The Materials Data Facility (MDF) is building data services to support materials science researchers in publishing, discovering, preserving, and sharing research datasets ranging in size from kilobytes to terabytes. MDF currently holds around 30 terabytes of data from over 150 authors, including millions of individual results. One effort using MDF is the Gr-ResQ project, which is developing a database of chemical vapor deposition recipes for synthesizing graphene with over 200 variables per sample, in order to improve the reproducibility and understanding of graphene synthesis experiments. Researchers can submit CVD recipes and subsequent analyses such as Raman spectroscopy or image data to the MDF database for publication and sharing with the materials science community.

1. Materials Data Facility as
Community Database to Share
Nano-manufacturing Recipes
Presented by Ben Galewsky (bengal1@Illinois.edu)

2. MDF Overview
Build data services to
• Empower researchers to publish data, regardless of
size, type, and location
• Automate data and metadata extraction and ingest
• Enable unified search and discovery across
disparate materials data sources
Deploy with APIs to simplify connection to
other data efforts and to enable automation
ML and Informatics in Materials

3. Materials Data Facility
Discovered
Searched
Browsed
Shared
Recommended
Accessed
The Materials Data Facility (MDF) is an effort to build a set of data services
to support materials science researchers. MDF offers a data publication
service where materials scientists can publish, discover, preserve, and
share research datasets ranging in size from kilobytes to terabytes.
Published
Identified
Described
Curated
Verifiable
Accessible
Preserved
AND

4. Materials Data Facility Processes
• Connect: Extract domain-relevant
metadata / transform the data
• Publish: Built to handle big data
(many TB, millions of files),
provides persistent identifier for
data, distributed storage enabled
• Discover: Programmatic search
index to aggregate and retrieve
data across hundreds of indexed
data sources
• Currently holds ~30TB of data
from over 150 authors, millions of
individual results
Beta testing
Open Open

5. Joshua Schiller, Matthew Robertson, Kristina Miller,
Ricardo Toro, Benjamin Galewsky, Darren Adams,
Kevin Liu, Daniel Katz, Kevin Cruse, Kaihao Zhang,
Elif Ertekin, Sameh Tawfick and Placid Ferreira
Department of Mechanical Engineering
University of Illinois at Urbana-Champaign
Gr-ResQ
Graphene Recipes for Synthesis
of High Quality Materials

6. Graphene
2-dimensional sheet of carbon atoms
“Honeycomb” lattice
Properties
• High strength
• Electrically conductive
• Lightweight
• Flexible
Applications
• Transparent conductor
• Flexible electronics
• Mechanical composites https://www.nature.com/news/graphene-the-quest-for-supercarbon-
1.14193

7. How Do We Make It?
http://www.plasma-electronics.com/chemical-vapor-deposition.html

8. CVD Recipes
Over 200 variables per sample
o Furnace temperature
o Furnace pressure
o Gas flow
o Sample position
o Annealing condition
o Growing condition
o Cooling condition
o Sample preparation
o Catalyst
o Furnace dimensions

9. Challenges With CVD
o Poor repeatability of experiments
o Lack of quantitative description of the parameter space
o Necessity for centralized database of existing experimental data

10. Gr-Res-Q
Analysis Tools
Raman Tool Image Tool
Submit Tool
User Info
Recipe
Data
Image Files
Raman
Files
Query Tool
Raw Query
Analysis
Query
MDF Database
Raw Set
Analysis
Validate
Globus Auth
Globus Search
Globus
Transfer

11. Publication to MDF
•We treat each recipe as a dataset
•Subsequent analyses are also datasets
•Collections of datasets can be bundled together
with a single, DOI minted for publication
•Datasets are searchable via Forge

12. bengal1@Ilinois.edu