V O L U M E 2 2 , NO, 2, F E B R U A R Y 1 9 5 0 ~-
~
‘Table 11.
Sairiple
I C
S
OH Group, iler Glucoae Unit
283
~
~~
__
~
___
-~~
gioups \vhich caniiot lie tletermined convenientl\r hy chemiral method5 the slwtioscopic nicxthod ma\ b(, adv~iit~tgeous.
Determination of Extinction Coefficient ‘loI
)
~
1~s I ’ ~ ui
lw (Io
I)OH log (10I)cii 0 0 0 0 0
500 505
490 490 433 0 228 0 292 0 3396
-,
log ( I o / ~ ) o H b* 1 13
1 21 1 11 0 814
0 0 0 0 0 ~
805
617 603 294 ~
~~~
07 11
1 13 1 05 1 05
295 ~
05 07
1 1 1 1
~
iCKYOW LEDG\ltYT
The author i b indebted t o Helen Johns fot pieparing tlic samples and operating thv spectrometer, to .\I. L. Wolfi.om, of Ohio Statr Universit),, for supplying most of the sugar Poiit cfe Srmours ~k (‘ornpsny for tllc-
(Bther or in water and then drl-ing again, t,he curves obtained no longer showed the carbonyl band. It, is possible that the ether and the water are also retained by the film, but they are not readily tletected from the infrared curves. Soaking the film in hexane and tlrying did not result in the removal of the ?thy1 acetate.
samples. and t o 1’. I. dii nitlocellulosc~saniplt~s.
The wsults o1)taiiicd fr,oni eight diffei.ent samples are giveii in Table 11. Thcsr films were soaked in chthcr. The variation in k’ which anitrunts to ahout SC;, is probnhly due to the variation in the amount of solvcsrit retained t)y the films. The nitrate group (*an be detc~rniirid c.oiivc.nic,iitly 1)y chcmical methods, but for
( 1 ) Barilea, It. B., Gore, H. C . , Liddel, Uriier., and l\’illiaiiis,\-, Z.,
LITEK4’I‘IJKE (;i‘l‘EI)
”Infrared SgectroscopJ-,” New Yolk, lieinhold Publidiing Corp., 1944. ( 2 ) Barnes, H . B., Gore, It. C . , Staffed, i t , W., u i d Willimiis, V. Z., .~s.LL.
CHICK,20, 402 (1948).
J, force, and aniplitutle over its operating range. The effects o f operational Iariables were iniestigated: frequencj, duration of test, and static and d j namic load. The ratio of “static” to “d?naniic” modulus depends more on aniplitude than on frequencj Contrar! t o usual ideas, the Etatic niodiilus actuall?
.
R
013,ICr (8)tlt~vc.lopeda m:whinr, for (1ett~i.niiiiingt h r hysteresis and dynamic modulus of rubber compounds to fill a need in t8he(krniaii program of synthetic ruhb-r development which, i n this country, n-as satisfied by a variety of generally less elaborate test methods ( 2 ) . The broad purpose of these t,esting proc d u r c s is t80 furnish an evaluation of rubberlike performance under conditions of low strain amplitude and high rates of &.formation, found in many impoltant applications of rubber, such as tires. The Roelig machine represents a highly specializcd German technic,al development and n-as not duplicated in t,his country during the war. At the conclusion of the war, interest in the machine led to the formulation of plans for securing scveral of them from Germany. These plans could not be consummated, hut. eventually a set, ot blueprints was secuwd, t,ogether with the dynamometer, which is the hcart of the measuring system.
m a ) be greaicr than the dynamic niodulus if the amplitude of the static deflection is sufficientl) smaller than the dynamic amplitude. Data were seciired for 11) steresis determinations for sj nthetic pol? tilers arid Hetea compounds. ’l’eniperature rise during some of the tests w a s measured b? means of a needle thermocouple to compare with h >steresis Ialues froni the loop. Tests on the same compoiinds were made b) otherniethods of h?steresis eialuation. Ueterniinatioiib with the ( h o d : ear \. ibrotester, reborincl pendulum, and flexometers were compared with results front the Hoelig machine. lXfferences in the results could he reasonablj interpreted o n the basis of the different conditions of iiieasiirenietit a n d calculation. ‘The importance of understanding the significance of a particular method of h? steresis determination for an? specific application is thereh? einphasimcl.
