It is shown that in the ergodic regime, the temperature dependence of Young’s modulus is solely determined by the magnetic properties of a material. For the large class of materials with paramagnetic or diamagnetic response, simple functional forms of the temperature derivative of Young’s modulus are derived and compared with experimental data and empirical results. Superconducting materials in the Meissner phase are ideal diamagnets. As such, they display remarkable elastic properties. Constant diamagnetic susceptibility gives rise to a temperature independent elastic modulus for ceramic and single crystalline superconductors alike. The thermodynamic approach established in this report, paves the way to tailor elastic material parameters through the design of magnetic properties.
Some of the most powerful properties of equilibrium thermodynamics originate from the fact that the formalism implicitly includes interaction between various degrees of freedom such as the elastic, electric, and magnetic ones. Exchange of energy between different modes is built-in through the requirement of thermal equilibrium without reference to a specific microscopic mechanism. Because of this, thermodynamics establishes rigorous relations among equilibrium properties such as Maxwell and Clausius Clapeyron relations rather than model dependent approximations1.
A particular form of Maxwell’s relation is widely employed in the field of magnetocaloric materials2. Isothermal magnetic field-induced entropy change of a material originates from magnetic and other degrees of freedom. Elastic degrees of freedom can significantly contribute to . Nevertheless, is completely determined by the temperature and field dependence of the magnetization without reference to elastic properties3, 4.
This report utilizes the powerful property of equilibrium thermodynamics to relate magnetic and elastic properties. Simple, generic expressions for the relative temperature variation of Young’s modulus, E, are provided for temperature regimes of reversible equilibrium where materials show simultaneously linear magnetization response of paramagnetic or diamagnetic type and linear stress-strain relation. Results derived for the case of paramagnetic response can also be applied in the high temperature limit of ferroic materials. Superconductivity resembles an ideal realization of diamagnetism giving rise to the remarkably simple result of a temperature invariant elastic modulus in the Meissner phase. The possibility to express the temperature dependence of elastic parameters solely in terms of magnetic material properties paves the way to tailor elasticity by magnetic design.
A prominent example for the technological relevance of elastic properties of materials is the space shuttle Challenger accident where low ambient temperature at launch day caused hardening of rubber O-rings in a solid rocket booster with subsequent catastrophic failure5. In the scientific context, interest in the T-dependence of elastic properties has revived in a variety of fields. Those range from observations of abnormal elastic properties in Earth’s lower mantel6 to the investigation of magnetic field dependence of phonons in diamagnetic materials7. They further include magnetocaloric, elastocaloric, and barocaloric phenomena with modern cooling applications, and encompass new frontiers in the design of artificial materials8,9,10,11,12,13,14,15,16,17,18. Recently, microarchitectures with effective negative thermal expansion have been designed through structural integration of constituents with dissimilar but positive thermal expansion coefficients19. Combining the reported advances in microarchitecture with thermal properties of materials such as Invar and anti-Invar alloys may broaden the spectrum of applications20. Thermodynamics is an important tool in understanding the effective temperature response, properties of new materials for advanced cooling technologies, and fundamental relations between elasticity and magnetism.
Thermodynamic relations between magnetic and elastic properties are well-established in the literature. For example, in materials exhibiting the giant magnetocaloric effect, a first order magnetic phase transition accompanies a structural transition. The magneto-structural transition is an extreme manifestation of interaction between magnetic and elastic degrees of freedom9, 21, 22. Pasquale et al. evaluated for the Heusler alloy Ni55Mn20Ga25 at the H-field induced first order transition23. At the transition, the alloy changes from a superelastic austenite to a pseudo-elastic martensite phase. A discontinuity, ΔM, in magnetization accompanies the lattice contraction. Pasquale et al. used the experimental ΔM-value and the magnetic Clausius-Clapeyron relation to calculate the isothermal entropy change11, 23. They found good agreement with the complementary evaluation of ΔS using the stress-strain data of Ni55Mn20Ga25. With the help of they numerically showed that ΔS can be determined from either the magnetic or the elastic Clausius Clapeyron equation. Their example implies and thus establishes a relation between magnetic and elastic properties. Here is the relative lattice contraction of Ni55Mn20Ga25.
