skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Formation enthalpies for transition metal alloys using machine learning

Authors:
; ; ;
Publication Date:
Sponsoring Org.:
USDOE
OSTI Identifier:
1361113
Grant/Contract Number:
SC0008877
Resource Type:
Journal Article: Publisher's Accepted Manuscript
Journal Name:
Physical Review B
Additional Journal Information:
Journal Volume: 95; Journal Issue: 21; Related Information: CHORUS Timestamp: 2017-06-01 22:10:40; Journal ID: ISSN 2469-9950
Publisher:
American Physical Society
Country of Publication:
United States
Language:
English

Citation Formats

Ubaru, Shashanka, Międlar, Agnieszka, Saad, Yousef, and Chelikowsky, James R. Formation enthalpies for transition metal alloys using machine learning. United States: N. p., 2017. Web. doi:10.1103/PhysRevB.95.214102.
Ubaru, Shashanka, Międlar, Agnieszka, Saad, Yousef, & Chelikowsky, James R. Formation enthalpies for transition metal alloys using machine learning. United States. doi:10.1103/PhysRevB.95.214102.
Ubaru, Shashanka, Międlar, Agnieszka, Saad, Yousef, and Chelikowsky, James R. Thu . "Formation enthalpies for transition metal alloys using machine learning". United States. doi:10.1103/PhysRevB.95.214102.
@article{osti_1361113,
title = {Formation enthalpies for transition metal alloys using machine learning},
author = {Ubaru, Shashanka and Międlar, Agnieszka and Saad, Yousef and Chelikowsky, James R.},
abstractNote = {},
doi = {10.1103/PhysRevB.95.214102},
journal = {Physical Review B},
number = 21,
volume = 95,
place = {United States},
year = {Thu Jun 01 00:00:00 EDT 2017},
month = {Thu Jun 01 00:00:00 EDT 2017}
}

Journal Article:
Free Publicly Available Full Text
This content will become publicly available on June 1, 2018
Publisher's Accepted Manuscript

