betaFeedback

WorksPhysics and Astronomy

AFLOW-CCE for the thermodynamics of ionic materials

Abstract

Accurate thermodynamic stability predictions enable data-driven computational materials design. Standard density functional theory (DFT) approximations have limited accuracy with average errors of a few hundred meV/atom for ionic materials such as oxides and nitrides. Thus, insightful correction schemes as given by the coordination corrected enthalpies (CCE) method, based on an intuitive parameterization of DFT errors with respect to coordination numbers and cation oxidation states present a simple, yet accurate solution to enable materials stability assessments. Here, we illustrate the computational capabilities of our AFLOW-CCE software by utilizing our previous results for oxides and introducing new results for nitrides. The implementation reduces the deviations between theory and experiment to the order of the room temperature thermal energy scale, i.e. ~25 meV/atom. The automated corrections for both materials classes are freely available within the AFLOW ecosystem via the AFLOW-CCE module, requiring only structural inputs.

In data-driven computational materials science, density functional theory (DFT) has become the standard method to characterize systems and an indispensable tool enabling insightful predictions. Over the last decades, John Perdew and his group have made DFT accurate enough for materials design while setting the standards in the field according to the functional developments related to LDA [1], PBE [2], and SCAN [3]. These approximations are used for the data generation of virtually all big materials databases with millions of entries [4, 5, 6, 7, 8, 9, 10, 11, 12, 13], and thus closely match the quest posed by Dirac for “approximate practical methods” to solve the quantum many-body solid state problem [14].

Thermodynamics is one of the most crucial aspects for materials design, since synthesizability can be ensured for thermodynamically stable systems. As such, the need for an accurate computational treatment of thermodynamic stability is pressing for many issues related to e.g. stability of competing phases [15], high-entropy materials [16, 17], solution enthalpies in liquid metals [18], and novel (2D) nanostructures when evaluating relative energies and with respect to the bulk parent compound [19, 20, 21, 22, 23]. The formation enthalpy rigorously quantifies the thermodynamic stability, as the enthalpy difference between the material and its elemental references. Its accurate calculation is a basic challenge. Standard (semi-)local and even currently available more advanced ab initio methods yield errors of several hundred meV/atom for ionic compounds such as oxides and nitrides [24, 25, 26, 27, 28, 29, 30, 31, 32] when comparing to validated standard collections for thermochemical data [33, 34, 35, 36]. The reason is that accurate total energies for all systems involved — compound and elemental references — are still not accessible with the standard functionals, and errors when calculating the enthalpy difference between chemically dissimilar systems do not cancel. The explicit treatment of self-interaction errors might be an avenue to further significant advances from ab initio [37, 38, 39, 40], but likely at an elevated computational cost infeasible for high-throughput materials design. Even efforts based on Quantum Monte Carlo calculations only partially remedy the discrepancies between theory and experiment [41, 42, 43].

A solution for this dilemma can be provided by physically motivated empirical enthalpy corrections. These parameterize (semi-)local DFT errors with respect to measured values, are hence feasible to enable materials design, and can include uncertainty quantification [44]. It turns out that correction methods, mixing schemes, and machine learning approaches based on only the composition of a material can already lead to significant improvements in accuracy reducing the mean absolute errors (MAEs) down to ∼50 meV/atom [24, 25, 26, 27, 45, 46, 47, 48]. In addition to the still limited accuracy, such an approach can, however, be problematic for the phase diagrams of certain systems due to only tilting the Gibbs energy landscape [32]. These schemes do also not allow for a correction of the relative stability of polymorphs whose energetic ordering is known to be incorrectly predicted by DFT in several cases [30].

To rectify these issues, we have recently introduced the method of cordination corrected enthalpies (CCE), based on the bonding topology in a material [32]. CCE avoids thermodynamic paradoxes by construction, reduces the MAE down to ∼25 meV/atom, and can correct the relative stability of polymorphs if they differ in the number of nearest neighbor bonds [32, 49]. So far, the method has been parameterized for oxides and a few halides calling for an extension to other anion classes such as nitrides. Most importantly, effective implementations are needed freely available to and easy to use by the computational materials science community. Here, we provide such a tool by our automated corrections within the AFLOW-CCE module.

In this article, we i. present some theoretical basics, a motivation of CCE, and its formalism; and ii. demonstrate the computational capabilities of our implementation using our previous results for oxides and new results for nitrides, including a discussion of the error bars of plain DFT for different standard functionals, as well as a demonstration of the performance of the AFLOW-CCE module for a test set of ternary nitrides.

Theory

Formation enthalpies. From DFT, one directly computes an estimate of the formation enthalpy ΔfEAx1Bx2…Yxn0,DFT\Delta_{\mathrm{f}}E_{A_{x_{1}}B_{x_{2}}\ldots Y_{x_{n}}}^{0,\mathrm{DFT}} of a compound Ax1Bx2…YxnA_{x_{1}}B_{x_{2}}\ldots Y_{x_{n}} at zero temperature and pressure without zero-point vibrations:

ΔfEAx1…Yxn0,DFT=UAx1…Yxn0,DFT−[∑i=1n−1xiUi0,DFT+xn2UY20,DFT]\Delta_{\mathrm{f}}E_{A_{x_{1}}\ldots Y_{x_{n}}}^{0,\mathrm{DFT}}=U_{A_{x_{1}}\ldots Y_{x_{n}}}^{0,\mathrm{DFT}}-\left[\sum_{i=1}^{n-1}x_{i}U_{i}^{0,\mathrm{DFT}}+\frac{x_{n}}{2}U_{Y_{2}}^{0,\mathrm{DFT}}\right]

where UAx1…Yxn0,DFTU_{A_{x_{1}}\ldots Y_{x_{n}}}^{0,\mathrm{DFT}}, Ui0,DFTU_{i}^{0,\mathrm{DFT}}, and UY20,DFTU_{Y_{2}}^{0,\mathrm{DFT}} are the total internal energies of the compound, the ii-element reference phase, and molecular Y2Y_{2}, respectively; and x1,…,xnx_{1},\ldots,x_{n} are stoichiometries. In this manuscript, YY stands for O or N and this notation can be extended to other anion species.

On the other hand, the tabulated measured standard (“∘\circ”) formation enthalpy at room temperature (Tr=298.15 KT_{\mathrm{r}}=298.15\ \mathrm{K}) corresponds to:

ΔfHAx1…Yxn∘,Tr,exp=HAx1…Yxn∘,Tr−[∑i=1n−1xiHi∘,Tr+xn2HY2∘,Tr]\Delta_{\mathrm{f}}H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\circ,T_{\mathrm{r}},\mathrm{exp}}=H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\circ,T_{\mathrm{r}}}-\left[\sum_{i=1}^{n-1}x_{i}H_{i}^{\circ,T_{\mathrm{r}}}+\frac{x_{n}}{2}H_{Y_{2}}^{\circ,T_{\mathrm{r}}}\right]

where HAx1…Yxn∘,TrH_{A_{x_{1}}\ldots Y_{x_{n}}}^{\circ,T_{\mathrm{r}}}, Hi∘,TrH_{i}^{\circ,T_{\mathrm{r}}}, and HY2∘,TrH_{Y_{2}}^{\circ,T_{\mathrm{r}}} are the standard enthalpies of the compound, the ii-element reference phase, and Y2Y_{2} at TrT_{\mathrm{r}}.

