” injection. The possibility of stripping was investigated by injecting one hundred times.5 ml of a mixture containing benzene, toluene, ethylbenzene ‘and isopropylbenzene (cumene) into a 20 m x g mm coiled glass column filled with. the coarse white support medium which we made ourselves and which is described in ref. 3. With this sample size the- column was overloaded and only a partial separation was achieved. This support did not have a very great capacity for the stationary phase, 25 oh loading with SE. 30 was the highest limit and this gave a sticky column material. The column was filled for 1/4,of its length with a, loading of 25 %‘SE. 30 and for 3/4 with a loading of X-O.%SE; 30;“details and reasons for this special way of filling the column will be published later. After the 100 injections the column was emptied in small portions and the amount of stationary phase in each portion .was determined by extraction and weighing; It was found that stripping was negligible’and that the stationary,phase distribution ‘had not changed. This could be expected since chromatographic results before and- after the, injections were practically the same. The separating’ power. of the column for small samples dropped’ slightly after a ,few 5 ml injections but remained afterwards remarkably w constant as can be seen in Table II/For large samples the “preparative-scale’ plate “.[b number” .was practically unchanged, taking into account the bad reproducibility of plate number determinations. !.
J.
Clwomatog., xg (1965) 504-51 I
506
M. VERZELE,
TABLE
II
ANALYTICAL AND PREPARATIVE-SyCALE AS A FUNCTION OF COLUMN AGE
Sam&& size
50
M. VERSTAPFE
Number of 5 ml i~jeci?ions at zG5’
1st
Pl
After After After
30 60 100
I ml After After After
Y& 30 60 IOO
* l?or explanation
Plalc
PLATE
NUMBERS
FOR
A 20 Ill
X
Q IllIll
ToEuene
ElJtylbenzcne
Cumene
952 792 796 680
1180
1471 1098
1530 1154
x220
1368
1098
112G
64 63
164 149 =42 172
70
GLASS
COLUMN
vmn~bers
Bcnrene
62
COILED
90.5 1032 841
346 279 309 317
57’5 443 468 491
see text.
It must of course be remembered that this experiment is for SE. 30 and that a more soluble stationary phase may give different results. : APPARATUS
CONSTRUCTION
The extra long columns discussed in this paper were housed in a long tubular ovenprovided at one end with a heater and a fan. A blower is used to push air over the heater and fan into the oven. By regulating the blower output it is possible to obtain’ either a chtomathermographic gradient or practically isothermal heating of the oven. “Programmed chromathermography” is obtained simply by switching on the heating current. This last temperature evolution in the column has a beneficial effect on effluent concentration and therefore on recovery of separated substances3. The details of the construction of an oven housing for a 75 m ‘x g mm coiled glass column are given below. The oven is made of thin sheet steel and has a double mantle to isolate the interior. No insulation material is provided and in general the heat capacity of the apparatus is kept as low as possible, Air can be blown between the mantles in order to promote rapid cooling. At one end of the oven there is an injection port. The heater of this injection port is of the analytical type used in Wilkens Aerograph Instruments and is only provided for work with very small samples. The other end of the oven is coupled to a conventional katharometer detector and further connection of column and’ detector to the collector table and electrical circuits of an Autoprep 700 (Wilkens Instruments) is arranged. This is shown in Pig. 1. ‘In order to make the glass column a I’m x 30 cm cylinder of 2 mm sheet steel was,mounted so that it could be rotated very slowly by means of a reducing gear box. Pyrex glass tubes of 1.5 x 8-g mm internal diameter and,zo-11 mm outer diameter were coiled on the cylinder and welded to each other until the cylinder was completely ‘covered with the glass coil. This gave a 75.m x g mm column. Thecolumn was conditioned against stress by annealing overnight in an oven at 500’. The dimensions of this annealing oven in fact determined the size of our instrument. We think’that much longer columns could be used with advantage, J. Clwomalog,,
19 (19%) 504-S~I
PREPARATIVE-SCALE
GAS
CHROMATOGRAPHY.
