Final Report Summary - AL-IN-WON (AlGaN and InAlN based microwave components)
Executive Summary:
The AL-IN-WON project was aimed at developing new generation of WBG materials and devices (AlInN/GaN) to address space borne application up to Ka band and evaluate existing 0.25 PHEMT AlGaN/GaN UMS source in its last stage before qualification for Ku band.
Behind the requirements of the space users, there is also a real necessity to catch our delay compared with the US and Japan technology companies and consequently, have an European technology free from ITAR regulations.
The huge activities performed during the project duration led to significant progresses and many outstanding outcomes:
For AlInN/GaN activities, at epitaxy level:
o Growth of 3 inch AlInN/GaN HEMT structure in multi-wafer reactor
o Carbon compensated buffer approach demonstrated
o Rsheet # 220 W/sq – ns # 1.7e13 cm-² - µ=1800cm²/Vs ( 300K)
At process level:
o 0.15µm gate technology improved with optimized passivation scheme.
o Low lag effects and low leakage currents (<< 1mA/mm) achieved
o Gmmax = 450mS/mm - Ids+ = 1.4A/mm
o Improvement in performance with 0.15µm vs 0.25µm devices (short channel effects mastered) on last generation realized during time extension
o 0.15µm process ready for MMIC realization (process shared between III-V Lab & UMS)
At process electrical performance level:
o Ft = 55 GHz & Fmax = 135GHz at 20V
o 3.2W/mm – 28% @ 17.5V on 6x50 µm @ 30GHz CW
o 6.2W/mm – 36% measured @18GHz using the last and optimized process variation
At circuit level, electrical simulations, based on transistors characterization and modeling, a Ka band HPA and LNA have been performed on an interim process variation. The results obtained during this phase, the electrical performance obtained on the last process version, the preliminary reliability and robustness evaluations confirm the InAlN/GaN interest and its capability to address Space requirements in terms of Output Power, PAE and robustness in harsh environment.
For AlGaN/GaN activities, Ku band HPA and LNA MMICs have been designed (2 iterations) based on the pre-qualified UMS GH25-10 process. The circuits have been characterized in test fixture to evaluate its performance and preliminary reliability and robustness. Results are very good and compliant with the targeted challenging specifications.
This demonstrates also the capability of the process to address Space borne applications in Ku band.
Important Dissemination activities have been done through 4 workshops, international conferences and 20 publications.
III-VLab has submitted to the French patent office two potential patents following the activities on the AlInN/Gan material and process.
Project Context and Objectives:
AL-IN-WON is a 3-years project, starts on November 1st 2010. 5 European Companies (UMS SAS, UMS GmbH, TAS, 3-5LAB and MEC) and 2 Academics (XLIM, UNIDP) are involved in this project funded by the European Commission under Seventh Framework Programme addressing Space technologies.
This project is focused on the development of a new generation of wide band gap (WBG) GaN technology and electronic devices for which strong impacts in terms of performance, reliability and robustness are expected.
New semiconductors such as gallium nitride (GaN) are being developed for both civilian and governmental applications which are expected to provide a higher level of robustness and performance for RF and microwave systems. Why is GaN so interesting? Indeed, it presents several basic characteristics which make it an almost ideal semiconductor: its breakdown electric field is about 10 times that of GaAs, it can withstand high radiation and high temperature (50°C to 100°C more than GaAs), its thermal conductivity (when grown on silicon carbide) is x10 that of GaAs and x3 that of Si and its electron mobility is comparable to GaAs allowing microwave applications.
GaN technology is an ideal candidate for several applications and especially for on board satellite payloads to address requirements more and more important in terms of transmitted power level, power efficiency, miniaturisation, life time,…etc in a harsh environment. Satellite equipment is exposed in space to severe radiations and sharp temperature variations over a wide range (typ. –55°/+65°C). Also, a successful design should decrease size and weight to a minimum, save a maximum of electrical power and ensure a long life time without failure (>15 years for a geostationary satellite).
Since the earliest demonstrations, a significant advance has been taken by US companies supported strongly by their government. Potential US sources are identified but this promising and very strategic technology will be impacted by export regulations (ITAR). Moreover Japanese sources have made remarked announcements with commercial devices already, which is not surprising for a country which was a leader in the field of power microwave transistors in the recent past.
The European situation is more contrasted with, as usual, many different actors. However, there is a real will in this application field of co-ordination now among the financing bodies such as the Ministries of Defence (MoD) and the Space Agencies. At present there is no established GaN space component supply chain in Europe and overseas component procurement is the subject of export restrictions.
In this context, AL-IN-WON project explored two main disrupting routes:
o New generation of WBG devices based on new epi material (InAlN/GaN) for strong improvements in terms of performance and reliability.
o High efficiency / High power generation in Ku / Ka frequency Bands.
This 3-5LAB AlInN/GaN epi material was planed to be evaluated through the development of the process and Microwave Monolithic Integrated Circuits (MMICs), CW Ku and Ka Band MMIC High Power Amplifiers (HPA) and Low Noise Amplifiers (LNA) for the evaluation for the maturity and the capability of the GaN processes to addressing space applications. As reference and for comparison purpose, part of the MMICs, was developed on the UMS AlGaN/GaN process which is more conventional and the first European qualified GaN Process in production.
