NSF sponsored program to bring High Performance Computing to America's high schools

This paper is prepared by Terry Koker and Barry Rowe of ChemViz. Revised June 1998 by Robert A. Miller.

High school chemistry teachers from the Champaign-Urbana, Illinois area, with the support of the National Science Foundation and the National Center for Supercomputing Applications (NCSA) have developed a method which combines the computing power of the supercomputer with the graphics and ease-of-use capabilities of the desktop computer to enable students to use visualization to understand abstract concepts in chemistry. From the inception of the project in the spring of 1991 until the beginning of 1995, the project made use of the CRAY supercomputer. Since then, the project has run all of the same programs on the SGI Challenge.

Using the supercomputer, teachers and students are able to create sophisticated and accurately calculated images of orbitals and electron densities for atoms and molecules. The images are calculated using a research level program called DISCO which uses the self consistent field theory (SCF) to evaluate wave functions for electrons in the atom or molecule which is being modeled. Orbital values or the electron densities for molecular orbitals are calculated for a set of points in a volume containing the atom or molecule. Using an Internet connection, browser software, and the Waltz interface, the calculated images are displayed using color to distinguish high from low electron density values.

The project began in the spring of 1991 when Dr. Nora Sabelli of NCSA recruited four area high school chemistry teachers: David Bergandine, University High School; Terry Koker, Urbana High School; Robert Miller, Champaign Central High School; and Barry Rowe, Champaign Centennial High School to develop a proposal for "Visualizing Chemistry". The project was funded by NSF and supported by Benjamin Cummings, Apple, and NCSA. Staff members who have worked on the project include: Lisa Bievenue, David Atkins, Juan Moran, Elaine Schulte, Keith Wessels, Savva Korolev, Ed Thomson, Matt Thomas, Todd Veltman, An Le and the original high school teachers. When the original Principal Investigator, Nora Sabelli left for NSF, Harrel Sellers, the author of DISCO, became involved in the project as principal investigator, and has given valuable insight into the interpretation of the meanings of the images generated by the group. When Harrel left NCSA, Ken Suslick and Steven Zumdahl agreed to become the principal investigators. Grant funding for the project ended in 1993. Since that time, ChemViz has been maintained by the staff at NCSA.

Much of the preliminary work was done during the 1991-92 school year, consisting of authoring software and support materials to aid in the creation of input files for DISCO and interpretation of the images generated. Much time was spent in exploration of other methods for incorporating computer visualization into the chemistry curriculum, and in preparation for summer workshops hosted by NCSA and put on by the ChemViz staff.

Two of the four chemistry teachers involved in the project brought their high school classes to NCSA during the 1990-91 school year. These classes had to do all their modeling using only the Cray supercomputer to generate data files. Learning to use UNICOS (the Cray operating system) was time consuming and error prone. The images were acceptable, but not many were generated. It became obvious that a user-friendly 'front end' had to be generated to make it easy for high school students to use high powered computing. Another problem was that students had to physically go to NCSA to make images.

All four chemistry teachers brought their high school classes to NCSA during the 1991-92 school year. Images generated benefited greatly from a user interface generated by the ChemViz programmers. The students needed only twenty minutes of introduction to start making pictures. Students started out with images that the teachers suggested, including H 2 . See Figure 1 for an image generated by Justin Hill of Champaign Centennial High School.

Figure 1: H molecule by Justin Hill.

Soon the students were asking questions of the computer on their own. Johanna Klaus and Kristie French of Champaign Centennial High School wondered why He 2 doesn't form, so they tried it. See Figure 2 for their images.

Figure 2: He 2 anti-bonding orbital and He 2 bonding orbital: generated by Johanna Klaus and Kristie French -- Champaign Centennial High School chemistry students

Some students even wanted to return on their own time, during the evening. No one told Jason Stiff and Chris Hamelberg of Champaign Centennial High School that three-atom molecules could be done, but since no one said they could not (they can be done); they decided to try hydrogen cyanide. See Figure 3 for their image.

Figure 3: Hydrogen Cyanide by Jason Stiff and Chris Hamelberg of the Champaign Centennial High School AP chemistry class.

The most exciting part of watching high school students use high performance computing is seeing them seek answers by using computational chemistry. They seek answers to their questions, not answers phrased by teachers 'for' them. They are doing real research.

After watching high school students work with images, it was indeed obvious that student-generated images do two things. Students are motivated to do their own research, answering their own questions; and they spend lots of time analyzing the images they get, applying the other knowledge they have acquired about atoms and molecules to those images. Both of these characteristics of student-generated images cause the student to do what is most important to any serious chemistry student -- think about chemistry

During the summer of 1992, two workshops were held for field testers with approximately 15 teachers from across the United States in attendance at each workshop. Each teacher left the workshop with the capability of generating images using the supercomputer over the Internet or by modem. They were provided with email support, Cray accounts, and training in using the ChemViz programs and NCSA tools. There are now over 3 high schools who have the capability to use high powered computing in their chemistry classrooms. The field tester's purpose was to help develop curricular uses of the newly developed easy-to-use modeling capabilities of the NCSA Cray-2.