V.
so7
Fig. Apparatus for preparative-scale gas chromatography on long columns (not to scale). A = Wilkens Autoprep 700; I3 = collector table; C = detector oven heater: D = katharometcr detector oven ; E = double mantle column oven ; F I: coiled column (actually the coils are as close as possible to one other) ; G = column oven Calrod heater: H = detachable heater fan motor block ; I = fan motor; I< = inlets for air under pressure: L = fan: M = injector.
The oven dimensions are I m x 33 cm. The heaters are of the “Calrod” type and have 2400 W capacity. The blower is of the centrifugal type and must be sufficiently powerful to heat the oven uniformly. A temperature of r50° .is,.reached in rg min and the maximum temperature, which is around 275”, is reached in less than an hour. Cooling is much quicker. The long glass columns are not sufficiently strong to be self-supporting and have to be supported by a frame. The longitudinal rods of the frame supporting the glass coils are provided with a sufficient number of hooks to immobilise the column completely. Filling the column with coarse support medium is a question of patience and is carried out, with the column in an upright position, with the aid of suction by a water-pump and a vibrator (applied to the frame). Because of the care which has to be taken not to break the column the filling operation lasts several hours. The same column has been filled and emptied three times without damage so far.. The 20 m x g mm coiled glass columns filled with Chromosorb W 30/60 mesh used in earlier work3 and also mentioned in this paper, were built and mounted in the same way. The disadvantages of glass columns are that they break easily and are more difficult to make. We .Nanted, however, to use glass at all costs because of the lability of some substances under investigation in this laboratory. We have also tried metal columns. To fill these, the columns were held out of a window of a 60 m high building and the filling was simply poured in; the filled column was then lowered down with a rope and coiled on a metal cylinder. Copper has to be avoided, but stainless steel is for most purposes as good as glass. Several 25 or 50 m sections can be joined together with swagelok fittings. Columns prepared in this way can be emptied and refilled again without uncoiling. For some time we have connected glass columns to the instruments with teflon swagelok fittings. In our experience this is, however, not better than connecting the columns with silicone rubber gaskets which also allow work at and with the latter column breakage is not so’ frequent as with the teflon #US 250°, IL1fittings. The pressure needed to obtain leak-proof connections with teflon is also rather high while this is not the case with silicone rubber gaskets. J. Chvomatog.,19 (1965) 504-51 z
50s
M. VERZELLE,
EXI’ERLMENTS’WLTH
M. VERSTAPPE
75 Zll X 9 l3lm COLLEl3 GLASS COLUMNS
One of the problems with regard to the necessarily coarse support material for filling the longer columns was that these are (or were) not commercially available. We made our own support as stated above, but this left much to be desired. Coarse grain supports which could most easily be produced in quantity are of the Chromosorb P type. This type of support, however, suffers from the drawback that many substances are destroyed on it. The Chromosorb W type supports, which do not show these properties, cannot be processed in coarse grain sizes. Johns Manville has now produced a new type of support with a pink colour like Chromosorb P, but which cauces less decomposition and which can be obtained in coarse grain sizes. We could successfully. chromatograph on this support humulene, a- and p-pinene and several dioxolan mixtures, which shows that it is considerably more inert than Chromosorb P. This support (Johns Manville No. 5641097) was available in mesh sizes IO--20, 20-30 and 30-40. Comparative experiments with these supports were carried out on the 75 m x g mm coiled glass column described in the previous section of this paper. The statiomary phase was LO yOSE. 30. The time taken to fill the column was about the same for each of the three supports (about 4 11); emptying took only about 1.5 to 2 h. The largest inlet. pressure which’ we accepted in this glass column was 3 kg/cm2 and with this pressure, hydrogen gas flow through the 20-30 and 30-40 mesh support columns was rather similar and around 250 and 150 ml/min respectively. This was not as high as we had hoped for. The retention time for the air peak was about 20 to 30 min and the analyses were very time-consuming. With the 10-20 mesh size supports results were much better. Hydrogen gas flows as high as 200 ml/min could be obtained with only L kg/cm2. The retention time’for the air peak was only 15 min. Comparative results of the plate numbers for these columns are found in Table III. TABLE PLATE
III NUMBERS
The temperature
Sampte
size of isopvopylbename
75 m x gmm COLUMN was 140~ and the partition
FOR
Support mesh
factor was about 5,
size
zoo ml Halmin __.. 30-40 20-30
500 ml N&ha IO-20
IO-20
5024 2304 560 229 I57
4736 2048
,-_-.--_ 25
250 1.25 2.5 5
Pl
Pl ml ml ml
r2416 317G 496 240 x7=
7X20 1944 416 256 =7o
::: 1x5 ‘,‘.
