Quantitative Temperature Measurement of Self-Heating in Non-Linear Devices Using Scanning Thermal Microscopy

Abstract

Self-heating in electronic devices relates to rich and unexplored transport physics of heat and charge. Phenomena range from hot spots causing device failure [1], to exploiting self-heating as a new platform for computing by utilizing electro-thermal processes [2]. Imaging and understanding thermal transport and thermal processes at the nanoscale enables thermal management and new ways of computing, making it highly relevant in today’s integrated microelectronic devices. Scanning Thermal Microscopy (SThM) has become a valuable method to investigate device states and failure mechanisms due to its high spacial and temperature resolution. However, quantitative SThM thermometry is so far mostly limited to devices with linear current-voltage characteristics [3]. We have extended the method to investigate non-linear devices, broadening the application range to include many of the functional device types used in today’s microelectronics. For instance, devices for logic or sensing involve complex interfaces that may lead to energy barriers and energy filtering. In this method, the non-linear device under study is driven to an elevated temperature with a periodic and constant part using an applied oscillating and offset voltage. The applied amplitude allows the simultaneous estimation of temperature and thermal resistance, while the applied offset is used to set the device state. These lead to an oscillating and constant temperature rise in the heated sensing element, which are detected through resistance changes via a Wheatstone-bridge and lock-in amplifier. The constant temperature rise in the device can then be inferred using $ ΔT_{DC} = T_{sensor,0}\space^{ *} ΔV_{AC,nω} / (ΔV_{AC,nω} - β_{n}\space^{ *} ΔV_{DC}) $(1) Where $ T_{sensor,0} $ is the heater temperature out of contact with the device, ΔV are the heater’s constant and oscillating voltage rise measured over the Wheatstone-bridge and n refers to the harmonic under investigation. The newly introduced non-linearity factor βn accounts for the changing device resistance and is estimated numerically from its I-V characteristics. Our new technique was verified mapping the temperature of exemplary linear (ohmic) and non-linear (memristive and logical) devices. In conclusion, the method developed here can be used to investigate non-linear devices and materials, including volatile and non-volatile phase-change materials. The combination of an AC voltage with a DC bias allows the separate investigation of different voltage-dependent device states. The method will therefore be crucial to understand dynamics in operating devices and to shine light on reliability issues and break-down mechanisms. REFERENCES 1. E. Pop, Nano Research 3, 147–169 (2010) 2. E. Corti et al., Solid-State Electronics 168, 107729 (2020) 3. F. Menges et al., Rev. Sci. Instrum. 87, 074902 (2016)