peroxide; however, as organic material and hydrogen peroxide form detonable mixtures, this source of oxygen should not be resorted to indiscriminately (3, 4). The method has bl2en in routine use more than three years. It is fast (15 minutes), and the equipment is relatively free of maintenance problems. ACKNOW LEIIGMENT
The authors are indebted to s. z. Perry and E. D. Peters for valuable
counsel and encouragement, and to the personnel of Shell Chemical Co. a t Norco, La., for contribution of the data in Table I11 and for practical evaluation of the method. LITERATURE CITED
(1) Brooks, F. R., A w z i , E. J., Shell Development C0.j Private COmmunic@tion, 1959. (2) Duswalt, A. A., Brandt, W. W., ANAL. CKEM.32, 272 (1960). (3) Monger, J. M., Sello, H., Lehwalder, D. C., J. Chem. Eng. Data 6, 23 (1961).
(4) Shell Chemical Corp., New York,
~ ; ~ ~ g , B , " ~ e ~ ~
search Data on Safety Limitations.~, (5) Streim, H. G., Boyce, E. A., U. S. Naval Air Rocket Test Station, Lake Denmark, Dover, N. Y., private communication, 1957. (6) Sundberg, 0. E., Maresh, C., ANAL. CREM. 32, 274 (1960). (7) Tenney, H. M., Harris, R. J., Zbid., 29, 317 (1957).
RECEIVED for review November 26, 1962. Accepted March 15, 1963.
S pect rop(7 otome t ric Determination of 1,4-Naphthoquinone in Phthalic Anhydride C. E. GONTER and JOHN J. PETTY Pittsburgh Chemical Co., Research and Development Department, Neville Island, Pittsburgh 25, Pa.
b The reaction of qLiinones with compounds containing active hydrogen to form intensely colored compounds is used as the basis for a method to determine microgram quantities of 1,4naphthoquinone in refined phthalic anhydride. When 1,4naphthoquinone i s reacted with malononitrile in aqueous or alcoholic solution followed b y the addition of ammoniiim hydroxide or other alkali, the solution becomes blue. The intensity of the color is dependent upon pH, the solvent, and the anion concentration. Maximum absorption occurs at 583 mH, Beer's law i s obeyed, and the color i s stable for 24 hours. 1,4-Naphthoquinone can be determined in refined phthialic anhydride in concentrations as low as 0.5 p.p.m., with a standard deviation of *0.12. Furthermore, the method has been adapted to crude phthalic anhydride.
B
1,4naphthoquinone (1,P NQ) and/or its adducts and decomposition products lire highly colored, their presence in refined phthalic anhydride (PAA) is part cularly objectionable. Numerous methods have been proposed for analyzing for 1 , 4 N Q ; however, most are limited by either sensitivity or specific ty. The polarographic procedure brtsed on the reduction of 1,PNQ to 1,4-dihydroxynaphthalene at a dropping mercury electrode is satisfacto1.y when the concentration of 1,PNQ is above 0.1% and it is the only quinone present. The ultraviolet spectrophotometric method of Peters (6) lacks sensitivity when the concentrati'sn of 1 , 4 N Q in PAA is below 1%. The colorimetric method proposed by Johnson and CritcMeld ( 1 ) employing the reaction of ECAUSE
2,Pdinitrophenylhydrazine with quinones is sufficiently sensitive. However, no practical way has been found to hydrolyze PAA to phthalic acid, which is necessary because the hydrazine reacts preferentially with the anhydride. Condensation reactions of quinones with amines to form colored compounds have been described by Lacoste, Covington, and Frisone (6). Of the amines tested with 1 , 4 N Q in the presence of PAA, propylamine, isopropylamine, and cyclohexyl amine gave stable, yellowcolored solutions. The limit of detectability of 1 , 4 N Q in 3 grams of PAA was between 1.2 and 1.3 p.p.m. Investigations described by Kesting (2-4) indicate that compounds which have one or two negative groups attached to a methylene group react with quinones to form colored compounds. The color and its intensity were dependent upon the type of quinone, the pH, the solvent, and the negative groups attached to the methylene groups. Kesting used the reaction between malononitrile and 1 , P N Q as a p H indicator. The following method for the determination of 1,4-KQ is an adaptation of this reaction. EXPERIMENTAL
Apparatus and Reagents. Malononitrile, 1% solution in methanol. Reagent solution for crude PAA or liquors, 80 grams of quinone-free PAA, 96 ml. of concentrated ammonium hydroxide, and 304 ml. of water diluted to 1 liter with methanol. 1,4-Naphthoquinone, purified by steam distillation and sublimation. 1,4-Kaphthoquinone standard solution, 0.01 gram of purified 1,4-NQ in 100 ml. of methanol. Store in a dark bottle. Phthalic anhydride, recrystallized from chloroform.