T h t w blucsprints \ v r i ~trttnslattd ~ iiit,othe‘ Eriglisli system wit,li a niinimum of rt%\vorkiiigto w i i f o i m to machining prwticcls in this country, and t h c x iiiachinc. vas built in .iki,on. Provision was made for photographing thtx hystcw l00ps in c~)iitrastto thc: German practice of manually tracing them on paper. Upon conipletion it was w t up arid put into satisfactory operation with lit1 I(difficulty. Th(x results secured stxm comparable ivith thoscs (l(’scribpd bv Roelig ( 7 , 9). h Roelig inachint, has been rctported as under procurenleiit in Germany for use in the Rubber Research Laboratority at Cro>-don, England ( b ) . DESCRIPI’IOS OF MACHISE
The Goodvear-Roelig machine is a rather large imd at~vngly built apparatus. The niwhanisni, which itsclf ncighs sc
A N A L Y T I C A L CHEMISTRY
284
The horizontal component of any point on the tmce d l , therefore, he proportionnl to the applied lorce a t that instant; and thc vertical component of the same point will be proportional to the displacement a t that instant. A photograph of tho optical System is shown in Figure 4. Part of the 2 2.25 :3 3 Sulfur 1..i 1.5 1 I 1 Captax 1 , .i :3 -1 4 Stearic acid 25% diphenyl0.8 guanidine master batell Phenyl-2napthylaininr Pine t a r Softener (Paraflux) Tetramethyl thiuram disulfide Samples 1 through 12 are of a si)et,ial goiylll?r series for wliii~hadditional information is gi\.en in Table 11. Samples 13, 14, a n d 1 9 are Herea. Sample 16 is regular GR-S a n d sample 17 i q GR-I. ~
~
~~~~~
~
~
~~
~~~~~
~
I
I
I
I
010. I
I
I
8
(REF.
1
Figure 10. Flow during Periodically Intervuptetl Cycle
~~~
‘Tahle 11. Polymerization Data. Sample No.
I’olynier Type GR-9 QR-9 GR-S GR-S (+R-S (iR-S OR-S CrR-S GR-S
l’olymerieatioii Temp.. E’. 122
I’olyisopren?
131
I’olyisoprrnr PolylsolJrtne
41 14
Hydrocarbon Conversion. fll
61
60 iU
H0
::
122 121
122 122
72 75
I22
100
122
73 38 ?2 .>8
131 41
Raw Moonry Viscosity (212O F.) .5 8 .4 i d3 48 39 43 60 48 47 43
Ts2.8MIN.
200
L0s
,050. i
l
n
t
t
i
t
t
l
REF.
,001
A-
E-
CD-
O T 0 3 SECS. . S MIN. 2 MINS. 5 MINS.
INCWCS
Figure 8.
Relation of Dynamic Hysteresis Loops to Static Stress-Strain Curve
lipure 9.
