New publication in Molecular Simulation

Posted January 03, 2018 at 06:28 AM | categories: publication, news | tags:

This paper is our latest work using neural networks in molecular simulation. In this work, we build a Behler-Parinello neural network potential of bulk zirconia. The potential can describe several polymorphs of zirconia, as well as oxygen vacancy defect formation energies and diffusion barriers. We show that we can use the potential to model oxygen vacancy diffusion using molecular dynamics at different temperatures, and to use that data to estimate the effective diffusion activation energy. This is further evidence of the general utility of the neural network-based potential for molecular simulations with DFT accuracy. 

@article{wang-2018-densit-funct,
  author =       {Chen Wang and Akshay Tharval and John R. Kitchin},
  title =        {A Density Functional Theory Parameterised Neural Network Model
                  of Zirconia},
  journal =      {Molecular Simulation},
  volume =       0,
  number =       0,
  pages =        {1-8},
  year =         2018,
  doi =          {10.1080/08927022.2017.1420185},
  url =          {https://doi.org/10.1080/08927022.2017.1420185},
  eprint =       { https://doi.org/10.1080/08927022.2017.1420185 },
  publisher =    {Taylor \& Francis},
}

Copyright (C) 2018 by John Kitchin. See the License for information about copying.

org-mode source

Org-mode version = 9.1.5

Discuss on Twitter

2017 in a nutshell for the Kitchin Research group

Posted December 31, 2017 at 01:21 PM | categories: news | tags:

Updated December 31, 2017 at 01:21 PM

Since the last update a lot of new things have happened in the Kitchin Research group. Below are some summaries of the group accomplishments, publications and activities for the past year.

1 Student accomplishments

Jacob Boes completed his PhD and began postdoctoral work with Thomas Bligaard at SLAC/Suncat at Stanford. Congratulations Jake!

Four new PhD students joined the group:

  1. Jenny Zhan will work on simulation of molten superalloys
  2. Mingjie Liu will work on the design of single atom alloy catalysts
  3. Yilin Yang will work on segregation in multicomponent alloys under reaction conditions
  4. Zhitao Guo is also joining the group and will be co-advised by Prof. Gellman. He will work on multicomponent alloy catalysts.

Welcome to the group!

2 Publications

Our publications and citation counts have continued to grow this year. Here is our current metrics according to Researcher ID.

We have eight new papers that are online, and two that are accepted, but not online yet. There are brief descriptions below.

2.1 Collaborative papers

larsen-2017-atomic-simul
This is a modern update on the Atomic Simulation Environment Python software. We have been using and contributing to this software for about 15 years now!
saravanan-2017-alchem-predic
This collaborative effort with the Keith group at UPitt and Anatole von Lilienfeld explored a novel approach to estimating adsorption energies on alloy surfaces.
xu-2017-first-princ
We used DFT calculations to understand epitaxial stabilization of titania films on strontium titanate surfaces.
wittkamper-2017-compet-growt
We previously predicted that tin oxide should be able to form in the columbite phase as an epitaxial film. In this paper our collaborators show that it can be done!
kitchin-2017-autom-data
This paper finally came out in print. It shows an automated approach to sharing data. Also, it may be the only paper with data hidden inside a picture of a library in the literature.

2.2 Papers on neural networks in molecular simulation

boes-2017-neural-networ
We used neural networks in conjunction with molecular dynamics and Monte Carlo simulations to model the coverage dependent adsorption of oxygen and initial oxidation of a Pd(111) surface.
boes-2017-model-segreg
We used neural networks in conjunction with Monte Carlo simulations to model segregation across composition space for a Au-Pd alloy.
geng-2017-first-princ
We used a cluster expansion with Monte Carlo simulations to resolve some inconsistencies in simulated Cu-Pd phase diagrams. There is an interesting transition from an fcc to bcc to fcc structure across the composition space that is subtle and difficult to compute.

2.3 Papers accepted in 2017 but not yet in press

  1. Chen Wang, Akshay Tharval, John R. Kitchin, A density functional theory parameterized neural network model of zirconia, Accepted in Molecular Simulation, July 2017.
  2. Hari Thirumalai, John R. Kitchin, Investigating the Reactivity of Single Atom Alloys using Density Functional Theory, Topics in Catalysis, Accepted November 2017.

3 New courses

After a five year stint of teaching Master's and PhD courses, I taught the undergraduate chemical engineering course again. This was the first time I taught the course using Python. All the lectures and assignments were in Jupyter notebooks. You can find the course here: https://github.com/jkitchin/s17-06364. The whole class basically ran from a browser using a Python Flask app to serve the syllabus, lectures and assignments. Assignments were submitted and returned by email through the Flask app. It was pretty interesting. I did not like it as much as using Emacs/org-mode like I have in the past, but it was easier to get 70 undergraduates up and running.

