:dso evident in the graphs presented by Hill and coworkers (3) on the oven drying of fresh and aged superphosphate. Two further experiments showed t h a t tlie calcium sulfate in a fresh sample of single superphosphate was mostly not in the form of dihydrate and t h a t i t \vas resistant to hydration. I n the first experiment a 100-gram sample was washed with 1.5 liters of water and dried in a n air stream a t 60” C. The dried sample contained 3.2% phosphorus pentoxide and yielded 1.0% moisture and only 2.57, water of crystallization on distilling with benzene and xylene, thus showing t h a t the calcium sulfate in the superphosphate included little or no dihydrate. I n the
second experiment, 5 nil. water followed by 150 ml. of benzene were added to 20 grams of the original superphosphate and allowed to stand for 48 hours. On distillation all the water added plus the original moisture of the sample were recovered, showing the “inactivity” of the calcium sulfate. Additional interesting information on tlie distribution of TT ater in superphosphate is included in an article by Llarshall, Hendrick-, and Hill (6). ACKNOWLEDGMENT
The authors are indebted to E. R. Herman for his helpful criticism and to Gad Stiassny for his work on the theoretical asprcts of the problem.
LITERATURE CITED
Fetzer, \T. R., , ~ N A L . CHEV. 23, 1062-9 (1951). Gericke, S., “A4nalytischeChemie der Dunneniittel.” Ferdinand Enke. Stutrgart, Germany, 1919. Hill, W. L., Caro, J. H., Ihmagai, R., J . Assoc. Ofic. Agr. Chetnzsts 34,
641-53 (1951). Horsley, L. H., Advances ztz Chettz. Ser. No. 6 , 9, 11 (19523. Iiarabinos. J. V.. Bartels. G. E.. Ballun, ’ ii. T.,‘ Chemist d n a l y s i 43,37-8 (1954). Marshall, H. L., Hendricks, S. B., Hill, W. L., Znd. Eng. C‘hem. 32, 1631-6 (1940).
RECEIYED for review March 4, 1957. hccepted June 3, 1957.
Fluorometric Determination of Chlortetracycline D. H. FELDMANl, H. S. KELSEY, and J. C. CAVAGNOL* lederle laboratories Division, American Cyanamid Co., Pearl River, N. Y.
,The alkaline degradation of chlortetracycline to isochlortetracycline at pH 7.5 i s the basis of the fluorometric method for process samples, such as fermentation mash. Variables studied were the high sensitivity of fluorescence intensity to changes in pH, the choice of primary and spectrally matched secondary filters, the effect of organic solvents on fluorescence, fluorescence standards and conditions for storage, and analyst-to-analyst precision. Statistical control of precision applied to routine determinations of chlortetracycline in fermentation mash produced 95% confidence limits of
+ 2.2%.
C
(Aureomycin) degrades in alkaline media t o form several products ( 2 , 3, 7 , 8) depending upon the p H and the presence or absence of oxygen. A t p H 7 . 5 , and in the absence of dissolved air, the principal product is isochlortetracycline (6). I n Figure 1 are the ultraviolet absorption spectra of chlortetracycline degraded under the conditions of the fluorometric procedure and pure isochlortetracycline a t the same concentration. I n Figure 2 are shown the two fluorescence emission spectra. I n all essential respects the findings of Levine, Garlock, and Fischbach (5) have been confirmed b y this laboratory HLORTETRACTCLISE
Present address, General Tire Co., Ashtabula. Ohio. * Present, address, General Foods Research Center, Tarrytown, N. Y.
