January, 1945
INDUSTRIAL AND ENGINEERING CHEMISTRY
67
Hycar compounds the value is 58', and with the Thiokol compound it is 60'. The temperature coefficient seems to be somewhat larger with the compounds which exhibit less swelling. The results with the Neoprene GN composition were compared with the reported values of Fraser ( 4 4 7 , and of Hanson (9) and an earlier study (3)from this laboratory (A-70 and B-70). These results are shown in Figure 6. Hanson used the maximum aniline point in his figures, but his results have been redrawn using the 50% aniline point. Since these compounds have various amounts of filler and other eompounding ingredients and since the conditions of cure vary somewhat, it is not surprising that the curves do not always coincide. It is surprieing that the slope of the curves is so nearly the same in almost every instance. This is in agreement with the finding above that the slope of the curve was independent of temperature, but also indicates that it is apparently independent of loading and normal variations of the degree of cure.
The hardness decreases in most instances with incrcased swelling, but there is a noticeable increase in some of the Perbunan and Hycar samples a t 100' C. This suggests further vulcaniration under the conditions of test.
PROPERTIESOF SWOLLEN SAMPLES
The authors wish to thank S. W. Eby for obtaining the swelling data and for determining the physical properties.
Tensile test pieces were immersed in the various liquids for 4 weeks, and tensile strength and hardness were determined in the swollen condition (Table 111). The tensile strength is computed on the dimensions of the unswollen sample to make the results more nearly comparable. It will be noted that the volatile swelling liquids generally reduce the tensile strength for a given amount of swelling much more than the nonvolatile oils. Low-molecular-weight plasticizers usually have a greater softening effect than those of high molecular weight. Fraser (6) showed k h a t volatile materials soon evaporate from Neoprene GN, the tensile strength reaching that of the unswollen composition. The tensile strength and elongation are not adversely affected by moderate swelling, but as swelling approaches 100% the values show an appreciable decrease.
CONCLUSIONS
The aniline point of hydrocarbon oils and solvents is a satisfactory index of the swelling of oil resistant synthetic rubber compositions. The logarithm of the per cent volume increase varies inversely with the aniline point up to 100% swelling. The slope of the swelling curve is apparently characteristic of each synthetic rubber and is not affected by loading, temperature or degree of cure. Slight swelling does not greatly decrease the tensile strength, but above 100% swelling the strength is greatly reduced. ACKNOWLEDGMENT
LITERATURE CITED
(1) Am. SOC.for Testing Materials, Standards on Rubber Products, p. 135,Designation D471-43T (Feb., 1944). (2)Ibid., Supplement, Part 111,p. 267,Designation 611-41 (1941). (3) Carman, F. H., Powers, P. O., and Robinson, H. A., IND.ENQ. CZ~EM., 32,1069 (1940). (4) Fraser, D.F., Zbid., 32, 320 (1940). (5) Ibid., 34, 1298 (1942). (6) Ibid., 35,948 (1943). (7) Fraser, D.F., Rubber C h m . Tech., 14,204(1941). ( 8 ) Geddes, B. W., Wilcox, L. Z., and McArdle, E. H., IND.ENG. CHEM,,ANAL.ED.,15,487(1943). (9) Hanson, A. C.,IND.ENQ.CHEM.,34, 1326 (1942). (10) Juve, A. E.,and Garvey, Jr., B. S., Ibid.,34,1316 (1942). (11) Son. of Automotive Engrs., Aeronautical Materials, Gpec. AM 322311 (1943).