Save / Share:
  • We present heats of formation of ordered fcc transition-metal binary alloys as predicted by the non-self-consistent electron density functional corrected effective-medium (CEM) theory. Two forms of the CEM theory are used. The first includes explicit numerical evaluation of the kinetic-exchange-correlation energy functional, a computationally intensive task. The second approximates this term as a local function of electron density and leads to a formulation that is very similar to the embedded-atom method. Calculated values of the energies of formation from the two forms of the CEM method are compared to each other and to experimental data. We demonstrate that both formsmore » of the CEM theory provide formation energies accurate to within about 0.05--0.1 eV/atom (5--10 kJ/mol) for 3[ital d] and 4[ital d] binary alloys.« less
  • A machine learning–based framework for modeling the error introduced by surrogate models of parameterized dynamical systems is proposed. The framework entails the use of high-dimensional regression techniques (eg, random forests, and LASSO) to map a large set of inexpensively computed “error indicators” (ie, features) produced by the surrogate model at a given time instance to a prediction of the surrogate-model error in a quantity of interest (QoI). This eliminates the need for the user to hand-select a small number of informative features. The methodology requires a training set of parameter instances at which the time-dependent surrogate-model error is computed bymore » simulating both the high-fidelity and surrogate models. Using these training data, the method first determines regression-model locality (via classification or clustering) and subsequently constructs a “local” regression model to predict the time-instantaneous error within each identified region of feature space. We consider 2 uses for the resulting error model: (1) as a correction to the surrogate-model QoI prediction at each time instance and (2) as a way to statistically model arbitrary functions of the time-dependent surrogate-model error (eg, time-integrated errors). We then apply the proposed framework to model errors in reduced-order models of nonlinear oil-water subsurface flow simulations, with time-varying well-control (bottom-hole pressure) parameters. The reduced-order models used in this work entail application of trajectory piecewise linearization in conjunction with proper orthogonal decomposition. Moreover, when the first use of the method is considered, numerical experiments demonstrate consistent improvement in accuracy in the time-instantaneous QoI prediction relative to the original surrogate model, across a large number of test cases. When the second use is considered, results show that the proposed method provides accurate statistical predictions of the time- and well-averaged errors.« less
  • The Standard enthalpies of formation of 14 neodymium alloys have been determined by direct synthesis calorimetry at 1,477 {+-} 2 K. The following values of {Delta}H{degree}{sub f} (kJ/g atom) are reported: NdNi{sub 5}, {minus}(26.2 {+-} 1.1); Nd{sub 5}Ru{sub 2}, {minus}(17.2 {+-} 1.9); NdRu{sub 2}, {minus}(18.8 {+-} 1.2); Nd{sub 5}Rh{sub 4}, {minus}(59.9 {+-} 2.5); NdRh, {minus} (64.2 {+-} 2.0); NdRh{sub 2}, {minus}(59.9 {+-} 1.1); NdRh{sub 3}, {minus}(44.4 {+-} 1.6); NdPd {minus}(77.2 {+-} 2.7); NdPd{sub 3}, {minus}(73.3 {+-} 2.3); Nd{sub 5}Ir{sub 3}, {minus}(59.7 {+-} 2.7); NdIr{sub 2}, {minus}(67.6 {+-} 1.5); NdPt, {minus}(104.4 {+-} 2.6); NdPt{sub 2}, {minus}(97.9 {+-} 2.4); and NdPt{sub 5},more » {minus}(55.0 {+-} 3.1). The results are compared with available literature data for some of the neodymium alloys and with predicted values from the Miedema model.« less
  • Enthalpies of formation of (Pd + In) alloys have been obtained by direct reaction calorimetry using a very high temperature calorimeter between 1,425 and 1,679 K in the concentration range 0 < x{sub Pd} < 0.66. They are very negative with a minimum {Delta}{sub mix}H{sub m}{degree} = {minus}59.6 {+-} 2.5 kJ mol at x{sub PD} = 0.59 and independent of temperature within the experimental error. The integral molar enthalpy of mixing is given by {Delta}{sub mix}H{sub m}{degree}/{l_brace}kJ/mol{r_brace} = x(1 {minus} x) ({minus}126.94 {minus} 92.653x {minus} 83.231x{sup 2} {minus} 734.49x{sup 3} + 949.07x{sup 4}), where x = x{sub PD}. The limitingmore » partial molar enthalpy of palladium in indium was calculated as {Delta}h{sub m} (Pd liquid in {infinity} liquid In) = {minus}127 {+-} 5 kJ/mol. The results are discussed and compared with the enthalpies of formation of solid alloys. The anomalous behavior of the partial enthalpy of Pd is assumed to be due to the charge transfer of, at most, two electrons of In to Pd.« less
  • The standard enthalpies of formation of 13 dysprosium alloys with late transition metals have been determined by direct synthesis calorimetry at 1474 {+-} 2 K. The following values of {Delta}H{sub f}{sup o} (kJ (mole atom){sup {minus}1}) are reported: DyNi, {minus}(35.2 {+-} 1.5); DyNi{sub 5}, {minus}(27.4 {+-} 0.7); DyRu{sub 2}, {minus}(27.3 {+-} 0.9); DyRh, {minus}(76.5 {+-} 2.0); DyRh{sub 2}, {minus}(62.3 {+-} 0.8); Dy{sub 7}Rh{sub 3}, {minus}(56.8 {+-} 2.2); DyPd, {minus}(83.3 {+-} 2.0); Dy{sub 3}Pd{sub 4}, {minus}(86.6 {+-} 2.1); DyPd{sub 3}, {minus}(76.2 {+-} 1.5); DyIr{sub 2}, {minus}(69.9 {+-} 2.1); DyPt, {minus}(109.4 {+-} 1.8); DyPt{sub 2}, {minus}(98.1 {+-} 2.8); and DyPt{sub 3}, {minus}(82.8more » {+-} 2.2). The results are compared with predicted values from the Miedema model and with available literature data for DyNi, DyNi{sub 5}, DyPd, and DyPt.« less