Leading data-driven device development: from concept to fully-funded partnership
Note: some information has been scrubbed from these charts and some names redacted, but the story told by the data is still intact.
While at the Georgia Tech Research Institute, as a member of a small lab, I initiated a literature review to investigate using carbon nanotubes (CNTs) as electron emitters in ion electric spacecraft propulsion. From the beginning of this project, I had a clear target of what I wanted to realize -- my initial report contained a Gantt chart outlining the project's path over the next four years, including collaborating with the state of the art. As project lead, I made this the timeline a reality and generated results an order of magnitude better than the state of the art at the time.
As technical lead, I received several rounds of internal funding. My results earned a government contract Phase I award, worth $1.125 million, and a subsequent government contract Phase II award, worth $6.5 million. My project would grow to be the lab's most lucrative project and, ultimately, the technology would be realized and launched into orbit.
Starting from scratch: literature review & internal funding
What started as a literature review quickly gained momentum towards proof of concept, funded by other projects that could use these results. The initial concept was proven by an early prototype, shown in figure 1. This device was a simple glass slide with a metal stack patterned by teflon tape. After CNTs were grown on top of the metal stack, the sample was placed in a test fixture - which I designed, in Figure 2 - that allowed it to function as a cathode. The anode, a phosphor screen, was placed on top, isolated by a teflon washer. After applying a bias, the CNTs would emit electrons illuminating the phosphor screen.
Figure 1: First prototype of CNT emitters that would eventually launch into orbit. The area of the square is approximately a 1 inch x 1 inch.
Figure 2: Test fixture I designed to evaluate samples
The results were exciting, but if this project had any hope of surviving by receiving funding independent of other projects, these results would need to be quantified. I characterized individual sample performance so that results by group would be meaningful. Figures 3 describes part of the process to create samples; figure 4 describes how they performed in terms of current generated as function of voltage applied.
Figure 3: Temperature-time chart of CNT growth cycle [time scrubbed]
Figure 4: IV curve of CNT emitter
One challenge was reporting results of prototypes to match fully-deployed parameters: fully-realized devices are often in a triode configuration (cathode/gate/anode), these samples were early stage development diode configuration (cathode/anode). Critical parameters for the funding I targeted included:
- Turn-on voltage - usually reported in terms of minimum voltage. For the samples, as the turn on voltage was a function of the distance between the cathode and gate, these results were reported in terms of turn on electric field, voltage normalized by distance between cathode and anode.
- Maximum current density - usually reported in terms of current by area. While the area of these CNTs could be determined by referencing the CAD file used for the photolithography mask, the patterns were on the micro scale, requiring tools such as a scanning electron microscope to characterize accurately. These types of characterizations were planned after subsequent funding. In the meantime, maximum current was presented.
Figure 5: Turn on electric field by sample, chart generated by Python modules MatPlotLib and numpy. Gist file below.
Figure 6: Maximum current by sample, chart generated by Python modules MatPlotLib and numpy. Gist file below.
The initial results, which helped guide further development, were substantial enough to gain independent internal funding. During this time, I networked with another group that had domain knowledge of the device application. Through building on these results, developing techniques further, and collaborating with other domain experts, the project was awarded a $1.125 million government contract Phase I award. This facilitated rapid project growth, including additional groups and more mature technology -- a direct response to the data generated and presented.
External funding, advanced equipment, and growing partnerships
Figure 7: Overview of process development yielding a triode device
For the Phase I Award, my lab officially partnered with another lab on campus and a remote lab from an Ivy League Institution. Following this, the device was developed from a single step process to a multi-step process, shown in figure 7. This involved using more advanced equipment capable of producing structures on the nanoscale. This equipment required precise control, with parameters driven by previous data analysis. Further refined methods were needed to accurately characterize the devices at this stage of funding. Even with a substantial budget on-hand, I remained resourceful -- open source software was also used for image processing.
After significant effort and many iterations, CNT current density -- the key metric -- was demonstrated at an order of magnitude higher than both previously measured and the state of art, figure 8 and figure 9, respectively. This indicated CNTs had the potential to provide more efficient space craft propulsion by ionizing propellant at a higher rate. These results secured another round of funding, allowing for the device to be developed even further.
Figure 8: Current density of various samples, chart generated by Python modules MatPlotLib and numpy. Gist file below.
Figure 9: Current density of samples against state of the art
From a single step process that provided primary results, I utilized project management and data analysis to develop the device to a reliable, multi-step process that would yield a chip that could be installed in an electric propulsion engine and fully-realized the project plan I established years prior.