Some preliminary work cited in the experimental part was carried out by Robert D. Goldberg. The design of the apparatus was suggested by Julius Chupak. The authors are especially indebted to Carol Thoren for checking the procedure and contributing some of the data given in Table I.
(1) Ephraim, F., “Inorganic Chemistry,”
5thed., p. 540, Throne, P. C. L., Roberts, E. R., eds., Gurney and Jackson, London, 1948. (2) Feigl, F.,,,“Qualitative Anal sis .by Spot Tests, 3rd ed., p. 72, &ewer, New York, 1947. (3) Handbook of Chemistry and Physics, 40th ed., p. 1740, Chemical Rubber Publ., Cleveland, Ohio, 1958.
(4)Leverenz, H. W., “Introduction to Luminescence of Solids,” p. 333, Wiley, New York, 1950. (5) National Bureau of Standards Handbook 66, “Safe Design and of Industrial Beta-Ray Sources, May 1958.
RECEIVED for review November 20, 1958. Accepted April 28, 1959.
CoIo rimetr ic Dete rmina tion of 0 rga nic
MERLE I. ElSS and PAUL GIESECKE Central Research Division, American Cyanamid Co., Stamford, Conn.
b Benzoyl leuco methylene blue i s a new reagent for the quantitative spectrophotometric determination of organic peroxides. In a benzenetrichloroacetic acid solution, it reacted with peroxides and hydroperoxides to form the characteristic methylene blue color. The reaction was sensitive to ultraviolet light and to a lesser degree to artificial light and heat, but the color i s stable for several days if kept in the dark at 24” C. Zirconium naphthenate was used to accelerate the peroxide decomposition and thereby increase the leuco dye reaction rate. O f the five peroxides tested, benzoyl peroxide, lauroyl peroxide, p-methane hydroperoxide, and cumene hydroperoxide followed Beer’s law. tert-Butyl hydroperoxide deviated somewhat, and a calibration curve of concentration vs. absorbance was necessary for the compound. The method was found to be simple and sensitive. As little as 0.5 mg. of active oxygen could be detected. An estimate of precision of the method was obtained by running replicate samples for each compound. The 95% confidence limits ranged from h l . 7 to
interest in polymers required a reliable and simple colorimetric method for the determination of traces of organic peroxides and active sites on oxidized monomers and polymers. A method using leuco methylene blue as a reagent for peroxides was proposed by Ueberreiter and Sorge in 1956 (9, 10). However, the leuco base was not available in a usable form because of its extreme instability. It was difficult to synthesize and troublesome to store. Benzoyl leuco methylene blue n as found to be st new sensitive colorimetric reagent for organic peroxides. I n a bPnzene-trirrill~~r,a.eTic acid s o h ESEARCH
3 Benzoyl leuco methylene blue
Methylene blue cation
Table 1. Effects of Metallic Naphthenates on Benzoyl Methylene Blue System
Metals5 Zirconium Lead Zinc Manganese Cerium Iron
Effects Excellent acceleration Some acceleration Some acceleration Excessive reagent blank Excessive reagent blank Highly colored, green Highly colored, green Copper Cobalt Yo acceleration Calcium No acceleration a One drop commercial naphthenate in 52 ml. of benzene-trichloroacetic acid solution. tion, it reacted with peroxides and hydroperoxides to form the characteristic methylene blue color. The reagent was stable in its crystalline form and storable in a refrigerator under normal conditions. It was not greatly affected by air, and in benzene solution could be stored in a brown bottle at room temperature. Benzoyl leuco methylene blue was affected by ultraviolet light (it turned blue rapidly and therefore had to be kept out of sunlight) and, to a lesser degree, by heat and artificial light. However, the benzoyl leuco dye-peroxide reaction was fairly slow. A hydroperoxide of the tert-butyl type required 36 hours to react at room temperature, and benzoyl peroxide took over 120 hours. An attempt was made to find a method of accelerating the color development by speeding up the decoiiiposition of the peroxide and, at the same time, maintaining a good level oi accuracy. Several heating variat’ions
, / , I , l , I
l , I , ! , ! , , , I
$W 20 40 €0 80 So0 20 40 60 BO 6W 20
1,l 40 50 BO
WAVE LENGTH IN MILLIMICRONS
Figure 1. Spectrophotometric curves of methylene blue system a. N o active oxygen
b. 3 y active oxygen
were attempted, but even though the reaction was accelerated, the results were not quantitative. Light was also found to increase the reaction rate, but it caused a marked decrease in reproducibility (Figure 1). It was thought that amines and cobalt naphthenate would increase the rate of color development by quickly decomposing the peroxides. Several amines and amides were tried ( 8 ) but did not have the required speed Cobalt naphthenate was reported to be a useful compound for decomposing peroxides (1-4). However. it did not increase the rate of this reaction Ar; investigation of all available metallit naphthenates was undertaken Table 1 lists the metals and theii effects 011 the reaction. Zirconium w i t founa T O $.we the fastest reaction time and still h a w
a reasonable reagent blank. Zinc and lead produced some increase of color development but were not as good as the zirconium. Cerium and manganese could possibly have been used in lower concentrations but, under these particular conditions, gave a n excessive reagent blank. In general, the metallic naphthenates are thought t o react with hydroperoxides and peroxides in a manner similar to the following lead reaction ( I ) :
Reaction Times and Absorbances for Peroxides Tested
Name tert-Butyl hydroperoxide
Absorbance, 1 Mg./100 Ml., 1-Cm. 0 at 25 C. Cell 17.8 30min. (36 hr.a) Approx. 16
% Complete Color Active Develofment
10.5 40mi11.(38hr.~)9 . 7
+ ROO. + H?O
The RO' and ROO' free radicals oxidize benzoyl leuco methylene blue to the methylene blue cation. However, it is not yet known whether zirconium reacts with peroxides in the same way as lead, because of its reluctance to exist in several valence states.