In the summer of 1993, another workshop was held in Champaign-Urbana for 18 of the original ChemViz field testers. During this workshop, improved ChemViz tools were used and curricular materials were developed by the users of the ChemViz materials. These curricular materials are available on the ChemViz website.

What impact will the chemistry visualization project have on the high school chemistry curriculum? We, the project members, believe the effect will be revolutionary. This project will allow high school chemistry students to use the supercomputer as a laboratory for designing experiments which will answer their questions concerning such abstract concepts as electrons, atoms, molecules, and chemical bonding. By generating images of the electron densities for various combinations of atoms, students will be able to understand in concrete terms the differences between: equal and unequal sharing of electrons; bonding and antibonding orbitals; strong and weak bonds; and the energy differences of atoms at appropriate and inappropriate bond distances and angles. Virtually any concept involving chemical bonding can be explored using this tool.

Currently, chemical bonding is taught in variety of ways at the high school level. In most high school classrooms, the Bohr model of the atom is used to introduce students to the concept of energy levels. Bonding is described as occurring when atoms transfer or share electrons from their outermost shells with the usual objective of filling these shells.

At some later point, the Bohr model is replaced by the electron cloud model in which electrons are no longer considered as being located at a given distance from the nucleus, but are instead described using probability regions.

High school students are more successful at visualizing the Bohr model of the atom because it is more concrete. Pictures can be drawn on the blackboard which obey a given set of rules and illustrate most simple examples of bonding. The electron cloud model is much more difficult for high school students to visualize. It often requires some understanding of probability, 3D geometry, wave-particle duality, and of quantum mechanics.

In our experience, high school chemistry students adopt the most sophisticated model of the atom that they can visualize . Sadly, in most cases this is the Bohr atom. This problem can be addressed and student understanding of bonding can be greatly enhanced by the use of images to illustrate the concept of electron clouds. An attempt to do this is made in most textbooks with pictures of s, p, d, and f orbitals, with some books showing similar representations of hybrid orbitals and molecular orbitals. Those of us who have taught high school chemistry know that we have a very difficult time "selling" these pictures to students. They ask tough questions like, "How do I know these pictures are real?", and "Where did these pictures come from?" At this point we usually kick ourselves for encouraging students to question, and then launch into a lame attempt to explain with references to Schrodinger's equations and statements like, "Those pictures were drawn on a computer". But are they really? Most of the orbital pictures are "artist's impressions" of chemists' understandings with no direct computational basis. The students simply ignore them because the Bohr atom's simplicity is so enticing and our explanations are so lacking.

The textbook images, despite the difficulty with explaining them, are still generally of much higher quality than the figure eight's and oddly shaped clouds we teachers attempt to draw on the blackboard.

Using the method developed by the ChemViz project, teachers and students are able to generate images illustrating the electron cloud model that are of much higher quality than those generally found in textbooks and on blackboards. Images can be generated to illustrate almost any aspect of the theory the teacher is addressing and the use of color enhances the fine details in each picture. Furthermore, using the Waltz interface, the student can begin using the supercomputer as a powerful laboratory in which variables such as type of atom and distances between atoms are varied, with differences between images generated analyzed and hypotheses generated and then tested. Not only are the students given visual images which "stick with them" better, but they are allowed to apply the scientific method in their interaction with the model.

One point needs to be made about the teacher who uses the ChemViz materials. Chemistry teachers may not be 'up to date' on the latest in computational chemistry; most have not used high powered computing; the software we use is complicated and research-level but user friendly; and some of the current high school teachers may not have used Macintoshes or PCs before. But the ChemViz project gives support to those teachers through electronic mail and from the ChemViz website. This is a very important part of the project, as questions can be answered quite quickly, and shared with more than the original questioner. As a collateral advantage, this type of support leads to participation in discussion lists about chemistry (and other subjects) so that the high school chemistry teacher becomes part of a world-wide electronic support group.

Students from the high schools involved in the project have benefited from the opportunity to learn about bonding using high powered computers and have generated images with positive results. The project was also well received by the field testers who attended the summer workshops, many of whom are applying supercomputer generated images to the teaching of their students.

Additional curricular materials are being developed that allow a much broader experience in computational chemistry for all of the ChemViz participants, and chemistry students in high schools around the United States . In addition, the Cambridge Structural Database (CSD) has recently become available at the ChemViz website. This database allows the student to search for molecules by formula or name (the database consists mainly of organic molecules), then view the molecules. The ChemViz group, with the help of those who find the CSD useful, will develop curricular materials incorporating this powerful research tool into the high school chemistry curriculum.