Coniparison of Dynamic and Static~\lodIlllrs
20 cvcles per sccond. Using thc Goodycar-Rocslig machiricb this range has been extended don-nn-ard to 8.6 cycles per scxond. 13). special manipulation frequencies of a fraction of a cycle pvr second may be obtained. I n agreement with Stambaugh’s work it !vas found that in normal operation there is very little dependence of modulus on frc,qucncy. However, essentially static. measurements made with the Roelig machine and also with othri, apparatus show the static modulus to be lower than the dvnamic
Figure 12. Change in Dj namic Properties during Test
modulus ( 3 ) . To study this phcsnoiiicnoii scvcral special t,csr s nc’re dcviscd. Using the pr~~compression spring adjustment, thr load on :I sample was periodically increasd and instantaneous esposures oi’ I he film wcrc made at intervals of 0.2 minute. Then, after rcturning to zero compression the path of this stress-strain c u r w was rctraced, and several small amplitude dynamic loops were photog r a p h ( ~ i . An enlargcsd copy of this film is show11 in Figurrs 6. Points for thcl static caondition h a w been conncctecl t o form a cont inuous curve. Thc dynamic modulus, as indicatcd hv thcx steepiic’ss of the loops, is visibly greater than the static modulus. A particular point on this static curve iras chosvn for further cwmparative tests as illustrated by Figurtl 9. .i very short esposurc., 0.01 sccond, was made of this point. Tht.11 the sample was cluickly comprwsed about 0,090 inch and a second very short esposure made n-ithin 1 sccond. Without further cornpression, subsequent exposurcs were made after intervals of 6, 18, and 60 seconds. The decreasing height of thrse points indicates a definite, tirmiy of stress with time. A comparison with the slow speed dynamic loop below s h o w the 1-second static modulus to bc, nearly equal to the dynamic modulus. Static moduli calculatt,tl for points of suhsclqucnt, esposurcs arc1 considc3mi)lv lnwc~r.
V O L U M E 2 2 , N O . 2, F E B R U A R Y 1 9 5 0 .Ilt,hough from this it appears that the invariably higher dyiianiic modulus may be merely the result of an insufficient time interval for stress relaxation to take place, additional experiment’s, cspecially r5ith different amplitudes, indicate that the phenomena arc’ more cornplicatcd. Further investigations of the effect were made using “interrupted eyclt~s.“ This is the ttwn used for the type of h,ysteresis loop shown below the normal loop in Figure 10. The procedure for obtaining such a loop was as follolvs: Beginning halfway through the compressive half of a cycle, cluic-k compressive strokes of about 0.010-inch length were made. 1,ess than a second elapsed during the motion. After each stroke, however, compression was halted for about 10 seconds. The tlo\\-nward movement of the trace during this halt is a definite iiitlication of a dewease in stress, as may be expected from the re- d t s obtained in the static test of Figure 9. During the retraction half of the cycle just the opposite effect occurs. A sharp decrease i i i stress during the second of mot,ion is follo~~-ed by a slow rise in >tress during t h r 1i:ilt in deformation.
the clffect of the duration, or running time, of the test. Pictures may be made a t intervals throughout the test and studied later. Figure 12 illustrates this point. K h a t appears to be a series of concentric loops a t the lrft of the figure is actually the trace tormed as the machine starts from rest and comes up to normal speed. The group of three loops to the right shows how this modulus and damping gradually dwrease as time passcss.
b
I
100
I50
DYNAMIC S T R E S S - P S I 38
Results such as hvre shown are illustrative of thci phenomena know1 as “delayed elasticity,” “time effect,’’ etc. They offer >motherapproach to a problem which is receiving the attcntion of workers in tho fic~lduf fibxs ( 6 ) as well as rubber ( I ) . .in inttwsting (:omparison of such an interrupted cycle to a vt’ry slow speed dynamic loop is shown in Figure 11. The loops are shown separ;Lt,cly and also in a superimposed position. This shows t h a t thc, dynamic loop represents an avcragc’ of the extrvnics in st,rosc:of th(, intvrruptrd cyclr. EFFECT O F DURATION O F T E S T
The method of obtaining data bv means of a photograph of the Iiysteresis loop makm tht, machine particularly useful in studying 2900
-m
t26
2800
IY
v)
3
2700
24
0
l 0
r
5
?
x
2 0
22
2600
g 4
20
2500
I
0
I
IO
I
I
I
20
30
DURATION -MINS.
Figure 13. Change in Dynamic Properties during Test
/ 4000
5
I 4 0
36
35 50
IO0 DYNAMIC STRESS-PSI
IS0
Figure 15. Effect of Increasing Dynamic Stress
Data from a similar series of loops have been plotted to form the curves of Figure 13. Ca1uc.s of both modulus and damping arc highest during the first fen. minutes of the test but decrease rapidly and approach equilibrium conditions after 10 t,o 20 minutes. .Ut,hough some of this change is undoubtedly due t o temperature increase in the sample, a considerable part is probably due to orientation or structural viscosity! a phenomenon similar in an estreme esnmplo to the thixotropic csffcct found in gels. Aside from such infoi,mation of theoretical in obtained from study of this type of test, there is thc practical consideration of the question of how long to run samples in rouiincl tcsting. I n the interest of cxconomy of time it is dtisirahlt. to vibrate each sample only as long as necessary t o get reliable results. Data from a numbcr of such tests indicatc that a running times of 10 minutes gives consistent results.
n
2400‘
a
37
37
I
I
IO0
I I I I50 200 250 STATIC STRESS-PSI
I 300
-
36
Fipirrr
I IO0
14.
