2. Future directions 🛣¶

Here we will share bits and pieces of our vision for how interactive computing environments (such as the one demonstrated in this digital textbook) can be extended to enhance materials science education more broadly. If you have ideas or suggestions and would like to collaborate, please do not hesitate to contact us. 😍

Note

If you were looking for content suggestions (i.e., what to study after going through this curriculum), please refer to What’s next?

2.1. Accessibility¶

Two of the main reasons we are such passionate educators are because we strongly believe in the enabling potential of a quality education and that education is not a zero-sum game. We enjoy sharing our knowledge with students in the hopes that it will open new doors of opportunity for them and help every student achieve their dreams. Our desire to impact large, diverse populations aligns well with the mission of public universities and highlights one of the advantages we see of computational materials science (CMS), which allows us to engage more students in MSE in a manner that transcends the walls of the physical laboratory to reach potentially more diverse audiences (though we are mindful of inequities with regards to computing technologies).

For these reasons, we designed this curriculum to be broadly accessible for everyone who wants to learn about MI, including those who are new to programming and those who are not students at UC Berkeley. We’re proud to publicly share as much content as possible (under a CC BY-SA License) and we’re grateful for the support from like-minded individuals (see Acknowledgements). We hope users find this to be a helpful resources, whether they’re students learning this material for the first time or instructors looking to adapt this work to their own curriculum.

In the first iteration of this module, we found this curriculum to be effective at teaching students with a variety of personal and academic backgrounds (four different engineering majors), although more scaffolding could be added after the first week so students continue to feel supported as they synthesize the concepts from the tutorials. Nevertheless, students found it helpful to have all of the reading and exercises neatly organized in a central location and they enjoyed learning the content in the interactive Jupyter environment. We hope to continue this work to enable MI to be another powerful tool in every materials scientist’s toolbox as we move forward in this increasingly data-driven world.

2.2. Scalability¶

In addition to accessibility, we designed for scalability in mind as we’re extremely excited for a future where computing education can be done at scale (e.g., MOOCs, Code in Place). In the field of MSE, which is traditionally limited to smaller cohorts in higher education, CMS and MI make it possible to teach MSE problem-solving skills and computational thinking [3] to a much broader audience. In particular, using the open-source Jupyter infrastructure enables us to create rich interactive content in a format that is familiar to us (as computational researchers) and yet scales in robust way that doesn’t sacrifice pedagogical value.

Having carried out the first iteration with members spread across three time zones and two countries, we are hopeful that this infrastructure can support future iterations of virtual programs that want to expand their reach. While the growing undergraduate MSE population [2] will certainly be seeking practical skills such as data science, there are many more students outside our home institutions and/or who may not even know what MSE/MI entails. Whether this material is used for an academic course or summer research program, it is designed to be easily scaled up so that any number of students can simultaneously access the materials and learn interactively at their own pace. This would, in theory, bring the joys of MI to anyone in the world! Of course, whether this will succeed in practice remains an open question—but if we don’t at least try, how will we ever know?

2.3. Integrate interactive computing experiences into other MSE courses¶

While it may seem obvious to include computational learning experiences into a module on MI, we believe CMS and the Jupyter infrastructure can be integrated more broadly across the entire MSE curriculum [3]. Just as CMS and MI will not replace scientists, these learning tools aren’t meant to replace more traditional pedagogy, but rather enhance understanding by illuminating new pathways to understanding. For example, many ideas that were previously confined to abstruse equations or static illustrations can now be modeled and simulated on a computer to provide validation of the theory and engage students with dynamic content. The growth of powerful and user-friendly programming languages like Python means that more and more students are able to learn sophisticated, transferable, and valuable skills such as data analysis and data visualization that will aid them in their experimental, computational, and other pursuits. The integration of computing into MSE will also introduce these important skills to students who, for a multitude of good reasons, may not normally take a CS course. In particular, by applying the same computational concepts and computational thinking to their domain of interest and expertise, we are able to make computing relevant to students and engage them in critical discussions of the broader ethics and societal implications of computing in their lives.

2.3.1. Thermodynamics and phase transformations¶

One subject where we see CMS playing a prominent role is thermodynamics and phase transformations, which has also been discussed by others [5]. In Mark’s MATSCI 103: Phase Transformations and Kinetics course at UCB, students are asked in each homework assignment to write a small piece of code to model a particular concept. Students gain insight into how thermodynamic equations can be simulated and visualized, particularly as the natural variables are changed. Moreover, the programming tasks can range in difficulty depending on the modeling assumptions and student proficiency, such as more rudimentary implementations like Gibbs free energy curves of ideal solutions to more advanced constructions like the common tangent line between two free energy curves. The combination of prose, equations, and programming exercises makes an interactive platform such as this Jupyter Book ideal for such a course to support student learning.