As we have shown in Ref. 32, the measured value can be written in a very good approximation (accuracy better than 1 meV/atom) as the sum of the internal energy contribution and a small vibrational term ΔfHAx1…Yxnvib\Delta_{\mathrm{f}}H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\mathrm{vib}} of the order of 10-20 meV/atom due to zero-point and thermal effects:

ΔfHAx1…Yxn∘,Tr,exp≈ΔfEAx1…Yxn0,DFT+ΔfHAx1…Yxnvib≈ΔfHAx1…Yxn∘,Tr,cal.(1)\Delta_{\mathrm{f}}H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\circ,T_{\mathrm{r}},\mathrm{exp}}\approx\Delta_{\mathrm{f}}E_{A_{x_{1}}\ldots Y_{x_{n}}}^{0,\mathrm{DFT}}+\Delta_{\mathrm{f}}H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\mathrm{vib}}\approx\Delta_{\mathrm{f}}H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\circ,T_{\mathrm{r}},\mathrm{cal}}. \tag*{(1)}

Both contributions in Eqn. (1) can be estimated computationally to yield ΔfHAx1…Yxn∘,Tr,cal\Delta_{\mathrm{f}}H_{A_{x_{1}}\ldots Y_{x_{n}}}^{\circ,T_{\mathrm{r}},\mathrm{cal}}, the calculated standard formation enthalpy at TrT_{\mathrm{r}} [32].

The vibrational term can be computed reliably within a quasi-harmonic Debye model as implemented within the AFLOW Automatic GIBBS Library (AGL) [50, 51, 52, 53, 54, 55] with an average error as small as ∼5\sim5 meV/atom [32]. The zero-point and thermal effects of O2_{2}/N2_{2} are accounted for based on reliable tabulated (spectroscopic) data [34] as well as assuming perfect gas behavior.

The large errors between calculated and measured formation enthalpies are due to estimating the internal energy contribution in Eqn. (1) from DFT.

Motivation of corrections per bond. To obtain accurate enthalpies, an insightful correction scheme is required. The idea behind CCE is illustrated in Figure 1. When forming perovskite CaTiO3_{3} out of CaO and rutile TiO2_{2}, the number of nearest neighbor cation-anion bonds, i.e. the coordination number of Ca, changes from six to eight. The number of bonds in a material is crucial for its thermodynamic stability and it is commonly discussed how well a specific approximation to DFT captures a certain type of chemical bond [56]. It thus seems plausible to assign an error per bond to the computed enthalpies making the coordination number an appropriate descriptor to parameterize DFT errors. For optimal transferability from fit to target compounds, the correction also needs to be specific to the oxidation number of the cation as a different self-interaction error is expected for different charge states of transition metal centers such as Fe2+^{2+} vs. Fe3+^{3+}.

Structural models of CaO, rutile TiO2, and perovskite CaTiO3

Figure 1. Corrections per bond. Structural models of (a) CaO, (b) rutile TiO2TiO_2, and (c) perovskite CaTiO3CaTiO_3. While the Ti coordination number is six in both rutile and perovskite, Ca exhibits a coordination change from sixfold to eightfold when going from CaO to CaTiO3CaTiO_3 (Ca–O bonds marked in yellow). The number of bonds is thus a crucial parameter for the thermodynamic stability of a material.

The CCE formalism. Taking binary compounds Ax1Yx2A_{x_{1}}Y_{x_{2}} as the fit set, the CCE corrections δHA−YT,A+α\delta H_{A-Y}^{T,A+\alpha} per cation-anion A−YA-Y bond and cation oxidation state +α+\alpha are obtained from the difference between DFT formation enthalpies and experimental standard formation enthalpies at temperature TT [32]:

ΔfEAx1Yx20,DFT−ΔfHAx1Yx2∘,T,exp=x1NA−YδHA−YT,A+α,\Delta_{\mathrm{f}}E_{A_{x_{1}}Y_{x_{2}}}^{0,\mathrm{DFT}}-\Delta_{\mathrm{f}}H_{A_{x_{1}}Y_{x_{2}}}^{\circ,T,\mathrm{exp}}=x_{1}N_{A-Y}\delta H_{A-Y}^{T,A+\alpha},

where NA−YN_{A-Y} is the number of nearest neighbor A−YA-Y bonds and xix_{i} are stoichiometries for the ii-species. TT can be 298.15 or 0 K, i.e. temperature effects are included in the corrections as 0 K estimates of experimental ΔfH\Delta_{\mathrm{f}}H values can be obtained by taking into account thermal effects only for the fit compounds according to Ref. [49]. As detailed in Refs. [32, 49], the corrections are fit to the formation enthalpies directly calculated from DFT, without taking into account thermal/vibrational contributions since these terms are to a very good approximation (average accuracy ∼1\sim1 meV/atom) implicitly included in the corrections.

The corrections can then be applied without additional computational cost compared to plain DFT to any multinary compound Ax1Bx2…YxnA_{x_{1}}B_{x_{2}}\ldots Y_{x_{n}} to obtain the CCE formation enthalpy ΔfHAx1Bx2…Yxn∘,T,CCE\Delta_{\mathrm{f}}H_{A_{x_{1}}B_{x_{2}}\ldots Y_{x_{n}}}^{\circ,T,\mathrm{CCE}}:

ΔfHAx1Bx2…Yxn∘,T,CCE=ΔfEAx1Bx2…Yxn0,DFT−∑i=1n−1xiNi−YδHi−YT,i+α,\Delta_{\mathrm{f}}H_{A_{x_{1}}B_{x_{2}}\ldots Y_{x_{n}}}^{\circ,T,\mathrm{CCE}}=\Delta_{\mathrm{f}}E_{A_{x_{1}}B_{x_{2}}\ldots Y_{x_{n}}}^{0,\mathrm{DFT}}-\sum_{i=1}^{n-1}x_{i}N_{i-Y}\delta H_{i-Y}^{T,i+\alpha},

where Ni−YN_{i-Y} is the number of nearest neighbor bonds between the cation ii and anion YY-species. The method has been implemented for automated enthalpy corrections requiring only an input structure as the AFLOW-CCE module available to the scientific community [49]. The implementation is also applicable to enthalpies computed from DFT+UU if the same settings as for the AFLOW-ICSD database are used [49, 57].

The ab-initio calculations of this study for the exchange-correlation functionals LDA [1, 58, 59], PBE [2], and SCAN [3] are performed with AFLOW [4, 5, 60, 61, 62, 63, 64, 65] and the Vienna Ab-initio Simulation Package (VASP) [66] with settings according to Refs. [32] and [57].

Results and Discussion

Implementation: corrections for oxides. We first demonstrate the capabilities of our AFLOW-CCE implementation on our previous data for oxides included in the Supporting Information of Ref. [32]. While absolute formation enthalpies are on the order of several eV/atom, Figure 2 depicts the difference between computed (DFT+AGL) and measured room temperature formation enthalpies for 79 binary (a) and 71 ternary (b) oxides also including a visualization of the vibrational contribution in the lower panel in both cases. The errors for all functionals are substantial with SCAN delivering the most accurate results still showing overall MAEs of at least about 100 meV/atom. The MAEs over the whole binary set are 235, 176, and 105 meV/atom for PBE, LDA, and SCAN, respectively.

Formation enthalpies and vibrational contribution for binary and ternary oxides

Figure 2. Formation enthalpies and vibrational contribution for binary oxides. (a) Differences between calculated and experimental room temperature formation enthalpies of 79 binary oxides (upper panel) and vibrational (zero-point + thermal) contribution to the calculated formation enthalpy (lower panel) for three standard DFT functionals. Vertical blue lines separate blocks with different ll-character of the cations. (b) Differences between calculated and experimental room temperature formation enthalpies of 71 ternary oxides (upper panel) and vibrational contribution (lower panel). Adapted from Ref. 32.

In general, PBE underestimates the (absolute) formation enthalpies meaning that they are less negative than the experimental values while LDA and SCAN mostly over-estimate them. For the binaries, the ll-character of the cation has, however, a strong influence. While SCAN yields constantly very accurate enthalpies with a MAE as small as 27 meV/atom for ss-element oxides, all functionals show a systematic decrease of the computed values for heavier pp-oxides which can be related to increasing covalency [32]. For the dd-oxides, the systematic trends of the functionals are diminished and large individual errors ranging up to ∼800 meV/atom are observed increasing also the MAEs to over 150 meV/atom. This drastic behavior for the rock salt structure dd-oxides VO, MnO, FeO, CoO, and NiO is well known [27, 32]. For the ternaries, mostly including transition metal elements, the large errors are confirmed with MAEs of at least 100 meV/atom for each functional.