Soectroohotometer. Beckman Model DK12. Absorption cells, silica, 1.0- and 10.0-cm. Calibration Procedure for Refined PAA. Dilute 5.0 ml. of 1,4-?uTQstandard t o 100 ml. with methanol. Pipet 0.0 (blank), 2.0, 4.0, 6.0, and 10.0 ml. of the dilute solution into 50-ml. volumetric flasks, each containing 5.0 grams of recrystallized PXA and 25 ml. of methanol. Add 2 ml. of malononitrile solution. Heat on a steam bath until all of the PAA is dissolved. Cool the solutions, and add 10 ml. of 9N ammonium hydroxide. Cool the solutions to room temperature. Dilute to volume with methanol. Determine the absorbance of the solutions a t 583 mp in matched, 10.0-cm., silica absorption cells. Use water as a reference and use a slit width of 0.043 mm. Refined PAA Samples. Accurately weigh 5 grams of sample into a 50-ml. volumetric flask. Add 25 ml. of methanol and 2 ml. of malononitrile solution. Treat according to the procedure for calibration. Calibration Procedure for Crude PAA. Dilute 25.0 ml. of 1,4-N& standard solution t o 100 ml. with methanol. Pipet 0.0 (blank), 3.0, 5.0, 10.0, and 15.0 ml. of the dilute solution into 50-ml. volumetric flasks. Add 2 ml. of malononitrile solution and 25 ml. of reagent solution. Heat from 5 to 10 minutes on a steam bath. Cool the solutions to room temperature. Dilute to volume with methanol. Determine the absorbance of the solutions a t 583 mp in matched, 1.0-cm., silica absorption cells. Use water as a reference, and use a slit width of 0.043 mm. Samples of Crude PAA. Accurately weigh 0.5 gram of sample into a 100ml. volumetric flask. Add 50 ml. of methanol. If the sample does not dissolve, heat on a steam bath. Dilute t o volume with methanol. Pipet A
VOL 35, NO. 6, MAY 1963
663
I
0.5
0.71
i-
w 0.4 0
z a 0.3 0
in
:0.2 0 .I
WAVELENGTH, rnp 1
1
475 500
I
I
550
600
&
Figure 2. Absorption spectra of the reaction products of 1,4-naphthoquinone
700
WAVELENGTH, mu
1.
Figure 1 . Absorption spectrum of the reaction product of 1,4-naphthoquinone and malononitrile
suitable aliquots into each of two 50ml. volumetric flasks. Add 2 ml. of malononitrile solution t o one flask. Add 25 nil. of reagent solution to each flask and heat on a steam b a t h for 5 t o 10 minutes. Cool t h e solutions t o room temperature. Dilute t o volume with methanol. Determine the absorbance of the solution containing the nitrile at 583 mp in matched 1.0-cm., silica absorption crlls. Tse the sample solution as reference, and use a slit width of 0.043 mm .
2. 3.
4. 5. 6.
To increase the sensitivity of the test the amount of PAA used mas increased to 5 grams. Conditions for optimum color development were determined experimentally. PAA is partially ester-
Table 1.
