is not n.nahed to avoid the possibility of peptizing some of it. Iron and aluminum must be quantitatively removed because they give positive interference b y precipitating with the 8quinolinol. The amount of original solution lost in the double precipitation is less than 0.017,. The second precipitation can be carried out a t p H 5.5 without adding sodiuni acetate if the p H adjustment to 3.2 for the calcium phosphate is carefully made in the first precipitation so that no calcium phosphate is coprecipitated with the iron and aluminum phosphates. Approximately 20 times the stoichioiiictric amount of 8-quinolinol is employed; therefore, the volume of this rcagent is changed proportionately nith the expected quantity of magnesium in the sample. Probably the most critical step in the entirr analysis is the pH adjustment following the addition of the 8-quinolinol rcagent. The pH should he checked and readjucted, if necessary, when the teniperaturt of the solution has dropped to aboiit 40" C. If an adjustment is made, stirring is continued for an additional 15 minutes. Thirty-minute stirring time is approximately the minimum required for the precipitate to form. The speed of the magnetic stirrer should be adjusted to form a cone inch deep in the solution. The precipitate of calcium oxalate and magnesium 8-quinolinolate must be w a s h d completely free of excess 8quinolinol because the final titration measures onlv 8-qiiinolinol. I n establishing a given analysis, the completen c v of this nashing step should be chtcked b y acidifying the final wash and
carrying out the titration as described for the sample itself. The final solution is deaerated with nitrogen to remove dissolved oxygen, which causes a measurable error a t the trace magnesium levels under investigation. Bromine is immediately liberated when bromate is first added to the titration mixture, and a n immediate deflection of the galvanometer needle is noted. -4s the bromine is used by the 8-quinolinol, the needle gradually returns to zero. Thus, bromate must be added until the deflection of the galvanoincter needle remains constant for a couple of minutes. Only then can the reaction between 8-quinolinol and bromine be considered complete. The electrodes are conditioned by sonking them in cleaning solution for j minutes, rinsing, then running several titrations n-ith standard reagents until consistent results are obtained. A potentiometer is used to adjust the voltage across the electrodes to 35 mv.
was found to be less than 2cc of the value for a series of eight analyses. Other results not described in detail here definitely show that phosphorus pentoxide interferes in the precipitation of trace quantities of magnesium Squinolinolate in the presence of ,1111monium ion if the concentration of pentoxide in the solution exceeds about 50. y. Although commercial lime coiltains sufficient phosphorus pentoxide to cause interference in the presence of ammonium ion, results showed that it can be satisfactorily analyzed h;v the procedure recommended here. illternate procedures employing prior remoi-a1 of phosphorus pentoxide with -4niberlite IR-4B anion exchange resin or as a ferric phosphate precipitate can be used, but are not recommended over the procedure outlined here because they are more involved and time consuming. ACKNOWLEDGMENT
The authors wish t o thank Editlia Karl-Kroupa and Larry Sacks for valuable suggestions made in the courqe of this work.
RESULTS
Table I summarizes typical results obtained with commercial lime, synthetic calcium phosphate, and commercial dicalcium phosphate dihydrate. It also includes results on the recovery of magnesium oxide added to these materials. The recoveries of added magnesium oxide are good in all cases, ranging from 98.5 to 102.870'0. The standard deviation of replicate determinations on a solution of commercial calcium phosphate to which magnesium was added
LITERATURE CITED
(1) Barnard, b. J., Jr., Broad, IT. C., Flaschka, H. Chemist Analyst 46, 76
(1957).
(2) Foulk, C. W., Bawden, A. T., J . Am. Chem. Sac. 48, 2045 (1926). (3) Lott, P. F., Cheng, K. L., Chemist Analyst 46, 30 (1957). (4) Redmond, J. C., Bur. Standards J . Research 10, 823 (1933).
(5) Van Thiel, H. E., Tucker, W. J., J. Agr. Food Chem. 5 , 442 (1957).
RECEIVEDfor review January 10, 1958. Accepted June 30, 1968.