The vibrational contribution is very small in all cases reaching a max about 20 meV/atom for Al2O3Al_2O_3 and kyanite Al2SiO5Al_2SiO_5 and as such are about two orders of magnitude smaller than absolute formation enthalpies.

The results from the binaries are then used to obtain the CCE corrections and to apply them to the ternary test set. Figure 3 shows the example of the corrected SCAN results where we also compare to an implementation of the FERE method for our dataset (quasi-FERE). Both methods significantly decrease the plain DFT errors by a factor of 2 to 4 with the CCE MAE of 27 meV/atom being about half of the quasi-FERE one of 44 meV/atom. The latter accuracy agrees well with the previously reported MAEs of about 50 meV/atom of the other correction schemes [26, 27]. The CCE errorbar is thus on the order of room temperature thermal energy and may as such enable more exact thermodynamic stability predictions. CCE achieves the same accuracy for ternary halides [32]. Since it is based on the bonding topology in materials, the method can also correct the relative stability of different polymorphs as demonstrated for several systems and can yield accurate defect energies as exemplified for Ti-O Magnéli phases [32]. For a few systems with large errors after correction in Figure 3 for both methods (CCE and quasi-FERE) such as FeAl2O4FeAl_2O_4, the experimental values might be inaccurate and should potentially be revised.

Corrected SCAN enthalpies for ternary oxides

Figure 3. Corrected SCAN enthalpies for ternary oxides. Differences between corrected and experimental room temperature formation enthalpies of 71 ternary oxides for SCAN. In addition to the CCE results, a comparison to an implementation of the FERE method [25] for our dataset (quasi-FERE) [32] is also included. The red lines at ±50 meV/atom indicate the typical MAE of previous correction schemes [26, 27]. Adapted from Ref. [32].

Implementation: corrections for nitrides. The study of a suitable set of binary materials will first allow uncovering the errors of uncorrected DFT for this materials class, as well as deriving the corrections. The predictive power of the corrections will then be validated for ternary nitrides. The computed and corrected data of this part are included in the Appendix.

For an overview, in Figure 4, elements forming binary nitrogen compounds with tabulated high quality thermochemical data in the standard collections [33, 34, 35, 36] are circled. Nitrides are much more scarce than oxides since there are only 20 binaries. The major reason is that the N2N_2 molecular reference phase is thermodynamically exceptionally stable which makes it difficult to form a nitrogen compound with a net energy gain. Recent computa-

Phase space of binary nitrides

Figure 4. Phase space of binary nitrides. Periodic table with elements forming binary nitrides for which validated thermochemical data are available in standard collections [33, 34, 35, 36] highlighted by circles. Blue indicates the formation of rock salt structures and red stands for other structural prototypes.

tional efforts have therefore also been devoted to predict metastable nitrides from suitable reactive nitrogen precursors [67] and to build large stability maps of the uncharted nitrides materials space [68]. We note that in the set of known binaries in Figure 4 there is a predominance of rock salt structures indicated in blue as all dd-element nitrides except Zn₃N₂ crystallize in this prototype. These are expected to be very suitable precursors for high-entropy systems since many high-entropy carbides and oxides have this structure [17, 69].

Figure 5 shows the difference between calculated DFT+AGL and experimental room temperature formation enthalpies, as well as the vibrational contribution in the lower panel. The MAEs for the nitrides are generally

larger as for the oxides with PBE again mostly underestimating the enthalpies with respect to experiment while LDA and SCAN typically lead to an overestimation. Although there has been some debate about whether the tabulated experimental ΔfH\Delta_{\mathrm{f}}H for GaN [33] should be updated with a more negative value [25, 70], we stick to the old result since only in this case it fits to the overall behavior that LDA and SCAN overestimate the formation enthalpy whereas PBE does the opposite as observed for BN, AlN, and InN. With the small vibrational contribution being at max of the order of 10-20 meV/atom with no significant differences between the approximations, the main error arises once more from the internal (DFT) energy contribution. Over the whole set, the three functionals exhibit significant MAEs of 161, 229, and 155 meV/atom for PBE, LDA, and SCAN, respectively outlining again the meta-GGA as most accurate. There is nevertheless a strong dependence on the ll-character of the cation species. While for the ss-element nitrides, SCAN exhibits the by far smallest MAE of 69 meV/atom, PBE performs best for the pp-compounds and yields about the same average error as SCAN for the dd-systems. LDA shows the largest errors for all three groups in strict contrast to the oxides where it is significantly better than PBE and, for the ternaries, even close to SCAN. Thus, this success of LDA is not systematic for all materials classes. It should be noted that, compared to the oxides, where the experimental errorbar is likely on the order of 10-20 meV/atom [32], the average error of the Kubaschewski values [33] given for all binaries except TiN is ∼33 meV/atom. The results from the binaries are then used to obtain the CCE corrections for nitrides which are listed in Table I including also the experimental formation enthalpies per bond CCE@exp [32, 49]. As shown by us previously in Refs. [32] and [49], the vibrational contribution does not need to be taken into account explicitly and the corrections can be directly fitted to the plain DFT formation enthalpies. This gives two types of corrections: (i) for room temperature (298.15 K) and (ii) for 0 K, meaning that when they are applied to the plain DFT enthalpies of target compounds, they yield corrected estimates for the respective temperature.

Formation enthalpies and vibrational contribution for binary nitrides

Figure 5. Formation enthalpies and vibrational contribution for binary nitrides. Differences between calculated and experimental room temperature formation enthalpies of 20 binary nitrides (upper panel) and vibrational (zero-point + thermal) contribution to the calculated formation enthalpy (lower panel) for three standard DFT functionals. Vertical blue lines separate blocks with different ll-character of the cations.

cation species AA+α+\alphaPBE 298.15 KPBE 0 KLDA 298.15 KLDA 0 KSCAN 298.15 KSCAN 0 KCCE@exp 298.15 K
Li+10.016630.00912−0.06850−0.07562−0.03588−0.04338−0.21350
Be+20.059080.05292−0.06500−0.07133−0.03433−0.04075−0.50917
B+30.00100−0.00867−0.15933−0.17000−0.11000−0.12067−0.86700
Mg+20.083000.07642−0.03475−0.04167−0.01242−0.01950−0.39858
Al+30.118750.10850−0.04575−0.05575−0.03850−0.04875−0.82500
Si+40.00000−0.01300−0.22408−0.23733−0.15183−0.16508−0.64325
Ca+20.039670.03400−0.09225−0.09792−0.04392−0.04942−0.37225
Sc+3−0.09850−0.10500−0.23150−0.23833−0.20667−0.21350−0.54200
Ti+30.00367−0.00283−0.14550−0.15233−0.08733−0.09400−0.58400
V+30.045000.03917−0.09483−0.10100−0.01683−0.02250−0.37650
Cr+30.084170.08167−0.02100−0.02583−0.01867−0.02233−0.20250
Zn+20.067330.06225−0.03492−0.039080.00417−0.00025−0.01950
Ga+30.047250.03950−0.11725−0.12550−0.04200−0.05075−0.28400
Y+3−0.06283−0.06750−0.18267−0.18783−0.16950−0.17450−0.51683
Zr+30.042500.03733−0.09933−0.10500−0.04067−0.04617−0.63100
Nb+30.074670.06967−0.06550−0.070830.026670.02150−0.40833
In+30.088250.08350−0.05700−0.06225−0.01125−0.01650−0.04450
La+30.083000.07983−0.00100−0.00433−0.01600−0.01950−0.52400
Hf+30.059830.05633−0.09283−0.09683−0.01433−0.01817−0.64533
Ta+30.152830.149330.006170.002330.131000.12733−0.43583

Table 1. CCE corrections for 298.15 and 0 K for nitrides. Corrections per bond δHA−YT,A+α\delta H^{T,A+\alpha}_{A-Y} of the CCE method for each cation species AA in oxidation states +α+\alpha for 298.15 and 0 K obtained from binary nitrides for the different functionals. The experimental formation enthalpies per bond that can be used to obtain a rough estimate of the ΔfH\Delta_\mathrm{f}H value of a material without DFT calculations according to Ref. 32 and [49], are given in the last column. All corrections are in eV/bond.