I I 150 200 250 STATIC S T R E S S -PSI
0
Effect of Increasing Static
300 ...
Stress
EFFECT O F STATIC LOAD
The static load, or preload, applied to the sample may be varicd over a wide range. In tcsting a stiff compound the static stress may be made as high as several hundred pounds per squarcl inch irithout undulv distorting the sample. Results of a test in which the stat,ic load was systematically increased while the dynamic load \\-as held constant are shon-n in Figure 14. Modulus is afftscttd much more than damping For this sample. EFFECT OF DYNAMIC L o A n
An increase of the dynamic load produces an effect, on modulus which is just opposite t,hat o b t a i n d by increasing the static load. This is shown in Figure 15. The &crease in modulus is characteristic of loaded stocks. The increase in damping shown by this sample does not occur in all compounds. Damping, which is a function of both internal friction arid modulus, may rithri, increase or decrease, depending on the rates of change, of these quantities. The decrease in modulus n-ith increase in dynamic stress, and consequent increase in amplitude, has been obsxved by other workers ( 3 , IO). Dynamic stress seems t,o have a more pronounced effect than does frequency. Evidence of this is shown in Figure 16. The small loop was traced n-hile the sample \\-as being disformed a t a linear velocity of approsimately 0.4 inch per minute, as compared t o the large loop which was t r w c d while t h r same
288
ANALYTICAL CHEMISTRY
sample Fas being deformed a t an average rate of over 100 inches per minute. For the dynamic loop, of course, the velocity is not uniform, Because of the large difference in amplitude thc static modulus is actually greater than the dynamic modulus, which is just the reverse of usual observations. The sample was a t its equilibrium running temperature when these loops n‘ere photographed.
REF.
.050“
Figure 16. Effect of Amplitude arid Frequency
For the latter it is
H , 0: (100 - R)EX* relative heat genrration R = dynamic resilience E = dynamic modulus F = dymamic stress X = amplitude
where H
=
For puipohes of comparison, a sample of Hevea tread stock having a modulus of 918 pounds per square inch and a resilieiicc, of 41.8y0was arbitrarily assigned a heat genelation value of 100 units. Other samples nere thcn compared to this standard. Table 111 lists the results of such calculations for the group of special synthetic stocks and one natural rubber compound for both the Roelig machine and the Vibrotester (4). In the same tablrl are also shown temperatuie rises measured during btandni d flexometer tests. Calculated relative values of heat generation have been plotted against actual measurements of tenipcratule rise in Figure 18. Temperature readings nere made using ihc needle thermocouple at the time the hysteresis loop was photographed. Thus data from tlic hysteresis loops niav bc used nith
The effect of dynamic stress, or amplitude of oscillation, whicli may include structural effects due to orientation viscosity must Ix* considered along with the “delayed elasticity” effect in forming any theories to explain the difference between dvnaniic and static> moduli.
0
HEAT GEZIERATIOh
One of the most important applications for hysteresis deteiminations is to enable an estimation of the temperature rise of a compound under various circumstances of use.
a
E
m E
a 100
80 60
0
10
20
30
40
50
60
70
80
T E M P E R A T U R E RISE - O F
Figure 18.
40
Calculated Heat Generation Temperature Rise
IS.