UMS had also the opportunity thanks to this program to evaluate the benefit of this new material for future evolution of its processes.
This project had also as major objectives to provide a secure alternative way for Europe to access to this strategic space technology. Following our information, “standard” approach like AlGaN/GaN HEMT technology underway in Europe is fully covered by US and Japan patents portfolio. A major risk exists in front of this situation.
Moreover, InAlN/GaN HEMT technology is protected by EU patent and thus offers a strategic and secure position to develop these technologies.
Underway U.S. and Japan R&D development on these items confirm these assessments.
Project Results:
2 main results can be reported at the end of the project:
- a first one concerning the development of the new generation of WBG devices based on a new epi material (AlInN/GaN on SiC)
- a second one concerning the development of Ku band MMICs addressing future space borne requirements based on the European GaN source from UMS (AlGaN/GaN 0.25 PHEMT process on SiC) .
A collaborative process on AlInN/GaN was set up between 3-5LAB and UMS SAS and UMS GmbH in order to be able to manufacture electronic devices. A first transistor mask-set was fabricated and process shared between UMS GmbH and 3-5LAB, were validated on 3 inch wafers. Two challenges have been took up during this phase: for one hand, the transfer of the InAlN/GaN growth conditions from a former 2 inch reactor to a new acquired multi-wafers reactor (from 3 to 6 inch), the compatibility with the UMS production line and for a second hand, the necessity to reduce the transistor gate length from 0.25µm/0.7µm down to 0.15µm in order to address high frequency operation.
The following sticking points or questions were addressed during the optimization of the process which were far heavier than expected:
o Ohmic contact resolution for precise ohmic contact positioning in regards with gate;
o Ohmic access resistance control;
o Difference in behaviour of 2D sheet resistance during processing;
o Gate metal adhesion;
o Pinch-off voltage;
o Lag effect control;
o Power gain improvement.
A large part of the project have been dedicated to investigate the origin of problems and solve them.This generates a six months extension time over the three initial years.
Latest results in November 2012 showed excellent progress with measurement of Ft and Fmax of 70GHz and 140GHz respectively. The best trans-conductance so far obtained reaches 450mS/mm. All these values were in line with electrical properties needed for realising Ka Band amplifiers.
During the last period of the project, excellent progresses were achieved by 3 5LAB to obtain 0.15µm AlInN/GaN HEMT showing very good DC, pulsed as well as small and large signal results.
An action plan has been set-up to adapt oneself to the difficulties met. This concerned essentially
o Development of a new mask set compatible of fast processing
o Development of carbon compensated GaN buffer layers.
o Realization and processing of HEMT including six different types of buffer layer.
o Evaluation of three different passivation layers and a large set of surface preparations
o Realization and process of 3’’ wafers grown in upgraded 2’’ mono wafer reactor (3 wafers) for reference
o Purchase and evaluation of commercial InAlN/GaN wafers (3 wafers processed)
The work was mostly dedicated to investigate the carbon compensation approach of the buffer layer to improve the confinement of the 2D-gas and thus the pinch-off voltage of the devices. Six types of buffer layers were grown and processed with a 0.15µm gate length technology. The other sticking point was the study of the passivation layers and their associated surface preparations to master dispersive effects and leakage currents. Three types of passivation realized on different tools were investigated with many different surface preparation schemes.
Progresses were constant and best results were achieved during the time extension thanks to improvement coming from our third passivation scheme. At the epitaxy level, low sheet resistance of 200/sq with carrier density of 1.7.1013 cm-² were obtained on 3 inch InAlN/GaN on SiC wafers. It allowed us to achieve HEMT transistors with drain current of 1.4A/mm with very low drain lag effects and low gate leakage current (<<1mA/mm). Extrinsic transconductance was maintained to excellent value of 450mS/mm. Small signal measurements showed Ft and Fmax of 55GHz and 135GHz respectively. Large signal measurements were performed on wafers showing 3.2W/mm 28% of power added efficiency (PAE) @ 30 GHz and 6.2W/mm 36% of PAE @ 18 GHz.
Despite the difficulties faced by changing the epitaxy reactor and the gate length shrinking, latest results achieved are now very good and reinforce the interest of such InAlN/GaN technology. During the full program, 28 InAlN/GaN on SiC wafers were grown in both reactors and 31 were processed instead of 14 in the initial work plan.
Different representative samples of transistors from interim process variations have been characterized and modelled by XLIM, 3-5LAB and MEC in order to evaluate by simulations the capability of the process to address Space borne requirements in Ka band frequency released by Thales Alenia Space (TAS).
o TAS has designed a first Ka band HPA based on a preliminary model of the 8x75µm InAlN/GaN transistor. Over 20.5GHz-21 GHz at 5 dB compression a gain around 12dB, an output power around 42 dBm and a PAE around 25% are simulated.
The late performance reported above about the last process variation let us be very optimistic about the potential outstanding performance which could be reach with a such process.
o Two versions of the LNA have been designed to reach performance as close as possible to targeted specifications. As for the HPA, the performance are below the target due to an transistor size limitation and interim process but encouraging. Between 27.5GHz and 31GHz a Gain higher than 22dB (target 25dB) and a Noise Figure around 3dB (target 1.9dB) is demonstrated.