Absolute! 3: identical conditions cannot be realised. While the ‘flow rate was about the same -for tha first three columns of figures in Table III, the pressure drop was much larger fclr the first two columns (about 3, kg/cm2 inlet pressure) than for the 10-20 mesh support of the third column (1.2 kg/cm2 inlet pressure). Deviations in Table XII from 01e cxlxcted values must be explained by this. For identical inlet J. Chromatog., 19 (~965)‘~&4--~~1
PREPARATIVE-SCALE
so9
GAS CWROMAT0GRAPHV.V.
the gas flow rate was much larger for the IO--20 mesh support (fourth column). Table III again shows that the ratio of plate num’bers for fine grain veyszcs coarse grain supports is higher for analytical samples than for .preparative samples. Ten ml of a mixture of equal volumes of berizene, toluene, ethylbenzene. and isopropylbenzene is easily separated on such columns. This is shown in Fig., 2.
p~~~~res
Fig. 2. Separation of IO ml benzene, toluene, coiled glass column filled with experimental qoo ml H,/min at 150~. Duration 70 min.
ethylbenzene, chromosorb
isopropylbenzene on a 75 m x g mm mesh, coated with IO o/O SE.30.
10-20
With decalin at about zgo* and with zoo ml H,/min as gas flow rate, the largest sample giving practically complete separation is 7.5 ml on the 30-40 mesh column and 6 ml on the 20-30 mesh column. On the IO-zo mesh column it is only 4 ml. Nevertheless the IO--20 mesh support is the one to be recommended for these very long columns. This support yields nearly the same separation as the other supports but has the advantage of being much faster.. With 3 kg inlet pressure (gas Aow rate 400-500 ml H,/min) and at Igo* it separates 4 ml decalin mixture in 45 minutes. Automatically repeated separation on a 6 m x g mm column of 4 ml decalin would take much longer (about 3 h). For shorter columns below 30 m length, the 20-30 mesh, would seem preferable. High gas flow rates are possible and the capacity is somewhat better than for the IO--20 mesh material. DISCUSSION
The advantages of very long narrow-bore columns are mainly the following: (a) The sample size can be increased proportionally to the column length. For example if IOO ~1 can be separated on the 6 m x 9 mm column of the Autoprep 700 instrument, a column of 60 m x g’ mm will separate. I ml of the mixture3. (b) Separations which are not possible on a preparative scale with normal length columns can be carried out on the longer columns with reasonable samples. (c) A long narrow column has high versatility, many types of mixtures can be separated on the same column. This avoids frequent column exchange. (d) .Long narrow columns compared to other preparative-scale gas chromatographic techniques are economic to run with regard to gas consumption and to stationu1 ary phase and support material needed. (e) The size of the instruments considering their possibilities remains reasonable. 2) .’ (f) Because ‘of the incretised sample size, recovery is better than on shorter . columns. J. Clwomatog.;.I9(w6s)',so1t_5r.x
M. VERZELE,
510
M. VERSTAPPE
The main disadvantage of long narrow-bore columns is the increase in time needed for a separation. At the same temperature, with an identical gas flow rate and a moderate pressure drop, a ten fold increase of column length will lead to about a ten fold increase of retention time. This disadvantage is minimised by using a very coarse grain support (10-20 mesh), a relatively low stationary phase percentage (10 %), a loose packing, high gas flow rates and high temperatures, as explained before. In many cases the longer columns are in fact faster than the shorter columns (see ref. 3). SEPARATIONS
