Schottky contact based diamond power devicestuning the interfacial properties to demonstrate diamond unleashed performance

  1. Cañas Fernández, Jesus
Supervised by:
  1. Marina Gutiérrez Peinado Director
  2. Etienne Gheeraert Co-director

Defence university: Universidad de Cádiz

Fecha de defensa: 28 January 2022

  1. Denis Dauvergne Chair
  2. Robert Nemanich Secretary
  3. Philippe Godignon Secretary
  4. Mariko Suzuki Committee member
  5. Eva Monroy Committee member
  1. Ciencia de los Materiales e Ingeniería Metalúrgica y Química Inorgánica

Type: Thesis

Teseo: 707193 DIALNET


The power electronic market is dominated by silicon devices, which are responsible for most of the energy losses over the whole energy chain. The performance of these devices is limited by the material properties of silicon. The on-state resistance, related to the material resistivity, dissipate heat due to Joule effect during current conduction. The maximum voltage blocked by the devices is related to the maximum electric field sustained by the material. Indeed, these two material properties are connected, resulting in a core trade-off for power electronics: on vs off state performance. Also, switching losses coming from the fact that a device requires time to switch from on to off state (and viceversa) are always present. With the current silicon technology reaching its theoretical limits, the only road to improve the performance is to use new materials with better properties, and thus, better trade-off between on and off states.Wide band gap materials have been under research for the last decades due to their immense potential to block a very high voltage while delivering a very high current. Between them, diamond is the ultimate semiconductor for power electronics due to its outstanding properties. Besides being the hardest material, diamond possesses the best properties among ultra-wide band gap semiconductors including its high electron-hole mobility, high critical electric field and low dielectric constant. Moreover, it surpasses all competitor materials in terms of thermal conductivity with the highest value reported for any material. All these properties make the ideal performance of diamond devices to be above any other material. However, diamond devices are still under development with several technological limitations and bottlenecks that need to be addressed in order to extract diamond’s full potential. The main drawback is that both p and n type dopants give rise to relatively deep levels causing partial ionization at room temperature. However, the partial ionization can be easily mitigated for p-type diamond by working at high temperatures (550K) leading to outstanding on-state device performances. Also, substrates are expensive and limited in size, although an encouraging progression have been shown in the last decade. Therefore, the future market of diamond devices seems to be focused to ultra-high power, high frequency and temperature and harsh environment applications.Concerning diamond field-effect transistors (FET), the majority have been fabricated using 2D hole gas that emerges when an H-terminated diamond surface comes into contact with a range of adsorbates and/or selected metal oxides. However, this unconventional manner of creating a p-type region in a semiconductor suffers from instability and reproducibility problems and low operational mobility issues, although recent progress with the use of metal oxide passivation/gate structures have considerably reduced these problems. Nonetheless, device fabrication strategies that rely upon a conventional doping approach have also been explored. Based on p-type boron doped diamond epilayers, normally-on metal semiconductor FET (MESFET) and metal oxide semiconductors FET have been demonstrated with blocking voltages up to 3 kV. These ensure optimized compatibility for ultimate device integration with existing technology along, with the full realization of the potential for operation in harsh environments. Also, they provide the best performance at elevated temperatures (550K) due to the ionization of boron impurities that can be exploited in power electronics thanks to the self-heating of devices.On the other hand, diamond Schottky diodes have been fabricated since the 90’s and are the most developed component nowadays. Most Schottky contacts are also fabricated based on boron-doped diamond using a pseudo-vertical architecture. Schottky diodes are very promising for its fast-switching capabilities and high voltages. In fact, optimized diamond Schottky diodes can ideally sustain up to 30 KV. Extensive research has led to impressive results reported over the las decade. Different surface treatments and metals (W, WC, Zr, Pt, Mo,etc) have been explored with the purpose of reducing leakage currents and thus obtaining the best possible on vs off trade-off. Diamond Schottky diodes able to sustain up to 10 kV and a record critical electric field of 7.7 MV/cm have been reported. On the other hand, heteroepitaxial Schottky diodes have also been fabricated showing promising results, as diminishing the cost of devices is fundamental for their commercialization. However, technological limitations in etching and epitaxy limits the ultimate performance of diamond Schottky diodes. Also, irreproducibility among samples and fabrication methods is one of the main constraints that has to be faced in the coming years.This thesis is dedicated to the development of Schottky barrier diodes and Schottky based transistors (MESFET) based on p-type diamond. The optimization of these components is crucial for the diamond future roadmap for power electronics, as a true demonstration of diamond potential still lacks, and it will definitely attract interest from research and companies into it further pushing the development of devices and eventually ending up in their commercialization. In order to get demonstrate diamond potential, various challenging objectives were set regarding Schottky contact based devices: i) The study and optimization of metal-diamond properties from an electronic, physico-chemical and material science point of view in order to extract the best possible performing Schottky contact. ii) The design, fabrication and characterization of an optimized high voltage diamond vertical Schottky diode. iii) The design, fabrication and characterization of a normally-off 1 kV diamond gate-drain Schottky MESFET. The development of such studies and devices is carried on within the institute NÉEL and the University of Cadiz under the frame of two different projects: The first one, the diamond-HVDC project. Its goal is the fabrication of HV diamond Schottky diodes for its integration inside a commutation cell and test its reliability. The second one is the Green diamond European project. My purpose inside this project was to coordinate and participate on the fabrication and characterize a normally-off 1 kV diamond gate-drain Schottky lateral MESFET. The ultimate goal for these projects is to fabricate ideal performing devices, to demonstrate diamond ultimate