Challenge 2 – Diamonds and ice: when is H2O no longer a molecule?

Garrett Granroth, Peter Peterson, Malcolm Guthrie||
Oak Ridge National Laboratory

What happens when you push atoms closer together? In the early 1900’s, Percy William Bridgman was the first person to begin to answer this question, inventing the field of high-pressure research and earning a Nobel prize in the process. We now know that simple mechanical pressure can transform graphite into diamond, polymerize carbon dioxide and transform oxygen into a shiny metal.

Amazingly, the pressures required to achieve these phenomena (which approach those of the outer core of the Earth), can be created in the laboratory. At Oak Ridge National Laboratory, we can then use our powerful neutron beams to study the exotic behavior observed under pressure with atomic-level precision. However, the only material we have that is strong enough to hold such enormous pressures is single-crystal diamond. When our neutron beams interact with the diamond container, it generates complex scattering patterns that lie superimposed on top of the crystallographic data we hope to measure.

One of Bridgman’s discoveries was that, under pressure, regular water ice is actually only one of many different crystalline forms that can be formed. This data challenge involves analyzing a neutron dataset collected from a special phase of ice measured at pressures so intense – almost 1,000,000 times higher than atmospheric pressure – that the water molecule itself is believed to have dissociated creating a highly-symmetric lattice of oxygen and hydrogen atoms. The key goal of this challenge is to carefully isolate the signal of the ice from that of the diamonds applying the pressure.

The dataset provided contains the scattered neutron intensity as a function of 3-dimensional scattering space. specific challenges are:

  • Load the scattered neutron intensity data, and view it as:
    1. a rotatable 3d image.
    2. 2d slices from the 3d volume showing interesting regions of signal. Often these slices are perpendicular to one of the data axes, but they do not need to be.
    3. Enable setting of the intensity range to readily separate very weak signals in the presence of very strong signals.
  • Use your results from (1) to identify spherical surfaces, centered on the origin. These are the signal from the sample of ice, that we wish to retain.
  • Also identify and classify other features, such as intense, highly localized spots of intensity, broad bands of intensity emanating from the localized spots, and other extended features that may not be centered on the origin. These originate from the diamonds.
  • Using what you learned in (2) and (3), develop an approach that labels voxels, according to the categories of signal you discovered.

The data set can be found at :