ON LONG
NARROW
COLUMNS
Separations actually carried out on the 20 m x 9 mm coiled glass column filled with Chromosorb W 30/60 mesh coated with 25 o/OSE. 30 are presented in Table IV. Separations on the 75 m x g mm column with more polar substances than the hydrocarbons already mentioned were also successful with larger samples. With analytical samples, ketones and esters showed strong tailing and poor separation. This was even more pronounced with alcohols and with the enones of Table IV. TABLE
IV
SEPARATION
OF VARlOUS
MIXTURES
ON
A 20 Xl-i X
9 IllI’ll COILED
GLASS
Mixi!ure
Xem$wralawe
trans-Dccalin ois-Dccalin
500 to xgo”
e-4-Mothoxy-Wt.-butylcyclohexane a-4-Methoxy-ten!.-butylcyclohexano
205~
750 c11
2100
500 1.11
4-Isopropylcyclohex-2-enone 4-Ethyl-4-mcthylcyclohex-2-enone
180~
150 Pl
&s-Octahydro-5-oxohydrindan Irans-Octahydro-5-oxohydrindsn
IGOO
4.00 lul
4-Isopropylcyclohex-2-enone 4-Isopropylcyclohex+enone
180~
4oo Pl
2,4,5-Trimethyl-2-phenyldioxolan (the three stereoisomers)
Sample
COLUMN
size
2 ml
Paradoxically, the analytical separation of these substances was impossible while the separation of ml samples proceeded satisfactorily. A remaining difficulty was that some delicate substances are destroyed on the support. ,This strong support activity may be largely due to the low percentage of stationary phase. With 20 oh Apiezon L, ,however, the situation was not much improved. The support was of the non-acid washed (NAW) type. According to Bulletin FF-114 of Johns Manville the surface activity of these supports can be reduced considerably by acid washing and by treatment with dimethyldichlorosilane. Experiments with such AW-DMCS supports are in progress. J.
C~JVOttZtZiOg.,
19 (1965)
504-511
FREFARATIVE-SCALE
GAS
CHROMATOGRAPHY.
5x1
V.
ACKNOWLEDGEMENTS I
We wish to thank the “Ponds voor bet Collectief Fundamenteel Wetenschappelijk OncE’erz.oek”~;for financial aid to the laboratory and the Johns Manville Corporation for aid with the support material problem. SUMMARY
Practical preparative-scale, gas chromatography on long narrow-bore columns (75 m x g mm) on coarse support material (IO--20 mesh) seems very promising. The main ,advantage may turn out to be high versatility. Large samples (in the ml range) of hydrocarbons, aromatics, ketones, esters and alcohols could be,‘separated on the same column with IO 0/0SE. 30 as stationary phase. This is even the case for a-values as low as 1.2 .to 1.1. Details of the construction and experimental handling of such apparatus are described. The main difficulty ,for the generalisation of this approach lies in the support material. Chromosorb ,W type supports are .preferable but cannot be obtained in a sufficiently coarse mesh size to make really long columns, Chromosorb P type supports and the new experimental Chromosorb supports described in the paper show support interactions which are more conspicuous in the long columns mentioned than on analytical columns. REFERENCES
I M. VERZELE, 2 M. VERZELE, 3 M.VER~ELE, 4 M. VICRZELE,
J. Cltromatog.,
J. Cltromatog.,
13 (1964) 15 (1964)
377. 482.
J, BOUCSE, A. DE BRUYNE AND M.~ERSTAPPE, J. Clzromatog., 18 (1965) PYOC. Ckromatog. Symfi. Bclg. SOG. Phavm. Sci., BYwsseZs, Sept. rg64. J. Cltromatog.,
19 (1965)
2.53.
504-511