Long-time dynamic compatibility of two ethylene-propylene elastomers

these elastomeric materials over long-time periods are discussed. Introduction ... elastomeric materials in liquid hydrazine during short-time tests t...
0 downloads 0 Views 2MB Size
Ind. Eng. Chem. Prod. Res. Dev. 1984, 23, 311-315 Robinson, W. E.; Heady, H. H.; Hubard, A. B. Ind. Eng. Chem. 1053, 45(4), 788-91. Robinson, W. E.; Lawlor, D. L.; Cummins, J. J.; Fester, J. J. “Oxidation of Colorado Oil Shale”, US. Bureau of Mines, Report of Investigations No. 6166, 1983. Saxby, J. D. “Chemical Separation and Characterization of Kerogen from Oil Shale”, I n “Oil Shale”; Yen, T. F.; Chllingarian, G. V., Ed., Elsevler: New York, 1976. SchmkltCdierus, J. J.; Bonomo, F.; Gala, K.; Leffler, L. ”Polycondensed A r e matlc Compounds and Carcinogens In the Shale Ash of Carbonaceous Spent Shale from Retorting of Oil Shale”, In ”Science and Technology of 0 1 1 Shale”, Yen, T. F., Ed., Ann Arbor Scientific: Ann Arbor, 1976.

311

Schmidt-Coiierus, J. J.; Prlen, C. H. “Investlgatlons of the Hydrocarbon Structure of K e r w n from 011 Shale of the Green River Formation”, I n “Science and Technology of Oil Shale”, Yen, T. F., Ed., Ann Arbor Sdentlflc: Ann Arbor, 1976. Smith, J. W.; Hlgby, E. W. Anal. Chem. 1060, 32, 1718-9. Yen, T. F.; Chlllngarlan, c. V. “Introduction to Oil Shales”, Chapter 1 In “Oil Shale”; Yen, T. F.; Chlllngarlan, G. V., Ed.; Elsevier: New York, 1976.

Received for review October 31,1983 Revised manuscript received December 8, 1983 Accepted December 27, 1983

Long-Time Dynamic Compatibility of Two Ethylene Propylene Elastomers with Hydrazinet Cllfford D. Coulberl, Edward F. Cuddlhy, and Roberl F. Fedorr’ Jet Propulsion Laboratory, California Institute of Technology, Pasadena, Callfornk 9 1 103

A test method is described for predictlng the long-time survivability of elastomers in hydrazine under dynamic stressing conditions. The method selected Is based upon the existence and approximate invariance of a “physical property surface” relating the mechanical response of the elastomer In terms of stress, strain, time, and temperature. The property surface was generated for two selected elastomers (EPT-10 and AF-E-332). A novel carousel testlng tank was designed to allow sequential testlng of eight tensile specimens immersed in IiquM hydradne within a constant-temperature water bath. The test procedure and data reduction methods used to generate the property surface are described. The utility and validlty of these results applled to fatigue and flexure loading to these elastomeric materials over long-time periods are discussed.

Introduction Elastomeric materials in contact with liquid hydrazine may experience changes in physical properties as manifested by swelling, reduction in tensile strength, and possible leaching of constituents (Coulbert and Yankura, 1972; Martin et al., 1971). When these elastomers are employed as components subject to cyclic loading such as propellant tank bladders or diaphragms, it is important also to be able to predict their long-time survivability under dynamic stressing conditions in hydrazine. This paper describes a technique for characterizing elastomeric materials in liquid hydrazine during short-time tests to provide a basis for predicting the useful life of bladders or diaphragms subject to multiple propellant expulsion cycles over long time periods. The method is applied to those candidate elastomers which have survived a preliminary compatibility examination, typically static immersion at various temperatures in hydrazine with inspection for visible evidence of incompatibilities, and measurements of any swelling or softening of the elastomer (Boyd et al., 1965). In normal practice, the elastomers surviving the compatibility screening are removed from the hydrazine and tested for changes in tensile strength, modulus, elongation-at-break, etc. However, this approach only evaluates changes from immersion, if any, by comparison of properties before and after, but does not provide information on the level of elastomeric properties while in the hydrazine ‘This paper representa one phase of research performed by the Jet Propulsion Laboratory, California Institute of Technology, sponsored by the Materials Division of the NASA National Space Flight Center, Huntsville, AL, and performed under NASA Contract No. NAS7-100. 0196-4321/84/1223-0311$01.50/0

environment, nor give information which directly allows for a prediction of dynamic lifetime.