HUGH WlNN AND J. REID SHELTON Case School of A p p l i e d Science, Cleveland, O h i o
HE fundamental nature of fatigue failure in rubber during flexing has been widely studied and is known to be complex. Among the more important factors involved (aside from variations in compounding) are state of cure, temperature of test, and oxidation. The object of this paper is to review and extend these studies as they apply to GR-S. Several investigators ( I , 6,6, 10, 11) have studied the effect of time of cure upon the flex cracking resistance of rubber. Busse (6) states that, as the degree of cure was advanced, the flex resistance reached a maximum and then decreased. Similar studies on GR-S (3,4,8,10) show that undercures flex better than either normal cures or overcures. Prettyman (IO), using a different type of machine, showed a maximum near the optimum cure. Cadwell and co-workers (8)and Rainier and Gerke (11) report that the rate of cracking of rubber increased with the temperature of flexing. Similar results have been demonstrated with GR-Sby Breckley (3)and by Carlton and Reinbold (4). Neal and Northam (9)flexed rubber in air, oxygen, and nitrogen on a Du Pont type flexing machine. They report no difference between the flexing behavior in air and oxygen. I n nitrogen, however, an uninhibited stock flexed five times longer
7
than in air before any failure was apparent, The presence of an, antioxidant did not change the behavior in nitrogen but doubled the normal flex life in air. These workers have concluded that the failure of rubber on flexing is due to oxidation rather than mechanical fatigue. No comparable work on GR-S has yet been reported. The present authors (1.9)previously showed that oxygen plays an important part in the hardening of GR-S vulcanizates during accelerated aging tests in the 100' C. air oven. Since high temperatures are developed during flexing which might well result in a similar hardening by oxidation, it seemed desirable to determine whether oxygen plays a part in the flex cracking of GR-S. For this purpose a typical tread stock was selected and flexed in air and in nitrogen which contained various small amounts of oxygen. The effect of state of cure and testing temperatures upon the resistance of GR-S to flex cracking was also included in the study. EXPERIMENTAL PROCEDURE
The tread type stock employed in this study was compounded according to the following formula from GR-S made according to
..
I N D U S T R I A L AND E N G I N E E R I N G C H E M I S T R Y
Vol. 37. No. 1
Tread type OR-S rtockt have been tested in (I De Metti. typc flexing machine. Curer included a marked undercure, an OV~ICYIO, and two intemrdieta CYI~S. Ti* were conducted at cabinet tempwetwes of ¶Os, N o ,W‘, and 80‘ C. The oxygem concentration ranged from that in eir to 0.05% by vdurnr in nitregen. Although a mahrd undersure Re& quite well, the Rex life of a GR-5 dock deerearn rapidly with longer curing time, until the optimum CUI. is reached and then remains esssantially constant. Flex life deoeares repidly with increasing temperetwq the 1.mperatuce coefRcirnt d this chansr C %.3 pet 10” C. over the m g r PO’ to 80’. The Rex life ir increased approximately 50% when oxygen i s present only in tr4cc11 the critical oxygen content seems to b e about 0.3% b y volume, in thatrlightly higherwluergive reruth cornperable to thorn obtained in air.
Figure 1.
Flexing Machine Equipped for Testing in Nitrogen
government specifications and oontaining spproximately 2 parts o i phenyl-B-naphthylamine: OR-S Channel black Bsrdol Fat *Oid
loa
50 5 1.5
Zinc oxide Sulf“, Gantoours
S 2 1.2
Regulation grooved test specimens were flexed in a De Mattia type flexing mnchine. Three specimens were used for each detennination. Vila (13) showed that the gr0wt.h of an initial cut is more reprnduoible than moasuroments of fiex life on unnicked speeimens. The present authors employed an automatio lance for the initial cqt to ensure uniformity. The rcsultiog nick was approximately 0.035 inch in length. In most of the work B duplicafe set of unnicked specimens w86 run simultaneaudy to determine the number of flexures required hoth for the sppearenee of the 6mt crack and for this crack to pragregs completely & c r m the one-inch specimen width. These values are reported in the t,ablesunder the headings “appearnnce of first cmek”, “first crack to failure”, and “total to failure”. In sli oaqes “fsilure” WBS taken as that point at which t,hocrack had progresed completely 8 ~ ~ the ~ xgroove. 8 When the variation between individuhl test ~pecimensexceeded B few thousand flexes,the average maximum difference is indicated in the tables. This number was obtained by dividing the maximum difference between specimens by the number included in the average. The flexing machine was enclosed in a case both for testing st elevated temperatures and in nitrogen atmaspheres as shown in Figure 1. Strip heaters and B fan were mounted inside the cabinet to maintain thc desired temperature. Thormumeteia mounted at specimen level were used to measure the temperature of the test. For those experimenls in which a reduced oxygen coneentrstion w m employed, nitrogen of appropriate purity was passed through the cabinet, at % constant rate, and analyses of the exit gas for oxygen weit’ made at frequent intervals by an adaptstion of the Winkler method (14). The average oxygen content reported for the test is the mean of the sepnrate analyses made during the test. An oxygen content of approximately 1% by
