Simon S.M. Chan

Polycrystalline diamond photoconductive device with high UV‐visible discrimination

Planar metal–diamond–metal photoconductive devices have been fabricated from free standing large grain (20–30 μm) polycrystalline thin film diamond. An interdigitated electrode design with spacings of 20 μm was used to produce effective UV photodetecting devices at bias values in the range 0.1–10 V. A methane‐air treatment has been used to modify the structures such that unprecedented performance characteristics have been recorded (106 higher response to 200 nm than visible wavelengths, <0.1 nA dark currents); spectral features similar to those observed in natural diamond crystals have been observed indicating that the treatment used led to near ideal electronic characteristics from polycrystalline material.

Photoconductive properties of thin film diamond

Abstract The origin of photoconductivity within diamond is briefly reviewed. Several applications for thin film diamond formed by chemical vapour deposition (CVD) could be realised if the extrinsic photoconductive response could be controlled, including the fabrication of deep UV photodetectors which are “blind” to visible light. It is possible to reduce the extrinsic photoconductive response otherwise apparent in CVD films by a combination of careful device design and well chosen post fabrication gaseous treatments; the nature of the treatments used and how they may be affecting defects within the diamond are discussed. Highly sensitive deep UV photodetectors can be produced in this manner, exhibiting external quantum efficiencies of greater than 500%.

Cleaning thin‐film diamond surfaces for device fabrication: An Auger electron spectroscopic study

Auger electron spectroscopy was used to analyze polycrystalline thin‐film diamond surfaces following the use of differing methods for the removal of unwanted nondiamond carbon. Exposing the film to a hydrogen plasma at the termination of the growth process is effective for producing a surface that gives an Auger spectrum typical of diamond with little contamination. Strongly oxidizing solutions involving sulfuric acid generate low concentrations of surface sulfur together with an oxide phase. However, in the case of an ammonium persulfate–sulfuric acid etchant solution, the Auger features associated with the diamond more closely resemble those of single crystal material suggesting that this treatment may offer better performance when used during the fabrication of thin‐film diamond electronic devices.

Thin film diamond photodiode for ultraviolet light detection

A photodiode has been constructed from lightly boron doped, Si supported, thin film chemically vapor deposited (CVD) diamond which shows over five orders of magnitude discrimination between deep UV (≤220 nm) and visible light. A thin (10 nm) gold Schottky barrier with an associated Ti–Ag–Au ohmic contact was used to create a rectifying device with low (≤2 pA) dark currents when reversed biased. This structure showed a sharp cut off in photoresponse at 220 nm, the band gap energy of diamond. Conversely, a photoconductive device fabricated from similar (nominally undoped) material gave a broader UV photoresponse and displayed high dark currents; the superior performance of the diode structure on fine grain material is discussed.

Thin film diamond UV photodetectors: Photodiodes compared with photoconductive devices for highly selective wavelength response

Abstract Silicon-supported and free-standing thin films of diamond have been used to fabricate photoconductive and photodiode structures for the detection of UV light. On free-standing (80 μm thick) material, a planar interdigitated design with 20-μm electrode spacings is found to offer unprecedented wavelength discrimination between deep UV and visible light, with dark currents

A thin film diamond p-channel field-effect transistor

A depletion-mode metal-insulator-semiconductor field-effect transistor has been fabricated from thin film polycrystalline diamond with a p-type (boron doped) channel and an insulating diamond gate. This device has been successfully operated at 300 °C displaying pinch off when in depletion and high levels of channel current modulation in enhancement. A transconductance value of 174 μS/mm has been measured, the highest reported value to date for this type of device.

High temperature polycrystalline diamond metal-insulator-semiconductor field-effect-transistor

Abstract High temperature p-type depletion-mode metal-insulator-semiconductor field-effect-transistors with intrinsic diamond gates have been fabricated from thin film polycrystalline diamond. They have been successfully operated at 300°C with low leakage currents, and exhibit complete channel current pinch-off and modulation. A transconductance of 174 μS mm −1 has been measured, the highest reported value to date for this type of device.

The effect of excimer laser etching on thin film diamond

Excimer laser projection patterning with an ArF (193 nm) source has been employed in the irradiation of thin diamond films. The effect of a number of process parameters including laser fluence and processing ambient on the quality of the etch product has been investigated; scanning electron microscopy shows that good control of etch quality may be achieved with excellent lateral reproduction of images down to 2 μm. Raman scattering and Auger electron spectroscopy of irradiated films have been correlated, and modifications in the diamond surface have been quantified according to processing parameters. Electrical tests on laser modified surfaces show that the reactivities of metals have a major role in the performance of contact metallizations on such a material. The viability of excimer laser etching of diamond as a manufacturing technique is considered.

Laser projection patterning for the formation of thin film diamond microstructures

Abstract Projection patterned excimer laser radiation at 193 nm has been used to etch free-standing polycrystalline diamond films. Excellent lateral reproduction of images down to 2 μm has been achieved, but features evolve with a side wall that slopes at around 25 ° to the normal; this limits the depth of each feature in relation to its lateral dimension. However, three-dimensional microgears, with a diameter of 930 μm, have been successfully fabricated using this technique. Patterning in air gives rise to etch rates as high as 31 nm per laser pulse, but in an evacuated environment this is reduced to around 0.6 nm; furthermore, Raman spectra indicate that patterning in vacuo leaves strongly modified surfaces, whilst air processing causes little degradation to the diamond film.

Continuous hyperparameter optimization for large-scale recommender systems

While the prediction accuracy of a large-scale recommender system can generally be improved by learning from more and more training data over time, it is unclear how well a fixed predictive model can handle the changing business dynamics in a real-world scenario. The adjustment of a predictive model is controlled by the hyperparameter settings of a selected algorithm. Although the problem of hyperparameter optimization has been studied for decades in various disciplines, the adaptiveness of the initially selected model is not as well understood. This paper presents an approach to continuously re-select hyperparameter settings of the algorithm in a large-scale retail recommender system. In particular, an automatic hyperparameter optimization technique is applied on collaborative filtering algorithms in order to improve prediction accuracy. Experiments have been conducted on a large-scale real retail dataset to challenge traditional assumption that a one-off initial hyperparameter optimization is sufficient. The proposed approach has been compared with a baseline approach and a widely used approach with two scalable collaborative filtering algorithms. The evaluations of our experiments are based on a 2-year real purchase transaction dataset of a large retail chain business, both its online e-commerce site and its offline retail stores. It is demonstrated that continuous hyperparameter optimization can effectively improve the prediction accuracy of a recommender system. This paper presents a new direction in improving the prediction performance of a large-scale recommender system.