The amount of ternary nitrides in particular with validated thermochemical and structural data which are also available in the AFLOW-ICSD database is, however, very scarce, with no entries in the standard collections that were used for the oxides. In Figure 6 we present both the uncorrected and corrected results for three compounds with measured formation enthalpies reported in Refs. [71, 72]. As expected, the plain DFT results show large errors for all three functionals (MAEs of 85, 162, and 88 meV/atom for PBE, LDA, and SCAN). When CCE is applied, the MAEs reduce to 16, 18, and 21 meV/atom again confirming the high accuracy of CCE enthalpies although of course average values for a set of only three entries must be taken with some care.

Uncorrected and corrected enthalpies for ternary nitrides

Figure 6. Uncorrected and corrected enthalpies for ternary nitrides. Differences between calculated (a) as well as corrected (b) and experimental room temperature formation enthalpies for seven ternary nitrides. The red lines at ±50 meV/atom indicate the typical MAE of previous correction schemes [26, 27].

The oxide and nitride corrections are available from the AFLOW-CCE module which automatically determines the oxidation numbers of all ions and only needs a structure as input [49]. It has been integrated into the AFLOW software [4] of version 3.2.7 or later for which the source code is available at http://aflow.org/installaflow/ and http://materials.duke.edu/AFLOW/. The implementation also includes a Python environment that is distributed with the AFLOW source and can be generated with the command aflow --cce --print=python. The CCE web tool [5] can be accessed via: http://aflow.org/aflow-online/. The main commands for the AFLOW-CCE command line tool are summarized in Table II. Instructions are available through the AFLOW-School: http://aflow.org/aflow-school/. All options and further details are described in Ref. [49].

CommandDescription
aflow --ccePrints user instructions.
aflow --cce=STRUCTURE\_FILE\_PATHProvides the output of the full CCE analysis, i.e. cation coordination numbers, oxidation numbers, and CCE corrections and formation enthalpies, for the structure in STRUCTURE_FILE_PATH.
aflow --get\_cce\_corrections < STRUCTURE\_FILE\_PATHReturns the CCE corrections and formation enthalpies for the structure in STRUCTURE_FILE_PATH.
aflow --get\_oxidation\_number < STRUCTURE\_FILE\_PATHGives the oxidation numbers for the structure in STRUCTURE_FILE_PATH.
aflow --get\_cation\_coord\_num < STRUCTURE\_FILE\_PATHDetermines the cation coordination numbers for the structure in STRUCTURE_FILE_PATH.

Table II. Main commands for the AFLOW-CCE command line tool.

Future implementations. The CCE parameterization will be extended to all other relevant anion classes. These include: (i) all halides, i.e. fluorides, chlorides, bromides, and iodides; (ii) all additional chalcogenides, i.e. sulfides, selenides, and tellurides; (iii) additional pnictides, i.e. phosphides, arsenides, and potentially antimonides; and (iv) possibly hydrides. The method will also be leveraged for the challenging design of disordered high-entropy ceramics by providing an ensemble of corrections for the partial occupation (AFLOW-POCC) [64] algorithm used to model these materials. It is also expected to become very useful for energetic corrections of nanostructures and in surface science based on the bonding topology, where for instance, the relative stability of different quasi 2D systems and with respect to bulk materials [19, 20, 21, 22, 23] must be estimated accurately. The temperature dependence can be implemented in a dynamic fashion allowing for enthalpy estimates at any given temperature on the fly.

Conclusions

We have presented the computational capabilities of the AFLOW-CCE software for thermodynamic stability predictions utilizing our previous oxide data and new results for nitrides. When calculating the formation enthalpies from plain DFT for several standard functionals such as LDA, PBE, and SCAN, average errors compared to high-quality experimental data of the order of several hundred meV/atom are found. For oxides, SCAN shows the best performance achieving ∼100 meV/atom closely followed by LDA. Also for nitrides, SCAN depicts the smallest mean errors slightly above 150 meV/atom almost matched by PBE with LDA revealing larger deviations.

Since the accuracy of these approximations is too low for materials design at room temperature (∼25 meV), we employ the AFLOW-CCE implementation of our previously developed coordination corrected enthalpies method for the accurate evaluation of the thermodynamic stability of ionic materials. CCE is based on an intuitive parametrization of DFT errors with respect to coordination numbers and cation oxidation states. The method has been benchmarked for the formation enthalpies of oxides and nitrides revealing mean errors on the order of ∼25 meV/atom. In addition, it can rectify errors in relative stability predictions of polymorphs from plain DFT. The corrections are freely available in the AFLOW-CCE module for the automatic correction of enthalpies based on only structural inputs. The method and its implementation can be valuable for correcting the results in large materials databases to enable advanced materials predictions.

Acknowledgments

The authors thank David Hicks, Cormac Toher, Ohad Levy, Simon Divilov, Hagen Eckert, Michael Mehl, Adam Zettel, Corey Oses, Arrigo Calzolari, Arkady V. Krasheninnikov, Thomas Heine, and Xiomara Campilongo for fruitful discussions. S.C. acknowledges the DoD SPICES MURI sponsored by the Office of Naval Research (Naval Research contract N00014-21-2515) for financial support. R.F. acknowledges funding for the “Autonomous Materials Thermodynamics” (AutoMaT) project by Technische Universität Dresden and Helmholtz-Zentrum Dresden-Rossendorf within the DRESDEN-concept alliance. R.F. acknowledges support from the Alexander von Humboldt foundation under the Feodor Lynen research fellowship. The authors thank the HZDR Computing Center, HLRS Stuttgart (HAWK cluster), the Paderborn Center for Parallel Computing (PC2, Noctua 2 cluster), and the DoD High Performance Computing Modernization Program for computational support.

Tables with numerical data

TABLE III: Structural data for binary and ternary nitrides. ICSD numbers, space group numbers, Pearson symbols, and AFLOW prototype labels [73, 74, 75] for the 20 binary and three ternary nitrides. Space-groups and Pearson symbols are calculated with AFLOW-SYM [76]. Note: the order of the Wyckoff position letters in the prototype column follows the alphabetic syntax to generate the prototype, while the associated web-link points to the standardized description in the AFLOW Encyclopedia of Crystallographic Prototypes [73, 74, 75].

formulaICSD #space group #Pearson symbolAFLOW prototype
Li3NLi_3N76944191hP4A3B_hP4_191_bc_a
Be3N2Be_3N_2616348206cI80A3B2_cI80_206_e_bd
Mg3N2Mg_3N_2411210206cI80A3B2_cI80_206_e_bd
Ca3N2Ca_3N_250991206cI80A3B2_cI80_206_e_bd
BN186246194hP4AB_hP4_194_d_c
AlN602459186hP4AB_hP4_186_b_b
Si3N4Si_3N_474744176hP14A4B3_hP14_176_ch_h
GaN156259186hP4AB_hP4_186_b_b
InN157515186hP4AB_hP4_186_b_b
ScN26948225cF8AB_cF8_225_b_a
TiN64907225cF8AB_cF8_225_b_a
VN76526225cF8AB_cF8_225_b_a
CrN41827225cF8AB_cF8_225_a_b
Zn3N2Zn_3N_284918206cI80A2B3_cI80_206_ad_e
YN76528225cF8AB_cF8_225_b_a
ZrN167851225cF8AB_cF8_225_a_b
NbN183423225cF8AB_cF8_225_b_a
LaN641462225cF8AB_cF8_225_a_b
HfN53025225cF8AB_cF8_225_a_b
TaN644727225cF8AB_cF8_225_a_b
Li3AlN2Li_3AlN_225565206cI96AB3C2_cI96_206_c_e_bd
LiCaN10730462oP12ABC_oP12_62_c_c
Ca2ZnN2Ca_2ZnN_269049139tI10A2B2C_tI10_139_e_e_a