In superelastic alloys such Ni-Ti doped with Cu, Co or Pd it is customary to calculate from integration of the Maxwell relation 24. Maxwell relations follow from the analytic behavior of thermodynamic potentials. Formally, the mixed partial derivatives of a thermodynamic potential depend on the sequence of differentiation at a first order phase transition. However, it has been shown that numerical integration of the Maxwell relation yields the correct entropy change when employing for magnetocaloric materials or for elastocaloric materials close to the first order transition25. In addition to the indirect derivation of entropy change from magnetization data, S(H, T) can be determined from heat capacity measurements via . Both methods have been successfully employed and critically compared21, 25, 26.
Atherton et al. showed the validity of , illustrating the correlation between magnetostrictive and magneto-mechanical effects27. Similarly, Bozoroth reports fair agreement between and when evaluating experimental data of Permalloy28. Such investigations show that elastic properties are intimately related to magnetic properties of materials. However, none of the relations makes quantitative predictions about the T-dependence of an elastic constant such as Young’s modulus solely based on magnetic input without reference to stress, strain, or lattice properties.
Modeling of the temperature dependence of Young’s modulus is involved. In contrast to approaches based on microscopic models and statistical mechanics it is shown here that universal results can be derived for materials with generic magnetic properties. No reference to lattice properties of the material, the interatomic potential, defects, grain size, stress or strain is required. The rich experience in characterization and tailoring of magnetic properties can potentially be exploited to design elastic properties in temperature regimes with ergodic magnetic behavior.
Thermodynamic derivation of the relative temperature change of Young’s modulus
As a starting point the differential, , of the Gibbs free energy density g = G/V is considered with V being the sample volume. Therefore, the validity of all subsequently derived relations is limited to the framework of reversible thermodynamics. To keep the notation simple and because experimental data of polycrystalline materials are considered, the elastic work differential is written in terms of scalar variables for stress, , and strain, . The simplified differential form reads
From the identity one obtains the established Maxwell relation , a thermodynamic relation for the Villari or inverse magnetostrictive effect, i.e. the change in magnetization due to applied stress28. From Legendre transformation of g one obtains and the differential form
with its three corresponding Maxwell relations
Integration of Eq. (2a) for the special case yields the well-known expression
frequently applied to calculate the isothermal entropy change of magnetocaloric materials. Integration of Eq. (2b) for the special case H = 0 yields the isothermal entropy change
applied in the case of elastocaloric materials. Using Eq. (4) with the general stress-strain relation in the Hooke limit yields
In analogy to the approach by Pasquale et al. one expects a relation between and . Specifically, one expects that the same magnitude of isothermal entropy change obtained from straining a material by a small amount in H = 0 can be induced under = 0 condition when applying a magnetic field H f such that . For materials with , such as paramagnets, is negative. For materials with , such as metamagnets below the critical temperature, the entropy change is positive29. In paramagnets, originates from the fact that a magnetic field has the tendency to align magnetic moments through minimization of the Zeeman energy. Similarly, the -induced entropy change can display both a positive and a negative sign. Although is less common, it is observed for instance in rubber where stretching of the material reduces the number of possible configurations of the polymer molecules and thus reduces entropy30.
In the case of an elastic material with paramagnetic magnetization response
and , can be found by equating the reversible magnetic and elastic work according to
The conditions under which Eq. (7) provides such that require some consideration. In fact, validity of Eq. (7) constraints the possible magnetic field responses to simple, ideal gas type, linear functional forms similar to the one given by Eq. (6). Next, general results of the formalism obtained for paramagnetic response are provided. Calculating with the help of Eq. (6), utilizing the -expression provided by Eq. (5), and applying the entropy condition together with the work condition of Eq. (7) yields
Eq. (8) can be compared with results from a model where E(T) is explicitly calculated from a given elastic equation of state . As a toy model, a material is considered which obeys the magnetic equation of state given by Eq. (6) and has the elastic properties of a rubber band which responds to tension according to
where is a constant, is the T-independent length of the rubber band at , and the length of the rubber band for applied tension 31. Linearization of the elastic equation of state (9) provides in agreement with the general result of Eq. (8) (see Supplemental Material for details).