As a demonstration of the previous ideas, we combine the Gibbs free energy common tangent construction with the corresponding phase diagram in one plot (see video on the right). While the connection between these two plots is typically described with a series of static images, we use the matplotlib Animation module to create a movie that traces out this correspondence in real time as the temperature of the system is varied. After completing the reading and attempting the foundational exercises for themselves, students can play through this video to establish what we hypothesize to be a more lucid understanding of the connection between free energy curves and phase diagrams. This example gives a glimpse into the new world of possibilities that is created when we integrate computing experiences to enrich existing curricula.

2.3.2. Materials characterization¶

Materials characterization is another domain within MSE that is being revolutioned by computing and informatics [6], and the classroom experience must also adapt to this changing reality if we are to fully prepare our students to be future leaders. Fortunately, there are many opportunities to integrate CMS into a materials characterization course, and here we give a few examples in the context of an old favorite: X-ray diffraction (XRD).

XRD is a great candidate for teaching students about tabular data and scientific computing. A classic assignment is to use the results from a powder XRD experiment to derive the crystal structure of a material through the application of Bragg’s law. While this is a rudimentary exercise that can be done by hand or using a spreadsheet application, it can also be a nice introduction to tabular data analysis in Python (e.g., pandas). Students will then have an opportunity to manipulate tabular data structures and apply algorithmic thinking to a relevant MSE problem. Further extensions outside the scope of a characterization course could involve applying ML techniques to the XRD spectra for automated structure search, fingerprinting, and more!

Another consideration for XRD is that while the above procedure is instructive and proper experimental procedure, it is actually the “inverse” of the actual physics that occurs. By this we mean (obviously) that it’s not the XRD spectra that determines the crystal structure, but rather the structure [and chemistry and experimental parameters] that determine the spectra. While this “forward direction” is implied when teaching the structure factor and intensity factor, it is rarely demonstrated explicitly for the students, even when they’re in the lab doing an XRD experiment, which not all students might even have access to.

With CMS, however, the resulting spectra can be modeled from the crystal structure (lattice + atomic basis) along with some mild assumptions, giving students an opportunity to see how the XRD spectra can actually be produced from the physical parameters. We have taken this idea one step further by using the ipywidgets package to create an interactive XRD spectra visualization where students can change several input parameters to see how they affect the resulting XRD spectra in real time. This tool can be used for lecture, homework assigments, exam review, and more, as was recently demonstrated in Prof. Andrew Minor’s MATSCI 104: Materials Characterization course at UCB. As students and instructors were unable to access the lab due to health and safety protocols, they found this tool to be instructive for understanding the inner workings of XRD. Note that the rendered version in Jupyter Book is static as it lacks a Python kernel, so we suggest you open this page in JupyterHub or Google Colaboratory to see it in action.

2.4. References¶

1: Jeannette M. Wing. Computational thinking. Communications of the ACM, 49(3):33–35, 2006. doi:10.1145/1118178.1118215.
2: R. Allen Kimel and Susan B. Sinnott. The materials science and engineering undergraduate enrollment floodgates are open. MRS Bulletin, 43(4):257–261, 2018. doi:10.1557/mrs.2018.78.
3: Raúl A. Enrique, Mark Asta, and Katsuyo Thornton. Computational materials science and engineering education: An updated survey of trends and needs. JOM, 70(9):1644–1651, 2018. doi:10.1007/s11837-018-2989-7.
4: John D. Hunter. Matplotlib: A 2D graphics environment. Computing in Science Engineering, 9(3):90–95, 2007. doi:10.1109/MCSE.2007.55.
5: Camilo Vieira, Alejandra J. Magana, R. Edwin García, Aniruddha Jana, and Matthew Krafcik. Integrating computational science tools into a thermodynamics course. Journal of Science Education and Technology, 27(4):322–333, 2018. doi:10.1007/s10956-017-9726-9.
6: Steven R. Spurgeon, Colin Ophus, Lewys Jones, Amanda Petford-Long, Sergei V. Kalinin, Matthew J. Olszta, Rafal E. Dunin-Borkowski, Norman Salmon, Khalid Hattar, Wei-Chang D. Yang, Renu Sharma, Yingge Du, Ann Chiaramonti, Haimei Zheng, Edgar C. Buck, Libor Kovarik, R. Lee Penn, Dongsheng Li, Xin Zhang, Mitsuhiro Murayama, and Mitra L. Taheri. Towards data-driven next-generation transmission electron microscopy. Nature Materials, 20(3):274–279, 2021. doi:10.1038/s41563-020-00833-z.

Introduction to Materials Informatics