664
0
ANALYTICAL CHEMISTRY
ified upon initial dissolution in methanol. The mono ester formed is very soluble in methanol. The ammonium salts of phthalic acid and the monoester, which are formed upon the addition of am-
1,4-Naphthoqvinone in Refined Phthalic Anhydride 1,4-N&, p.p.m. Analyst A B C
Day
RESULTS AND DISCUSSION
To determine if the reaction between 1,4-?;Q and malononitrile was quantitative in the presence of PBA, a solution containing 10 pug. 1,4-?\'& per milliliter of methanol and a 1% methanol solution of malononitrile were prepared. Three-gram samples of recrystallized PAX and 50-1111. total volumes were used for the preliminary investigation. The PAX was dissolved in 25 ml. of methanol by heating on a steam bath. Five milliliters of the quinone solution, 2 ml. of the nitrile solution, and 5 ml. of concentrated ammonium hydroxide were added. The solution was cooled and diluted to volume with methanol. -1 scan of the 740- to 450-mp region on a Beckman Model DK-2 spectrophotometer indicated that maximum absorption is between 580 and 585 mp (Figure 1). Similar solutions were prepared containing the same concentration of 1,4-N& and reagents, but varying the order of addition and heating before the ammonium hydroxide was added. No difference in absorption was observed. Solutions were prepared varying the amounts of 1 , 4 N Q solution added. The color obeyed Beer's law and mas stable for 24 hours.
Malononitrile Cyanoacetic acid 2-Cyanoacetamide Acetyl acetone 1 -Phenyl-l,3-butanedione Ethyl malonate
Sample 1 (6.2p.p.m. 6.09 6.17 5.90 5.95
1
2
theoretical) 6.02 5.88 6.16 6.39
6.30 6.31 6.47 6.12 Mean 6.15
Sample 2 (0.8p.p.m. theoretical) 0.84 0.83 0.63
1 2
Source Analysts Samples Time
0.70
SS
A-S A-T S-T A-S-1'
Duplicates Total Interactions Duplicates
0.1536 171.6815 0.0030 0.0275 0.1596 0,0012 0.0319 0,1048 172.630 0.2202 0.1048
0.74 0.73 0.84 0.82
ANALYSISO F VARIANCE df h!fS 0.0768 21 171.6815 1
2 2 1 2 12 23 7 12
0.81 0.81 0.91 0.91 Mean 0.80
2E U'E
gTU'ST U2AS
f
TU'ST
0,0030 0,0138 0.0798 0.0012 0.0159 0.0087
U23
0.0315 0.0087
4u21
f
++ -
U2E U'E
Estimate of:
+
U2E
f + +
CU'AT gU'AS CU'AT
+ +
f
CgU'A TgU'S
gUzAS CU'AT TU28T
U'P
U'D
+
U*D
U'D
- 0.0315 - 0.0087 - o,oo5, 4 u21 U'D = 0,0057 0,0087 = 0.0144 variance of single measurement s = dU2I U 2 D = *0.12 U I
+
+
+
TCU'T
moniuin hydroxide, are not readily soluble in organic solwnts ; therefore i t was necessary to add n ater. The mater had some effect on the color development as well as or the solubility. Between 17 and 21'2 (by volume) water was satisfactory when 5 grams of PA9 was used. The p H range for iuasimum color development under conditions of the test was found to be between 9.1 and 9.7. From tlic, results iipon application of the standardization procedure, the absorptivity values for 1,4-NQ in the presence of P-1.1 were calculated t'o be Agm,I . = 63 and egm. n l o ~ e . ' ~ . = 9960. On the haeis of R 5-grain sample of PAA the limit of detectabilit,~of 1,4-9& is 0.3 p.p.m. It was also established that 1,4-dihydroxynaphthal~nereacts in a similar manner and has equiyalent absorptivity values. Benzoquinone and hydroquinone under conditions of thi: t'est formed red solutioiis iihicli absorbed a t 517 and 480 m p . I-Ionever, the rea.ction is more pH dependent and the colcr less stable than in the case of 1,4XQ. The reaction b e t w e n malononitrile and I . 4 S Q appears tct be analogous to the reaction between unsaturated diketones and compounds containing active methylene hydrogen. Solutions of a number of compcunds which contain active hydrogen were added to solutioii,i containing a lcnon-n amount of 1,4-SQ. Varying amounts of ammonium hydroxide were added to vary t'he pII 9 through 113. Colored solutions which absorhed betn-ern 560 and 590 mfi were obtained with acetylacetone, diethylninlonate. cyanoacetamide, cyanoacetic acid, and l-pheny1-1,3-butanedione. Except for cyanoacetic acid none of the compound: was as sensitive as nialononit'rile. When the compounds were used in the presence of PI.1 under test conditions, the wavelengths a t which the maximum intensity appeared were lower in some tabes (Figure 2 ) . Cyanoacetic acid lost its sensitivit,r and the color formed was not stable. A number of quinone. were reacted with malononitrile a i d all of the previously mentioned compounds. The solutions prepared contained 5 mg. of PX.4 per milliliter and 0.09 mole of X H 4 0 H per milliliter. Xethanol was used as a solvent. Y o attempt was made t'o obtain conditicm for maximum color intensity. All of .;he quinones and their corresponding hydroxy compounds which reacted readily with malononitrile (Figure 3) also reacted with the other compounds containing ,ictive hydrogen. Compounds which react with malononitrile to give colored solutions which absorb in the 380- to 425-mu region in-
clude 2,3-butanedione, crotonaldehyde, cinnamaldehyde, methyl ethyl ketone, acetone, and maleic anhydride. Anthraquinones, acenaphthenequinone, and mono ethers of hydroquinone do not react with malononitrile. Under identical conditions the absorbances of solutions containing both PAA and 1,4-SQ were greater than those containing only 1.4-NQ. Solutions were prepared containing acetic, benzoic, hydrochloric, and sulfuric acids in place of the PAX. The absorbances of these solutions were equal to those obtained with PA4 present. If it is assumed t h a t the colored compound
Table II.