in adapting a fluorometric method t o routine use for the determination of chlortetracycline in process samples. The fluorescence is directly proportional to the concentration of chlortetracycline prior t o degradation. Furthermore, the rate of development of fluorescence depends upon the p H and temperature. From the 7 ien-point of blank stability and reasonable fluorescrnee intensity, the optimal p H is near 7 . 5 , and from the standpoint of time the rapid development of fluorescence a t 100” C. is desirable. Boiling also deaerates the reaction mixture. The fluorescent degradation products are stable foi several hours. Control of Precision. T h e determination was examined for sources of error while in routine use b y several analysts using several instruments. Good precision required attention t o the human factor as ne11 as the use of matched secondary filters. sperial mash pipets, and adequate p H control. For example, different analysts read the galvanometer differently, and this accounted for 7 to 307, of the number of variations. Data were collected on 180 batches of fermentation mash after a standard method for reading the galvanometer v a s put into use. Each batch was analyzed in duplicate by pairs of analysts on different n-ork shifts using different fluorophotometers. The averages of the pairs were compared and gave a n over-all mean difference of 2.5%, with 957c confidence limits of &6%, and 90% confidence limits of i57cvariation. Further data were collected on 143
batches. K l i t i i the variation bet\\ wii analysts exceeded the 907, confidence limits, their techniques were examined. Where errors were found, they were corrected. I n this series the mean difference fell to 1.9%, the 95% confidence limits to +5%, and the 90Vc confidelice limits to =t4.1%. The foregoing method has been applied t o each successive group of 100 hatches. The present mean difference is 1.2%, 95Yc confidence limits are &2.2%. and 90% confidence limits are i1.&% variation between analysts. Thus, statistical control of precision is effective in significantly decreasing the perqon-to-person variation. REAGENTS A N D APPARATUS
i l n aqueous chlortetracycline hydrochloride standard solution (50 y per ml.) was prepared from crystals of at least 99.573 purity, and was stored under refrigeration prior to use. -4 maximum 3-week storage provided a good safety factor, which was established b y using the quinine sulfate as a secondary standard and making up chlortetracycline standard solutions a t frequent intervals, rlqueous phosphate buffer. 0.34011f in potassium monohydrogen phosphate, and 0.0545M in potassium dihydrogen phosphate, was prepared m ith reagent grade salts. It yielded a p H of 7.50 rt 0.03 when 15 ml. was mixed with 2 ml. of 0.1N hydrochloric acid and diluted to 100 nil. Aqueous Versene [disodiuin (ethylenedinitrilo) tetraacetate], solution 57c \v./v., and 0.1N hydrochloric acid were prepared from reagent grade chemicals. Fluorophotometers (Pfaltz and Bauer, VOL. 29, NO. 11, NOVEMBER 1957
1697
Inc., N e v York, N. Y.) were fitted with Corning KO. 5840 primary glass filters, a metal reflection mask painted flat black placed between the incident beam and the cuvette, Corning Nos. 3389 and 5113 secondary filter combinations selected to have a maximum transmittancy at 438 nih and a halfpeak band width of 60 mN, and another metal reflection mask b e h e e n the cuvette and the barrier layer cell. T h e fluorophotonieter used a G. E. %-watt mercury vapor lamp as the light source, an iris-type diaphragm, rectangular cuvettes of about 18-nil. capacity, a transformer and voltage stabilizer hetween the 110-volt alternating current line and the lamp, a selenium-iron barrier layer cell as the photoreceptor, and a lorn resistance multiple mirror galvanometer of about 800 ohms. The latter had a sensitivity of 2 X 10-9 ampere per nim., and a 200-mm. ruled scale. The cuvette chamber, painted a flat black, mas fitted with a refill funnel (Figure 3) which facilitated rapid filling and emptying of the cuvette. The nipple of the funnel n a s connected to a vacuum line in series nith a 2-liter filter flask n-hich received the cuvette samples. The special pipets used for fermentation mash had an over-all length of 33 to 37 cm. and a delivery orifilce 2.0 =k 0.2 mm. in diameter. They were calibrated to contain 5.00 =t 0.02 ml. at 23' C. (room temperature) betiyeen the tip arid a full circle graduation. PREPARATION OF SAMPLES
Mash, Dilute fermentation mash t o a chlortetracycline concentration of 10 to 35 y per nil. as follom: For each 10 nil. of mash add 1 drop of 2-octanol (capryl alcohol) to aid in defoaming the sample, Mix well through inversion of the sample vessel n-ith a gentle rocking motion, and then free it of bubbles by tapping the vessel on a bench top. Remix by inversion just before pipetting. Discharge the mash ( 5 ml.) into an appropriate dilution flask, flush the pipet clean into the flask and dilute to volume with 0.1N hydrochloric acid. Other Samples. Dilute 5 ml. of t h e sample t o a chlortetracycline concentration of 10 to 70 y per ml. with 0.1N hydrochloric acid. Where oil is present, add enough methanol to the dilution flask to dissolve it, prior to dilution. When aqueous extracts contain high concentrations of calciuni ion, add 1 ml. of 5% Yersene solution to the treatment and blank flasks before niaking the dilution.
