A Study on the Effects of High Temperature Annealing of Borazine Based hBN Films
Open Access
- Author:
- Kronz, Jeffrey
- Graduate Program:
- Materials Science and Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- August 18, 2021
- Committee Members:
- Nasim Alem, Major Field Member
Mark Horn, Outside Unit & Field Member
Joan Redwing, Co-Chair & Dissertation Advisor
David Snyder, Chair & Dissertation Advisor
John Mauro, Program Head/Chair - Keywords:
- Boron Nitride
hBN
Borazine
Chemical Vapor Deposition
2D Materials - Abstract:
- hBN is a 2D ultra-wide bandgap insulator of interest for use in 2D semiconductor, UV optoeletronic, thermal transport, and III-Nitride systems. Use of hBN in these systems typically requires high crystal quality, wafer scale films to achieve the performance of material properties that are desired for use in these systems. Growth routes such as molecular beam epitaxy (MBE) and metal organic chemical vapor deposition (MOCVD) have demonstrated their potential to generate such hBN films, though limitations with both of these approaches due to precursor toxicity or expensive ultra-high vacuum setups can be avoided through the use of borazine in a CVD growth. This thesis investigates the influences of the high temperature anneal conditions needed to convert borazine films to hBN and how this process affects the final hBN film properties. Initial work was done to shortly understand the sublimation conditions of ammonia borane to produce borazine to maximize its production during the deposition. Increasing the sublimation temperature from 100 ℃ to 120 ℃ to 140 ℃ reveals that borazine is most produced when both the expected reaction temperatures of ammonia borane at 110 ℃ and 130 ℃ are passed. The other undesired reaction products were found to be mostly produced during the first 110 ℃ reaction so increasing beyond this to the experimental temperature of 140 ℃ increases the borazine production without adding equal amounts of additional unwanted byproducts. Changing the sublimation ramp rate from 2 ℃/min to 4 ℃/min to 10 ℃/min displayed a tradeoff in both borazine production and undesired byproduct production. The slowest ramp time increased the total time of species production to over 30 minutes, while the fastest ramp rate simply spiked in production over twice what was seen for the middle ramp rate of 4 ℃/min. Since these effects were seen for both borazine production and the undesired species production, the middle rate of 4 ℃/min was the best choice to allow for an increase in borazine production without the addition of significant amounts of undesired products. The modulation of borazine source amounts during the deposition was found to be the most influential on final film thickness prior to annealing. Increasing the pre-anneal thickess to above 200 nm was found to be critical to allow for proper characterization and residual film survival of the high temperature anneals. Doubling the growth time saw an increase from the baseline 178 nm thickness to only 220 nm. This suggested that the amount of source material was limiting the growth of the films. While doubling the gas concentration of the carrier gas for the sublimation increased the film thickness to an average of 216 nm, doubling the amount of ammonia borane powder resulted in an average thickness of 324. While this route also increased the variance from run to run to give a standard deviation of about 66 nm, the most of all the parameters changed, this was found to be less of a concern. This was due to the fact that even with the large variation in thickness run to run, the overall thickness being over 200 nm was the most important factor in gathering consistent characterization data after annealing, and none of the other parameters were able to achieve a film thickness over 200 nm reliably each deposition. The first set of anneals were done in a low temperature regime from 400 ℃, the growth temperature, to 1000 ℃, just before the end of the second weight loss of polyborazylene to form hBN. FTIR of these films revealed that the measurable amount of hydrogen bonded species in the system were removed by 800 ℃, as expected. However, an interesting peak was found in both FTIR and Raman for the 700 ℃ and 800 ℃ annealed samples, which is believed to be caused by this initial reaction to dehydrogenate the film and reaction with residual oxygen to produce boric acid, B(OH)3 in the film. Otherwise, the increased temperatures up to 1000 ℃ showed an improvement in the films conversion to hBN measured by the reduction in FWHM of the E2G peak for hBN at 1369 cm-1. Increasing the anneal temperature into the high temperature regime from 1000 ℃ to 1600 ℃ continued to show the improvements in conversion to hBN via Raman and the reduction of FHWM in the same E2G peak. This temperature increase also revealed the severe reduction in film thickness after anneal, even to the point of total film removal for 1500 ℃ and 1600 ℃. The surfaces of these films in SEM revealed etched sapphire surfaces, and the surviving film at 1400 ℃, while still able to produce a signal in Raman was damaged with large holes found in the film, and hill-like features that were determined to be reaction points with the sapphire surface underneath that became the etch pits seen at 1500 ℃. This was believed to the result of hydrogen present in the film during the dwell at high temperatures, where either hydrogen that was not sufficiently diffused out of the film or was present in the gas phase above the film surface that etched the film. This reaction suggests that kinetics of the dehydrogenation should be utilized to minimize this damage. Differing heating rates of the anneal from 5 ℃/min to 20 ℃/min did not appear to significantly alter the results observed in the previous high temperature study. Damage to the hBN films was still occurring at 1300 ℃ and higher, with the same conversion improvements seen from the FWHM reduction being observed as well. Changing the dwell time at temperatures below 1300 ℃, however, did display the effects of the dehydrogenation rate. It was found that increased dwell time at lower temperatures, 4 hours at 1000 ℃- 1300 ℃, resulted in a reduction in FWHM of the hBN E2G peak in Raman over the films that were only annealed for 1 hour, while the surfaces did not display significant damage that was previously seen. Extending the anneal dwell time to 8 hours continued the improvements in FWHM in Raman, though now damage was found in the 1300 ℃. These results suggest that there is a balance between the annealing temperature to enable the conversion and dehydrogenation, and the time at which the dehydrogenation takes place at a safe temperature where the free hydrogen will not have enough energy to react with the system. This balance of dehydrogenation and conversion rate was demonstrated by utilizing a two-part anneal approach. Using a lower temperature anneal, 900 ℃ - 1000 ℃, for a longer initial dwell would allow for the dehydrogenation to take place safely. Once enough hydrogen is removed, the film should be stable to withstand the higher temperatures over 1400 ℃ where conversion should take place readily and form a crystalline hBN. Initial anneals of 1000 ℃ for 1 hour and 900 ℃ for 4 hours were both able to produce a film that survived at 1500 ℃ for the first time. This difference in initial anneal dwell time is explained by the difference in the anneal temperature, since the 1000 ℃ dwell would require less time to remove hydrogen given the rate should be higher than at 900 ℃. Additionally, the FWHM of the hBN E2G in Raman was the lowest measured FWHM of any film previously annealed in this work, also supporting the need for the highest anneal temperature to facilitate conversion of hBN. XRD was also taken for these samples and found to have a broad (0002) peak, indicating that the films were crystalline enough to be measured, but still nanocrystalline due to this peak breadth. Despite increased initial anneal dwells up to 8 hours at both 900 ℃ and 1000 ℃, there was not a film that could be made to survive at 1600 ℃, indicating further dehydrogenation could be required. Overall, the conversion rate from polyborazylene to hBN was found as the limiting factor in producing crystalline hBN. Due to the necessity of the dehydrogenation in conversion, increased temperatures lead to increased hBN conversion, but also enabled the hydrogen in the system to remove the BN films and etch the sapphire surface. By allowing these reactions to occur over longer time at reduced rates from lower temperatures, the BN films can successfully be converted to a polycrystalline hBN. These trends observed for anneal time and temperature in this thesis will allow for further development of the borazine based hBN film process and improvements to the production of wafer scale crystalline hBN.