1,4-Naphthoquinone in Crude Phthalic Anhydride
Analysts Day
1,4-N&, %
__
B
C
-4
Sample 1 1
2
0 0 0 0
15 15 14 14
0 0 0 0
14 13 11 12
0 0 0 0
0 14 0.14 0.14 14 0.13 Mean 0.14
14 14 14
Sample 2 1 2
0 0 0 0
14 13 13 12
Mean
0.14 0 14 0.11 0 11 0 13
Sample 3
5 7;
0.37 0.35 0.37 0.37 0.36 0.36
1
2
3
Mean 0.36 Sample 4 0.36 0.37 0.35 0.40 0.36 0.36
1
2 I 1
450
475 500
,
550
'\ , ' \ '\\
600
\ \
3
700
Alean 0 . 3 7
WAVELENGTH, mv
Figure 3. Absorption spectra of the reaction products of some quinones and malononitrile I. 2. 3. 4.
1,2-Naphthoquinone Methyl-p-benzoquinone 2,5-Dimethyl-p-benzoquinone Benzoquinone
can exist in both anhydrous and hydrated form, the anhydrous state being deeper in color, then i t may be considered that the salts formed upon addition of ammonium hydroxide shift the equilibrium toward the anhydrous state by combining with available water. To determine the precision of the method, two synthetic samples of l+XQ in recrystallized PAA were prepared. The procedure wis followed in duplicate by three analysts a t two different times (Table I). The analysis of variance estimates the standard deviation for a single measurement a t 1 0 . 1 2 . It is emphasized that the reported standard deviation represents not merely the duplication error (which almost invariably underestimates the true error), but rather the experimental error between one operator using the method on one day and another operator using the method on a different day.
For samples of crude PALAcontaining from 0.1 to 0.4y0 1,4-NQ plus 1,4-dihydroxynaphthalene, the standard deviation from the analysis of variance is +0.01 and at 95% confidence is *0.02 which amounts to between 5 to 10% relative standard deviation, Table 11. ACKNOWLEDGMENT
The authors acknowledge the technical assistance of J. R. Lutchko, Milton Manes, and Raita Rlurnieks. LITERATURE CITED
(1) Johnson, D. P., Critchfield, F. E., ANAL.CHEW34, 1389 (1962). ( 2 ) Kesting, W., Ber., 62, 1422 (1929). (3) Kesting, W., J . Prakt. Chem. 138, 215 (1933): (4) Kesting, W, Z.Angew. Chem. 41, 358, 745 (1928). ( 5 ) Lacoste, R. J., Covington, J. R., Frisone. G. J.. A s . 4 ~ .CHEM.32, 990 (1960). (6) Peters, H., ANAL. CHEM. 31, 1326 (1959).
RECEIVED for review November 15, 1962. Accepted February 6, 1963. Presented before the Division of Analytical Chemistry, 141st Meeting, .4CS, Washington, D. C., March 1962.
VOL.
35, NO. 6 , MAY 1963
e
665