During all the process development, several versions of materials and devices have been evaluated in terms of reliability by UNIPD and 3-5LAB.
After preliminary DC characterizations, several tests have been performed:
o Trapping analysis to understand mechanism which affect current collapse and define the traps nature, their activation energy and find a correlation with the device structure.
o Radiation hardness (Proton radiations) to understand the device robustness, effects on DC and pulsed parameters and effect of process variations and gate length.
o Off State Step Stress, Reverse Bias Step Stress to define breakdown robustness and understand degradation mechanism.
Two traps have been detected. Both are thermally activated (Ea=0.85eV and 0.94eV). Both traps behave as line defects and a high Carbon buffer doping induces an increase in both traps amplitude but no variation in activation energy.
A good robustness to proton radiations at 3MeV has been demonstrated for very high fluence (1.E+14 and 3.E+14) and consistency with previous works reported in literature, both concerning robustness and degradations mechanism. The main impacts are:
o a IDSS decrease, RON increase, VTH shift and slight gm decrease.
o a Current collapse increase only with high VGDQ value and not when VGSQ=VGDQ.
The device with low C doping exhibits a high current collapse variation after radiation when a high VGDQ is applied. Also, a very good correlation is observed between IDSS decrease and the radiation fluence.
Off state step stress and constant voltage stress demonstrate a Good robustness. Although different mechanisms, different VG exhibit similar robustness (almost 50V except for VG=0V due to thermal aspects) and degradation mechanisms, namely gate current increase, IDSS and max gm decrease, VTH non monotonic trend
After Reverse bias step stress, the devices show a good robustness and repeatability. Although gate current increase, no catastrophic failure is reported up to VG=-98V.
Significant spots are noticed in emission microscopy. For higher Vgstress (-40V) a monotonic decrease of both max gm and IDSS is noticed; VTH shows a non monotonic decrease.
III-VLab has submitted to the French patent office a potential patent entitled: "Couche tampon optimisée pour transistor à effet de champ à haute mobilité", by Jean-Claude Jacquet, Raphaël Aubry, Piero Gamarra, Olivier Jardel, Stéphane Piotrowicz", in March 2014. The positive appraisal by the French patent office is necessary to get a patent number.
Moreover, at the date of report redaction, an innovation is to be submitted to the III-VLab patent committees. It is dedicated to an innovative passive approach to decrease potential deep electrical level effects and to improve long term reliability. It should be presented on the first week of June 2014.
2. Relating to the AlGaN/GaN activities, on the contrary to the previous task, the AlGaN/GaN 0.25 PHEMT process already exists. It is not finalized at the beginning of the project.
Based on Thales Alenia Space (TAS) targeted specifications corresponding to their future needs in Ku band, Low Noise Amplifier (LNA) and High Power Amplifier (HPA) are designed by TAS, MEC and UMS. The required performance are challenging. Two Design iterations are planned. For each one several versions of MMICs are designed and realized on 3 inch wafer. Best candidates are selected based on on wafer pulsed measurements ([S] parameters, Power and DC) and some of them characterized in test fixture.
The second iteration allowed us to refine the design and improve some performance thanks to transistor models tuning.
The performance are good close to the requirements and very reproducible. This demonstrates the capability of the the UMS GaN source to address such applications.
o For a 2-stages HPA within the operating frequency band (12.2GHz-12.7GHz) Gain higher than 16dB, Output Power higher than 41 dBm and PAE higher than 30% (at 4dB compression) up to 85°C are demonstrated with a good reproducibility from piece to piece. Due to Space derating, these performance have been obtained for 22.5V supply voltage.
Comparing with the requirements, Output Power and PAE (respectively 43dBm and 40%) are lower than expected at 85°C but in specification at 25°C at 6 dB compression.
o For a 3-stages LNA within the operating frequency band (12.75GHz-14.8GHz) gain higher than 22dB, Noise Figure better than 1.75dB for a P1dB higher than 21.5dB and over an Input overdrive up to 23dBm are demonstrated. These performance are obtained at 25°C for 10V supply voltage.
Comparing with the requirements, Noise Figure is higher than expected (1.4dB) but nevertheless acceptable for a process not optimized for a such performance. We can also notice the very good robustness in terms of overdrive which is a advantage compared with GaAs devices.
The reliability and the robustness have been evaluated on the LNA and HPA MMICs from the first iteration. Several tests have been performed to determine some design boundaries (RF step stress, RF life test). The HPA present no significative drifts on DC or RF parameters up to 8dB compression. A life test is still on going (1000 hours). The LNA present no drift up to 6dB. At 15dBm Input Power (10dB compression) -15% is observe on DC ID3 with negligible impact on the RF performance.