Figure 1 illustrates the general concept of the thermodynamic approach. The blue and green lines (color online) associated with the left axes in Fig. 1a and b show and . The horizontal dashed line elucidates that for a given value there is a field such that . The work condition which relates and is obtained by equating the magnetic and the elastic reversible work. Dashed regions of identical area are enclosed by the linear functions and . They depict the magnitude of the reversible magnetic (Fig. 1a) and elastic (Fig. 1b) work. The geometric construction of equal area relates and in accordance with Eq. (7) and establishes together with the entropy condition the temperature dependence of Young’s modulus. Eq. (7) provides in the absence of spin-spin correlations (see Supplemental Material). A potential generalization to cases which include spin correlation requires a refinement of the work criterion given by Eq. (7).
An important application of the thermodynamic approach includes materials with a magnetic equation of state given by Eq. (6), linear stress-strain relation, and . These conditions are fulfilled, e.g., for ferromagnetic metals significantly above . Without any additional input application of
as a special case of Eq. (8). Equation (11) is expected to hold for all materials with in the Hooke limit and paramagnetic magnetization response. Prominent examples are 3d metals for . Reciprocity suggests that elastic properties affect the magnetic properties of a material. Subsequent discussion of Eq. (11) shows that this is the case. In the limit , Eq. (11) loses validity as a consequence of Nernst’s theorem, i.e., for . For , the low temperature limit of E/T diverges. Resolution of this inconsistency requires that the magnetic equation of state deviates from the paramagnetic form of Eq. (6) in the limit . In fact, quantum ground states of solids deviate from paramagnetism. For example, simple crystalline solids tend to either order magnetically or condensate into a superconducting ground state. In both cases the magnetization response is no longer given by Eq. (6). The case of superconducting material in the Meissner phase is discussed below in detail.
Not all materials show paramagnetic response at high temperature. However, all materials have diamagnetic response either exclusively or in linear superposition with other susceptibility contributions. The diamagnetic susceptibility gives rise to linear response in magnetization on an applied magnetic field in a wide field range. Again, the linear response and absence of spin-spin correlation ensures applicability of the thermodynamic approach in conjunction with the reversible work condition of Eq. (7). Results for the case of diamagnetic response are applicable for virtually all non-magnetic elastic materials. A general relation which includes Eq. (11) as a special case and is also applicable in the case of diamagnetic susceptibilities is derived from and the use of Eq. (7) for . It yields
Temperature independence of Young’s modulus of superconductors in the Meissner state
The temperature dependence of the diamagnetic susceptibility of most materials is small and often experimentally difficult to separate from other contributions. In addition, theoretical understanding of diamagnetism in solids is incomplete. Superconducting materials in the Meissner state are a prominent exception. Here diamagnetism, with in the ideal case or in the presence of defects, is a consequence of magnetic flux expulsion. With follows and Eq. (12) yields for all temperatures sufficiently below the transition into the superconducting state. Obviously, is consistent with Nernst’s heat theorem. Remarkably, to the best of the author’s knowledge, the simple but profound result has not been discussed in the literature.
Figure 2 shows the temperature dependence vs T (circles) of the cuprate superconducting ceramic Bi2Sr2CaCu2O8+γ with a critical temperature of 8 K. The data are adapted from Lin et al.32. The predicted temperature dependence (dashed line in Fig. 2) is virtually perfectly fulfilled in the entire temperature range K. Remarkably, as a special case of Eq. (12) holds despite the fact that cuprates are ceramic materials. Their ill-defined structural details are expected to affect the absolute value of , but they do not affect the temperature dependence of the elastic modulus in the Meissner phase. Similarly, the measured compressional sound velocities in YBa2Cu3O7−δ measured by Toulouse et al.33 clearly plateau for 90K.
