Periodic Reporting for period 1 - UNOGAN (Unveiling down to 0-dimensional confinements in GaN devices for RF power application)
Berichtszeitraum: 2020-10-10 bis 2022-10-09
GaN is rapidly becoming the first choice of semiconductor for power conversion and their presence is emerging in a wide range of applications from fast charging of batteries, electric vehicles etc. to high-end solutions like light detection and ranging devices. Despite exhibiting properties that allow for higher breakdown strength, faster switching speed and low switching losses, the technological challenges in their processing impacts the final device performance and hinders the replacement of incumbent Si-based switching devices by more efficient ones. Overcoming such challenges will increase the incorporation of these energy efficient GaN devices in widespread applications, which advances our effort in reducing energy consumption and greenhouse gas emission. UNOGAN has identified that one such challenge is related to the difficulty in assessing defects and doping issues induced in a fully processed device. Its overall objective was to develop a set of methodologies built on existing scanning probe microscopy (SPM) techniques to provide a visualizing aid to pinpoint the spatial location of these issues in a fully processed device stack. Achieving this target strictly required the completion of the following actions:
(i) Identifying bottlenecks of the methodologies (electrical contacts, specimen preparation and type of probes) and overcoming them (WP1,2)
(ii) Establishing the methodology on GaN device stacks (WP2,4)
The project also created a platform for in-operando measurements, that could be used to probe fundamental workings of an active device with novel architectures. The analyses supported by TCAD computation led to generation of knowledge establishing device-specific contrast analysis, doping dependence of the detection signal, classification of defects according to their atom core structures and trap densities and identification of dislocations as leakage paths. The project has achieved most of its objectives, milestones, and deliverables within the planned period with relatively minor deviations
(i) Identifying bottlenecks of the methodologies (electrical contacts, specimen preparation and type of probes) and overcoming them (WP1,2)
(ii) Establishing the methodology on GaN device stacks (WP2,4)
The project also created a platform for in-operando measurements, that could be used to probe fundamental workings of an active device with novel architectures. The analyses supported by TCAD computation led to generation of knowledge establishing device-specific contrast analysis, doping dependence of the detection signal, classification of defects according to their atom core structures and trap densities and identification of dislocations as leakage paths. The project has achieved most of its objectives, milestones, and deliverables within the planned period with relatively minor deviations
WP1: Optimization of ohmic contacts on n-GaN and p-GaN
In this WP, different types of metallic contacts were systematically investigated for their use as back contacts to enable different electrical modes of SPM. The desired contacts achieved on Ti/Al stacks presented an ohmic behaviour on 2.8 µm thick AlGaN buffer layers. As desired, it only required a small thermal budget avoiding an alteration of the material resulting from this thermal step.
WP2: Carrier, charge and band profiling
The main task of this WP was to establish SPM methodology consisting of conductive-atomic force microscopy, scanning capacitance microscopy (SCM) and kelvin probe force microscopy (KPFM) that can independently provide information on conductivity, local charges and local band bendings in device stacks. Accordingly, various probes (metallic and doped-diamond tips) were evaluated for their efficacy. The following device stacks were successfully investigated using these methodologies:
(i) semi-vertical p-n diodes grown on Si substrate, which is used as test-vehicle for module development, and on QST® for devices covering up to 1.2 kV breakdown voltage
(ii) 200 V and 650 V p-GaN HEMT devices
For HEMTs, the methodology not only confirms the expected carrier distribution in the active layers through its direct visualization but also provides first of evidence of n-type carriers (1014 - 1015 cm-3) in strain management layers. The latter concerns associated RF loss. In case of p-n diodes, the contrast analysis was more complex as it was overwhelmed by the space charge of dislocations but it improved understanding of the role of dislocations in such devices.
KPFM simulations were carried out to test the a priori assumption of the analytical dependence of contact potential difference on the Fermi level, which resulted in compilation of a database with calculated values for varying surface states and doping. As the calculated relation deviated significantly from the analytical expression, it was concluded that the technique cannot be straightaway used for band profiling without using the corrections from the database. This technique was further advanced to visualize evolution of the surface potential of the channel when switching from its OFF state to ON state in p-GaN HEMTs, which was in good conformity with the calculated potential distribution of the channel.
WP3: Surface characterization of the recessed regions
This WP provided an in-depth understanding of the properties of dislocations. The atom core-structures of different types of dislocations were first identified, which led to their classification. By combining TCAD simulation with experimental SCM/KPFM analyses, their trap densities were determined in n-type GaN. It was also found that none of these dislocations can be associated with the leakage paths, which was in strong contradiction with what has been reported in the literature. However, in p-n diodes and HEMT stacks, up to 5% of the dislocations were associated with leakage current. Based on a comparative statistical analysis conducted using electron microscopy, this study suggested that the detrimental dislocations are likely to be pure-screw type.
WP4: Training activity for atomic resolution STM analysis
The fellow carried out his training on STM on: NX-Hivac Park Systems SPM installed at imec site and UNISOKU ultra-high vacuum system at KU Leuven. This has led to a close collaboration between the two groups for defectivity study in 2D materials yielding an undertanding of water intercalation process and their correlation with defects.
WP5: Project management, dissemination and communication
The progress of the project was regularly monitored and received continuous support from the supervisor, senior members of the epitaxy and SPM experts. The most relevant research outcomes of the project for both GaN and 2D programs have led to or will lead to publication in peer-reviewed journals. The results were disseminated within the industrial partners through IMEC’s Industrial Affiliation Program at internal PTW symposium and four international conferences. As a keynote speaker at International SPM symposium “Failure Analysis and Material Testing”, the fellow promoted the activities of the project to a broader audience of the electronic industry. UNOGAN contributed to the public outreach activity with a demo explaining the demand for nanotechnology on open company day.
In this WP, different types of metallic contacts were systematically investigated for their use as back contacts to enable different electrical modes of SPM. The desired contacts achieved on Ti/Al stacks presented an ohmic behaviour on 2.8 µm thick AlGaN buffer layers. As desired, it only required a small thermal budget avoiding an alteration of the material resulting from this thermal step.
WP2: Carrier, charge and band profiling
The main task of this WP was to establish SPM methodology consisting of conductive-atomic force microscopy, scanning capacitance microscopy (SCM) and kelvin probe force microscopy (KPFM) that can independently provide information on conductivity, local charges and local band bendings in device stacks. Accordingly, various probes (metallic and doped-diamond tips) were evaluated for their efficacy. The following device stacks were successfully investigated using these methodologies:
(i) semi-vertical p-n diodes grown on Si substrate, which is used as test-vehicle for module development, and on QST® for devices covering up to 1.2 kV breakdown voltage
(ii) 200 V and 650 V p-GaN HEMT devices
For HEMTs, the methodology not only confirms the expected carrier distribution in the active layers through its direct visualization but also provides first of evidence of n-type carriers (1014 - 1015 cm-3) in strain management layers. The latter concerns associated RF loss. In case of p-n diodes, the contrast analysis was more complex as it was overwhelmed by the space charge of dislocations but it improved understanding of the role of dislocations in such devices.
KPFM simulations were carried out to test the a priori assumption of the analytical dependence of contact potential difference on the Fermi level, which resulted in compilation of a database with calculated values for varying surface states and doping. As the calculated relation deviated significantly from the analytical expression, it was concluded that the technique cannot be straightaway used for band profiling without using the corrections from the database. This technique was further advanced to visualize evolution of the surface potential of the channel when switching from its OFF state to ON state in p-GaN HEMTs, which was in good conformity with the calculated potential distribution of the channel.
WP3: Surface characterization of the recessed regions
This WP provided an in-depth understanding of the properties of dislocations. The atom core-structures of different types of dislocations were first identified, which led to their classification. By combining TCAD simulation with experimental SCM/KPFM analyses, their trap densities were determined in n-type GaN. It was also found that none of these dislocations can be associated with the leakage paths, which was in strong contradiction with what has been reported in the literature. However, in p-n diodes and HEMT stacks, up to 5% of the dislocations were associated with leakage current. Based on a comparative statistical analysis conducted using electron microscopy, this study suggested that the detrimental dislocations are likely to be pure-screw type.
WP4: Training activity for atomic resolution STM analysis
The fellow carried out his training on STM on: NX-Hivac Park Systems SPM installed at imec site and UNISOKU ultra-high vacuum system at KU Leuven. This has led to a close collaboration between the two groups for defectivity study in 2D materials yielding an undertanding of water intercalation process and their correlation with defects.
WP5: Project management, dissemination and communication
The progress of the project was regularly monitored and received continuous support from the supervisor, senior members of the epitaxy and SPM experts. The most relevant research outcomes of the project for both GaN and 2D programs have led to or will lead to publication in peer-reviewed journals. The results were disseminated within the industrial partners through IMEC’s Industrial Affiliation Program at internal PTW symposium and four international conferences. As a keynote speaker at International SPM symposium “Failure Analysis and Material Testing”, the fellow promoted the activities of the project to a broader audience of the electronic industry. UNOGAN contributed to the public outreach activity with a demo explaining the demand for nanotechnology on open company day.
1. Created a platform for in-operando measurement to study fundamental aspect of an active device in commercial SPMs
2. Established a robust method for correlation study between SPM methodology and electron microscopy on devices
3. Computed contact potential difference and differential capacitance to extract trap densities of defects
4. Classified dislocations in GaN based on their core structures
2. Established a robust method for correlation study between SPM methodology and electron microscopy on devices
3. Computed contact potential difference and differential capacitance to extract trap densities of defects
4. Classified dislocations in GaN based on their core structures