Relation of Elastic Modulus to Thermal Expansion Coefficient in Elastic and Viscoelastic Materials R. J. Arenz, S.J., Jesuit Community, Loyola Marymount Univer ty, Los Angeles, CA 90045 ABSTRACT The relationship of elastic modulus to thermal expansion coefficient is examined for a variety of solid materials. Initially, thermodynamic theory is used to obtain a qualitave relationship that is then given quantitative evaluation from experimental data. An approximate linear logog inverse relation appears to represent a useful guide for a wide range of materials, but a significant deviation is discussed for soft elastomeric substances. INTRODUCTION The interrelation of material behavior properties of solids is of continuing interest to the engineering feld. In some applications (for example, a spacecraft communication antenna) dimensional stabilty and precise positioning of elements under varying load and temperature conditions is of design significance. This study examines the relation of the elastic (Young's) modulus to Uniaxial thermal expansion coefficient in a variety of solid materials, While some correlation of data has been mentioned Previously [1], here energy well theory as well as thermodynamic analysis are presented intially to show the quaitative relationship to be expected. This s then given quantitative evaluation from experimental data, ENERGY WELL CONCEPT The energy well concept for solids, Fig. 1, ilustrates the qualitative type of relation to be expected between the modulus of elasticity and coefficient of thermal expansion. The figure might represent the potential energy of an atom as a function of the position ofall nearey atoms, with the binding energy being indicated by the depth of the minimum energy point of the well below an arbitrary reference level corresponding to infinite separation trom all other atoms. For all types of interatomic bonding a similar graph is found; the precise shape of the well, which depends on the type of bonding, is significant in determining both elastic properties and thermal expansion behavior. [tis well known that the elastic modulus for a piece of material subjected to an external force is given by the curvature ofthe potential wel tits minimum point [2]. A narrow steep well corresponds to a high modulus and a broad shallow well Toprosents a low modulus. Ifthe configuration is that of a thin bar under a uniaxial load in simple tension, and the force and displacement are in terms of stress and strain, he curvature is equal to the Young's modulus E of the material. The width of the wall at any energy level determines the amplitude of vibrations of the atoms about their equllbrium positions. ‘As the temperature of the material is increased, the level of energy rises and the amplitude of vibration (as indicated by the width ofthe well) also increases. Due to the asymmetric shape of the wel, the mean separation distance of the atoms also increases, producing thermal expansion. Again, the precise shape of the well is important as a narrow energy well produces a ‘small thermal expansion coefficient while a broad well has greater asymmetry and gives 2 relatively large coefficient Comparison of these effects of the shape of the energy well thus indicates a general inverse relationship between elastic ‘modulus and thermal expansion coefficient. (It is noted in passing that the inverse character of the variation should apply not ‘only when comparing one isotropic material with anather but also in comparing the properties in diferent directions for an anisotropic material) It remains to investigate the madulus-hermal expansion relationship quantitatively by application of thermadynamic principles. THERMODYNAMIC ANALYSIS Since the quantities of prime interest in our comparison are the Young's modulus (E) and linear thermal expansion coetfcient (a), We consider a rod of length L in simple tension subjected to a uniaxial load F. The resulting axial incremental deformation is dL. Tho first law of thermodynamics for the rod as an isolated system is IU = d+ aW = 4Q—PaV + Fat, a ‘where U is the intemal energy in the rod, Q is the heat added, W is the work done on the system, Vis the volume of the specimen, and P is the hydrostatic pressure. For a reversible system, the second law of thermodynamics defines the entropy SbyTas = 4a @ where T is the absolute temperature, For our comparison, a constant (atmospheric) pressure process is appropriate, involving the enthalpy, H = U + PV, making the Gibbs free energy, G = H - TS, the convenient energy parameter, Thus, G becomes G=U+PV-TS @ Incorporating (1) and (2), the differential change of Gibbs free energy can be written as dG = -SdT+VdP + Fal. @ Fora simple tension situation, G = G(T,P,L). Using partial dervatives to express this dependence, we have in conjunction with (4) that (GI Tee 2-8, (GP =, (GI Ure =F 6) Hence, in the usual way (5) shows that (SI Ler = (FI Tee © and CFP Phe = (W Ure a Since F = F(T,P,L), using partial derivatives and (7), we have dF = (FIT) dT + (Wi Lip dP + (FI Leal ® Differentiating (8) partially with respect to L and combining with (5) and (6) gives ( F Tp, which is substituted into (8) to give = ((FIT)- (HY UF Ure (PITY. W UprpldT + (WE Lined + ( FI Led ° Ina test with constant F, dF = 0, so that (9), upon rearranging, becomes. (CFI Ure (U Tee ATIC UI Use FL + (PMT) =( PE TAM Ure (10), In terms of the uniaxial stress « = FIA (where A is the cross-sectional area of the material) and strain ¢ (with e = aL/L), the thermodynamic quantities need to be expressed per unit volume, and (10) becomes (ol chee (ef Tew (UT) (40) ( UF 2.2 ~0) + (1V) [(PIT)—( PP Taal (ME eee an The product on the left-hand side is Ea. so that the generally inverse nature of the relationship between E and « is verified The precise determination of the quantitative variation depends on the evaluation ofthe right-hand side, specifically on the change of intemal energy with strain and any change of volume with strain, It should be noted that the relation isin general limited to conditions of small strain The type of situation envisaged here is constant pressure so that dP = 0. Moreover, solids are nearly incompresible under hydrostatic compression, especially at low applied pressures (such as atmospheric), so that the PdV contribution is generally negligible compared to the internal energy contribution; that is, one can use U rather than H = U + PV in the analysis (2). (This might have to be modified in the case of elastomers where the major effect of elongation is a change of entropy [3]. However, in he case of ideal isotropic rubbery response, the condition of incompressiblty usually applies; thus, there is no change of Volume with strain and the second term ofthe right-hand side of (11) s zero.) In these situations, then, @ good approximation for (11) becomes. E AfaT) (1M) ( UF ee - 6) (12) (On a logarithmic basis, the variation of E with a can be written as flog EVdlloga) = dlog (VT ((AM)( UW edie - ol dlloga) - 4 (13)where, since « and U are both dependent on interatomic bonding, there can be an interdependency between ( UI e}y» and a The reason for expressing it asin (13) isto facilitate correlation with experiments of uniaxial specimens where T, V, and o in @ given test situation would be known. Thus, the nature of the dependency on the internal energy should appear from a series of tests and a plot of log E vs. log aif Ur were not related to a, such a plot would have a slope of 1 EXPERIMENTAL CORRELATION For a series of isotropic hard solids, E and «as well as ( U! c)rp would vary from material o material. The type of variation that results is ilustrated in Fig. 2 where log E has been plotted vs. log « fora wide range of metalic, ceramic, and hard and soft polymeric solids [24,5]. (The symbols in Fig. 2 are defined as follows: open symbols are for metals and compounds filled symbols are for polymers; circles are from Ref. 2, squares are from Ref. 4, triangles are from Ref. 5 with inverted triangles representing the elastomers.) There is the expected large scatter but a generally satisfactory straight line can be drawn through the data for the harder materials. The slope d{log EVid(og a) is approximately -2.3, indicating a strong dependency on the UI ¢ term involved in the right-hand side of (13) The data at the bottom of Fig. 2 for soft elastomers are typical of rubbery response in which the effect of internal energy cchange with strain is usually small compared to the change in the entropy term [3]. It consequently appears that thelr behavior is sufficiently afferent that these data do not fal on the same general curve associated with the hard crystaline and glassy polymer materials DISCUSSION In accordance with Fig. 2, many hard solids respond approximately to the empirical relation E = 4.5a"°*, This correlation may be useful for some engineering design purposes in choosing materials where dimensional stability is required under varying Concitions of load and temperature, The data shown here are for isotropic materials. However, in the case of anisotropic solids, the graphing technique involved in Fig, 2 can possibly be used to estimate material properties in one of the transverse directions. For example, orthotropic rmalerials will exhibit three principal moduli and associated thermal expansion coefficients, conditions that wil establish a locus (of points that cross over the isotropic line. Such correlation has been used to establish the thickness-direction modulus in application of thin polymeric fms for recording tape usage. While thickness-direction modulus is difficult to measure in thin films, experimental data in longitudinal and transverse directions plus the thermal expansion coefficient in the thickness. direction allowed a line crossing the isotropic curve to give a reasonable estimate of the thickness-direction modulus, This information allowed tape-manufacture design parameters to be optimized for improved reliabilty in spacecraft applications. REFERENCES [1] Barker, 8 E., Jr, “An Approximate Relation Between Elastic Moduli and Thermal Expansivites," Journal af Appliod Physics, 34(%), 107-116, 1963 [2] Barrett, R., Nix, W. D. and Tetelman, A. S., The Principles of Engineering Materials, Prentice Hal, New York, p. 35, 1973. [3] Tretoar, LR. G., The Physics of Rubber Elasticity, 2™ Ed., Oxford University Press, London, 1958 {4] Richards, C. W., Engineering Materials Science, Chapmar and Hall Ltd, London, 1964 [5] Polymer Handbook, 4" Ed., John Wiley and Sons, New York, 2002.MODULUS OF ELASTICITY (DYNE/CN2) INCREASING TEMPERATURE MEAN ATOM POSTION POTENTIAL ENERGY Z ‘FOMIC SEPARATION DISTANCE Fig. 1. Energy wall for atomic spacing in solids tors TTT TT TTT TTT Diamons Wat Diamond oAlg0s ISOTROPIC SOLIDS 12 wr Ey 4 Gloss ron B 4 108 4 a tof 4 roe E 4 2 R so Beat a EN Boe te 8k ge ge THERMAL EXPANSION COEFFICIENT (0C-1) Fig. 2. Modulus vs. thermal expansion coefficient: symbols defined tn text