Characterisation and Reliability Testing of THz Schottky Diodes

1. Characterisation and Reliability Testing of THz Schottky Diodes Chris PriceUniversity of Birmingham, UKhttp://www.sr.bham.ac.uk/yr4pasr/project06/CP/schottkydiodes/mainpage.html Physics Background • Characteristics of a Schottky diode can be derived from the energy band diagram of the system • If a material is in equilibrium, its energy levels are constant, Fig 3 • When the two materials are brought into contact and remain in equilibrium, the energy band diagram changes subject to 3 laws, Fig 4: • Fermi level,Ef, must remain constant throughout the system • The electron affinity, χ, must remain constant • The free space energy level, E0, must remain continuous • The motivation for this project is to develop Schottky diodes which are sensitive to the requirements of Europe: To do this the fabrication process needs to be optimised and the reliability of the devices tested • A Schottky diode is formed by the intimate contact between a metal and a doped semiconductor • Schottky diodes produced at RAL are planar devices, shown in Fig 1. The Schottky junction is made from titanium and n type gallium – arsenide • Schottky diodes are specifically used for their non-linear current voltage relationship, shown in Fig 2, as frequency mixers and multipliers Fig 3: Energy band diagram of a metal and semiconductor in equilibrium but not in contact Fig 4: Energy band diagram of metal and semiconductor in equilibrium and in contact • As the Fermi levels are different in the two materials the electrons with the highest level spontaneously move to the lowest level. This creates a depletion zone in one material as all of its free electrons move • These effects create a potential barrier at the junction the height of which is the difference between the two work functions • This barrier can be lowered or raised by applying a forward or reverse bias voltage, Fig 5a & 5b Fig 5a: Energy band diagram when a forward bias is applied Fig 1: Schematic of a planar Schottky Diode Fig 2: A graph showing the non-linear properties of a Schottky diode Fig 5b: Energy band diagram when a reverse bias is applied Derivation of the I-V relation • Electron conduction across the junction occurs primarily by thermionic emission • Charge density is given by the Maxwell-Boltzmann distribution: Evaluation of Schottky diodes WHY - As with any product manufactured, the end product is never as good as the ideal design. Therefore methods of measuring the quality of each device are used to ascertain how good or how close to the ideal product each end product is where Current voltage characteristics IDEALITY - In real life, diodes do not behave in a completely ideal manner, so a parameter needs to be added to compensate for their non-ideal behaviour The ideality factor is a number around 1.0, typically between 1.1 and 1.2 for real diodes. As the ideality factor increases, the non-linearity behaviour of the diode decreases changing the gradient of its I – V plot, Fig 2; the effect of this is to reduce all aspects of a Schottky diode’s performance whether it is being used for harmonic generation or RF mixing. • Current across the junction is directly proportional to the change in charge density from zero bias to forward bias conditions • It thereforefollows: where Where n1is the charge density under zero bias, n2 is the charge density under forward bias, nd is the doping density, kB is Boltzmann constant, E is energy, T is the absolute temperature, q is the charge on an electron, Φbi is the built in potential, Φm & Φs is the work function of the metal and semiconductor respectively, V is the voltage applied, I0 is the reverse saturation current, A** is the Richardson constant and W is the junction area Ideality equation Modified I-V equation SERIES RESISTANCE - The series resistance within a Schottky diode structure is mainly formed from highly doped semiconductor regions at semiconductor–metal interfaces. This affects the diodes’ performance when the voltage across this region is comparable to the voltage across the junction and also varies as a function of current Fig 6 Development Progress WHY – optimise the fabrication process to produce the best quality devices; as the smaller the anode size, the more sensitive the diodes are to the fabrication process HOW – Measuring and comparing the ideality and series resistance properties of each diode batch Fig 6: A graph to show the series resistance as a function of measurement current Series resistance equation HOW - The calculations of the I – V characteristics are done by applying a current over several decades and measuring the voltage across the diode Fig 9: Ideality against series resistance scatter plot of diodes produced at RAL Fig 7: Simplified circuit diagram of how the voltage is measured Fig 8: Table of the currents used to evaluate the DUT Fig 10: Average ideality against series resistance plot for same fabrication process Reliability testing • Reliability can be defined simply as the quality over time • Repeatability • Repeatedly measuring the diode ideality and series resistance may cause them to change due to a burn-in effect • Tested by contacting a diode once and repeatedly measuring its properties • The result of this experiment can be seen in Fig 15 • Repeatedly contacting the same pad may lead to inconsistent connections • Tested by repeatedly contacting the same diode • The result of this experiment can be seen in Fig 16 • The conclusion was that these effects are negligible and would not bias the results Fig 9 shows a scatter plot comparing ideality and series resistance of two generations of diodes Fig 10 shows how 2 batches of diodes compare by showing the average value and standard deviation RESULTS – bunching of data points shows uniformity in the fabrication process and the motion towards the bottom left corner shows the improvement in quality of the devices Fig 15: A graph showing how ideality and series resistance vary with multiple measurements Surface mapping As the diodes are produced in an array, Fig 11, a measurement of each one can be taken and plotted as a map, Fig 12. Fig 11: Picture of a wafer produced at RAL Fig 12: Surface map of how ideality varies across a wafer (10 x 10 array) RESULTS – Ideality across a wafer is fairly uniform with no obvious areas of an inconsistent fabrication process USES – Potential to test several fabrication condition across a wafer, saving time and resources and see visually how the diodes’ properties change • Short term thermal reliability • The most common method for integrating these devices into circuits/waveguides is to use solder, however, little is known about how this heating changes the diodes’ properties • Several diodes with similar idealities were heated for 30 seconds, cooled and had their characteristics measured; this was done over a temperature range between 80-200°C and 90-190°C in 20°C steps • The average ideality for the two sets of experiments can be seen in Fig 17 • It can be seen that there is a threshold temperature at around 130°C after which the diodes begin to degrade • The most probable reason for this degradation is material inter-diffusion at the anode Fig 16: A graph showing how ideality and series resistance vary with multiple contacts Native oxide growth Oxide growth on the anode is a major cause of degradation in a diodes’ performance. It is formed simply by exposing the cleaned anode to air To measure the effect of short term oxide growth a wafer was split into 3 sections, Fig 13: left 0 mins, centre 5 mins, right 15mins RESULTS – The average ideality increased from 1.171 (left) to 1.189 (centre) to 1.205 (right). Fig 13: A diagram showing the partitioning technique Left Centre Right Fig 17: A graph showing how ideality changes with increasing temperature Fig 14: Surface map showing the increased ideality as exposure to air increases