JZeasured
20 Table 111. Heat Generation IO 20 30 DURATION M I N S . Figure 17. Temperature Rise in Tread Stocks during Test 0
-
Temperature rise in the sample in the Roelig machine may be easily measured while the test is in progress by inserting a needletype thermocouple. In Figure 17 are shown curves giving the temperature rise during tests of four compounds. The increase in heat generation brought about by the addition of carbon black is evident in the two curves for Hevea, where the only difference between the two compounds is the addition of 50 parts of EPC black to one. The butyl compound, which has very low resilience, shows the largest temperature increase. This type of curve has been thoroughly d k u s s e d by Springer (9). Because of the difficulties and uncertainties in calculating the values of quantities to be used in Springer’s equation, a relation for relative heat generation based upon theoretical considerations and measurements of the hysteresis loops may be used instead. Two expressions are used, depending upon whether the samples are vibrated a t constant dynamic stress or constant dynamic amplitude. For the former the expression is (100 - R ) F2 H/ o: E
Batnlile S o . 1 2 3 4 5 6 7 8 9 IO 11 12 13 Standard
Calculated Relative Heat Measured Temperature Rise, Generation, N, Flexometer, I?. Roelig Yihrotestrr Goodyear Goodrich 224 224 141 .5 5 396 63 266 3’62 iis ?! 242 260 135 aa 238 254 145 56 221 249 137 56 229 254 135 57 247 298 158 236 258 147 rjb 168 I96 121 56 221 233 153 60 200 219 125 54 137 192 100 40 100 100
Table IV. Dynamic nlodulus Sample S o . 1 2 3 4 5 6 7 8 9 10 11 12 13
-
Roelig, L b . / S w a r e I n r h Constant Constant stress amplitude 2690 1830 9240 2980 3540 2360 2880 2010 2930 1940 2110 1845 2800 1830 2490 1780 2660 1960 1265 1270 1650 1520 1900 1725 1730 1600
Vibrotester, Pendulum Lb./Square Rebound, Inch Inch Indentation 2030 0.193 0.143 2495 0.179 2155 0.188 2020 0.187 1915 0.193 1960 0.193 2200 0.184 2015 0.191 1445 0.222 1665 0.204 1780 0,210 1715 0.229
_______
V O L U M E 2 2 , NO. 2, F E B R U A R Y 1 9 5 0
289 A4greenient between Roelig and Vibrotester results is good for
'lahle Saniplf. So. 1
2 3
4 6 7 8 R
10 11 12 18
V. Dynamic Resilience
Roelie, % Constant Constant stre57 amplitude 33. b 32.6 33.4 26.6 40 0 37.6 36.8 33.7 32.4 32.6 32 8 33.8 26.9 30.9 23.8 23.1 33.4 31.7 26 9 28.8 21 3 19.7 34.6 .52,1
36.0 i S , '1
Yibrotester,
70
33.6
3i:z
Pendulum Rebound, c; Rebound Dynamic resilience resilience 29.8 18.8
28.0
29.2 30.9 29.8 29.2 28.t "3.8 28.1 23 8
22.: 32.0 88.2
1'1 (i 51 4 :3 !I , 1
.33.3 30.4 28.1 28.6 25.1 29.3
'lahle \ 1. I n ternal Friction Sample S o . 1 2 3
I
8 9 10 11 I:! 13
Vihrot es t er , Kilopoises 62.7 77:8 66.9 68.0 68.3 71.1 85.8
Roelii.,
1Glopoi;i.300 .i7 .5 334 317 318 29 1 313
378 812 243 858 2 .r,i 149
59.3 .,6, 6 f i g , (i 57.0 4li 7
modulus when amplitude effects are taken int'o consideration. Sarrow loops obtained for high resilience stocks such as sample 13 are difficult, to measure and consequently increase the error i n resilience values. Certain differences in the methods of expressing results must be considered in comparing rebound valurs. Although no exact modulus is calculated for rebound tests, tlic gtanerally used concept-that a smaller indentation indicates a higher modulus-may be used to get qualitative comparisons. Resilience as determined with a rebound test must be converted to a complete cycle basis before comparing it to resilience va1ut.s from thc dynamic testers. This is done bl- squaring the original result. When this is done, good agreement is obtained. Internal friction may be calculatc,d for both Roelig and Tibrotester tests, but is not usually found for rebound tests. The differenws sho~vnin the table are clue principally to the difference in frequent?.. Internal friction varies inversijly with frequcncy to a cloac appr(~xiination. ACKSOWLEDGJIENT
The authors wish to express their thanks to the Goodyear Tirc and Rubber Company and L. B. Sebrell for permission to publish this work. LITERATURE CITED
(1)
Aleksandov, A. P., and Lazurliin, Yu. S.,Rubber Chrm. and TechnoZ., 1 3 , 8 8 6 (1940).