volume w m obtained by passing commeroial nitrogen through the cabinet at a rate of 0.25 cubic foot per minute. A purified nitro gen, analyzing O.OS% oxygen and -ing at a rate of 0.3to 0.4 cubic fwt per minute, gave an avemge content of0.4% oxygen; a rate of 1 cubio foot per minute reduced the average content to 0.2%. For still lower oxygen contents it WBS found desirable to employ B specidly prepurified nitrogen, supplied by the General Electric Company, containing 1- than 5 psrts per million of oxygen. Using this nitrogen at B rate of 1 cubic foot per minute, the average oxygen oontent of the exit gas w m 0.05% by uolme. Since this was sufficiently low to show 8 definite effect upon the flexing results, no attempt was made to reach alowar value. All ssmples which were to be flexed in nitrogen, together with their oontrols, were placed (immediately after vulcanization) in B desiccator over alkaline pyrogallol 80 as to remove, in 80 far as possible, the oxygen from the surface of the samples, STATE (x CURE
The flexing data for foiv different cures, including a marked undercure and an overcure, are presented in Table I and plotted in Figure 2. In the oaae of the unnioked 15-minute cure specimens, the teat waa terminated after 100,ooO flexes; at thst point two of the samples were only one half and one third failed, respectively. One hnd failed at S5,ooO flexes. The dashed portions of the ouwes in Figure 2 are estimated from the degree of failure at 100,ooO aod measured r a t e of crack growth. The % m e ashow that GRS has good flexing resistance when undercured, but that, as the cure is advanced, the flex resistance decreases rspidly and levels off at whet is frequently o m sidered an optimum cure. An overcure, however, produced little additional change in the flexing resistance. These results are consistent with those obtained by the workers previously cited. Although Gehmsn, Jones, and Woodford (7) did not m-m flex cracking direotly, they showed that for both rubber and GR-9 the heat generatimiduring flexing under compressiveload at constant defleation startg with a lower value for sn undercure and spproachea a limiting value ~9 time of cure increases. T h e lower temperature rise for theZlndercure may in part account for the better flexing properties. Another faetnr may he the small
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Yanuary, 1945
-
Table
Cure, Min. at
1.
Effect of State of Cure on Flexing No. of Flexurea in Thousands
Firat crack Total to To appearance to failure failure of first crack 0 15 17 42 * b 0 ao 12 7*4 18 * 6 2s * e 60 8 a 13 16 11 80 8 3 8 One of,the three strips failed at 86,000; one wan half failed and the third w.88 one third fatled after 100,000Besee.
2 9 8 O F.
Initial cut to failure
Temp. of Test, C.
II.
Effect of Testing Temperature on Flexing (5O.Mlnute Cure at 298' F.) 7 No. of Flexures in Thousands Initial cut To appearance First crack Total to
Table
-
40
to failure 30 4 17
80
7
23
eo
'I0
of Brat crack
to failure
failure
e
48 * 8
64 8 87 21 10
a2 1s
6
a 2
8
A different lot of GR-S was employed for the last three batches. This difference is reflected in both the hardness and flexing data. Small differences in cure also make it impossible to compare the different control samples directly. Since each batch had its control run in air, these variations do not affect the validitp of the results from the standpoint of the effect of oxygen concentration upon flex cracking. No uninhibited stock was included in this series. It has been repeatedly demonstrated in this laboratory, however, that the presence or absence of the antioxidant phenyl-&naphthylamine makes little difference in the flexing results obtained in air, and consequently none would be expected in nitrogen. A typical example from these unpublished data is presented in the following tabulation; the uninhibited polymer was prepared in the manner previously described (19): Uninhibited GR-8 Tread EItock
Same Stook with 2 Part. Antioxidant
10
13 13 26 37 80
No. of flexures in thowands Initial out to fdure To appearance of 5mt crack
amount of flow or permanent set which resulta in a lower s t r w during the testing of an undercured sample.
69
9 19 28 21 68
First crack to failure Total to failure Cure at 298O F., min. Shore hardneas
b?