I did not teach in the Fall, because I was on Sabbatical!

4 Sabbatical at Google

In August 2017 I started my first sabbatical! I am spending a year in the Accelerated Science group at Google in Mountain View, California. I am learning about machine learning applications in engineering and science. This is a pivotal year in my research program, so stay tuned for our new work!

It has been great for my family, who moved out here with me. We have been seeing a lot of California. I have been biking to work almost every day, usually 15-20 miles. I have logged over 1200 commuting miles already since August.

5 Emacs and org-mode

org-ref remains in the top 15% of downloaded MELPA packages, with more than 24,000 downloads since it was released. It has been pretty stable lately. It remains a cornerstone of my technical writing toolbox.

I have spent some time improving org-mode/ipython interactions including inline images, asynchronous execution and export to jupyter notebooks. It is still a work in progress.

I spent a fair bit of time learning about dynamic modules for writing compiled extensions to Emacs to bring features like linear algebra, numerical methods and database access to it. I wish I had more time to work on this. I think it will be useful to make org-mode even better for scientific research and documentation.

6 Social media

I have continued exploring the use of social media to share my work. It still seems like a worthwhile use of time, but we need continued efforts to make this really useful for science.

6.1 kitchingroup.cheme.cmu.edu

I use my blog to share technical knowledge and news about the group. We had 48 blog posts in 2017. A lot of them were on some use of org-mode and Emacs. I also introduced a new exporter for org-mode to make jupyter notebooks. I spent November exploring automatic differentiation and applications of it to engineering problems. Visits to the site continue to grow. Here is the growth over the past two years. The big spike in Oct 2017 is from this article on Hacker News about one of my posts!

I continue to think that technical blogging is a valuable way to communicate technical knowledge. It provides an easy way to practice writing, and with comments enabled to get feedback on your ideas. It has taken several years to develop a style for doing this effectively that is useful to me, and to others. I have integrated my blog into Twitter so that new posts are automatically tweeted, which helps publicize the new posts.

It has some limitations, e.g. it is not obvious how to cite them in ways that are compatible with the current bibliometric driven assessment tools used in promotion and tenure. Overall, I find it very complementary to formal publications though, and I wish more people did it.

6.2 Github

I was a little less active on Github this year than last year, especially this fall as I started my sabbatical. Github remains my goto version control service though, and we continue using it for everything from code development and paper writing to course serving.

scimax finally has more Github stars than jmax does!

6.3 Youtube

Another year with over 100,000 minutes of Youtube watch time on our videos. org-mode is awesome was most popular, with almost 50,000 views. We have six videos with over 2500 views for the past year!

I have not made too many new videos this year. Hopefully there will be some new ones on the new features in scimax in the next year.

Copyright (C) 2017 by John Kitchin. See the License for information about copying.

org-mode source

Org-mode version = 9.1.3

Discuss on Twitter

New publication in J. Phys. Chem. Lett.

Posted September 25, 2017 at 08:24 PM | categories: publication, news | tags:

Updated September 26, 2017 at 07:16 AM

DFT calculations are extensively used to predict the chemical properties of metal alloy surfaces, but they are expensive which limits the number of calculations that can be practically be calculated. In this paper, we explore a perturbation approach known as alchemy to take previously calculated results and extend them to new compositions. We use oxygen reduction as a prototype reaction, and show that alchemy is often much faster than DFT, with an accuracy within 0.1 eV of the DFT. There are cases where the accuracy is not as good suggesting that further improvements to the perturbation model could be beneficial. Overall, alchemy appears to be a useful tool in high-throughput screening research. 

@article{saravanan-2017-alchem-predic,
  title =        {Alchemical Predictions for Computational Catalysis: Potential
                  and Limitations},
  year =         2017,
  Author =       {Saravanan, Karthikeyan and Kitchin, John R. and von
                  Lilienfeld, O. Anatole and Keith, John A.},
  Bdsk-Url-1 =   {https://doi.org/10.1021/acs.jpclett.7b01974},
  Doi =          {10.1021/acs.jpclett.7b01974},
  Eprint =       {https://doi.org/10.1021/acs.jpclett.7b01974},
  Journal =      {The Journal of Physical Chemistry Letters},
  Note =         {PMID: 28938798},
  Number =       {ja},
  Pages =        {null},
  Url =          {https://doi.org/10.1021/acs.jpclett.7b01974},
  Volume =       0,
}

Copyright (C) 2017 by John Kitchin. See the License for information about copying.

org-mode source

Org-mode version = 9.0.7

Discuss on Twitter

New publication in Journal of Physics Condensed Matter

Posted August 06, 2017 at 11:04 AM | categories: publication, news | tags:

Updated August 06, 2017 at 11:05 AM

The Atomic Simulation Environment is a powerful python library for setting up, running and analyzing molecular simulations. I have been using it and contributing to it since around 2002 when I used the ASE-2 version in Python 1.5! The new ase-3 version is much simpler to use, and much more powerful. This paper describes some of its design principles and capabilities. If you use ASE, please cite this paper! 