capabilities, and their integration into functional power electronic components. The diamond epitaxy, power electronic knowledge and capabilities from the SC2G group at NEEL institute in France and the powerful material science analysis techniques and the expertise from the UCA are put together during my PHD to address these challenges.This works stats by setting the theoretical framework underlying Schottky contact-based devices. Diamond Schottky contacts modelization is presented based on an analytical description of Schottky barriers and the equations governing the forward and reverse conduction mechanisms through the Schottky barrier are introduced and solved numerically. The state of the art diamond resistivity and breakdown models are as well presented and discussed. These physical models are then used for the quantitative comprehension of diamond diodes firstly and diamond lateral MESFET’s secondly. On one hand, the different diode structures and their fabrication constraints and technologic limitations are introduced. The impact of these constraints is illustrated as well as the optimal performance of diamond diodes. On the other hand, the RBMESFET concept is presented and a 1D model for the on-state current under the gradual channel approximation is developed. Its performance is analyzed relying on the developed physical models.Once the framework is presented, a dedicated experimental investigation of lateral molybdenum and zirconium Schottky diodes follows. The TEM and XPS extracted compositional profile is correlated to zirconium and molybdenum Schottky contact electrical performance. The ozone passivation treatment, demonstrated to be crucial in reducing leakage currents, is shown to oxidize the molybdenum and zirconium layers. While this did not greatly affect the molybdenum on-state characteristics, the degradation of the zirconium contacts is explained as a result of the high concentration of oxide in the interface. The potential profiles along the contacts edges, extracted from Kelvin probe force microscopy, showed the great passivation capabilities of the oxygen terminated diamond and its possible role in the leakage currents through the edge of the contacts in zirconium. On the other hand, molybdenum built-in voltage is shown to be lower than the oxygen-terminated band bending potential. This plays an important role to avoid leakage currents through the edge of the contacts and support the impressive off-state characteristics extracted. Lastly, a theoretical approach to understand both oxygen-terminated surface and its interface with molybdenum was taken. The results showed qualitative agreement with the experimental results yielding a higher surface band bending potential than molybdenum built-in voltage while revealing the structure of molybdenum and oxygen-terminated diamond.Concerning vertical diodes, two different Schottky diode samples were fabricated, one in order to evaluate molybdenum and zirconium Schottky diodes and another one for developing a kV range diode based on the findings. The first sample demonstrated the success of the fabrication process used, based on molybdenum, with the crucial importance of the surface passivation treatment for the high blocking voltage capabilities. The superiority of this molybdenum versus zirconium as Schottky contact was demonstrated as the former had higher Schottky barrier, lower ideality factor, better thermal stability and better blocking voltage capabilities. Over half of the contacts presented in the first sample were able to sustain 2.25 MV/cm with at least 4 magnitude orders of rectification. Its application in low doped diamond epilayers could consistently sustain as much as a 60% of low doped diamond breakdown field.The second sample implemented the developed fabrication process of molybdenum based Schottky contact into an HV design. It successfully demonstrated >1kV breakdown voltage capabilities with contacts sustaining (a low estimate) electric field of 2.2 MV/cm. This sample drift layer epitaxy produced numerous hillocks and round defects which makes even more noteworthy the contacts performance. The role of these defects was studied and only the hillocks contributed to the leakage currents through the contacts. Although the sample irregularity and high concentration of defects, the irreproducibility in the breakdown voltages was attributed to the electric field crowding on the edge of the contacts and this hypothesis was supported by EBIC measurements.Therefore, the molybdenum Schottky contacts developed in this work demonstrated consistently suitable performances for extracting diamond HV potential at high temperature. The leakage current problematic was addressed and by following the fabrication process developed in this thesis, future diamond Schottky diodes will be able to unleash its full potential. In chapter 5, a >1kV normally-off lateral reverse-blocking (Schottky Drain) MESFETs based on molybdenum Schottky contacts on O-terminated boron doped diamond have been fabricated. Precise control over epilayer thickness and doping level have been achieved such that full depletion of the channel arises without an applied gate bias with the possibility to open the channel at negative VGS. The devices display undetectable leakage current and a current of ~1.5 µA/mm in the on state at RT. Importantly, the transistor´s blocking capabilities were evaluated to be >1kV even at its maximum working temperature of 425K displaying no leakage current and after the HV stress, presenting non-changed reproducible characteristics. Higher temperature characterization (425K) reveals augmented transistor on-state capabilities due to fuller ionization of the boron dopant atoms, with the transistor reaching a current level of 70 µA/mm and a transconductance of 147 µS/mm. However, its working temperature is limited to 425K, below the optimal 500-550K, due to the non-grounded p-n junction between the p-type drift layer and the semi-insulating Ib substrate. Anyways, the high breakdown voltage, the relatively high temperature operation capabilities, normally-off behavior and diamond’s inherent radiation hardness as a semiconductor, make these devices an excellent approach towards the goal of achieving diamond transistors for high temperature, high voltage and harsh environment applications.As summary, the progress made during this thesis in the optimization of Schottky contacts allowed to push the off-state and reproducibility of vertical diamond Schottky diodes. Future diodes will benefit from the results shown, as the application of the Schottky contacts developed in this thesis together with thicker diamond drift layers could achieve the benchmark of sustaining 10 kV. On the other hand, the demonstration of a >1kV normally-off lateral reverse-blocking (Schottky Drain) MESFETs was the first of its kind. The unusual characteristics of this transistor show the precision acquired in diamond epitaxy technology. A future transition toward vertical transistors using the developed technology will unleash diamond MESFET potential performance.