n ould then
RESULTS AND DISCUSSION
Table I1 indicates the various per-
(8 8 i 0.15)b
( 4 . 6 i 0.10)b
0 6.6 30 hr. (120 hr.a) ( 6 . 2
0 . 12)b
Time required for complete color development wit,hout use of zirconium naphthenate. 95% confidence limit.
Standard peroxide and hydroperoxide solutions are prepared by dissolving weighed amounts of the pure materials in benzene. To obtain a calibration, aliquot portions of peroxide solution are pipetted into a 25-ml. volumetric flask which contains 15 to 20 ml. of 0.5% trichloroacetic acid-benzene solution, and 0.3 ml. of 0.24a/, zirconium naphthenate and 1 ml. of leuco dye are added. The flask is filled to the mark with benzene, mixed thoroughly, and protected from light immediately. The flasks are allowed to stand in the dark for a designated time (Table 11) at 24' =t '1 C. The transmittance is measured against water at 662 mp on a spectrophotometer using I-cm. cells. A reagent blank is run with each set of standards. The same procedure is used for sample analysis. If the peroxide species is unknown, the time for the peroxide decomposition to reach completion must be checked experimentally by measuring the absorbance until i t no longer increases. Results are then calculated as per cent active oxygen.
' 2 hours
REAGENTS AND APPARATUS
Benzene, reagent grade. Zirconium naphthenate, 0.24%. Dilute 1 ml. of commercial (6%) zirconium naphthenate to 25 ml. with benzene. Trichloroacetic acid, reagent grade, 0.5% in benzene. Benzoyl leuco methylene blue (obtained from the Natiolial Cash Register Go., Dayton, Ohio). Dissolve 0.05 gram in 100 ml. of benzene. Store in dark bottle. All spectrophotometric measurements were made with a General Electric recording spectrophotometer with I-cm. cells. A colorimeter can also be used with Corning filters 2403 and 3962.
Analyses of Peroxide Standards
I - B U T I L UIDROPEROXJIDE I - B U T I L UIDROPEROXJIDE
10.9 16.8 21.6 10.2 20.0 35.8
Benzoyl peroxide p-Menthane hydroperoxide
9.9 19.8 36.0 13.1 26.3 32.8
MINUTES (24" C.)
Name Lauroyl peroxide
Rate of color development
oxides and hydroperoxides tested and the time required for each reaction t o reach completion. It is clearly seen that zirconium aids in the decomposition of the peroxides, thereby accelerating the color reaction. The color development time of tert-butyl hydroperoxide was reduced from 36 hours to 30 minutes and the benzoyl peroxide from 120 to 30
3,8 7.6 15.2 tert-Butyl hydroperoxide 5 . 9 11.8
25.9 33.5 3.6 7.3
hours. Benzoyl peroxide decompm 3 much more slowly than the other peroxides tested, even with the addition of zirconium. For the faster reacting hydroperoxides (cumene and tertbutyl), most of the color was developed within the first 10 minutes (Figure 2). The reaction product of benzoyl leuco methylene blue with benzoyl peroxide, '401. 31, NO. 9, SEPTEMBER 1959
auroyl peroxide, cumene hydroperoxide, and p-menthane hydroperoxide followed Beer’s law u p to 1 p.p.m. However, tert-butyl hydroperoxide deviated somewhat, and a calibration curve of concentration us. absorbance was necessarv for this compound. The validity of this method should be verified experimentally before applying it to peroxides other than those tested. For example, dialkyl peroxides, such as di-tert-butyl peroxide ( I O ) , do not react with benzoyl leuco methylene blue. The reaction was heat sensitive, and therefore, all work was done a t 24” i~ 1’ C. At 30” C., the results became very erratic. Because the color was also light sensitive, the solutions were stored in the dark while the color was developing. ilrtificial light caused irregular results, and sunlight ruined the determination completely by causing very excessive reagent blanks. The reagents were all stable in benzene solution and were usable for 3 to 4 weeks. The leuco dye was kept in a brown bottle
an-ay from direct light. The only reagent concentration that appeared to be critical was the zirconium naphthenate. If too much was added, the reagent blank became excessive. iilthough there was a waiting period for color development, it was simple to run 6 to 10 determinations a t the same time. The reaction \vas sensitive, and active oxygen was determined down t o less than 0.5 mg. Table I11 lists analyses of the five standard peroxides. -411 commercial peroxides which were used as standards were analyzed iodometrically. The precision of the method was obtained by running 7 to 16 replicate samples for each peroxide. For the four compounds which follow Beer’s law, the 95% confidence limits range from =!= 2.6 to =!= 1.7%.