02
\
-
0 . 220
I
I
I
240
260
280
300
I
I
320
340
1698
ANALYTICAL CHEMISTRY
360
380
I~
I
400
420
\\ t 440
WAVE LENGTH (tip) Figure 1. Ultraviolet absorption spectra of chlortetracycline, alkaline-degraded chlortetracycline, and isochlortetracycline
Isochlortetracycline, 3.5 y per ml. in phosphate buffer us. phosphate buffer in 100-mm. cells _ _ _ - Chlortetracycline, same conditions . . . . . . Alkaline degraded chlortetracycline, same conditions 2 ml. of 0.1'1.' hydrochloric acid to the treatment flask before heating, and to the blank flask before dilution. Treat one aliquot by adding 15 ml. of phosphate buffer to the flask, and heating it at 100" C. for 6 minutes. Permit i t to cool to room temperature before diluting, to volume with distilled water. Flush the cuvette and then fill with this solution and observe the masimum galvanometer deflection. Fill about three quarters of the duplicate flask with distilled water before adding the 15 nil. of buffer and diluting to volume. Read this unheated blank within 2 minutes after adding the buffer. Calculate the concentration of chlortetracycline hydrochloride in micrograms per milliliter in the sample by dividing the net galvanometer reading b y 2 and multiplying the quotient b y the total dilution. Instrument Check. Prepare aqueous solutions of 10 to 90 y per ml. of chlortetracycline, in 10 y per ml. increments, and treat as above. Set the galvanometer to a net reading of 100.0 mm., using the treated 50 y per nil. solution.
PROCEDURE
Set the galvanometer to a net reading of 100.0 mm. b y adjusting the iris diaphragm and using the 50 y per nil. chlortetracycline hydrochloride standard solution treated and read as below. Pipet 2-ml. aliquots of the acid-diluted sample into duplicate 100-ml. volumetric flasks. When no preliminary acid dilution can be made on a sample, or when a standard is to be treated, add
I
DlSCUSSlON
Standards. A number of standards are available, from glass blocks t o quinine sulfate, which fluoresce in approximately t h e same spectral region as isochlortetracycline ( I ) . Levine (6)used quinine sulfate at concentrations of 0.5 y per mi. in 0.1N sulfuric
w
0
z 4
m
a
% m a
o.2
t
0 400
450
500
10
WAVE LENGTH (m Figure 2. Fluorescence emission spectra of isochlortetracycline, alkaline-degraded chlortetracycline, and quinine sulfate -- Isochlortetracycline, 10.5 y per ml. in phosphate buf-
fer in 10-mm. cell
_ _ _ _ Alkaline-degraded
chlortetracycline, same conditions , . . . . . guinine sulfate, 4 y per ml. in 0.1N H&O4 in 10-mm. cell
GRISOUATED FUNNEL
-
FILLING POSITION
-,-/
' 8
Figure
1
' CUVETTE CHAMBER PLATE
,
3. Exploded view of refill funnel
I
acid. I t s emission maximum is near 450 mp, while that of isochlortetracycline is near 425 mp (Figure 2). I n this laboratory two Pfalta and Bauer fluorophotometers were fitted with spectrally matched secondary filters and Cali-
I
I
brated with alkaline-degraded chlortetracycline. Readings were taken with both instruments on a quinine sulfate solution and a treated mash sample by a number of analysts over several days. The average of differences between in-
struments u-as 2.3% for quinine sulfate and 0.35% for the mash. This difference in averages is statistically significant and shows that alkaline-degraded chlortetracycline is superior to quinine sulfate as a standard. Solvents. T h e fluorescence of isochlortetracycline is enhanced in organic solvents, b u t t h e blanks in these solvents are also increased t o a point where no advantage over water is obtained through their use. Indeed, the solvents (50% aqueous) and the corresponding blanks expressed as percentages of gross readings were: water, 6.5; acetone, 6.3; 95% ethyl alcohol, 49; ethyl alcohol denatured with meth:mol, 15; and Cellosolve, 11. I n addition, the buffer reduces the solubility of the sample in these solvents. Storage of Standard Chlortetracycline Solutions. A study was made of t h e stability during