January 1953
INDUSTRIAL AND ENGINEERING CHEMISTRY
215
LITERATURE CITED
Table I. During these tests the material in the pot contained Q4.5yoof 1-methylnaphthalene. The performance of the same packing when evaluated by the authors, with other test mixtures developed in this laboratory, is shown in Table 11.
Berg, L., and Popovac, C. O., Chem. Eng. Prog., 45, 683 (1949). Bragg, L. B., and Richards, A. R., IND. E N G . CHEY.,34, 1088 (1942).
Cannon, M. R., Ibid., 41, 1953 (1949). Feldman, J., thesis, University of Pittsburgh, 1950. Feldman, J., Mylea, M., Wender, I . , and Orohin, M.,IND. ESG. CHEM.,41, 1032 (1949). Feldman, J., and Orchin, M., Ibid., 44, 2909 (1952). Feldman, J., and Orchin, M . , U. 8.Patents 2,581.398; 2,583,554
.Iprevious report from this laboratory compared the separating efficiency of various kinds of packing materials (IS). The results of some of the earlier tests are compared with those from the present study in Table 111. Results with a protruded packing (S), 0.16 X 0.16 inch made from 316 stainless steel, are also included.
(1952).
Fenske, M. R., Myers, H. S.,and Quiggle, D., I N D .ENG.CHEY.,
CONCLUSIONS
. 1
The results of the present study show t h a t the mixture of 1-methylnaphthalene and 2-methylnaphthalene is a useful standard test mixture for evaluating the performance of distillation columns in the pressure range 20 t o 760 mm. of mercury and in the separating-efficiency range of 5 to 60 theoretical plates. Within experimental error, evaluation of the same column with different test mixtures gave the same results. Heli-Pak packing is excellent for column operation at reduced pressure. I t s separating efficiency is less sensitive t o changes in throughput than Heligrid packing.
42, 649 (1950).
Friedel, R. 9., and Orchin, M., "Ultraviolet Spectra of Aromatic Compounds," p. 29, New York, John Wiley & Sons, 1951. Griswold, J., IND.ENG.CHEM.,35, 247 (1943). Hawkins, J. E., and Brent, J. A., Ibid., 43, 2611 (1951). Morton, R. 8., and Gouveia, A. J. 4 . de, J . Chem. Soc., 1934, 916.
Myles, hf.,Feldman, J., Wender, I . , and Orchin, M., IND. ENG. CHEM.,43, 1452 (1951). (14) Willingham, C. B , and Sedlak, V. A., J . Research .VatZ. Bur.
(13)
Standards, 45, 315 (1950) ; Research Paper 2140. RECEIVED for review March 27, 1952.
ACCEPTED September 8, 1932
Effect of Wilev Mill Cutting on. J
Solubilitv of Cellulose Fibers ARTHUR F. JOHNSON AND WILLIAM A. RlUELLER Buckeye Cotton Oil Co., Memphis, Tenn.
THE
< solubility of cellulose in alkaline solutions is widely employed in characterizing cellulose pulps for commercial uses. The influence of fiber structure on cellulose solubility does not appear to have received critical attention. Mechanical action on cellulose fibers t o destroy the gross fiber structure has been largely confined t o milling proccdures with considerable accompanying degradation (3). Wiley mill cutting has been shown t o affect the or-cellulose determination with little viscosity decrease ( 1 ) . A more detailed investigation of the effect of Wiley mill cutting on cellulose solubility is the purpose of the present work.