Potential Impact:
[1] As introduction, the main topic of this project was related to the development of new alternative of wideband gap semiconductor based on the InAlN system. This one is foreseen in the next year as a replacement to the standard AlGaN based, used today by all semiconductor manufacturers around the world. The purpose of this project is to address analog RF / microwave applications roughly above the X- Band and more precisely frequencies up to Ka band. This part of the bandwidth covers the up and down links for high speed data transmission by satellite. In the next 10 years, we expect a tremendous increasing of the mobile data transmission by satellite. It will address all transmission supports (flight, automotive, train ….) and also an improved interconnection in between network data centers and e-electronic system. This project contributes at the research and development level to provide an independency to Europe concerning the access to a key technology used in the design of microwave transmitters. To fulfill this objective, this project has involved and mixed a lot of scientific domains to progressively develop a better understanding of this new technology. These domains covered from the nm scale up to the cm scale with the control and the analysis of atomic layer up to final characterization at the module level. To develop such activities, an intensive exchange of information was necessary in between the scientists and the engineers. Publications, Forums and conferences were the better platforms to have access to accurate and extensive information. The two main channels of communication used the EuMA (European Microwave Association) and IEEE (Institute of Electrical and Electronics Engineers) By this way, different publications promoted the different results obtained though the project. Most of the results obtained on the optimization of the process were published into a paper named “InAlN/GaN Devices for Microwave Applications” at the IMS 2013 Conference at SEATTLE, USA. It is probably the most prestigious conference in the field the microwave around the world. In this paper, some details about the epitaxy used, especially the interest of a carbon-doped buffer are given. It is also mentioned the critical aspect on the device performances of the interfacial layer in between the InAlN and the surface passivation. Some communications have been done concerning the methodology put in place for the characterizations : which strategy of measurement on which sensitive electrical parameters and for which test set-up. It was detailed into the publications “Trap Characterization of Microwave GaN HEMTs Based on Frequency Dispersion of the Output-Admittance”; accepted for the conference EuMIC 2014 (October) and "Modeling of trap induced dispersion of large signal dynamic Characteristics of GaN HEMTs, also published at the IMS conference 2013.
One important part of this project was related to the evaluation of the reliability which is critical for technology planned for space application. The foreground developed for this activity consisted :
– to evaluate and develop a strategy of measurement in order to extract the sensitive parameters of degradation. Reference to : GaN HEMT Reliability: From Time Dependent Gate Degradation to On-state Failure Mechanisms”. Invited, 2012 MRS, San Francisco, April 2012.
- to analyze the root cause of the degradation by putting in place a methodology of failure analysis.
- to evaluate the robustness of the technology against physical aggression (radiation) or from the environment (temperature). Reference to : “Degradation of AlInN/AlN/GaN High Electron Mobility Transistors under Proton Irradiation”, RADECS 2012 and “Measurement of the thermal impedance of GaN HEMTs using the 3ω method.”; Electronics Letters, juin 2012.
Finally, different designs (microwave circuits) were designed based on this technology to evaluate the performances and the capabilities for space applications. The methodology of design, the architecture and the results obtained from this activity was published or shared through publications and workshops. These exchanges were really helpful to disseminate into the community of space users the possibilities of integration of the InAlN and the AlGaN technologies.
The main elements of communications are shared into “A robust Ku-band low noise amplifier using an industrial 0.25-µm AlGaN/GaN on SiC process”, EuMIC conference, 2013 and “GaN for Space Applications: ALINWON main results ”, EuMW 2012 – Amsterdam.
The AlinWon project was related to the development and the evaluation of a new technology for semiconductor. It means that from the R&D phase to industrialization up to a final use (qualified product), we are speaking generally of 5 to 10 years long time scaling. If we consider moreover the space requirement to be evaluated and qualified for a flight model, we have to add 3 to 5 years. We consider to be at the mid period of development, an extra period of 3 to 4 years being to be considered. This period corresponds to the transfer and the re-evaluation (optimization) of this technology at the industrial level.
Obviously, the main foreground planned to be exploited is related to the optimization of the material (epitaxial scheme) considering the interaction with the manufacturing process. Discussions are underway with a Eu based SME company having the capacity to reproduce such material on their industrial tool. The exact context (funding, support) has to be considered in parallel with options offered into the Horizon 2020 program. This point is very important and contributes to set up a sustainable European Supply chain. We note that this transfer will be based on the use of SiC substrate as support to the epitaxy. We expect to get access to a industrial source in 2016 / 2017.
The second is critical aspect is related to the optimization of the process development. Along this project, a continuous and intensive phase of experiment has improved a lot the electrical behavior of the device (PHEMT) by optimization of the building blocks : cleaning steps, ohmic contact, passsivation layers and associated properties, Schottky contact, annealing phase …. All these results are part of a foreground to be re-used inside a new phase dedicated to the industrialization and re-optimization if necessary. Such material (III-N semiconductor) being sensitive to the equipment (process), it is necessary to re-evaluate the recipes set up during the R&D phase. This activity is planned with one of the partner of this program UMS. The exact context (funding, support) has to be considered in parallel with options offered into the Horizon 2020 program. This second step would be initiated in 2015.
The AlInWon program has contributed to the development and evaluation of InAlGaN technology in preparation to the development of Ka band Satcom market. This one is foreseen as a key support around the world for transmission of high data rate. In consequence, by supporting a free access to strategic technologies, Europe will confirm his leadership in this new area.