storage of 30 y per ml. chlortetracycline hydrorhloride solutions in water and in 0.OlN and 0.1N hydrochloric acid over n 3-week period. Half of the solutions in each solvent Tvere refrigerated, and half m-ere frozen. The 35 samples dudied in each of the six combinations demonstrated that aqueous refrigerated solutions gave the smallest range of duplicates and smallest over-all range of galranometer readings. Instrument. T h e Pfaltz and Bauer fluorophotometer yielded a linear relation between concentration and galranometer reading, n i t h zero intercept. Results were reproducible and t h e instrument n as stable when used with a voltage stabilizer. Corning glass filter 5840 was selected to isolate the 365-nip mercury line. Filter 5860 had a much Ion-er transmittancy, thereby reducing the intensity of the incident beam. T o compensate for the lo~vintensity, very large diaphragm openings were used. A s the lamp efficiency fell with use, maximum diaphragm openings soon proved inadequate. One method for increasing the intensity of the lamp IT-asto boost the 1-oltage through a T-ariac placpd betn-een the line and the transformer. Reflections from the cuvette chamber, or from the liquid surface, can provide troublesome errors and high blanks. To eliniinate these the chamber was painted a flat black and two metal surface reflection masks were used. Distilled water gave galvanometer deflections of 0.5 mm. Tvhen these precautions were taken, but n ithout them readings of 6 to 30 mm. were obtained. The problems of high and inconsistent blanks for mash samples, and of large instrument-to-instrument variation, were solved by selecting the proper combination of secondary filters, and matching sets of these filters for use in several instruments. Kelsey and Goldman (4)have reported the useof thiamine VOL. 29, NO. 1 1 , NOVEMBER 1957
1699
assay secondary filters (Corning 3389 and 4308) for the fluorometric determination of chortetracycline. This combination of filters has a transmittance maximum a t about 405 mp, and transmits doivn to 360 mp. They permit the transmittance of scattered incident light, and of any fluorescent light emitted by impurities down to 360 nip, Another combination of Corning filters, Kos. 3359 and 5113, has a transmittance maximum near the emission maximum of isochlortetracycline and transmits down t o about 405 nip. Comparison of the blanks of mash samples obtained using the two filter combinations (Table I) reveals a marked reduction of blanks v i t h the use of So.. 3389 and 51 13. Table I.
Blanks of Mash with Two Filter Combinations
Range of Blank Reading!, hlm., Using Samplc
I3ilutio11 1 :250 1 : 500 1:1,000
1:2,500
1 : 5,000 1 : 10,000
1OS. _-
+
3389 4308 48-60 36-50 26-40 16-22 12-20 10-14
+
3389 5113
10-30
10-14 5-10 5-10
5-6 1-4
Cary recording spectrophotometer curves of the absorption spectra of different sets of Kos. 3389 and 5113 filters showed the existence of differences in the wave length of the minimum, and of the half-peak band u-idths. These
sets gave different fluorescence readings with the same samples. Curves taken on permutations of the combinations revealed that S o . 3389 determined the shape of the curve and the wave length of the minimuni for the combination. The absorbance a t the minimum is determined by No. 5113. Pairs of filters uere then matched to have minimum absorbance or maximum transmittance a t 435 mp, and a half-peak band width of 60 mp. Vatched pairs on different instruments gave identical fluorescence readings for samples, and made the blank readings consistent n ith dilution, and reproducible. Control of pH. T h e variation of the fluorescence intensity of alkalinedegraded chlortetracycline with pH is shown in Figure 4. A change of 0.1 pH unit near 7.5 will produce a 7 to 8% relative change in fluorescence intensity. This vas confirmed in the analysis of mash samples which were diluted n-ith both water and O.1N hydrochloric acid. The buffers in mach did not affect the pH when qamples were diluted about 100-fold with O . l A r hydrochloric acid. Blanks were read within 2 minutes after addition of buffer becauqc of the slow conversion of chlortetracycline to ieochlortetracyclinp a t pH 7.5 and room temperature. There is no evidence that chlortetracycline fluoresces a t this pH. Where calcium ion has been concentrated in some aqueous extract samples, i t may precipitate out during buffer treatmmt as the phosphate and change the pH of the reaction mixture. T o prevent this precipitation and the resulting