The authors thank Marjorie C. Lyon and Thomas Gordon of the National
Cash Register Co. for supplying the benzoyl leuco methylene blue reagent. REFERENCES
( I ) AIIen, L. H., Paint Technol. 22 (248),
161 (1958). (2) Ibid., 22 (249), 205 (1958). (3) Ibid., 22 (250), 241 (1958). (4) Bawn, C. E. H., Mellish, S.F., Trans. Faraday SOC.52, 1216 (1956). (5) Egerton, A. S., Everett, A . J., Anal. Cham. Acla 10, 422 (1954). (6) Miller, G. L., “Zirconium,” Academic Press, New York, 1954. (7) Robey, R. F., Wiese, H. X., Ind. Eng. Chem. 17, 425 (1945). (8) Tobolsky, A. V., Mesrobian, R. B., “Organic Peroxides.” Interscience. Neiv York 1954. (9) Ueberreiter, K., Sorge, G., Angew. Chem. 68 (lo), 352 (1956). (10) Ibid., 68 (15), 486 (1956). (11) Venable. F. P.. “Zirconium and Its Compounds,” Chemical Catalog Co., New York, 1922. (12) Wagner, C. D., Clever, H. L., Peters, E. D., A X A L . CHEM.19,980 (1947). RECEIVED for review March 5, 1959. Accepted May 26, 1959. Division of Analytical Chemistry, 135th Meeting, ACS, Boston, llass., rZpril 1959.
Determination of Boric Oxide in Glass by Pyrohydrolysis Separation J. P. WILLIAMS, E.
Glass Research and Developmenf Division, Corning Glass Works, Corning, The boric oxide in many glass compositions can be separated by a pyrohydrolysis-type reaction in which steam is passed over a mixture of glass sample, uranium oxide (UaOs), and sodium metasilicate nonahydrate (No2S i 0 3 . 9 H 2 0 ) at 1300’ to 1350°C. in a platinum tube. The boric acid in the distillate is determined by the usual mannitol-sodium hydroxide titration. The effects of glass sample size and composition, catalyst, temperature, and distillation rate are discussed. Quantitative separation can be expected from most glass compositions except those containing greater than 10% lead, 5% zinc, or 1% phosphorus oxides. A complete determination can be carried out in about 90 minutes. Pyrohydrolysis results compare favorably with more conventional methods of analysis for glasses containing from about 0.2 to 3070 boric oxide. ORIC OXIDE in
glass has been determined by opening u p the glass by carbonate fusion, dissolving the melt in acid. and separating the boron by dis1560 *
tillation of methyl borate from methanol or controlled p H precipitation. The boron is then usually determined by titrating the mannitol-boric acid complex with standard sodium hydroxide (5). The American Society for Testing Materials ( 2 ) uses the methanol distillation procedure originally reported by Wherry and Chapin (I$), while Hollander and Rieman (6) and Webster (11) recommend controlled hydrolysis for separating the boron. Pyrohydrolysis separation of the halides has been investigated by a number of workers. Early mention of this process by Warf (9) was expanded in a later report (10). Most investigators have employed uranium oxide (U308) as an accelerator or catalyst. Powell and Menis ( 7 ) have reported both micro- and macroseparations of fluoride from inorganic materials and the effect of various catalysts, as well as the use of moist oxygen instead of steam. Susano, White, and Lee (8) suggested a nickel tube to replace platinum; Gillies, Keen, Lister, and Rees (3) described a silica apparatus, and Adams and
Williams (I) used Corning Code i900 96% silica glass. The high volatility of boric oxide a t glass melting temperatures, coupled with the knowledge that boric acid readily steamdistills (4) suggested that pyrohydrolysis so successfully applied to the separation of fluoride and other halides from inorganic materials might be useful for separating boric oxide from borosilicate glasses. APPARATUS
The apparatus (Figure 1) is a modification of other typical pyrohydrolysis setups. The reactor tube is fabricated from 1 5 4 1 platinum sheet shaped into a tube 1.00 inch in outer diameter and 17 inches long. Both ends of the platinum tube extend beyond the combustion furnace and after assembly are wrapped with asbestos to minimize heat loss. One end of the reactor tube is attached to a T-connector tube, which consists of a &inch horizontal length and a 17-inch vertical arm made from 1.0-inch outer diameter Corning Code 7900 tubing. The two horizontal end openings on the cross arm are 29/42 standard taper malt joints and the opening at the lower mi-