Ca rbamate-Am mo nia Assay for Novobiocin F. A. BACHER, G. V. DOWNING, Jr., and J. S. WOOD, Jr. Merck Sharp & Dohme Research laboratories, Division of Merck & Co., Inc., Rahway, ,Novobiocin may be determined rapidly by a method based on the release of ammonia from the carbamyl group by alkaline hydrolysis. The ammonia is collected by steam distillation and titrated. Both novobiocin and inorganic salts of novobiocin have been assayed by this method with a precision of f1%.
analytical method for novobiocin should be capable of differentiating it from descarbamyl novobiocin, the degradation product of novobiocin most likely to be present SATISFACTORY
in process samples and pharmaceutical formulations. Since this paper was originally submitted, a n inactive isomer of novobiocin, isonovobiocin, has been reported ( 3 ) . As isonovobiocin has a carbamyl group, i t cannot be distinguished from novobiocin by the procedure given here. How widely this isonier occurs has not been established; in the authors' experience agreement betn-een bioassays and chemical assay results has been satisfactory, indicating that the maximum amount of isonovobiocin in the samples discussed herein would be on the order of 5 to 10%. The carbamate ester group is the most
N. 1. labile part of the novobiocin molecule (4, 8, 11). Examination of existing methods for the determination of norobiocin discloses no procedure that is specific, can be carried out in a reasonable length of time, and yields satisfactorily precise results. Mcrobial assays (2,7,20-12) on novobiocin are not very precise. Bchieiement of a fair degree of accuracy requires a large number of assays and consequent expenditure of a great deal of time. Colorimetric assays (1) generally depend on the phenolic character of the molecule and are subject to interferences from the descarbamyl deg-
VOL. 30, NO. 12, DECEMBER 1958
1993
radation product, as well as from siniilar phenols. The descarbamyl compound also interferes with assays based on acid-base titrations and ultraviolet absorption spectra (4, 6). Paper chromatographic procedures (9, 13) require a long time for resolution and are not precise enough to be ideal analytical methods. I n alkaline solution novobiocin has been observed to degrade easily to the descarbamyl product with concomitant release of ammonia. However, the yield of ammonia from steam distillation of the alkaline solution is not stoichiometric, although the formation of descarbamyl norobiocin is quantitative. A plausible explanation is that the hydrolysis of the carbamyl ester can yield either ammonia or carbaniate ion, the latter being stable in alkaline solution. Upon acidification of the reaction mixture the carbamate ion is deconiposed and, after again making the solution alkaline, the ammonia may be quantitatively steam distilled. This removal and titration of the ammonia is the basis of a precise assay for novobiocin. Descarbamyl novobiocin does not interfere. A correction may be applied for small amounts of ammonium ion and other nonextractable interfering substances b y extracting the novobiocin into ethyl acetate and carrying the residual water phase through the assay procedure. Application of the blank correction is regarded as more informative than direct assay of the ethyl acetate layer. I n addition, the presence of the ethyl acetate leads to some difficulties with the steam distillation.
V A
Figure 1, Distillation apparatus
Table 1. Comparison of Assay Values Obtained by Carbamate-Ammonia Chemical Method and Microbial
Sample ?To.
Methods 70Sodium Novobiocin Corrected for Loss on Drying Chemical Microbial assay assay"
97.0.97.4 98.5 2 95.6:96.2 98.8 3 96.5,96. 97.0 4 98.2,97.6 96.i a Performed by Chemical Control Division of Chemical Division, Merck & Co., 1
Inc.