errors due to light scattering, the calcium was sequestered with Versene. ACKNOWLEDGMENT
Credit must be given to J. Norman Laing for initially adapting the Levine method to the Pfaltz and Bauer fluorophotometer. The authors wish to thank Frank Wilcoxon and Charles 137. Dunnett for their help in planning the statistically designed experiments and interpreting the results. LITERATURE CITED
( 1j Boltz, D. F., ed., “Selected Topics in
Modern Instrumental Analysis,“
p. 87, Prentice-Hall, Sew York,
1952. (2) Hutchings, B. L., Waller, C. W.. Broschard. R. W..Wolf. C. F.. Fryth, P. ’W., Wilfiams, J. H., J,’ Chem. Ana. Soc. 74,4980 (1952). Hutchings, B. L., \Taller, C. W., Gordon, S.,Broschard, R. W., Wolf, C. F., Goldman, A. A , , Williams, J. H., I b i d . , 74,3710 (1952j. Kelsey, H. S.,Goldman, L., J . Clin. Invest. 28, 1048 (1949). Levine, J., Garlock, E. A., Jr., Fischbach, H., J . Bm. Pharm. Assoc.. Sei. Ed. 38, 473 (1949). Pruess, L. M., Demos, C. H., “Encyclopedia of Chemical Technology,” T’ol. SIII, p. 776, Int,erscience, S e n . I-ork, 1951. Waller, C. W.,Hut,chings, B. I,., Wolf, C. F., Broschard, R. IT., Goldman, A. A , , Williams, J. H., J . Am. Chem. Soc. 74, 49% (1952 1. Kaller, C. IT.,Hutchings, B. L., Wolf, C. F., Goldman, A. -4.,Broschard, R. W.,Williams, J. H., Ibid., 74, 4981 (1952). RECEIVEDfor review June 26, 1956. hccepted June 26, 1957.
Squalane: A Standard KARL J. SAX and FRED
H. STROSS
Shell Development Co., Emeryville, Calif.
b Squalane has been prepared by the hydrogenation of highly purified squalene. The properties of the product were determined and a synthetic squalane sample was prepared for comparison. Squalane was found useful as a standard for carbonhydrogen, molecular weight, and viscosity determinations.
M
ASSLYTICAL PROCEDURES require standard substances of reproducible purity for development of the method and calibration of the instruments. Such generally used pure hydrocarbons as triphenylmethane and n-hexadecane are solids or .melt just below room temperature; no convenient
1700
AST
ANALYTICAL CHEMISTRY
standards with more than 20 carbon atoms are available. During a n investigation of liquid standards of moderately high molecular weight, highly purified squalane (2, 6, 10, 15, 19, 23hexamethyltetracosane) was prepared by the hydrogenation of squalene. A synthetic squalane sample was prepared for comparison. I t s branched structure and the number of diastereomers make squalane one of the few readily available, stable, liquid hydrocarbons in its molecular weight range. The proportions of isomers formed during the hydrogenation of squalene under standard conditions should not vary greatly. No significant variation in properties was found. Tests show that squalane is a useful
and often superior standardizing and calibrating substance in such widely different applications as the determination of molecular weight, of viscosity, of refractive index, of carbon and hydrogen b y combustion, and as a stationary liquid in gas-liquid partition chromatography (4). This work suggests the use of squalane in the latter and other applications. EXPERIMENTAL
Purification. A 100-ml. quantity of squalene (Distillation Products, Inc.) was dissolved in 140 ml. of acetone, T h e solution was cooled in a dry ice bath and seeded with crystalline squalene. T h e product was collected in a chilled funnel,