Method for Predicting Long-Time Behavior The approach selected as a bisis for predicting long-time behavior was based on the existence and approximate invariance of the tensile property surface for elastomers. The concept of the property surface is based on the fact that the mechanical response of an elastomer can be considered in terms of the three variables, viz., stress (a), strain (E), and time (t),and hence a convenient representation is an analytic surface in stress, strain, and time space. The effect of temperature is incorporated into the time scale by the relationship log t / u T , where the value of log uT is the experimentally determined time-temperature shift factor. The development of a tensile property surface which relates these parameters by a single analytic surface with a single rupture failure boundary is described by Landel and Fedors (1964). A typical tensile property surface for Viton B rubber in air taken from the reference is shown in Figure 1. The surface is conveniently generated by measuring the uniaxial stress-strain response of the given elastomer as a function of strain rate and temperature. Typically, it is necessary to test the elastomer at about 10 different strain rates at each of 10 to 15 test temperatures. Although this represents a considerable expenditure of time and effort, knowledge of the surface is of fundamental importance in understanding and predicting the mechanical response of an elastomer. The response of an elastomer to any uniaxial input will simply be a path traced out on the surface. For example, the path traced out during a creep experiment (i.e., constant load) starting a t point A in Figure 1and terminating a t the point B will be the curve AB which is generated by the intersection of this surface with a plane parallel to the 0 1984 Amerlcan Chemical Society

312

Ind. Eng. Chem. Rod. R B ~ Oev.. . Vol. 23, No. 2, 1984

In many applications, the elastomer is subjected to biaxial deformations (e.g., flexing and folding) while the experience presented above was limited to tests carried out in uniaxial deformation. This again is an area where more work can profitably be carried out. However, it can be said that the failure time in biaxial deformation can never exceed that under conditions of uniaxial deformation. Hence, the results obtained with the uniaxial test can be used to assess the promise of a candidate material. For example, if the candidate material does not have sufficient lifetime in a uniaxial test, it should not be a candidate for an application involving biaxial deformation. With test data obtained over a sufficient range of temperatures and strain rates, the time dependencies of both the stress and strain-at-break may he readily determined for up to 13 decades of time (i.e., times from less than 1 s to more than 20 years). The test temperatures must be above the glass transition temperature but below the temperature at which chemical degradation would be rapid compared to the time scale of the experiment.

Figure I.

strain, time plane and which passes through both A and B. The projection of this path to the strain-time plane is depicted by the lower curve A B this latter curve corresponds to creep data as normally plotted. Similar consideration can be used for experiments carried out under other test modes. Thus, the path traced out on the surface by a streea relaxation (i.e., fixed strain) experiment carried out by starting from the same point A used for the creep experiment is shown in Figure 1 as the curve AC. Rupture on this surface is represented graphically as a discontinuity or boundary which can be considered to result from the intersection of the physical property surface with another surface representing failure. In Figure 1this rupture condition is represented by the terminating line on the right. Hence, the specimen undergoing creep and following the surface along path AB will break when this path intersects the boundary (at point C). From the general nature of the surface, it is evident that starting from the same initial point A, the creep path will reach the boundary before the stress-relaxation path and hence, the lifetime under creep conditions will be less than that under stress relaxation conditions. It has also been found experimentally (Landel and Fedon, 1964) that fatigue life in uniaxial tensile tests bears the same relationship to the property surface as the other modes of testing. In Figure 1a fatigue trace would have a saw-tooth waveform that moves a c r m the surface until it reaches the failure boundary. The implication of this criterion is that fatigue is less dependent on the number of cycles than on the total time under load. The effect of changes in the cross-link density of the elastomer on the location of the property surface has a h been studied for three chemically different types of rubber, viz., a hydrocarbon, a fluorocarbon, and a fluorosilicone rubber. For all three types, the effect of a change in cross-link density is simply to change the time scale of the experiment analogously to the change in time scale observed when the temperature is changed. For example, if the cross-link density were inadvertently increased by 20%. the lifetime under stress would decrease by a factor of 4. Such changes in cross-link density could occur due to variations in material formulation and processing conditions.