Table III.

formulaPBE+AGLPBELDA+AGLLDASCAN+AGLSCANExp.NA−YN_{\mathrm{A-Y}}+α+\alpha
Li3NLi_{3}N−0.389−0.394−0.560−0.564−0.494−0.499−0.427 [33]8+1
Be3N2Be_{3}N_{2}−1.069−1.080−1.365−1.378−1.291−1.304−1.222 [33]12+2
Mg3N2Mg_{3}N_{2}−0.759−0.757−1.039−1.040−0.986−0.986−0.957 [33]12+2
Ca3N2Ca_{3}N_{2}−0.806−0.798−1.122−1.115−1.006−0.999−0.893 [35]12+2
BNBN−1.297−1.299−1.532−1.539−1.459−1.466−1.300 [34]3+3
AlNAlN−1.405−1.413−1.735−1.742−1.719−1.727−1.650 [33]4+3
Si3N4Si_{3}N_{4}−1.094−1.103−1.476−1.487−1.352−1.363−1.103 [33]12+4
GaNGaN−0.481−0.473−0.808−0.802−0.658−0.652−0.568 [33]4+3
InNInN0.0740.087−0.215−0.203−0.124−0.111−0.089 [35]4+3
ScNScN−1.922−1.922−2.318−2.320−2.245−2.246−1.626 [33]6+3
TiNTiN−1.738−1.741−2.182−2.188−2.009−2.014−1.752 [33]6+3
VNVN−0.992−0.994−1.408−1.414−1.180−1.180−1.129 [33]6+3
CrNCrN−0.366−0.355−0.672−0.670−0.670−0.663−0.608 [33]6+3
Zn3N2Zn_{3}N_{2}0.1090.115−0.138−0.131−0.044−0.037−0.047 [33]12+2
YNYN−1.749−1.739−2.107−2.098−2.068−2.059−1.550 [33]6+3
ZrNZrN−1.771−1.765−2.195−2.191−2.020−2.015−1.893 [34]6+3
NbNNbN−1.006−1.001−1.424−1.421−1.149−1.145−1.225 [33]6+3
LaNLaN−1.337−1.323−1.589−1.575−1.634−1.620−1.572 [35]6+3
HfNHfN−1.768−1.757−2.225−2.215−1.990−1.979−1.936 [33]6+3
TaNTaN−0.860−0.849−1.299−1.289−0.924−0.914−1.308 [33]6+3
Li3AlN2Li_{3}AlN_{2}−0.853−0.865−1.109−1.120−1.053−1.065−0.981 [72]12, 4+1, +3
LiCaNLiCaN−0.667−0.663−0.929−0.926−0.830−0.827−0.562 [71]3, 4+1, +2
Ca2ZnN2Ca_{2}ZnN_{2}−0.718−0.711−1.006−0.999−0.914−0.907−0.785 [71]10, 2+2, +2

Table IV. Formation enthalpies for binary and ternary nitrides. Calculated DFT+AGL, plain DFT and experimental room temperature formation enthalpies, number of cation-anion bonds NA−YN_{\mathrm{A-Y}} per formula unit and oxidation states +α+\alpha of the cation(s) for binary and ternary nitrides. In case of ternaries, for the cation-nitrogen bonds- and oxidation numbers, the first (second) number in the column refers to the first (second) element in the formula. Experimental values are from Kubaschewski et al. [33], NIST-JANAF [34], Barin [35], and McHale et al. [71, 72]. Enthalpies are in eV/atom; corrections are in eV/bond.

formulaPBELDASCAN
AGL-TVCAGL-ZPCAGL-TCAGL-TVCAGL-ZPCAGL-TCAGL-TVCAGL-ZPCAGL-TC
Li3NLi_{3}N0.0040.019−0.0150.0040.018−0.0140.0050.020−0.015
Be3N2Be_{3}N_{2}0.0110.026−0.0150.0130.028−0.0150.0140.029−0.015
Mg3N2Mg_{3}N_{2}−0.0010.015−0.0160.0010.017−0.0170.0010.017−0.017
Ca3N2Ca_{3}N_{2}−0.0070.006−0.014−0.0070.006−0.014−0.0070.006−0.013
BNBN0.0020.017−0.0150.0080.024−0.0160.0070.022−0.016
AlNAlN0.0070.028−0.0210.0070.027−0.0200.0070.028−0.021
Si3N4Si_{3}N_{4}0.0090.031−0.0220.0110.034−0.0230.0110.034−0.023
GaNGaN−0.0070.008−0.016−0.0060.010−0.016−0.0060.012−0.018
InNInN−0.014−0.004−0.009−0.013−0.002−0.010−0.013−0.002−0.011
ScNScN0.0000.019−0.0200.0020.023−0.0210.0010.021−0.020
TiNTiN0.0030.022−0.0190.0060.026−0.0200.0050.025−0.020
VNVN0.0020.020−0.0180.0060.024−0.0190.0000.017−0.017
CrNCrN−0.011−0.004−0.008−0.0010.014−0.015−0.0060.005−0.011
Zn3N2Zn_{3}N_{2}−0.0060.006−0.012−0.0070.003−0.010−0.0070.004−0.011
YNYN−0.0100.004−0.014−0.0080.007−0.015−0.0090.006−0.015
ZrNZrN−0.0060.010−0.016−0.0040.013−0.017−0.0050.011−0.016
NbNNbN−0.0050.010−0.015−0.0030.013−0.016−0.0040.012−0.016
LaNLaN−0.014−0.005−0.009−0.014−0.003−0.010−0.014−0.003−0.011
HfNHfN−0.012−0.001−0.011−0.0110.001−0.012−0.0110.001−0.012
TaNTaN−0.011−0.001−0.010−0.0100.002−0.011−0.0100.001−0.011
Li3AlN2Li_{3}AlN_{2}0.0120.032−0.0210.0110.031−0.0200.0120.033−0.021
LiCaNLiCaN−0.0030.010−0.013−0.0030.010−0.013−0.0030.010−0.013
Ca2ZnN2Ca_{2}ZnN_{2}−0.0070.006−0.013−0.0070.005−0.012−0.0070.005−0.012

Table V. AGL contributions to the formation enthalpies for binary and ternary nitrides. Total vibrational (TVC), zero-point (ZPC) and thermal (TC) contributions to the calculated formation enthalpies obtained from AGL [50, 51, 52, 53, 54] for each functional for binary and ternary nitrides. The sum of ZPC and TC might not match exactly the total AGL contribution listed due to rounding. All values are in eV/atom.

formulaCCE@ PBECCE@ LDACCE@ SCANExp.
Li3AlN2Li_{3}AlN_{2}−0.978-0.978−0.953-0.953−0.967-0.967−0.981-0.981 [72]
LiCaNLiCaN−0.550-0.550−0.551-0.551−0.549-0.549−0.562-0.562 [71]
Ca2ZnN2Ca_{2}ZnN_{2}−0.818-0.818−0.800-0.800−0.821-0.821−0.785-0.785 [71]

Table VI. CCE formation enthalpies for ternary nitrides. CCE formation enthalpies (using the corrections from Table I) and experimental room temperature values for ternary nitrides. Experimental data are from McHale et al. [71, 72]. All enthalpies are in eV/atom.