EXPERIMENTAL
Reagents. Sodium hydroxide, approximately 2 5 . Sulfuric acid, approximately G S . The 6s acid and the 2 S sodium hydroxide concentrations should be adjusted carefully enough so that addition of 2 ml. of the acid to 5 nil. of the sodium hydroxide yields a strongly acid solution. Silicone antifoam agent GE-66 (General Electric Co.) diluted 1 to 1 with water and heated to achieve a smooth mixture. Mixed indicator, 2 parts by volume of a 0.1% ethanolic methyl red solution to 10 parts of a 0.1% ethanolic bromocresol green solution. Apparatus, The steam distillation equipment (Figure 1) is essentially that of Jenden and Taylor (5), except that the volunie of the unit was increased about twofold. Flask C has a full volume of about 60 ml. Procedure. Weigh accurately an amount of sample estimated t o contain about 0.09 nimole of novobiocin into round-bottomed flask C. Add two or three glass beads and 7 or 8 drops of GE-66 silicone antifoam agent. Lubricate ground-glass joint E and ball joint F liberally with 10N sodium 1994
ANALYTICAL CHEMISTRY
hydroxide before each run. Place receiver H , containing approximately 20 ml. of 2% boric acid, so that the condenser tube is below the solution surface. Connect the flask containing the novobiocin to the apparatus and add 5 ml. of 2147 sodium hydroxide through funnel A and stopcock D. Heat the contents of the flask a t the boiling point for 10 minutes without appreciable distillation of liquid. With the heat still applied, pipet 2 ml. of 6A' sulfuric acid into flask C through -4. Heat the resulting suspension until the sample is thoroughly mixed with the acid. Pipet 5 ml. of 2 5 sodium hydroxide through A . Remove receiver H as soon as the liquid begins to suck back. Replace the receiver and immediately start to paqs steam into the system. Remove the heat source a t C. Collect approximately 30 ml. of steam distillate (total volume in receiver, 50 ml,), The distillation should take 5 to 10 minutes. Titrate the distillate potentiometrically with standard 0.0lX hydrochloric acid to an end point p H of 5.2 or with the mixed indicator to a color change from blue-green to orangered.
A reagent blank should be run with each new batch of reagents by collecting 30 ml. of steam distillate from 10 ml. of the 2;1' alkali and 2 ml. of the 6N acid. This blank can be used for comparison of indicator end point color with the sample solution. Titrated potentiometrically, the reagent blank should be essentially zero. One mole of ammonia is obtained from 1 mole of novobiocin. The molecular weight of novobiocin is 612. Correction for Ammonium Ion and Nonextractable Amides and Amines. Weigh accurately about 0.5 mmole of the novobiocin sample into a 15-ml., stoppered centrifuge tube. Add 5 ml. of 0 . 2 s hydrochloric acid and 5 ml. of ethyl acetate. Stopper and shake 1 to 2 minutes. Centrifuge or allow the layers to separate. Draw off the bottom aqueous layer and put it into one of the steam distillation flasks. Extract the ethyl acetate phase with a n additional 5 ml. of 0.2N acid and add this extract to the flask. V i t h the flask attached to the apparatus and the receiver in place, pipet 5 ml. of 2N sodium hydroxide through A . Proceed with the steam distillation and titration with 0.01N standard acid as above. The quantities of sample specified above were chosen to permit correction for ammonium salts equivalent to 1% or more of the amount of novobiocin in the sample. DISCUSSION
A comparison of duplicate chemical assays with microbial assays on typical samples of sodium novobiocin is shown in Table I. The assay procedure can be performed in approximately 30 minutes A and yields a precision of =tl%. tenfold increase in the size of sample can be handled n-ith increased amounts of reagents and an increase in the size of the apparatus, A tenfold decrease in the size of sample causes considerable loss in precision. A sample of descarbamyl novobiocin gave no ammonia when carried through the procedure. I n general, there has been no reagent blank, and in the samples investigated there has been no ammonium ion correction. The method is designed for relatively pure samples of novobiocin and its salts, such as are found in the later stages of processing and in simple pharmaceutical formulations. The presence of other ethyl acetate-extractable carbamates, amides, or volatile amines might require different methods of separation. LITERATURE CITED
(1) Boxer, G. E., Shonk, C. E., Antibzotics & Chenzothempy 6 , 589 (1956). (2) Frost, B. I f . , \-ahant, RI. E., Ibid., 6 , 648 (1956). (3) Hinman, J. IT'., Caron, E. L., Hoeksema, H., J . Anz. Chem. Soc. 79, 5321 (1957). (4) Hoeksema, H., Bergy, ?I. E., Jackson,
.G Fonke \T
Shell .T.
IT.. Hlnman. J. W.,
E. L., Ford, J. H., DeVries, IT. H., Crum, G. F., Antibiotics & Chemotherapy 6, 143 (1956). (5) Jenden, D. J., Taylor, D. B., ANAL. CHEX 25, 685 (1953). (6) Kaczka, E. A,, Kolf, F. J., Rathe, F. P., Folkers, I