Experimental Program Approach. The objective of this program was to determine the property surface in hydrazine for two candidate elastomen in order to m s s and rank their potential for a long term dynamic application. The selected elastomers were two ethylene propylene terpolymers designated EPT-10 and AF-E-332. The ability to develop a complete property surface in hydrazine was limited by the limited range of experimental test temperatures available for conducting the tensile tests in liquid hydrazine, which were temperatures between the freezing point, 274.5 K, and the boiling point, 386.5 K. For reasons of safety and limited test experience, the actual test temperatures used were limited to values between 275 and 341 K. The experimental approach was to establish the property surface for each material first in air over a larger temperature range (233 to 453 K) to provide a basis for evaluating the influence of hydrazine immersion and, if possible, a basis for extrapolating the hydrazine property surface data. Thus, rupture data obtained in less than 2 h per specimen at a maximum of 341 K in hydrazine would be used in determining the time dependence of rupture for 10 years of operation a t temperatures between 275 and 348 K in hydrazine. Specimen Preparation. A single batch of each elastomer to be evaluated was ordered from the manufacturer in the form of 152 X 152 X 1.96 mm sheets of a sufficient quantity to complete all the contemplated tensile testing and other related tests such as swelling, stress relaxation, and folding tests. The materials were purchased from TRW Systems of Redondo Beach, CA. The test specimen rings were cut from the sheets by use of a rotating two-bladed cutter which cut the inside and outside circumferences simultaneously. The test specimen ring size was 19.2 mm i.d. X 22.3 mm 0.d. After cutting, the rings were individually measured, weighed, and bagged for subsequent testing. Approximately 300 rings of each material were thus prepared. Tension Tests in Air. Tension tests in air were conducted in a controlled temperature enclosure mounted in an Instron tensile test machine. Twelve test temperatures were selected between 233 and 453 K and controlled by circulating heated or cooled air through the enclosure. The cross head speeds were set at ten values from 8.5 to 0.0085 mm f s. During testing the test rings were mounted over hooks formed from 4.7 mm diameter steel rods. The recorded data consisted of load/time curves on a strip chart. The initial point on each load/time curve was determined

Ind. Eng. Chem. Prod. Res. Dev., VoI. 23,NO. 2, 1984 313 I

CONSTANT TEMP. BATH AROUND

-

2

b

- 2 . 1 Ikq 0 I\', 1 mn2

,

,

I

I

I

I

,

,

,

/

I

L-L 1kq I~l',', mo

Figure 3. Illustration of the superpositionof strew rupture data by shifting horizontally by factar log (IT.

Figure 2. Turret-action hydrazine ring test fixture.

by extending the initial straight portion of the curve backward to the zero load line. This correction was employed because the rings started out circular and the force required to straighten the ring caused an indeterminant s t a t point for the load/time curve. Since the elongations a t rupture were in the 250 t o 400% range, the effect of initial differences in i.d. and 0.d. stresses were assumed to be negligible. Selected noints from the loadhime curves were tabulated and used in subsequent computer data reduction calculations. A total of 120 rings were tested to rupture for each material. Tension Tests i n Hydrazine. Conducting a large number of tensile rupture tests in liquid hydrazine at elevated temperatures required special test planning considerations. Liquid hydrazine (N,H,) is a reactive compound used as a monopropellant and also a a bipropellant in combination with an oxidizer such as nitrogen tetroxide. It is thermodynamically unstable and is very sensitive to trace amounts of catalyst which cause it to exist in a state of continuous decomposition. This decomposition rate is a function of both temperature and the type or amount of catalyst which may be present. At ambient temperatures -293 K and in the absence of a catalyst, however, the average decomposition rate of NzH4is minimal. Although liquid hydrazine has been heated under .pressure to temperatures above 533 K with very little decomposition, the hydrazine vapors can present a hazard. Air saturated with hydrazine at any temperature above 313 K is a flammable mixture. Therefore, the testing approach which was developed utilized a specially designed carousel, or turret, arrangement (Figure 2) which provided the capability for testing eight ring specimens in hydrazine as a single operation at one temperature with minimum handling and exposure of the hydrazine. The design consisted of eight individual AISI 304 steel test chambers filled with hydrazine and contained inside a larger water-filled tank mounted in the Instron tensile test machine. In this test setup, ring specimens were mounted individually on AISI 304 steel hooks inside each of the eight test chamhers. The shafts of the upper books extended through caps on the top of each chamber to permit attachment to the Instron loading grip as each specimen chamber was successively rotated into test position. The eight chambers were mutually interconnected by a common hydrazine feed manifold which had an external entrance and exit port. After the eight spec-