List of Websites:
website address: http://www.alinwon-fp7.eu
Contact person: Didier Floriot / didier.floriot@ums-gaas.com
The AL-IN-WON project was aimed at developing new generation of WBG materials and devices (AlInN/GaN) to address space borne application up to Ka band and evaluate existing 0.25 PHEMT AlGaN/GaN UMS source in its last stage before qualification for Ku band.
Behind the requirements of the space users, there is also a real necessity to catch our delay compared with the US and Japan technology companies and consequently, have an European technology free from ITAR regulations.
The huge activities performed during the project duration led to significant progresses and many outstanding outcomes:
For AlInN/GaN activities, at epitaxy level:
o Growth of 3 inch AlInN/GaN HEMT structure in multi-wafer reactor
o Carbon compensated buffer approach demonstrated
o Rsheet # 220 W/sq – ns # 1.7e13 cm-² - µ=1800cm²/Vs ( 300K)
At process level:
o 0.15µm gate technology improved with optimized passivation scheme.
o Low lag effects and low leakage currents (<< 1mA/mm) achieved
o Gmmax = 450mS/mm - Ids+ = 1.4A/mm
o Improvement in performance with 0.15µm vs 0.25µm devices (short channel effects mastered) on last generation realized during time extension
o 0.15µm process ready for MMIC realization (process shared between III-V Lab & UMS)
At process electrical performance level:
o Ft = 55 GHz & Fmax = 135GHz at 20V
o 3.2W/mm – 28% @ 17.5V on 6x50 µm @ 30GHz CW
o 6.2W/mm – 36% measured @18GHz using the last and optimized process variation
At circuit level, electrical simulations, based on transistors characterization and modeling, a Ka band HPA and LNA have been performed on an interim process variation. The results obtained during this phase, the electrical performance obtained on the last process version, the preliminary reliability and robustness evaluations confirm the InAlN/GaN interest and its capability to address Space requirements in terms of Output Power, PAE and robustness in harsh environment.
For AlGaN/GaN activities, Ku band HPA and LNA MMICs have been designed (2 iterations) based on the pre-qualified UMS GH25-10 process. The circuits have been characterized in test fixture to evaluate its performance and preliminary reliability and robustness. Results are very good and compliant with the targeted challenging specifications.
This demonstrates also the capability of the process to address Space borne applications in Ku band.
Important Dissemination activities have been done through 4 workshops, international conferences and 20 publications.
III-VLab has submitted to the French patent office two potential patents following the activities on the AlInN/Gan material and process.
Project Context and Objectives:
AL-IN-WON is a 3-years project, starts on November 1st 2010. 5 European Companies (UMS SAS, UMS GmbH, TAS, 3-5LAB and MEC) and 2 Academics (XLIM, UNIDP) are involved in this project funded by the European Commission under Seventh Framework Programme addressing Space technologies.
This project is focused on the development of a new generation of wide band gap (WBG) GaN technology and electronic devices for which strong impacts in terms of performance, reliability and robustness are expected.
New semiconductors such as gallium nitride (GaN) are being developed for both civilian and governmental applications which are expected to provide a higher level of robustness and performance for RF and microwave systems. Why is GaN so interesting? Indeed, it presents several basic characteristics which make it an almost ideal semiconductor: its breakdown electric field is about 10 times that of GaAs, it can withstand high radiation and high temperature (50°C to 100°C more than GaAs), its thermal conductivity (when grown on silicon carbide) is x10 that of GaAs and x3 that of Si and its electron mobility is comparable to GaAs allowing microwave applications.
GaN technology is an ideal candidate for several applications and especially for on board satellite payloads to address requirements more and more important in terms of transmitted power level, power efficiency, miniaturisation, life time,…etc in a harsh environment. Satellite equipment is exposed in space to severe radiations and sharp temperature variations over a wide range (typ. –55°/+65°C). Also, a successful design should decrease size and weight to a minimum, save a maximum of electrical power and ensure a long life time without failure (>15 years for a geostationary satellite).
Since the earliest demonstrations, a significant advance has been taken by US companies supported strongly by their government. Potential US sources are identified but this promising and very strategic technology will be impacted by export regulations (ITAR). Moreover Japanese sources have made remarked announcements with commercial devices already, which is not surprising for a country which was a leader in the field of power microwave transistors in the recent past.
The European situation is more contrasted with, as usual, many different actors. However, there is a real will in this application field of co-ordination now among the financing bodies such as the Ministries of Defence (MoD) and the Space Agencies. At present there is no established GaN space component supply chain in Europe and overseas component procurement is the subject of export restrictions.
In this context, AL-IN-WON project explored two main disrupting routes:
o New generation of WBG devices based on new epi material (InAlN/GaN) for strong improvements in terms of performance and reliability.
o High efficiency / High power generation in Ku / Ka frequency Bands.
This 3-5LAB AlInN/GaN epi material was planed to be evaluated through the development of the process and Microwave Monolithic Integrated Circuits (MMICs), CW Ku and Ka Band MMIC High Power Amplifiers (HPA) and Low Noise Amplifiers (LNA) for the evaluation for the maturity and the capability of the GaN processes to addressing space applications. As reference and for comparison purpose, part of the MMICs, was developed on the UMS AlGaN/GaN process which is more conventional and the first European qualified GaN Process in production.
