Reproducibility of Rf and Correlation of Chromatographic Patterns on Paper and Thin Layer Plates T. F. BRODASKY Research laboratories, The Upjohn Co., Kalamazoo, Mich.
b In considering the use of thin layer chromatography in place of, or in conjunction with, conventional paper chromatography for the identification of antibiotics, it has been necessary to compare the techniques with respect to reproducibility of R, and correlation of chromatographic patterns. The present investigation undertakes such a comparison using silica gel and cellulose. Several antibiotic substances were chromatographed on all three matrices using the same mobile phases. An analysis of variance proved paper to b e the most reliable chromatographic system and only limited correlation of paper and thin layer chromatographic patterns was found to exist. The greatest precision in reproducing patterns was achieved on paper and cellulose. Recognition of patterns on cellulose and silica gel, compared with paper, both visually and by a sorting technique using IBM cards is discussed.
ducible R, values were on routinely prepared thin layer plates and what, if any, correlation existed between chromatographic patterns obtained with six standard mobile phases on paper and thin layer plates. These patterns, coupled with biological data, are the principal source of information used in the comparison of new antibiotics with known substances. The enormity of the task of reproducing all these patterns on even one thin layer matrix makes a study of pattern correlation necessary. Three randomly selected antibiotics were used in the R f study. In addition to these, five others were investigated with respect to correlation. The “pattern index,” a numerical description of Rj patterns employed in the correlation study, is described. The use of an automated technique for comparing patterns of the antibiotics on the different chromatographic matrices is also discussed. EXPERIMENTAL
T
since its introduction, has become a powerful adjunct to the other chromatographic methods employed in the analytical laboratory. The applications have been varied (3, 6, 6, I $ ) , and seem limited only by the imagination of the investigator. The literature is growing rapidly in this field, but although the papers concerning technique and application are numerous, there is a paucity of information on the statistical aspects of thin layer chromatographyLe., the reproducibility of Rj values (4, 7 , 9). Of secondary interest is the possibility of visual correlation of patterns on paper and the thin layer matrices. This laboratory is engaged in the identification of antibiotics based on a file of thousands of papergrams representing several hundred authentic compounds. Thin layer chromatography is playing an increasingly important role in the solution of problems which, a t best, are solved only with great difficulty through paper chromatography. The present study was undertaken to determine how repro996
HIN LAYER CHROMATOGRAPHY,
ANALYTICAL CHEMISTRY
Chromatography. Degreased glass plates (10 x 20 em.) mere coated with cellulose (Excorna, Pharm. Praparate 0. HG., Mainz) or silica gel (E. Merck AG., Darmstadt Ger.) prepared according to standard procedures. The coatings were applied with a Camag apparatus (Arthur H. Thomas Co.) set for a film thickness of 3 x 10-2 em. The plates were airdried and, without further treatment, stored in suitable containers. The paper used in this study was ST’hatman No. 1 chromatographic paper cut into Fheets 19 X 54 em. All chromatography was carried out using the following mobile phases: I, 1-butano1:water: :84:16 (v./v.); 11, I-butano1:water: :84:16 (v./v.) plus 0.2501, p-toluenesulfonic acid (w./v.) ; II1,l-butano1:acetic acid:water: :2:1:1 (v./v./v.); IV, 1-butano1:water: :84: 16 (v./v.) plus 0.2% (v.) piperidine; V, 1-butano1:water: :4:96 (v./v.); VI, 1-butano1:water: :4:96 (v./v.) plus 0.25’7, p-toluenesulfonic acid (tv.,’~.). Systems I through IT- required 16 hours for full development on paper, FT-hile 5’ and VI required 5 hours. The thin lager plates were developed until the solvent front had traveled about 15 cm. Thiq did not require more than 2 hours for any system. Descending
paper chromatography was carried out in developing tanks 61 X 30 em., whereas thin layer plates were developed ascending in tanks 31 X 14 em. No wicks were used in any developing chambers. All chromatography was carried out a t 25’ 2’ C. The antibiotics used in the Rf study (actithiazic acid, chloramphenicol, and chartreusin) were chromatographed a t 10 to 20 pg. on a single sheet. Six sheets were prepared to cover the six mobile phases. Three such sets were developed each day for 3 days to form a 3 X 3 matrix of values. The thin layer matrices were handled in a similar manner. The remaining antibiotics used for the correlation study were chromatographed only once. Detection. Several methods of detection were used depending on the chromatographic matrix involved. Actithiazic acid was detected on all systems with periodate-permanganate reagent (IO), while chartreusin was located by its blue-white fluorescence under 366-mp light. Chloramphenicol was detected on paper and cellulose thin layer plates by its adsorption of ultration light. 011silica gel it was observed as a yellow zone after treatment with periodate-permanganate reagent. On paper, the additional antibiotics used for the correlation study were detected by bioautography on the following organisms : hygromycin, Klebsiella pneitmoniae; nucleocidin, Proteus vulgaris; grisein, Bacillus subtilis; bamicetin, -1Iycobacterium tubercukxis var. avium; cycloserine, Escherichia coli. On silica gel they were all detected with permanganate-periodate reagent, while on cellulme the following methods were used: hygromycin, fluorescence; nqcleocidin, ultraviolet absorption; grisein, fluorescence; barnicetin, fluorescence; cycloserine, periodate-permanganate reagent. RESULTS A N D DISCUSSION
R j Reproducibility. The relative standard deviations of R, values encountered when chromatographin,0. certain antibiotics on paper and thin layer chromatoplates appear in Table I. The values showing a significant statistical day-to-day variation in addition to within-the-day experimental variance are so designated; the other values contain no day to day variation and,
therefore, reflect the experiment to experiment variance, Since each antibiotic was run three times a day for three days, the R, values obtained for each substance form a set (8) (Figure 1). In each set, the row and column designations represent experiments and days, respectively. Each R, is assumed to be composed of three components, an overall average, a day-to-day component, and an ex 2eriment-to-experiment component. The R, values in Table I are an estimate of the overall average. The analysis of variance was made according to the hierarchal design of a single variable c1:tssification. The standard deviation of a single observation is then given by
s = l/a2
+
kl
k2
k3
x11
x21
x3 1
x12
x22
x32
x13
x23
x33
T1+
T2+
T3+
T,
-
-
-
= days
-X
= f l + t T2+ = T++/N
The individual U'S are estimated using the following equaticns: mean square for days
x3
ki
Where
, T ue2
-x2
X1
(Grand Total)
X
(Grand Mean)
t T3+
N = Total number o f values in the set. ni = Number o f experiments in day i.
=
Figure 1.
x: Ti+-' T++2 "."=E,-
Rf
value matrix table