References

References

  1. [1]J. P. Perdew and A. Zunger, Self-interaction correction to density-functional approximations for many-electron systems, Phys. Rev. B 23, 5048–5079 (1981), doi: 10.1103/PhysRevB.23.5048.DOI
  2. [2]J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized Gradient Approximation Made Simple, Phys. Rev. Lett. 77, 3865–3868 (1996), doi:10.1103/PhysRevLett.77.3865.DOI
  3. [3]J. Sun, A. Ruzsinszky, and J. P. Perdew, Strongly Constrained and Appropriately Normed Semilocal Density Functional, Phys. Rev. Lett. 115, 036402 (2015), doi: 10.1103/PhysRevLett.115.036402.DOI
  4. [4]C. Oses, M. Esters, D. Hicks, S. Divilov, H. Eckert, R. Friedrich, M. J. Mehl, A. Smolanyuk, X. Campilongo, A. van de Walle, J. Schroers, A. G. Kusne, I. Takeuchi, E. Zurek, M. Buongiorno Nardelli, M. Fornari, Y. Lederer, O. Levy, C. Toher, and S. Curtarolo, aflow++: A C++ framework for autonomous materials design, Comput. Mater. Sci. 217, 111889 (2023), doi:10.1016/j.commatsci.2022.111889.DOI
  5. [5]M. Esters, C. Oses, S. Divilov, H. Eckert, R. Friedrich, D. Hicks, M. J. Mehl, F. Rose, A. Smolanyuk, A. Calzolari, X. Campilongo, C. Toher, and S. Curtarolo, aflow.org: A web ecosystem of databases, software and tools, Comput. Mater. Sci. 216, 111808 (2023), doi: 10.1016/j.commatsci.2022.111808.DOI
  6. [6]S. Curtarolo, W. Setyawan, S. Wang, J. Xue, K. Yang, R. H. Taylor, L. J. Nelson, G. L. W. Hart, S. Sanvito, M. Buongiorno Nardelli, N. Mingo, and O. Levy, AFLOWLIB.ORG: A distributed materials properties repository from high-throughput ab initio calculations, Comput. Mater. Sci. 58, 227–235 (2012), doi: 10.1016/j.commatsci.2012.02.002.
  7. [7]A. Jain, G. Hautier, C. J. Moore, S. P. Ong, C. C. Fischer, T. Mueller, K. A. Persson, and G. Ceder, A high-throughput infrastructure for density functional theory calculations, Comput. Mater. Sci. 50, 2295–2310 (2011), doi: 10.1016/j.commatsci.2011.02.023.DOI
  8. [8]J. E. Saal, S. Kirklin, M. Aykol, B. Meredig, and C. Wolverton, Materials Design and Discovery with High-Throughput Density Functional Theory: The Open Quantum Materials Database (OQMD), JOM 65, 1501–1509 (2013), doi:10.1007/s11837-013-0755-4.DOI
  9. [9]S. Kirklin, J. E. Saal, B. Meredig, A. Thompson, J. W. Doak, M. Aykol, S. Rühl, and C. Wolverton, The Open Quantum Materials Database (OQMD): assessing the accuracy of DFT formation energies, npj Comput. Mater. 1, 15010 (2015), doi:10.1038/npjcompumats.2015.10.DOI
  10. [10]C. Draxl and M. Scheffler, NOMAD: The FAIR concept for big data-driven materials science, MRS Bull. 43, 676–682 (2018), doi:10.1557/mrs.2018.208.DOI
  11. [11]S. R. Bahn and K. W. Jacobsen, An object-oriented scripting interface to a legacy electronic structure code, Comput. Sci. Eng. 4, 56–66 (2002), doi:10.1109/5992.998641.DOI
  12. [12]D. R. Landis, J. S. Hummelshøj, N. Nestorov, J. Greeley, M. Dulak, T. Bligaard, J. K. Nørskov, and K. W. Jacobsen, The Computational Materials Repository, Comput. Sci. Eng. 14, 51–57 (2012), doi:10.1109/MCSE.2012.16.DOI
  13. [13]G. Pizzi, A. Cepellotti, R. Sabatini, N. Marzari, and B. Kozinsky, AiiDA: automated interactive infrastructure and database for computational science, Comput. Mater. Sci. 111, 218–230 (2016).DOI
  14. [14]P. A. M. Dirac, Quantum Mechanics of Many-Electron Systems, Proc. R. Soc. A Math. Phys. Eng. Sci. 123, 714–733 (1929), doi:10.1098/rspa.1929.0094.DOI
  15. [15]M. J. Mehl, M. Ronquillo, D. Hicks, M. Esters, C. Oses, R. Friedrich, A. Smolanyuk, E. Gossett, D. Finkenstadt, and S. Curtarolo, The Tin Pest Problem as a Test of Density Functionals Using High-Throughput Calculations, Phys. Rev. Materials 5, 083608 (2021), doi: 10.1103/PhysRevMaterials.5.083608.DOI
  16. [16]P. Sarker, T. Harrington, et al., High-entropy high-hardness metal carbides discovered by entropy descriptors, Nat. Commun. 9, 4980 (2018), doi:10.1038/s41467-018-07160-7.DOI
  17. [17]C. Oses, C. Toher, and S. Curtarolo, High-entropy ceramics, Nat. Rev. Mater. 5, 295–309 (2020), doi: 10.1038/s41578-019-0170-8.DOI
  18. [18]J. Gil and T. Oda, Correction methods for first-principles calculations of the solution enthalpy of gases and compounds in liquid metals, Phys. Chem. Chem. Phys. 24, 757–770 (2022), doi:10.1039/D1CP02450G.DOI
  19. [19]A. Puthirath Balan, S. Radhakrishnan, C. F. Woellner, S. K. Sinha, L. Deng, C. d. l. Reyes, B. M. Rao, M. Paulose, R. Neupane, A. Apte, V. Kochat, R. Vajtai, A. R. Harutyunyan, C.-W. Chu, G. Costin, D. S. Galvao, A. A. Martì, P. A. van Aken, O. K. Varghese, C. S. Tiwary, A. Malie Madom Ramaswamy Iyer, and P. M. Ajayan, Exfoliation of a non-van der Waals material from iron ore hematite, Nat. Nanotechnol. 13, 602–609 (2018), doi:10.1038/s41565-018-0134-y.
  20. [20]R. Friedrich, M. Ghorbani-Asl, S. Curtarolo, and A. V. Krasheninnikov, Data-Driven Quest for Two-Dimensional Non-van der Waals Materials, Nano Lett. 22, 989–997 (2022), doi:10.1021/acs.nanolett.1c03841.DOI
  21. [21]T. Barnowsky, A. V. Krasheninnikov, and R. Friedrich, A New Group of 2D Non-van der Waals Materials with Ultra Low Exfoliation Energies, Adv. Electron. Mater. 9, 2201112 (2023), doi:10.1002/aelm.202201112.DOI
  22. [22]A. P. Balan, A. B. Puthirath, S. Roy, G. Costin, E. F. Oliveira, M. A. S. R. Saadi, V. Sreepal, R. Friedrich, P. Serles, A. Biswas, S. A. Iyengar, N. Chakingal, S. Bhat-tacharya, S. K. Saju, S. C. Pardo, L. M. Sassi, T. Fil- leter, A. Krasheninnikov, D. S. Galvao, R. Vajtai, R. R. Nair, and P. M. Ajayan, Non-van der Waals quasi-2D materials: recent advances in synthesis, emergent properties and applications, Mater. Today 58, 164 (2022), doi: 10.1016/j.mattod.2022.07.007.
  23. [23]H. Kaur and J. N. Coleman, Liquid-Phase Exfoliation of Nonlayered Non-Van-Der-Waals Crystals into Nanoplatelets, Adv. Mater. 34, 2202164 (2022), doi: 10.1002/adma.202202164.DOI
  24. [24]L. Wang, T. Maxisch, and G. Ceder, Oxidation energies of transition metal oxides within the GGA+U framework, Phys. Rev. B 73, 195107 (2006), doi: 10.1103/PhysRevB.73.195107.
  25. [25]S. Lany, Semiconductor thermochemistry in density functional calculations, Phys. Rev. B 78, 245207 (2008), doi: 10.1103/PhysRevB.78.245207.DOI