UMS had also the opportunity thanks to this program to evaluate the benefit of this new material for future evolution of its processes.
This project had also as major objectives to provide a secure alternative way for Europe to access to this strategic space technology. Following our information, “standard” approach like AlGaN/GaN HEMT technology underway in Europe is fully covered by US and Japan patents portfolio. A major risk exists in front of this situation.
Moreover, InAlN/GaN HEMT technology is protected by EU patent and thus offers a strategic and secure position to develop these technologies.
Underway U.S. and Japan R&D development on these items confirm these assessments.
Project Results:
2 main results can be reported at the end of the project:
- a first one concerning the development of the new generation of WBG devices based on a new epi material (AlInN/GaN on SiC)
- a second one concerning the development of Ku band MMICs addressing future space borne requirements based on the European GaN source from UMS (AlGaN/GaN 0.25 PHEMT process on SiC) .
A collaborative process on AlInN/GaN was set up between 3-5LAB and UMS SAS and UMS GmbH in order to be able to manufacture electronic devices. A first transistor mask-set was fabricated and process shared between UMS GmbH and 3-5LAB, were validated on 3 inch wafers. Two challenges have been took up during this phase: for one hand, the transfer of the InAlN/GaN growth conditions from a former 2 inch reactor to a new acquired multi-wafers reactor (from 3 to 6 inch), the compatibility with the UMS production line and for a second hand, the necessity to reduce the transistor gate length from 0.25µm/0.7µm down to 0.15µm in order to address high frequency operation.
The following sticking points or questions were addressed during the optimization of the process which were far heavier than expected:
o Ohmic contact resolution for precise ohmic contact positioning in regards with gate;
o Ohmic access resistance control;
o Difference in behaviour of 2D sheet resistance during processing;
o Gate metal adhesion;
o Pinch-off voltage;
o Lag effect control;
o Power gain improvement.
A large part of the project have been dedicated to investigate the origin of problems and solve them.This generates a six months extension time over the three initial years.
Latest results in November 2012 showed excellent progress with measurement of Ft and Fmax of 70GHz and 140GHz respectively. The best trans-conductance so far obtained reaches 450mS/mm. All these values were in line with electrical properties needed for realising Ka Band amplifiers.
During the last period of the project, excellent progresses were achieved by 3 5LAB to obtain 0.15µm AlInN/GaN HEMT showing very good DC, pulsed as well as small and large signal results.
An action plan has been set-up to adapt oneself to the difficulties met. This concerned essentially
o Development of a new mask set compatible of fast processing
o Development of carbon compensated GaN buffer layers.
o Realization and processing of HEMT including six different types of buffer layer.
o Evaluation of three different passivation layers and a large set of surface preparations
o Realization and process of 3’’ wafers grown in upgraded 2’’ mono wafer reactor (3 wafers) for reference
o Purchase and evaluation of commercial InAlN/GaN wafers (3 wafers processed)
The work was mostly dedicated to investigate the carbon compensation approach of the buffer layer to improve the confinement of the 2D-gas and thus the pinch-off voltage of the devices. Six types of buffer layers were grown and processed with a 0.15µm gate length technology. The other sticking point was the study of the passivation layers and their associated surface preparations to master dispersive effects and leakage currents. Three types of passivation realized on different tools were investigated with many different surface preparation schemes.
Progresses were constant and best results were achieved during the time extension thanks to improvement coming from our third passivation scheme. At the epitaxy level, low sheet resistance of 200/sq with carrier density of 1.7.1013 cm-² were obtained on 3 inch InAlN/GaN on SiC wafers. It allowed us to achieve HEMT transistors with drain current of 1.4A/mm with very low drain lag effects and low gate leakage current (<<1mA/mm). Extrinsic transconductance was maintained to excellent value of 450mS/mm. Small signal measurements showed Ft and Fmax of 55GHz and 135GHz respectively. Large signal measurements were performed on wafers showing 3.2W/mm 28% of power added efficiency (PAE) @ 30 GHz and 6.2W/mm 36% of PAE @ 18 GHz.
Despite the difficulties faced by changing the epitaxy reactor and the gate length shrinking, latest results achieved are now very good and reinforce the interest of such InAlN/GaN technology. During the full program, 28 InAlN/GaN on SiC wafers were grown in both reactors and 31 were processed instead of 14 in the initial work plan.
Different representative samples of transistors from interim process variations have been characterized and modelled by XLIM, 3-5LAB and MEC in order to evaluate by simulations the capability of the process to address Space borne requirements in Ka band frequency released by Thales Alenia Space (TAS).
o TAS has designed a first Ka band HPA based on a preliminary model of the 8x75µm InAlN/GaN transistor. Over 20.5GHz-21 GHz at 5 dB compression a gain around 12dB, an output power around 42 dBm and a PAE around 25% are simulated.
The late performance reported above about the last process variation let us be very optimistic about the potential outstanding performance which could be reach with a such process.
o Two versions of the LNA have been designed to reach performance as close as possible to targeted specifications. As for the HPA, the performance are below the target due to an transistor size limitation and interim process but encouraging. Between 27.5GHz and 31GHz a Gain higher than 22dB (target 25dB) and a Noise Figure around 3dB (target 1.9dB) is demonstrated.
During all the process development, several versions of materials and devices have been evaluated in terms of reliability by UNIPD and 3-5LAB.
After preliminary DC characterizations, several tests have been performed:
o Trapping analysis to understand mechanism which affect current collapse and define the traps nature, their activation energy and find a correlation with the device structure.
o Radiation hardness (Proton radiations) to understand the device robustness, effects on DC and pulsed parameters and effect of process variations and gate length.
o Off State Step Stress, Reverse Bias Step Stress to define breakdown robustness and understand degradation mechanism.
Two traps have been detected. Both are thermally activated (Ea=0.85eV and 0.94eV). Both traps behave as line defects and a high Carbon buffer doping induces an increase in both traps amplitude but no variation in activation energy.
A good robustness to proton radiations at 3MeV has been demonstrated for very high fluence (1.E+14 and 3.E+14) and consistency with previous works reported in literature, both concerning robustness and degradations mechanism. The main impacts are:
o a IDSS decrease, RON increase, VTH shift and slight gm decrease.
o a Current collapse increase only with high VGDQ value and not when VGSQ=VGDQ.
The device with low C doping exhibits a high current collapse variation after radiation when a high VGDQ is applied. Also, a very good correlation is observed between IDSS decrease and the radiation fluence.
Off state step stress and constant voltage stress demonstrate a Good robustness. Although different mechanisms, different VG exhibit similar robustness (almost 50V except for VG=0V due to thermal aspects) and degradation mechanisms, namely gate current increase, IDSS and max gm decrease, VTH non monotonic trend
After Reverse bias step stress, the devices show a good robustness and repeatability. Although gate current increase, no catastrophic failure is reported up to VG=-98V.
Significant spots are noticed in emission microscopy. For higher Vgstress (-40V) a monotonic decrease of both max gm and IDSS is noticed; VTH shows a non monotonic decrease.
III-VLab has submitted to the French patent office a potential patent entitled: "Couche tampon optimisée pour transistor à effet de champ à haute mobilité", by Jean-Claude Jacquet, Raphaël Aubry, Piero Gamarra, Olivier Jardel, Stéphane Piotrowicz", in March 2014. The positive appraisal by the French patent office is necessary to get a patent number.
Moreover, at the date of report redaction, an innovation is to be submitted to the III-VLab patent committees. It is dedicated to an innovative passive approach to decrease potential deep electrical level effects and to improve long term reliability. It should be presented on the first week of June 2014.
2. Relating to the AlGaN/GaN activities, on the contrary to the previous task, the AlGaN/GaN 0.25 PHEMT process already exists. It is not finalized at the beginning of the project.
Based on Thales Alenia Space (TAS) targeted specifications corresponding to their future needs in Ku band, Low Noise Amplifier (LNA) and High Power Amplifier (HPA) are designed by TAS, MEC and UMS. The required performance are challenging. Two Design iterations are planned. For each one several versions of MMICs are designed and realized on 3 inch wafer. Best candidates are selected based on on wafer pulsed measurements ([S] parameters, Power and DC) and some of them characterized in test fixture.
The second iteration allowed us to refine the design and improve some performance thanks to transistor models tuning.
The performance are good close to the requirements and very reproducible. This demonstrates the capability of the the UMS GaN source to address such applications.
o For a 2-stages HPA within the operating frequency band (12.2GHz-12.7GHz) Gain higher than 16dB, Output Power higher than 41 dBm and PAE higher than 30% (at 4dB compression) up to 85°C are demonstrated with a good reproducibility from piece to piece. Due to Space derating, these performance have been obtained for 22.5V supply voltage.
Comparing with the requirements, Output Power and PAE (respectively 43dBm and 40%) are lower than expected at 85°C but in specification at 25°C at 6 dB compression.
o For a 3-stages LNA within the operating frequency band (12.75GHz-14.8GHz) gain higher than 22dB, Noise Figure better than 1.75dB for a P1dB higher than 21.5dB and over an Input overdrive up to 23dBm are demonstrated. These performance are obtained at 25°C for 10V supply voltage.
Comparing with the requirements, Noise Figure is higher than expected (1.4dB) but nevertheless acceptable for a process not optimized for a such performance. We can also notice the very good robustness in terms of overdrive which is a advantage compared with GaAs devices.
The reliability and the robustness have been evaluated on the LNA and HPA MMICs from the first iteration. Several tests have been performed to determine some design boundaries (RF step stress, RF life test). The HPA present no significative drifts on DC or RF parameters up to 8dB compression. A life test is still on going (1000 hours). The LNA present no drift up to 6dB. At 15dBm Input Power (10dB compression) -15% is observe on DC ID3 with negligible impact on the RF performance.