@article{larsen-2017-atomic-simul,
  author =       {Ask Hjorth Larsen and Jens J{\o}rgen Mortensen and Jakob
                  Blomqvist and Ivano E Castelli and Rune Christensen and
                  Marcin Dułak and Jesper Friis and Michael N Groves and
                  Bj{\o}rk Hammer and Cory Hargus and Eric D Hermes and Paul C
                  Jennings and Peter Bjerre Jensen and James Kermode and John
                  R Kitchin and Esben Leonhard Kolsbjerg and Joseph Kubal and
                  Kristen Kaasbjerg and Steen Lysgaard and J{\'o}n Bergmann
                  Maronsson and Tristan Maxson and Thomas Olsen and Lars
                  Pastewka and Andrew Peterson and Carsten Rostgaard and Jakob
                  Schi{\o}tz and Ole Sch{\"u}tt and Mikkel Strange and Kristian
                  S Thygesen and Tejs Vegge and Lasse Vilhelmsen and Michael
                  Walter and Zhenhua Zeng and Karsten W Jacobsen},
  title =        {The Atomic Simulation Environment-A Python Library for Working
                  With Atoms},
  journal =      {Journal of Physics: Condensed Matter},
  volume =       29,
  number =       27,
  pages =        273002,
  year =         2017,
  url =          {http://stacks.iop.org/0953-8984/29/i=27/a=273002},
  abstract =     {The atomic simulation environment (ASE) is a software package
                  written in the Python programming language with the aim of
                  setting up, steering, and analyzing atomistic simulations. In
                  ASE, tasks are fully scripted in Python. The powerful syntax
                  of Python combined with the NumPy array library make it
                  possible to perform very complex simulation tasks. For
                  example, a sequence of calculations may be performed with the
                  use of a simple 'for-loop' construction. Calculations of
                  energy, forces, stresses and other quantities are performed
                  through interfaces to many external electronic structure codes
                  or force fields using a uniform interface. On top of this
                  calculator interface, ASE provides modules for performing many
                  standard simulation tasks such as structure optimization,
                  molecular dynamics, handling of constraints and performing
                  nudged elastic band calculations.},
}

Copyright (C) 2017 by John Kitchin. See the License for information about copying.

org-mode source

Org-mode version = 9.0.7

Discuss on Twitter

New publication in Crystal Growth & Design

Posted August 05, 2017 at 01:11 PM | categories: publication, news | tags:

Updated August 05, 2017 at 01:16 PM

Usually, metal oxides grow in a single, most stable crystal structure at a particular set of conditions. For example, TiO2 grows in the rutile structure for a large range of pressure and temperature conditions, but under some conditions it can also grow in the anatase structure. In this paper we show that epitaxial stabilization can be used to influence which crystal structures are observed for the growth of tin oxide. Tin oxide is normally only observed in the rutile structure. We grew tin oxide as an epitaxial film on a poly-crystalline substrate of CoNb2O6 which has an α-PbO2 crystal structure. We found that both rutile and α-PbO2 structures could be found in the film, and that the structure correlated with the orientation of the underlying grains. In other words, the orientation of a substrate can influence the structure of an epitaxial film, enabling one to grow films in crystal structures that may be metastable, and unobtainable in bulk samples. 

@article{wittkamper-2017-compet-growt,
  author =       {Wittkamper, Julia and Xu, Zhongnan and Kombaiah, Boopathy and
                  Ram, Farangis and De Graef, Marc and Kitchin, John R. and
                  Rohrer, Gregory S. and Salvador, Paul A.},
  title =        {Competitive Growth of Scrutinyite ($\alpha$-PbO2) and Rutile
                  Polymorphs of \ce{SnO2} on All Orientations of Columbite
                  \ce{CoNb2O6} Substrates},
  journal =      {Crystal Growth \& Design},
  volume =       17,
  number =       7,
  pages =        {3929-3939},
  year =         2017,
  doi =          {10.1021/acs.cgd.7b00569},
  url =          {https://doi.org/10.1021/acs.cgd.7b00569},
  eprint =       { https://doi.org/10.1021/acs.cgd.7b00569 },
}

Copyright (C) 2017 by John Kitchin. See the License for information about copying.

org-mode source

Org-mode version = 9.0.7

Discuss on Twitter
« Previous Page -- Next Page »