  26. [26]A. Jain, G. Hautier, S. P. Ong, C. J. Moore, C. C. Fischer, K. A. Persson, and G. Ceder, Formation enthalpies by mixing GGA and GGA+U calculations, Phys. Rev. B 84, 045115 (2011), doi:10.1103/PhysRevB.84.045115.DOI
  27. [27]V. Stevanović, S. Lany, X. Zhang, and A. Zunger, Correcting density functional theory for accurate predictions of compound enthalpies of formation: Fitted elemental-phase reference energies, Phys. Rev. B 85, 115104 (2012), doi:10.1103/PhysRevB.85.115104.DOI
  28. [28]J. Yan, J. S. Hummelshøj, and J. K. Nørskov, Formation energies of group I and II metal oxides using random phase approximation, Phys. Rev. B 87, 075207 (2013), doi:10.1103/PhysRevB.87.075207.
  29. [29]T. S. Jauho, T. Olsen, T. Bligaard, and K. S. Thygesen, Improved description of metal oxide stability: Beyond the random phase approximation with renormalized kernels, Phys. Rev. B 92, 115140 (2015), doi:10.1103/PhysRevB.92.115140.DOI
  30. [30]Y. Zhang, D. A. Kitchae,v J. Yang, T. Chen, S. T. Dacek, R. A. Sarmiento-Pérez, M. A. L. Marques, H. Peng, G. Ceder, J. P. Perdew, and J. Sun, Efficient first-principles prediction of solid stability: Towards chemical accuracy, npj Comput. Mater. 4, 9 (2018), doi: 10.1038/s41524-018-0065-z.DOI
  31. [31]E. B. Isaacs and C. Wolverton, Performance of the strongly constrained and appropriately normed density functional for solid-state materials, Phys. Rev. Materials 2, 063801 (2018), doi: 10.1103/PhysRevMaterials.2.063801.DOI
  32. [32]R. Friedrich, D. Usanmaz, C. Oses, A. Supka, M. Fornari, M. Buongiorno Nardelli, C. Toher, and S. Curtarolo, Coordination corrected ab initio formation enthalpies, npj Comput. Mater. 5, 59 (2019), doi:10.1038/s41524-019-0192-1.DOI
  33. [33]O. Kubaschewski, C. B. Alcock, and P. J. Spencer, Materials Thermochemistry (Pergamon Press, Oxford, UK, 1993), 6th edn.
  34. [34]M. W. Chase, Jr., NIST-JANAF Thermochemical Tables (American Chemical Society and American Institute of Physics for the National Institute of Standards and Technology, Woodbury, NY, 1998), 4th edn.
  35. [35]I. Barin, Thermochemical Data of Pure Substances (VCH, Weinheim, 1995), 3rd edn.DOI
  36. [36]D. D. Wagman, W. H. Evans, V. B. Parker, R. H. Schumm, I. Halow, S. M. Bailey, K. L. Churney, and R. L. Nuttall, The NBS tables of chemical thermodynamic prop- erties, J. Phys. Chem. Ref. Data 11, Supplement No. 2 (1982).
  37. [37]M. R. Pederson, A. Ruzsinszky, and J. P. Perdew, Communication: Self-interaction correction with unitary invariance in density functional theory, J. Chem. Phys. 140, 121103 (2014), doi:10.1063/1.4869581.DOI
  38. [38]Z.-h. Yang, M. R. Pederson, and J. P. Perdew, Full self-consistency in the Fermi-orbital self-interaction correction, Phys. Rev. A 95, 052505 (2017), doi: 10.1103/PhysRevA.95.052505.DOI
  39. [39]D.-y. Kao, K. Withanage, T. Hahn, J. Batool, J. Kortus, and K. Jackson, Self-consistent self-interaction corrected density functional theory calculations for atoms using Fermi-Löwdin orbitals: Optimized Fermi-orbital descriptors for Li-Kr, J. Chem. Phys. 147, 164107 (2017), doi:10.1063/1.4996498.DOI
  40. [40]S. Schwalbe, T. Hahn, S. Liebling, K. Trepte, and J. Kortus, Fermi-Löwdin Orbital Self-interaction Corrected Density Functional Theory: Ionization Potentials and Enthalpies of Formation, J. Comput. Chem. 39, 2463–2471 (2018), doi:10.1002/jcc.25886.DOI
  41. [41]M. Pozzo and D. Alfè, Structural properties and enthalpy of formation of magnesium hydride from quantum Monte Carlo calculations, Phys. Rev. B 77, 104103 (2008), doi: 10.1103/PhysRevB.77.104103.DOI
  42. [42]G. Mao, X. Hu, X. Wu, Y. Dai, S. Chu, and J. Deng, Benchmark Quantum Monte Carlo calculation of the enthalpy of formation of MgH₂, Int. J. of Hydrogen Energy 36, 8388–8391 (2011), doi: 10.1016/j.ijhydene.2011.04.093.DOI
  43. [43]E. B. Isaacs, H. Shin, A. Annaberdiyev, C. Wolverton, L. Mitas, A. Benali, and O. Heinonen, Assessing the accuracy of compound formation energies with quantum Monte Carlo, Phys. Rev. B 105, 224110 (2022), doi: 10.1103/PhysRevB.105.224110.DOI
  44. [44]A. Wang, R. Kingsbury, M. McDermott, M. Horton, A. Jain, S. P. Ong, S. Dwarakanath, and K. A. Persson, A framework for quantifying uncertainty in DFT energy corrections, Sci. Rep. 11, 15496 (2021), doi:10.1038/s41598-021-94550-5.DOI
  45. [45]M. Aykol and C. Wolverton, Local environment dependent GGA+U method for accurate thermochemistry of transition metal compounds, Phys. Rev. B 90, 115105 (2014), doi:10.1103/PhysRevB.90.115105.DOI
  46. [46]N. Artith, J. A. Garrido Torres, A. Urban, and M. S. H. Ybertsen, Data-driven approach to parameterize SCAN+U for an accurate description of 3d transition metal oxide thermochemistry, Phys. Rev. Materials 6, 035003 (2022), doi:10.1103/PhysRevMaterials.6.035003.
  47. [47]R. S. Kingsbury, A. S. Rosen, A. S. Gupta, J. M. Munro, S. P. Ong, A. Jain, S. Dwarakanath, M. K. Horton, and K. A. Persson, A flexible and scalable scheme for mixing computed formation energies from different levels of theory, npj Comput. Mater. 8, 195 (2022), doi: 10.1038/s41524-022-00881-w.DOI
  48. [48]S. Gong, S. Wang, T. Xie, W. H. Chae, R. Liu, Y. Shao-Horn, and J. C. Grossman, Calibrating DFT Formation Enthalpy Calculations by Multifidelity Machine Learning, JACS Au 2, 1964–1977 (2022), doi: 10.1021/jacsau.2c00235.DOI
  49. [49]R. Friedrich, M. Esters, C. Oses, S. Ki, M. J. Brenner, D. Hicks, M. J. Mehl, C. Toher, and S. Curtarolo, Automated coordination corrected enthalpies with AFLOW-CCΕ, Phys. Rev. Materials 5, 043803 (2021), doi:10.1103/PhysRevMaterials.5.043803.
  50. [50]M. A. Blanco, A. M. Pendaś, E. Francisco, J. M. Recio, and R. Franco, Thermodynamical properties of solids from microscopic theory: Applications to MgF₂ and Al₂O₃, J. Mol. Struct.: THEOCHEM 368, 245–255 (1996), doi: 10.1016/S0166-1280(96)90571-0.DOI
  51. [51]M. A. Blanco, E. Francisco, and V. Luaña, GIBBS: isothermal-isobaric thermodynamics of solids from energy curves using a quasi-harmonic Debye model, Comput. Phys. Commun. 158, 57–72 (2004), doi: 10.1016/j.cpc.2003.12.001.
  52. [52]C. Toher, J. J. Plata, O. Levy, M. de Jong, M. Asta, M. Buongiorno Nardelli, and S. Curtarolo, High-throughput computational screening of thermal conductivity, Debye temperature, and Grüneisen parameter using a quasiharmonic Debye model, Phys. Rev. B 90, 174107 (2014), doi:10.1103/PhysRevB.90.174107.DOI