The small difference in flexing behavior noted here may be attributed to the differente in cure resulting from the slower curing rate of the inhibited stock. I n other cases in which the inhibited stock waa cured long enough to give a slightly higher cure, the crack growth value waa actually lower than for the uninhibited stock. Similarly, the inclusion of 5 additional parts of antioxidant in a GR-S stock failed to affect the flexing results. This failure of phenyl-p-naphthylamine to affect the flexing resistance of GR-S is in contrast to the behavior with natural rubber (9). This difference may be explained in part by the much shorter duration of the test resulting from both the poorer flexing resistance of the GR-S and the type of flexing machine employed. The percentage of improvement in flexing resistance obtained by flexing in an atmosphere low in oxygen as compared to that in air is shown in Table IV. Since the flex life was the same with 0.4% by volume of oxygen as with air, whereas 0.2% showed a definite improvement, the critical oxygen content would seem to be about 0.3%. At the lowest oxygen content obtained, 0.05%, still greater improvement was noted in the total flex life although the resistance to the growth of an initial cut showed no additional improvement. Apparently, then, in a completely oxygen-free atmosphere, still greater improvement would be obtained.
TESTING TEMPERATURE
The result of flexing identical samples a t four different temperatures is shown in Table 11. These data are plotted as the log of the number of flexurek against testing temperature in Figure 3. Inasmuch as crack growth data are most reproducible, the straight lines drawn through the various points for the other measurements are parallel to the best straight line through the points representing the crack growth data. It is evident that testing temperature is an important factor and that the higher the temperature, the poorer the flexing properties become. The temperature coefficient obtained from the lines of Figure 3 is 1.3 per 10' C. change in temperature over the range 20" to 80" C. Although no values for this particular temperature coefficient have been reported previously, calculations based on the data of Carlton and Reinbold (4) show a temperature coefficient for this change of 1.5 per 10' C. change over arangeof approximately25' to 80' C. An estimate of the coefficient based on the work of Breckley (8) is 1.3 to 1.5 per 10"C. between room temperature and 100' C. The agreement is good, especially when it is remembered that Carlton and Reinbold used a Du Pont type machine whereas Breckley and the present authors used a De Mattia type. A coefficient of 1.3 would indicate a rather low energy of activation for a simple chemical reaction. Although a Table 111. Effect of Oxygen Concentration on Flex Life -N o . of Flexuree in Thousandmay be inFlaxed Oa Conon Tp& Initial aut Ap earanob of Fimt oraok Total to Cure Min cluded in this change, other in: % b y voi,' to failure &t crack to failure failure at 268' F : factors (physical in nature) Air 27 11 40 Nt 1:O' ao 9 would appear to be rate deterAir 27 10 40 MI 1.0 30 9 mining. 80
... ... ... ...
...
Alr
.
FLEXING IN AIR AND NITROGEN
E
N? Au
N¶
037 0:2i
d .'db
29 80 30 30 ao 80
The effect of oxygen upon the flex cracking rate of GR-S is shown in Table 111. I n each case identical samples were flexed, one set in air and the other in nitrogen having the aPPr0priate oxygen content. Duplicate batches were run a t the 1% oxygen level to show the reproducibility of results. No improvement in flexing characteristics was found 8t thie oxygen eoncentration.
-
lb
b
7 ia
16 23 30 22 28
18
18 7 7 12 Table
IV.
0, Content % by Vol:
0.a7 0.21 0.06
... ... ... ... 21
27 40 6 3b 6b
...... ... ...