imens were mounted in the chambers, the chamhers were evacuated and hydrazine was drawn into the chambers filling them all simultaneously to the same level from a common hydrazine supply. During testing, dry nitrogen gas was passed through the hydrazine manifold and the ullage space in each chamber. Temperature control was achieved by circulating water through the outer tank from a thermally regulated externally located separate water bath. Temwratures could thus be set at values between 275 and 368 K. This test arraneement allowed the test chambers to be ffled once each diy with hydrazine and heated (or cooled) to a selected temperature. Then the eight specimens were tested in tension successively at eight selected cross-head speeds from 8.5 to 0.085 mm/s. All of the ring specimens tested in hydrazine were presoaked for a minimum of 48 h in hydrazine at room temperature in order to reach equilibrium swelling. Preceding swelling studies showed this to be a sufficient swelling time. The recorded data were handled similarly to those for the air test data. The five selected temperatures were 275, 291,311,325, and 341 K. Thus 40 ring specimens of each material were tested to rupture in hydrazine. D a t a Analysis The load/time data from the Instron strip charts were tabulated for each ring and used as the input to a data analysis computer program for all subsequent calculations. From the computer program output, plots were prepared to display graphically on translucent graph paper the relationships of log (ob297/T) vs. log tb and log tb vs. log t b for each specified temperature, where ub = tensile stress at failure, N/mZ,T = test temperature, tb = time to failure, min, and cb = strain at failure. After the individual plots for each test temperature were prepared, the series of plots for each material were placed on a light table on top of each other and shifted along the time (log tb) axis until a single (best fit) master curve was derived (Landel and Fedors, 1964). As with similar data for other polymeric materials, it was found that a number of iterations were required to achieve a reasonable fit because of the considerable data scatter. To illustrate the principle involved, the two plots in Figure 3 may be studied. The plots on the left represent the strains and times at failure for each temperature level. The lowest curve represents data taken a t the highest temperature. All test times lie between 10 s and 100 min. In the figure, a t the right the data have been shifted horizontally with respect to the reference curve at 310 K to form a single curve. The amount of shift is log aT,so that the final cnme is plotted as log % vs. log &,/aT). The final

Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 2, 1984

314

I 1 I

I

I

1

14-a

I

I

I

I

I

I

1

l

l

A F - E - 3 3 2 IN AIR A F - E - 3 3 2 IN HYDRAZINE

0

I

I

1

0

0

*"# om

- " I

1 1 , - 4 - 2

l

0

l

I

/ 2

l

4

-

l

I

1

I

l

1

l

l

- 4 - 2

6

l

3

0

%*

l i

1

I 1 I I I l EPT-10 IN A#? E P T - I O IN HYDRAZINE

I 1

14-bi

l

I 2

0

I

I

~

,

l

6

4

lq !tbfaT), min

Iq Ib/aTl, min

Figure 4. Tensile stress failure envelope for AF-E-332 and EPT-10 elastomers tested in air and in hydrazine at constant strain rates. 15-bl

E P T - 1 0 IN AIH E P T - 1 0 IN HYDRAZINE C FIXED STRAIN IN AIR

0

A F - E - 3 3 2 IN HYDRAZINE 0.8

0.2F

l - 4 - 2

l

,

l

O

2

,

,*3i a O'Po~,l

4

6

c

"1

-

- 4 - 2

0

2

4

6

(bfaTl,min Figure 5. Tensile strain failure envelope for AF-E-332 and EPT-10 elastomers tested in air and in hydrazine at constant strain rates. lq !biaTl min

lcq

values of log aT were obtained by shifting both the strain and stress data with the requirement that the relationship between log a T and test temperature be a continuous smooth function of the form

were To= the selected reference temperature, K, T = test temperature, K, and C = experimentally determined constant for the material tested. The actual data taken for EPT-10 in hydrazine showed less scatter than the data taken in air and less than that for the AF-E-332 material in either air or hydrazine. Figure 3 presents the time dependence for the breaking strain for EPT-10 before and after shifting by the log UT factor. The final plots of the time dependence of the breaking stress and strain for both EPT-10 and AF-E-332 in air and in hydrazine are presented in Figures 4 and 5. A single equation for the value of the temperature shift factor, log UT,found to be satisfactory for correlating both materials in air and in hydrazine was