  53. [53]C. Toher, C. Oses, J. J. Plata, D. Hicks, F. Rose, O. Levy, M. de Jong, M. Asta, M. Fornari, M. Buongiorno Nardelli, and S. Curtarolo, Combining the AFLOW GIBBS and elastic libraries to efficiently and robustly screen thermomechanical properties of solids, Phys. Rev. Materials 1, 015401 (2017), doi:10.1103/PhysRevMaterials.1.015401.DOI
  54. [54]J.-P. Poirier, Introduction to the Physics of the Earth’s Interior (Cambridge University Press, 2000), 2nd edn.
  55. [55]J. J. Plata, P. Nath, D. Usanmaz, J. Carrete, C. Toher, M. de Jong, M. D. Asta, M. Fornari, M. Buongiorno Nardelli, and S. Curtarolo, An efficient and accurate framework for calculating lattice thermal conductivity of solids: AFLOW-AAPL Automatic Anharmonic Phonon Library, npj Comput. Mater. 3, 45 (2017), doi: 10.1038/s41524-017-0046-7.DOI
  56. [56]J. Sun, R. C. Remsing, Y. Zhang, Z. Sun, A. Ruzsinszky, H. Peng, Z. Yang, A. Paul, U. Waghmare, X. Wu, M. L. Klein, and J. P. Perdew, Accurate first-principles structures and energies of diversely bonded systems from an efficient density functional, Nat. Chem. 8, 9 (2016), doi:10.1038/nchem.2535.DOI
  57. [57]C. E. Calderon, J. J. Plata, C. Toher, C. Oses, O. Levy, M. Fornari, A. Natan, M. J. Mehl, G. L. W. Hart, M. Buongiorno Nardelli, and S. Curtarolo, The AFLOW standard for high-throughput materials science calculations, Comput. Mater. Sci. 108, 233–238 (2015), doi: 10.1016/j.commatsci.2015.07.019.DOI
  58. [58]W. Kohn and L. J. Sham, Self-Consistent Equations Including Exchange and Correlation Effects, Phys. Rev. 140, A1133 (1965), doi:10.1103/PhysRev.140.A1133.DOI
  59. [59]U. von Barth and L. Hedin, A local exchange-correlation potential for the spin polarized case: I, J. Phys. C: Solid State Phys. 5, 1629 (1972), doi:10.1088/0022-3719/5/13/012.
  60. [60]O. Levy, R. V. Chepulskii, G. L. W. Hart, and S. Curtarolo, The New Face of Rhodium Alloys: Revealing Ordered Structures from First Principles, J. Am. Chem. Soc. 132, 833–837 (2010), doi:10.1021/ja908789y.DOI
  61. [61]O. Levy, G. L. W. Hart, and S. Curtarolo, Structure maps for hcp metals from first-principles calculations, Phys. Rev. B 81, 174106 (2010), doi:10.1103/PhysRevB.81.174106.DOI
  62. [62]O. Levy, M. Jahnátek, R. V. Chepulskii, G. L. W. Hart, and S. Curtarolo, Ordered Structures in Rhenium Binary Alloys from First-Principles Calculations, J. Am. Chem. Soc. 133, 158–163 (2011), doi:10.1021/ja1091672.DOI
  63. [63]S. Curtarolo, W. Setyawan, G. L. W. Hart, M. Jahnátek, R. V. Chepulskii, R. H. Taylor, S. Wang, J. Xue, K. Yang, O. Levy, M. J. Mehl, H. T. Stokes, D. O. Demchenko, and D. Morgan, AFLOW: An automatic framework for high-throughput materials discovery, Comput. Mater. Sci. 58, 218–226 (2012), doi:10.1016/j.commatsci.2012.02.005.DOI
  64. [64]K. Yang, C. Oses, and S. Curtarolo, Modeling Off-Stoichiometry Materials with a High-Throughput Ab-Initio Approach, Chem. Mater. 28, 6484–6492 (2016), doi: 10.1021/acs.chemmater.6b01449.DOI
  65. [65]A. R. Supka, T. E. Lyons, L. S. I. Liyanage, P. D’Amico, R. Al Rahal Al Orabi, S. Mahatara, P. Gopal, C. Toher, D. Ceresoli, A. Calzolari, S. Curtarolo, M. Buongiorno Nardelli, and M. Fornari, AFLOWπ: A minimalist approach to high-throughput ab initio calculations including the generation of tight-binding hamiltonians, Comput. Mater. Sci. 136, 76–84 (2017), doi: 10.1016/j.commatsci.2017.03.055.DOI
  66. [66]G. Kresse and J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54, 11169–11186 (1996), doi: 10.1103/PhysRevB.54.11169.DOI
  67. [67]W. Sun, A. Holder, B. Orvañanos, E. Arca, A. Zakutayev, S. Lany, and G. Ceder, Thermodynamic Routes to Novel Metastable Nitrogen-Rich Nitrides, Chem. Mater. 29, 6936–6946 (2017), doi:10.1021/acs.chemmater.7b02399.DOI
  68. [68]W. Sun, C. J. Bartel, E. Arca, S. R. Bauers, B. Matthews, B. Orvañanos, R.-R. Chen, M. F. Toney, L. T. Schelhas, W. Tumas, J. Tate, A. Zakutayev, S. Lany, A. M. Holder, and G. Ceder, A map of the inorganic ternary metal nitrides, Nat. Mater. 18, 732–739 (2019), doi: 10.1038/s41563-019-0396-2.DOI
  69. [69]P. Sarker, T. Harrington, C. Toher, C. Oses, M. Samiee, J.-P. Maria, D. W. Bremner, K. S. Vecchio, and S. Curtarolo, High-entropy high-hardness metal carbides discovered by entropy descriptors, Nat. Commun. 9, 4980 (2018), doi:10.1038/s41467-018-07160-7.DOI
  70. [70]M. R. Ranade, F. Tessier, A. Navrotsky, V. J. Leppert, S. H. Risdub, F. J. DiSalvo, and C. M. Balkas, Enthalpy of Formation of Gallium Nitride, J. Phys. Chem. B 104, 4060–4063 (2000), doi:10.1021/jp993752s.DOI
  71. [71]J. M. McHale, A. Navrotsky, G. R. Kowach, V. E. Balbarin, and F. J. DiSalvo, Energetics of Ternary Nitrides: Li-Ca-Zn-N and Ca-Ta-N Systems, Chem. Mater. 9, 1538–1546 (1997), doi:10.1021/cm970244r.DOI
  72. [72]J. M. McHale, A. Navrotsky, and F. J. DiSalvo, Energetics of Ternary Nitride Formation in the (Li,Ca)-(B,Al)-N System, Chem. Mater. 11, 1148–1152 (1999), doi:10.1021/cm981096n.DOI
  73. [73]M. J. Mehl, D. Hicks, C. Toher, O. Levy, R. M. Hanson, G. L. W. Hart, and S. Curtarolo, The AFLOW Library of Crystallographic Prototypes: Part 1, Comput. Mater. Sci. 136, S1–S828 (2017), doi:10.1016/j.commatsci.2017.01.017.DOI
  74. [74]D. Hicks, M. J. Mehl, E. Gossett, C. Toher, O. Levy, R. M. Hanson, G. L. W. Hart, and S. Curtarolo, The AFLOW Library of Crystallographic Prototypes: Part 2, Comput. Mater. Sci. 161, S1–S1011 (2019), doi:10.1016/j.commatsci.2018.10.043.DOI
  75. [75]D. Hicks, M. J. Mehl, M. Esters, C. Oses, O. Levy, G. L. W. Hart, C. Toher, and S. Curtarolo, The AFLOW Library of Crystallographic Prototypes: Part 3, Comput. Mater. Sci. 199, 110450 (2021), doi:10.1016/j.commatsci.2021.110450.DOI
  76. [76]D. Hicks, C. Oses, E. Gossett, G. Gomez, R. H. Taylor, C. Toher, M. J. Mehl, O. Levy, and S. Curtarolo, AFLOW-SYM: platform for the complete, automatic and self-consistent symmetry analysis of crystals, Acta Crystallogr. Sect. A 74, 184–203 (2018), doi: 10.1107/S2053273318003066.DOI

Paper details

vlnc:W_6kd7kg8t
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents