SLUSCHI is a fully automated code which calculates melting points based on first-principles molecular dynamics simulations, with interface to the first-principles code VASP. Starting from the crystal structure of a solid (which the user inputs), SLUSCHI will automatically build a supercell of a proper size, prepare solid-liquid coexistence, and then employ the small-cell coexistence method to calculate the melting temperature. SLUSCHI is applicable to a wide variety of materials, thanks to the fact that density functional theory calculations are highly generalizable.

Systems Tested

Ta (0 and 200 GPa, bcc), Na (30-120 GPa, bcc/fcc), NaCl, La2Zr2O7 (La2O3-2ZrO2), the Hf-Ta-C-N systems, Al, Si, HfO2, Fe (330 GPa), Ti (bcc) and many more…

m.p.: melting point calculated by SLUSCHI.
If not specified, PAW-PBE is employed in the calculations.
HSE: melting point after HSE correction. (important for semiconductors)

MD: total number of MD trajectories

CPU hours are measured as / converted to TACC Stampede.
days: the physical time it takes. SLUSCHI is currently optimized to lower the CPU cost, rather the physical time. In order to reduce the physical time, user may manually explore temperature of interest. However, this may increase CPU hours.
“early runs” are calculations performed at early stage of method development. Hence the efficiency is relatively low compared to “sluschi”.

poor results and reason | good results

systemsm.p./KHSE/Kexpt./KDFT PPrad / Åkmesh# MDcpu hoursdaysnote
Al999±21 1054933Al10Auto 305816,00015sluschi
Ti_v1811±471941Ti_v9Auto 2049 15,50021sluschi
Ti_v1750±2519711941Ti_v10Auto 20267,90017sluschi
Ti_pv1952±451941Ti_pv10(1/4,1/4,1/4) 4820,00020sluschi
Ti_pv 60GPa2505±47?Ti_pv10(1/4,1/4,1/4) 4639,50017sluschi
Si724±471687Si10Auto 10251,0002sluschi
Si1378±2417851687Si10Auto 20197,50015el. DOS change, require HSE
Si1364±371687Si12Auto 202027,70027sluschi
diamond,100GPa4307±84250C10 gamma45313,000168sluschi
Ru_v2435±322607Ru_v10 Auto 20 3088,00037sluschi
Ru_pv2550±342607Ru_pv10(1/4,1/4,1/4) 33139,00051sluschi
Ru ternary alloys2564±40n/a pv10(1/4,1/4,1/4) 23247,00080sluschi
Hf,bcc 2562±312506Hf10(1/2,1/2,1/2)11514,90012sluschi
Hf,hcp 2122±50n/aHf10(1/2,1/2,1/2)5025,80021sluschi, hcp not stable at MT
HfO2 2327±473031valence10gamma3379,000sluschi
HfO2 PBE+U 3486±913031Hf_pv10gamma3499,600sluschi
HfO2 PBE+U 3313±813031Hf_pv12gamma25runningionic, use large $rad
ZrO22988valence10Auto 20running
Ta_v2986±413290Ta_v10(0,1/4,1/4)3832,00023early runs, low efficiency
Ta_pv3194±403290Ta_pv10(0,1/4,1/4)3854,00024early runs, low efficiency
Ta_pv_PW913066±513290Ta_pv10(0,1/4,1/4)3838,00068PW91 [1]
Ta_pv,200GPa7953±69n/aTa_pv10(1/4,1/4,1/4)80150,00048sluschi, high efficiency [4]
Na 15 GPa657±8810, 698 ?Na_pv10.4(0,1/4,1/4)5547,00024bcc, expt under dispute, e.g.,
Na 26 GPa750±16706970 ?Na_pv9.8(0,1/4,1/4)5226,000 17Zha, Boehler vs. Gregoryanz
Na 40 GPa742±17950 ?Na_pv9.4(0,1/4,1/4)7454,00037SLUSCHI results agree with
Na 55 GPa716±12810 ?Na_pv9.0(0,1/4,1/4)5664,000 31Eshet and Desjarlais (theory).
Na 80 GPa674±20700 ?Na_pv10.9(0,1/4,1/4)55170,00067fcc, expt under dispute
Na 100 GPa662±14450 ?Na_pv10.6(0,1/4,1/4)4857,00028fcc, expt under dispute
Na 120 GPa579±27?Na_pv10.4(0,1/4,1/4)88150,00077fcc, expt under dispute
NaCl1014±181074valence11gamma5924,00050early runs, low efficiency
La2Zr2O72420±2726302530Zr_v, La_sv10gamma64 575,000210[2]
Hf-Ta-C-Nvalence10 – – –[3]

1. Qi-Jun Hong and Axel van de Walle, Solid-liquid coexistence in small systems: A statistical method to calculate melting temperatures. Journal of Chemical Physics 139 (9), 094114 (2013). [DOI]
2. Qi-Jun Hong, Sergey V. Ushakov, Alexandra Navrotsky, Axel van de Walle, Combined computational and experimental investigation of the refractory properties of La2Zr2O7Acta Materialia 84, 275-282 (2015). [DOI]

  1. Qi-Jun Hong and Axel van de Walle, Prediction of the material with highest melting temperature from quantum mechanics. Physical Review B Rapid Communications, 92, 020104(R) (2015). [DOI]
  2. Ljubomir Miljacic, Steven Demers, Qi-Jun Hong and Axel van de Walle, Equation of state of solid, liquid and gaseous tantalum from first principles. Calphad: Computer Coupling of Phase Diagrams and Thermochemistry51, 133-143 (2015). [DOI (open access)].


This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License.

Please contact me at [email protected] if you have questions.

Several Notes for Users:

  1. A typical calculation with SLUSCHI takes a few days/weeks, with CPU cost ranging from 5,000 to 100,000 CPU hours. In this page, I list the systems I have studied.
  2. Try “kmesh = -1” in “”. This activates the use of a special kpoint (1/4,1/4,1/4). This provides a low-cost option (only twice the cost of gamma version), and yet it often generates reliable results for melting point.
  3. An example of SLUSCHI run is available at the repositories at the National Institute of Standards and Technology. Or follow this link on Google Drive for the same copy.

What’s new:

  1. Add feature to detect failed VASP jobs and restart them. It takes hundreds of VASP MD runs to calculate a melting temperature, and some of them may accidentally fail (e.g., running beyond walltime limit, failed to start due to queue/disk/network issue, etc.). Now SLUSCHI checks VASP MD runs and restarts failed jobs. Set “detectfail” and “maxwaithour” in
  2. Bug fixes and various improvements.

What’s new:

  1. Bug fixes and various improvements.

What’s new:

  1. “Bold sampling” mode: SLUSCHI launches MD duplicates aggressively. This strategy significantly reduces the physical time of calculations, though it may slightly increase computer cost.
  2. Heat of fusion calculations
  3. Bug fixes and various improvements.

User guide at the CALPHAD Journal.
What’s new:

  1. intpol: user does not need to guess a melting point. See “intpol”.
  2. adj_bmix: let SLUSCHI decide the value of BMIX.
  3. Bug fixes and various improvements.