EXPERIMENTAL
SAMPLEPREPARATION. A series of pulp samples of different histories (see Table I ) was selected. The samples were cut in a S o . 2 Wiley mill using a 0.5-mm. screen and a blade spacing of 0.005 inch. OF CELLULOSE IY 2.75 N SODIUM TABLE I. SOLUBILITY HYDROXIDE
Solubility of Sam lesb, Wt. % ' Cellulose Removed by Wlkaline Treatment .4 A-Hy A-Ht A-Bt B B-L C C-Ht 1.61 2.08 1.12 1.41 5.4 7.1 0.79 1.48 0 1.96 3.83 2.44 196 1.87 7.1 8.3 3.37 1.24 2.46 4.66 2.90 2.64 8.5 10.1 2.32 6.71 1.49 3.41 5.67 3.66 10.7 14.3 2.96 3.93 13.51 2.48 4.67 4.28 6.45 12.4 17.6 3.86 4.78 20.83 3.14 6 d = sieve mesh opening in millimeters through which the fraction passed. The initial pulp has l(d = 0. b Sam le identification: A cotton linter pulp [ q ] = 7.5 commercial acetate gra&: A-Hy acid hydroiysed A [ q ] = 3.5,' A-Ht, A heated at 150' C. for 30 minutes: A-Bt, A beaten from 5 to 20 seconds freeness; E , cotton = 3.3 commercial viscose grade: B-L, 1 hour exposure to linter pulp ultraviolet'( zalid machine) * C wood pulp, commercial acetate grade, sulfite process; C-HI, C heated'at k O 0 C. for 30 minutes.
-d1'
61
The cut material was separated into portions of different particle size with standard sieves. The sieved portions (characterized by l / d where d is the sieve mesh opening in millimeters through which the portion passed) were: through 50- to 100mesh, l / d = 3.37; through 100- t o aOO-mesh, l / d = 6.71; through 200- t o 3O0-meshJ l / d = 13.51; and through 300-mesh, l / d = 20.83. Moisture content by oven drying (3 hours a t 105" C.) was not affected by the Wiley mill treatment.
EXTRACTION. Solubility, defined as per cent by weight of cellulose removed by alkaline treatment, was determined by treating 0.2-gram samples with 20 ml. of 2.75 N sodium hydroxide a t room temperature. Pulp suspensions were filtered after 15 minutes through medium fritted-glass filters. ANALYSIS.The filtrates from the extractions were analyzed in duplicate for cellulose by an anthrone procedure similar to t h a t of Viles and Silverman ( 4 ) . Aliquots of cellulose solutions containing 100 micrograms of cellulose are made up to 2.5 ml. with 72% sulfuric acid. Five milliliters of 0.1% anthrone in concentrated sulfuric acid are added and the solutions are heated 15 minutes a t 80" C. Blanks and glucose standards are run concurrently. Color intensities are measured, after cooling, a t 625 mp using a Beckman DU spectrophotometer. Unknown concentrations are calculated using the Beer's law relationship. The data are given in Table I. STATISTICAL ANALYSIS. The data for each sample were fitted to the straight line
% solubility
= a
+b (l/d)
by the least-squares procedure. The calculated parameters are given in Table 11. Also in Table I1 are given the correlation coefficients together with tkeir levels of significance and the standard deviations of "scatter s7, and of the slope, Sb. The statistical procedures are given in ( 2 ) . The least-squares lines are plotted in Figures 1 and 2.
Vol. 45, No. I
INDUSTRIAL AND ENGINEERING CHEMISTRY
That the increase in solubility with Wiley milling is not principally due to degradation can be seen from sample A , where [ 7 ] = 7.56, and sample A - H y , where [ 7 ] = 3 . 5 . The solubility of the uncut pulp ( l / d = 0) increased 87% as the intrinsic viscosity dropped 50% from A t o A-Hu. For sample A , the solubility increased on ST7iley milling 323% ( l / d = 20.83) while the intrinsic viscosity dropped only about 9% at the same point (Table 111). The increase in solubility on Wiley milling is thus of an entirely different order of magnitude than the increase accompanying degradation in the intact pulp.
PULP A
%
/
P U L P B AND C /
15
/ /
-sz! I-
I
I
I
I
I
4
8
I2
16
20
(PARTICLE
A-HT
/
/
/'
IO
m
3
I/D
-I
0 v)
SIZE)-'
Figure 1. Regression Lines of Per Cent Solubility on (Particle Size) -l
0
0
,~'B-L
0 '
5
For a cotton linter pulp ( A ) , hydrolyzed ( A - H y ) , beaten ( A - B t ) , and heated ( A - H t )
TABLE 11. Level of Significance, Samplea A A-Hy A-Ht A-Bt
B
B-L
C
C-Ht a
%
%
bC
ffic
87 d
SbE
0.1 0.1 1.0 0.1
0.115
0,097 0,076 0.30 0.12
1.o 0.1 0.1
0.192 0.331 0.523
0.81 1.46 1.58 1.47 1.58 2.83 5.88 6.83
0.0058 0.0048 0.018 0.0072 0.0035 0.037 0.028 0.022
rb
0.996 0.998 0.983 0.997 0.999 0.948 0.989 0.997
STATISTICAL ANALYSIS
0.1
0.131
0.059 0.62 0.47 0.36
See Table I for sample identification.
b Correlation coefficient. c
0.116 0.164 0.157
Parameters i n least-squares line:
% solubility
= ffi
+,b ( l / d ) .
d Standard deviation of scatter a b o u t the least-squares line. e
I
I
4
8
(PARTICLE Figure 2.
I
1
12
16
I
20
I/D
SIZE)-'
Regression Lines of Per Cent Solubility on (Particle Size)-'
For a cotton linter pulp ( B ) , after irradiation with light ( B - L ) , and a wood pulp (C), after heating
Standard deviation of the slope.
VISCO~ITY h'iEASUREMENTS. The extent of degradation during
Wiley mill cutting was estimated from viscosity measurements for one of the samples. Intrinsic viscosities (cupriethylenediamine) for the original pulp and cut fractions are given in Table 111.
T TABLE 111. IKTRINSIC
T OF WILEY ~ MILLED ~ PULPS' ~
Sample Original < A ) T o t a l Wiley milled On 100-mesh Through 100- to 300-mesh a Kramer-Martin method (cupriethylenediamine).
[VI
7.56 7.21
7.46
7.26
RESULT§ AND DISCUSSION
The data (Table I) and the statistical treatment (Table 11) show t h a t the solubility of cellulose is a linear function of (sieve If a given sieve mesh opening is considered mesh opening)-' an estimate of particle size of the subsequent fraction of the material, it may be stated that the solubility is a linear function of the surface area of the sample.
The effect of the various treatments on the solubility versus ( l j d ) curves will be considered briefly. The effect of hydrolysis is seen in Figure 1 ( A and A - H y ) t o increase the solubility of the intact pulp while leaving the slope of the curve the same. The two lines are parallel over the range investigated. Mechanical treatment (A-Bt) and heating ( A - H y ) have increased not only the~ solubility ~ of the ~intact pulp ~ but also ~ caused a statistically significant increase in the slope of the curve. Figure 2 shows that heating of a more soluble pulp (C and C-Ht) resulted in the same type of change as observed with sample A . The effect of irradiation with ultraviolet light ( B and B-L) has also been to increase the solubility of the intact pulp as well as the slope of the curve. Although the number of instances of each treatment tested does not permit generalization as t o the specific effects of any one treatment, the general conclusions may be drawn that the history of a cellulose sample may affect not only the solubility of the intact pulp but also the slope of the solubility versus ( l / d ) curve. The pronounced effect of pulp history on the solubility versus (particle size)-' curve indicates the futility of the usual procedure of comparing pulp samples on the basis of the solubility of the
lanuary 1953
I N D U S T R I A L A N D ENGINEERING CHEMISTRY
fibrous material. It is essential t o measure both the slope and the Intercept of the curve. Three cases can be selected: Case I. Pulps may have different solubilities before milling with the same slope-e.g., A and A-Hy, Figure 1. Case 11. Pulps may have different solubihtles before milling and different slopes without intersection of the curves-e.g., A and A-Ht, Figure 1. Case 111. Pulps may have the same solubility before milling and different slopes-e.g., A-Hy and A-Bt, Figure 1.
*
4
It is also possible that a fourth case, different solubilities before milling with intersecting curves, may be anticipated with appropriate treatment combinations. I t is generally assumed that alkaline solutions of a given concentration are selective for a certain range of degree of polymerization, If this is true, it would be expected t h a t the solubility versus (particle-size) -1 curves should level off as the entire amount of material of given degree of polymerization becomes available for extraction. No tendency toward leveling off was noted. Either there is much more soluble (low degree of polymerization) material in these samples than would be expected from t h e intrinsic viscosity-Le., the curves would level off at smaller particle sizes-or the assumption of selectivity of alkali for a certain degree of polymerization range may need modification.
217
Since the solubility of a cellulose sample in sodium hydroxide solution is dependent on particle size, it may be expected t h a t solubility in other solvents would be similarly affected. Preliminary measurements show t h a t summative fractionation curves obtained with cupriethylenediamine by a fractional solution method show differences in the low degree of polymerization range between intact and Wiley milled cellulose samples. ACKNOWLEDGMENT
The authors wish to acknowledge the able assistance of Karl Jurbergs with the experimental work. LITERATURE CITED
(1) Browning, B. L., Paper Trade J., 124, No. 15, 54-5 (April 10, 1947). ( 2 ) Brownlee, K. A., “Industrial Experimentation,” 3rd ed., pp. 61-74, Brooklyn, N. Y., Chemical Publishing Co., 1949.
(3) Forsiati, F. H., Stone, W. K., Rowen, J. W., and Appel. W. D., J . Researeh Natl. B u r . Standards, 45, 2, 109-13 (1950) (4) Viles, F. J., and Silverman, L., Anal. Chem., 21, 950 (1949). RECEIVED for review May 26. 1952. ACCEPTED August 11. 1962. Presented before the Division of Cellulose Chemistry at the 121st Meeting of the AMERICAN CHEMICAL SOCIETY, Milwaukee, Wis.
Identification and Vapor Phase Oxidation of 07374Bicvclononadiene J
F. T. WADSWORTH AND FRANK J. SMITH’ Pan American Rejining Corp., Texas City, Tex.
I
T HAS been found during this investigation that phthalic
anhydride can be produced in high yield by the oxidation of 0,3,4-bicyclonoqane and 0,3,4-bicyclononadiene (19) with air over a vanadium pentoxide catalyst.
0
0,3,4-Bicyclononane
Phthalic anhydride
0
0,3,4BicycIononadiene
Phthalic anhydride
Compounds of this structure are of particular interest, because substantial quantities of 0,3,4-bicyclononadiene have been found in a n unsaturated stock produced from high temperature 1 Present address, Pan American Chemicals Division, 122 Eaet 42nd St., New York, N. Y.
cracking operations. The 0,3,4bicyclononadiene, which ias a ring system free of aromatic rings, was found to be readily converted t o phthalic anhydride in high yield by oxidation with air over a vanadium pentoxide catalyst. Such a reaction necesaitates the dehydrogenation of the six-membered ring t o an aromatic ring accompanied by the oxidation of the five-membered ring t o the anhydride structure. Hexahydroindane (0,3,4bicyclononane) produced by hydrogenating t h e corresponding diene was oxidized as indicated above t o phthalic anhydride but in considerably lower yield than the 0,3,4-bicyclononadiene. These data show certain naphthenes t o be satisfactory charge stocks for producing phthalic anhydride in high yield. Numerous references in the literature describe the vapor phase oxidation of such aromatic compounds as o-xylene (9), naphthalene (11), and methylnaphthalenes ( 1 1 ) t o phthalic anhydride. These oxidation reactions are carried out using air as a source of oxygen and vanadium pentoxide or a modified vanadium pentoxide as a catalyst a t temperatures ranging from 275’ t o 500” C. The reported yields of phthalic anhydride from the oxidation of such aromatic compounds range from 40 t o 90 weight % ’ or 28 t o 78 mole 70,depending on the particular catalyst and conditions of reaction. Tetralin (6) is likewise reported t o yield phthalic anhydride when oxidized in the vapor phase over a vanadium pentoxide catalyst. Such a reaction is of interest in t h a t phthalic anhydride would reault from t h e direct oxidation of the nonaromatic ring t o the anhydride structure or by oxidation of naphthalene re-