good precision to estimate the relativv heat generation of conipounds in particular applications. CORREI,&'I'IO\ O F TESTS
.4lthough each of the many types of hvstere-ic t($stiiig ninchincs ha- unique characteristics which may make it niost suitable f o r use in obtaining particular information, i t is of interest, wheic. possible, to compare similar quantities measuied by them. Thi, has been done for the Roelig machine, Vibrotester, and rebound pendulum with the results listed in Tables I V , V, and VI. All tests n i t h the Roelig machine were made using a static compression of 9.6%. Constant amplitude tcsts nere run a t an aniplitude of =+=3.2%.Constant dynamic load tests \\-ere run under load of *41 pounds per square inch. All tests with the Vibrotester were run under static compression of 8%. -4 constant amplitude of *2.1% !vas maintained. 1-ibration frequencies were 12 cycles per second for the Roelig machine and 60 cycles per second for the Vibrotester. Pendulum Ic'bound testi: were made using standard procedure.
(2) Dillon, J. H., and Gehmaii,
e. U., India
R d h e r World, 115, 1
(1946).
( 3 ) Gehman, S. D., J . Applicrl Phys., 61, 94 (1942). (4) Gehnian, 9. D., Woodford, D. E., and Stambaugh, 13. B.. Incl. Eng. Chem., 33, 1032 ( 1 9 4 1 ) . ( 5 ) India Rubber J . , 114, 739 (1945).
Leadelman, H., "Elastic and Creep Propertie3 of Filamentous Jlaterials," Washington, D. C., Textiie Foundation, 1943. (7) lioelig, H., K a u t s c h u k , 19, 47 (1943); Rubber Chem. and TechnoZ., 18, 62 (1945). is) Roelig, H., "Proceedings of Rubber Technology Conference," p. 521, Cambridge, England, W.Heffer & Sons, 193s. ( ! I ) Springer, A . , K a u t s c h u k , 19, 55 (1943); Rztbber Chem. u n d Techt G)
nol., 18, 7 1 (1945). (10)
Stambaugh, R. B., Ind. Eng. Chem., 34, 1355 (1942).
RECEIVED .Trine i , 1949. i-try,
AxERIrAs
Presented before the Dix-ision of Rubber Chem-
CHEVICAL SOCIETY,a t Boston, Axass., May 1949. Con-
tribution 165, Research Laboratory, Goodyear Tire a n d Rubber Company. Investigation carried out under t h e sponsorship of the Office of Rubber Reconstruction Finance Corporation, in connection x i t h the governnthetic rubber program.
Acid-Base Determinations of Petroleum Products H. P. FERGUSON The Standard Oil Company (Ohio), Cleveland, Ohio Extensive data are presented to show that single-phase rather than two-phase equilibrium titration for determining the acidic and basic characteristies ot petroleum products is preferable.
F
OR years the determination of acidic and basic compounds in petroleum products has been a difficult analytical problem. Originally, the determination was made to measure the effectiveness of treating processes for removing the strong acids and bases used in refining. Later it became of value for determining acidic oxidation products formed during usage and addition agents of either acid or basic character. Hence the analyst is now confronted with the acid-base titration of nonaqueous solutions
which may contain organic or inorganic acids or base,, esters, lactones, phenols, soaps, resins, etc. The conventional analytical method has been extraction, without actual segregation of the layers, from the petroleum to an aqueous or alcohol layer and titration of the acid or base in the separated layer (1, 3 ) . Because of the poor reproducibility of this method, Committee D-2 on Petroleum Products of the American Society for Te-ting Iraterials has carried out an exteniive co-