2e 27 40 I 8 L 12 42 67
26
30
Shore Hardnma 64 64
61
67 59
Improvement in Flex Rerlrtmce in Low-Oxygen Atmosphere -Percentage Improvement : n i Initial O u t TO appearance Fmt crack Total to to failure of 5nt crack to failure failure 0 0 0 0 ao $9 48 4b a7 70 67 eo
70
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 37, No. 1 pleasure to give special thanks to 0. D. Cole of Firestone and to C. F. Prutton of Case School of App l i e d S c i e n c e for their many valuable suggestions. LITERATURE CITED
(1) Bierer, J. M.,and Davis, C. C., T r a n s . Znst. Rubber Id., 3, 151 (1927); Rubber Chem. Tech., 1, 148 (1928). (2) B ' r e c k l e y , J . , Rubber Age (N., Y.),53, 331 (1943); Rubber 3.2l I I 1 20 40 60 80 C h e m . Tech., Temperature, 'C. 16,901 (1943). (3) Cadwell, S. M., Figure 3. Effect of Testing Temperature Merrill, R. A., Minutes Cure At 298'E Sloman, C. M., and Y0st.F. L., Figure 2. Effect of Cure IND. ENCI.CHEF., ANAL.ED., 12, 19 (1940); Rubber Chem. Tech., 13,304 (1940). Although a n improvement of 60% falls far short of that re(4) Carlton, C. A,, and Reinbold, E. B., India Rubber World, 108, 141 (1943); Rubber Chem. Tech., 16,897 (1943). quired to make a GR-S stock comparable to natural rubber, (5) Caasie, A. B.D., Jones, M., and Naunton, W. J. S., Trans. Inst. these results do show that, even in the relatively small time reRubber Znd., 12, 49 (1936); Rubber Chem. Tech., 10, 29 quired for the De Mattia flexing test, oxygen reacts with GR-S a t a (1937). sufficiently rapid rate to affect the results markedly. With a less (6) Davis, C. C., and Blake, J. T., "Chemistry and Technology of Rubber", A.C.S.Monograph 74,p. 199,New York, Reinhold drastic test-for example, in actual service-more time would be Pub. Corp., 1937. available for oxidation and the effect upon flex resistance would be (7) GehmaA, S. D.,Jones, P. J., and Woodford, D . E., IND.ENQ. correspondingly greater. The reaction is probably localized a t CHEM., 35,964(1943). the points of mmimum stress where high temperatures are (8) Juve, A . E.,and Garvey, B. S.,Jr., Ibid., 36,212 (1944). (9) Neal, A. M.,and Northam, A. J . , Zbid., 23, 1449 (1931); Rubproduced by flexing. This behavior is consistent with the work ber Chem. Tech., 5,90 (1932). previously reported ( I $ ) , where it was demonstrated that oxygen (IO) Prettyman, I. B . , IND.ENO.CHEM., 36,29 (1944). played a n important part in the hardening of GR-S stocks at (11) Rainier, E . T . , and Gerke, R . H., IND.ENQ.CHEM.,ANAL.ED., elevated temperatures. 7,368 (1935);Rubber Chem. Tach., 9,178 (1936). (12) Shelton, J. R.,and Winn, H., IND.ENO.CHEM.,36,728 (1944). (13) Vila, G. R., Ibid., 34, 1269 (1942); Rubber Chem. Tech.. 16, ACKNOWLEDGMENT 184 (1943). (14) Winkler, L.W., Ber., 21,2843(1888). The authors are pleased to express thanks to The Firestone
Tire and Rubber Company for sponsoring this work and for permission to publish certain portions of the data. Hugh Winn is Firestone Fellow a t Case School of Applied Science. It is also a
PBE~ENTED M a part of a Sympoeium on Synthetic Rubber before the Engineering Beotion of the American Association for the Advancement of Science at Cleveland, Ohio.
LIGNIN ETHERS AND ESTERS Preparation from Lead and Other Metallic Derivatives of Lignin
N
0 GENERAL attempt has been made to prepare ethers
F. E. BRAUNS AND H. F. LEWIS The Institute of Paper Chemistry, Appleton, Wis.
lignin by diasoethane and diethyl sulfate and the preparation of mixed methyl-ethyl ethers. The of lignin as such, except for E. B. BROOKBANK usefulness of these reigents as those needed in the study of the The Mead Corporation, Chillicothe, Ohio g e n e r a l a l k y l a t i n g agents is structure of lignin. Methyl ethers havd been used as resins or limited by the fact that diasoalkyl fillers in redwood molding compositions and appear to be superior derivatives higher than diazoethane and dialkyl sulfates above the butyl derivatives are hard ,to prepare. Other means must to lignin alone; likewise, the reactions of methylation have been studied to determine the nature and number of hydroxyl groups therefore be utilized in making long-chain lignin ethers. The alcohol lignins may be considered RS lignin ethers. They in the lignin structure. Diazomethane has been used to methylate acidic hydroxyl groups; dimethyl sulfate and sodium hydroxdiffer, however, from the methyl and ethyl ethers mentioned beide methylate all other hydroxyl groups except the carboxyl fore in that two alkyl groups have entered the lignin building unit in an acetal-type reaction, and they are split off again when hydroxyls. Recently Jones and Brauns (8) described the ethylation of the alcohol lignin is treated with strong mineral acid. Tho fol-