For the single crystalline BCS superconductor Nb, Eq. (12) is also applicable. The inset of Fig. 2 shows vs T of Nb measured under slow cooling condition in zero magnetic field (open squares) and in an applied magnetic field of H larger than the upper critical field H c2 = 376 kA/m (solid squares) (data are adapted from Pal-Val et al.34). Applicability of equilibrium thermodynamics requires a protocol of slow cooling. Therefore, only the data obtained on slow cooling and in zero field condition (inset Fig. 2 open squares) exhibit in good approximation plateau-like behavior for 9.3 K as predicted by Eq. (12) (dashed line). The data taken in the presence of a strong magnetic field (inset Fig. 2 solid squares) continue to increase with decreasing temperature below because the Meissner phase is destroyed and the magnetic susceptibility is no longer constant.
Comparison with experimental data and empirical T-dependencies of elastic constants
In order to obtain an explicit temperature dependence of Young’s modulus for diamagnetic materials other than superconductors, the temperature dependence of the diamagnetic susceptibility has to be known. Experimental and theoretical data concerning the temperature dependence of the diamagnetic susceptibility are scares. It is worth mentioning however, that the Langevin type contribution of valence electrons to the diamagnetic susceptibility, , is related to the thermal expansion, . The T-dependence of has been associated with via the approximate functional form 35. The T-dependence of lattice parameters resembles in good approximation the T-dependence of the internal energy, , of the lattice giving rise to with and the heat capacity at constant volume. When substituting, , and into Eq. (12) one obtains
this expression holds for the large class of elastic materials with diamagnetic susceptibility. Note that the connection between and has been made to compare Eq. (13) with results from the literature. It is important to stress that is determined by the diamagnetic susceptibility in accordance with Eq. (12).
Eq. (13) allows for comparison with empirical expressions for the T-dependence of elastic constants such as Wachtman’s equation 36. Su et al. put Wachtman’s empirical formula on a microscopic basis37. Their work shows that can be identified with the Debye or Einstein temperature, . For comparison between Wachtman’s expression and Eq. (13), has to be considered in the Einstein approximation (see Supplemental Material). The Debye model provides a better approximation of especially for the low temperature limit with . Substituting into Eq. (13) implies for . In fact, the alternative empirical T-dependence with has been successfully applied to fit various experimental E vs T data38. It finds theoretical support by Born and Huang39. Limitations are discussed by Yarshni40. The asymptotic expression reflects Eq. (13) within the Debye approximation for .
The above analysis shows that the T-dependence of Young’s modulus can be understood in terms of the magnetic field response of diamagnetic materials. The validity of the result is evidenced by comparison with empirical functional forms which have been successfully applied for data analysis in the literature. The empirical expressions have in part been motivated via thermodynamic and microscopic derivations which, however, refer to lattice properties of solids. The approach presented here relates E(T) to magnetic properties. They can be measured via magnetometry and potentially tuned via the vast toolbox available in the science of magnetic materials paving the way to magnetically tunable elastic properties. Next, Eq. (11), which has been derived for materials with paramagnetic response, is compared to data of Young’s modulus obtained from stress-strain relations in the literature.
Experimental data for Young’s modulus at high temperature and in particular data for 3d metals at temperatures substantially above are scarce. Perhaps the most comprehensive data collection is given in the review article by Ledbetter and Reed on elastic properties of metals and alloys, 1. Iron Nickel and Iron-Nickel alloys38. Figure 3 shows E vs T for Ni adapted from Ledbetter and Reed38. As expected, the magnetic state of a sample dramatically affects E at . This is clearly seen when comparing data obtained with (Fig. 3 solid squares) and without (Fig. 3 open squares) applied magnetic field. Below , the -effect is observed which quantifies the difference between E in a demagnetized and the magnetized state41,42,43,44. For there is virtually perfect collapse of and 44, 45. The ΔE-effect of polycrystalline Ni has been modelled by Hubert et al. using the linear temperature dependence of E vs. T in the reversible equilibrium regime at T > TC as phenomenological input46. The entropy approach introduced in this manuscript is not applicable in the non-ergodic regime below and thus cannot describe the -effect. However, for , i.e., in the absence of spontaneous magnetization when equilibrium thermodynamics is applicable, E becomes virtually independent from the applied magnetic field.
Similar to the disappearing of the -effect for , differences in E between samples of different mechanical treatment become far less pronounced at . Particularly conclusive is the perfect merging of E vs T at for the three cold worked samples shown in Fig. 3. The samples have been subject to different annealing protocols and show substantial variation in grain sizes ranging from 0.037 (Fig. 3 blue dashed line) to 0.0105 (red dotted line) and 0.0095 mm (green solid line). The data collapse for strongly supports the conclusion of a universal asymptotic high temperature limit of the relative temperature derivative of Young’s modulus. It is remarkable that in accordance with the thermodynamic prediction approaches a high temperature limit, which is independent from structural details as significant as grain sizes.
The temperature dependence of Young’s modulus of Ni is best suited for a quantitative comparison with Eq. (11). Open circles in Fig. 3 show E vs T data up to T = 1300 K taken by Armstrong and Brown47. A linear fit (line) of E vs T between K provides dyn/cm2K. The modulus at T = 1250 K, in the center of the fitting interval, reads E(T=1250 K) = dyn/cm2 which yields dyn/cm2K. Comparison of the numerical values for and validates Eq. (11) within an error of 14%. The residual discrepancy is expected when considering that the underlying assumption of correlation-free paramagnetic behavior is only approximately given at with .
The inset of Fig. 3 shows the T-dependence of the shear modulus, G, for fcc iron at temperatures K (triangles). Data are adapted from Ledbetter and Reed38. Assuming that the Poisson ratio, , is approximately constant in this temperature regime, is expected for polycrystalline Fe in accordance with for isotropic materials. Within this approximation, Eq. (11) can be applied when substituting E by G. From a linear fit of G vs T at K (line in inset of Fig. 3) one obtains dyn/cm2K. With dyn/cm2 we obtain dyn/cm2K. Although and are of the same order of magnitude, the discrepancy between the numerical values of and is much higher for Fe than in the case of Ni. This is expected due to the fact that is significantly higher than . Consequently, the deviation from uncorrelated paramagnetism is much more pronounced in the temperature regime available for Fe than it is for Ni.
Thermodynamics relates the temperature dependence of the elastic modulus of a material solely to its magnetic properties. General results for the temperature derivative of Young’s modulus can be derived for paramagnetic materials, diamagnetic materials, and superconducting materials in the Meissner phase. The temperature independent diamagnetic susceptibility of a superconducting material below its critical temperature gives rise to the remarkably simple result that Young’s modulus is constant below the superconducting transition temperature for ceramic and single crystalline superconducting materials alike. For normal conducting diamagnetic materials, it is possible to find explicit functional forms for the temperature derivative of the elastic modulus. These findings and the generality of the formalism lead to the conclusion that, in the framework of reversible equilibrium thermodynamics, the temperature dependencies of elastic moduli are fully encoded in the magnetic properties of a material. Energy exchange between various degrees of freedom, which is implicitly taken into account in the framework of equilibrium thermodynamics, makes it possible that magnetic properties determine elastic moduli. This insight paves the way to tailor elastic properties by tuning magnetic properties beyond the known -effect. The results presented in this manuscript are obtained for simple magnetic equations of state where spin-spin correlation is absent. A refined approach, which includes spin correlation, could unfold its full potential for applications aiming at design of elastic properties. Future work of particular interest includes microscopic theoretical approaches, which explore quantitatively the absolute temperature dependencies of Young’s modulus for magnetic and superconducting materials.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by the National Science Foundation (NSF) through the Nebraska Materials Research Science and Engineering Center (MRSEC) (grant No. DMR-1420645). The author is grateful to Robert J. Hardy and Sitaram S. Jaswal for very inspiring and useful discussions.
Electronic supplementary material
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.