I t is possible that slightly different constants might be established for each material if the data scatter could be reduced. Supplementary Testing One of the more striking observations made from comparing the EPT-10 data taken in air with those in hydrazine was that while there was a definite reduction in

tensile strength and a concomitant increase in the breaking strain in hydrazine, for longer times and higher temperatures the strain data in hydrazine coincided with the air test data. The significance of this is that the long time strain to failure for EPT-10 is the same in both air and hydrazine. These data, therefore, provided a potential basis for long time failure prediction. However, to assure the validity of these relationships, selected additional tests were designed to produce failures in somewhat longer test times to check the effect of temperature and low strains on the location of the lower end of the q, vs. (tb/aT)curve. Two types of supplementary tests were conducted. These tests were conducted in air since the same t b vs. ( t b / a T ) relationship was obtained in both air and hydrazine. In one test series performed in air, ring specimens of EPT-10 were stretched between a series of fixed rods mounted on an aluminum block in such a way that fixed strains of 100,125,150,200, and 250% could be imposed on the rings. The rings were stretched between the rods, several a t each strain value, and the whole assembly was placed in a preheated air oven at temperatures of 343,373, and 403 K. The combinations of strains and oven temperature were chosen so that the predicted failure times would range between one minute and a few hours. The failure times for such a test mode can be predicted directly from the plots of log vs. log (tb/aT). The data from these oven tests are indicated in Figure 5 by the flagged data points. The second type of test conducted was to form double folds of each material from strips 25 X 75 mm and clamp them in the folded position with a steel spring binder clip. These were placed initially in boiling water for times up to 12 h without a failure. Another set of double folds was placed in the oven a t temperatures from 373 to 403K for 48 h without failure. Although the maximum surface biaxial strain which occurred at the nose of the double fold was loo%, failures were never initiated at this location. Failures in these double fold specimens for both the EPT-10 and AF-E-332 were induced finally by reversing them in the oven at 403 K. Since these were supplementary and exploratory type tests, the test conditions were not precisely controlled. These initial results did confirm the fact that the severely folded elastomers would not fail in times shorter than those predicted by the uniaxial tensile failure data for 100% strain.

Discussion A working graph derived from the failure data for EPT-10 in hydrazine showing the strain to failure as a function of time and temperature (in " C ) is shown in Figure 6. In the use of such a working graph one must keep in mind the scatter in the experimental data (see Figure 5) which in some instances spanned two or three decades of time. However, the plot is very useful in demonstrating the sensitivity of the life of these materials to temperature and strain levels. For instance, elastomers which would fail in 16 h at room temperature (-20 " C ) a t 225% strain would fail in 20 min at 50 "C. Also, material dynamically stressed a t room temperature in hydrazine to 100% strain or less could last for more than 10 years. These results will be particularly useful in selecting test parameters and designing further experiments to be conducted in hydrazine, such as longtime creep or fatigue tests. For the application of these results to predicting the life of a propellant tank bladder or diaphragm for a long life multicycle mission, an experiment should be designed to produce failure during cyclic folding in a controlled manner

315

Ind. Eng. Chem. Prod. Res. Dev. 1904, 23,315-317

Conclusion

TIME TO RUPTURE ,E

E

i

I

I

I

E

a

imh 2 22

h ‘

I

I

I

i

EPT-10 IN HYDRAZINE

b.7

-g

2 2

,E

m&

0.3-

i 0.2 p: 4 =; 0.1-

Acknowledgment

0-

-0.2

-

-0.3

-

-0.1

This program has demonstrated that the typical elastomer tensile property surface does exist for the EPT-10 and AF-E-332 materials in hydrazine and a method for defining these characteristics has been outlined. Additional experimental work is required to refine the failure curves for long operating times at low-strain levels and to establish the relationship between folding failures and the experimental tensile strain failure limits.

I -

I 4

I -

2

I

I

I

I

0

2

log I$,),

min

I

I 4

6

Figure 6. Tensile strain vs. time to rupture in hydrazine as a function of temperature for EPT-10 elastomer.

in hydrazine. The results of this present program would indicate that failures of EPT-10 could be produced in a few weeks if the tests were conducted a t temperatures above 343 K. In this paper, no attempt has been made to specifically compare the performance of EPT-10 with AF-E-332 because the purpose here was to develop and evaluate a test method. Any performance differences which are apparent from these data should be confirmed by additional specific tests that have a more direct relationship to the intended use of the material.

We wish to thank John T. Schell of the Materials Division of NASA, George C. Marshall Space Flight Center (MSFC) Astronautics Laboratory for permission to publish this account. This program was performed by the Propulsion Division of the Jet Propulsion Laboratory under NASA Contract No. NAS 7-100. Registry No. Hydrazine, 302-01-2.

Literature Cited Boyd, W. K., et ai. “CompatlbliHy of Materiais with Rocket Propellants and OxMizers”. DMIC Memorandum 201, Defense Metals Information Center, Batteile Memorial Institute, Columbus, OH, 1965. Couibert, C. D.; Yankura, G. “Survey of Materials for Hydrazine Propulsion Systems in Multicycle Extended Life Applications”, Technical Memorandum 33-561, Jet Propulsion Laboratory, Pasadena, CA, 1972. Landei, R. F.; Fedors, R. F. “Fracture Processes In Polymeric Solids”, B. Rosen, Ed., Interscience: New York, 1964; Chapter 111 B, pp 361-485. Martin J. W.; Jones, J. F.; Meyers, R. A. “Elastomers for Llquid Rocket Propellant Contalnment”, Technical Report AFML-TR-71-59, Part I, Air Force Materials Laboratory, Dayton, OH, 1971.

Received for review August 12, 1983 Accepted October 20, 1983

COMMUNICATIONS Silicon Phosphates as a New Hardener for Alkali Silicate Solutions Silicon phosphates, 3Si02-2PO5 and Si02.P,05, were prepared by calcination of silica gel impregnated with H,PO, at temperatures up to 1000 C and tested as a hardener for an alkali silicate binder. The kind and properties of the products were found to vary with the preparation conditions, and a well-crystallized 3Si02*2P,05 exhibited a high adhesive strength with sodium silicate binder. The most important feature was that the hardened silicate materials obtained in such a system showed sufficient water resistance.

a

Introduction Alkali silicate solutions generally referred to as water glass are extensively utilized as a binder in adhesives (Vail, 1952; Falcone, 1982), and commonly used together with hardeners and fillers. A number of substances such as mineral acids, metal powders, oxides, hydroxides, fluorides, borates, phosphates, etc. are now being employed as a hardener for silicate binders, and the adhesive properties of the hardened silicate materials are found to be predominantly dependent on the kind of hardeners used (Kimura and Motegi, 1976,1982). In general the silicate adhesives have the advantages of being heat-stable, incombustible, and solvent-resistant. However, the water solubility of the silicate adhesives limits their further utilization. Therefore, special attention has been recently devoted to the development of a new hardener or additive which can provide the water insensitivity to the silicate

adhesives (Boberski et al., 1982; Campbell, 1975; Doi et al., 1972; Mclaughlin and Ramos, 1969; Nakajima et al., 1976). One of the present authors has previously shown that silicon phosphates, which were prepared by calcination of silica gel impregnated with a small amount of phosphoric acid a t high temperatures, can be used as a hardener, and the hardened silicate materials obtained in such systems show water resistance (Sugawara et al., 1977). Since silicon phosphates were first synthesized by Hautefeille et al. in 1883, some informations about the preparation methods, crystal form, and chemical composition of the products have been reported by earlier workers (Boulle and Jary, 1953; Jary, 1957; Lelong, 1964; Liebau et al., 1968; Makart, 1967). However, no information is a t present available on the preparation method based on the calcination process and the chemical properties of the products. In this study, we investigated the relationship 0 1984 American Chemical Soclety