Potential Impact:
[1] As introduction, the main topic of this project was related to the development of new alternative of wideband gap semiconductor based on the InAlN system. This one is foreseen in the next year as a replacement to the standard AlGaN based, used today by all semiconductor manufacturers around the world. The purpose of this project is to address analog RF / microwave applications roughly above the X- Band and more precisely frequencies up to Ka band. This part of the bandwidth covers the up and down links for high speed data transmission by satellite. In the next 10 years, we expect a tremendous increasing of the mobile data transmission by satellite. It will address all transmission supports (flight, automotive, train ….) and also an improved interconnection in between network data centers and e-electronic system. This project contributes at the research and development level to provide an independency to Europe concerning the access to a key technology used in the design of microwave transmitters. To fulfill this objective, this project has involved and mixed a lot of scientific domains to progressively develop a better understanding of this new technology. These domains covered from the nm scale up to the cm scale with the control and the analysis of atomic layer up to final characterization at the module level. To develop such activities, an intensive exchange of information was necessary in between the scientists and the engineers. Publications, Forums and conferences were the better platforms to have access to accurate and extensive information. The two main channels of communication used the EuMA (European Microwave Association) and IEEE (Institute of Electrical and Electronics Engineers) By this way, different publications promoted the different results obtained though the project. Most of the results obtained on the optimization of the process were published into a paper named “InAlN/GaN Devices for Microwave Applications” at the IMS 2013 Conference at SEATTLE, USA. It is probably the most prestigious conference in the field the microwave around the world. In this paper, some details about the epitaxy used, especially the interest of a carbon-doped buffer are given. It is also mentioned the critical aspect on the device performances of the interfacial layer in between the InAlN and the surface passivation. Some communications have been done concerning the methodology put in place for the characterizations : which strategy of measurement on which sensitive electrical parameters and for which test set-up. It was detailed into the publications “Trap Characterization of Microwave GaN HEMTs Based on Frequency Dispersion of the Output-Admittance”; accepted for the conference EuMIC 2014 (October) and "Modeling of trap induced dispersion of large signal dynamic Characteristics of GaN HEMTs, also published at the IMS conference 2013.
One important part of this project was related to the evaluation of the reliability which is critical for technology planned for space application. The foreground developed for this activity consisted :
– to evaluate and develop a strategy of measurement in order to extract the sensitive parameters of degradation. Reference to : GaN HEMT Reliability: From Time Dependent Gate Degradation to On-state Failure Mechanisms”. Invited, 2012 MRS, San Francisco, April 2012.
- to analyze the root cause of the degradation by putting in place a methodology of failure analysis.
- to evaluate the robustness of the technology against physical aggression (radiation) or from the environment (temperature). Reference to : “Degradation of AlInN/AlN/GaN High Electron Mobility Transistors under Proton Irradiation”, RADECS 2012 and “Measurement of the thermal impedance of GaN HEMTs using the 3ω method.”; Electronics Letters, juin 2012.
Finally, different designs (microwave circuits) were designed based on this technology to evaluate the performances and the capabilities for space applications. The methodology of design, the architecture and the results obtained from this activity was published or shared through publications and workshops. These exchanges were really helpful to disseminate into the community of space users the possibilities of integration of the InAlN and the AlGaN technologies.
The main elements of communications are shared into “A robust Ku-band low noise amplifier using an industrial 0.25-µm AlGaN/GaN on SiC process”, EuMIC conference, 2013 and “GaN for Space Applications: ALINWON main results ”, EuMW 2012 – Amsterdam.
The AlinWon project was related to the development and the evaluation of a new technology for semiconductor. It means that from the R&D phase to industrialization up to a final use (qualified product), we are speaking generally of 5 to 10 years long time scaling. If we consider moreover the space requirement to be evaluated and qualified for a flight model, we have to add 3 to 5 years. We consider to be at the mid period of development, an extra period of 3 to 4 years being to be considered. This period corresponds to the transfer and the re-evaluation (optimization) of this technology at the industrial level.
Obviously, the main foreground planned to be exploited is related to the optimization of the material (epitaxial scheme) considering the interaction with the manufacturing process. Discussions are underway with a Eu based SME company having the capacity to reproduce such material on their industrial tool. The exact context (funding, support) has to be considered in parallel with options offered into the Horizon 2020 program. This point is very important and contributes to set up a sustainable European Supply chain. We note that this transfer will be based on the use of SiC substrate as support to the epitaxy. We expect to get access to a industrial source in 2016 / 2017.
The second is critical aspect is related to the optimization of the process development. Along this project, a continuous and intensive phase of experiment has improved a lot the electrical behavior of the device (PHEMT) by optimization of the building blocks : cleaning steps, ohmic contact, passsivation layers and associated properties, Schottky contact, annealing phase …. All these results are part of a foreground to be re-used inside a new phase dedicated to the industrialization and re-optimization if necessary. Such material (III-N semiconductor) being sensitive to the equipment (process), it is necessary to re-evaluate the recipes set up during the R&D phase. This activity is planned with one of the partner of this program UMS. The exact context (funding, support) has to be considered in parallel with options offered into the Horizon 2020 program. This second step would be initiated in 2015.
The AlInWon program has contributed to the development and evaluation of InAlGaN technology in preparation to the development of Ka band Satcom market. This one is foreseen as a key support around the world for transmission of high data rate. In consequence, by supporting a free access to strategic technologies, Europe will confirm his leadership in this new area.
List of Websites:
website address: http://www.alinwon-fp7.eu
Contact person: Didier Floriot / didier.floriot@ums-gaas.com