Per Mårtensson

Drift correction for gas sensors using multivariate methods

Drift is one of the most serious impairments afflicting gas sensors. It can be seen as a gradual change in the sensor response over a long period of time when the external conditions are constant. This paper presents a new simple drift counteraction method based on PCA and PLS. The basic idea is to remove the drift direction component from the measurements. The direction of the drift, p, is calculated from measurements for a reference gas. Projecting the sample gas measurements on this vector gives the score vector t. The drift component tp T can then be removed from the sample gas data, which we call component correction (CC). The method is tested on a data set based on a reduced factorial design with four gases and a concentration gradient of hydrogen. It is found that the method works efficiently for both cases. Copyright

In situ substrate preparation for high-quality SiC chemical vapour deposition

Abstract In situ preparation of 4H and 6H silicon carbide substrate surfaces in hydrogen and hydrogen-propane etching systems has been studied. The etching of on-axis(0001) 6H-SiC substrates resulted in regular straight terraces and one unit high steps. The etching of on-axis (0001) 4H-SiC substrates resulted in broad terraces interrupted by large step formations. The 4H- and 6H-SiC (0001) off-axis substrates (3.5° towards 〈112¯0〉 yield smooth etched surfaces with the exception of stripe-like defects on the 4H polytype which are shown to be related to stacking-faults. The stacking faults are suggested to be a cause for step-bunching and surface roughening. Hydrogen-etching prior to growth has been shown to improve the epitaxial layer quality both concerning defect formation and step-bunching.

Removal of polishing-induced damage from 6H-SiC(0001) substrates by hydrogen etching

We use atomic force microscopy and ultra-high vacuum scanning tunneling microscopy to show that polishing-induced damage on 6H-SiC(0001) on-axis wafers is efficiently removed by hydrogen etching in a hot-wall chemical vapor deposition reactor. The obtained surfaces exhibit a highly regular step structure with typically 1500 A wide terraces separated by steps with the height of a single unit cell (15 A). The corresponding low-energy electron diffraction pattern is threefold symmetric as expected for a surface with a single preferred domain. These results are compared with results obtained for as-polished and sublimation etched SiC(0001) wafers.

Selection of variables for interpreting multivariate gas sensor data

In this work, methods to select relevant variables from a large set of available gas sensor parameters were examined. Two data sets, containing a large number of variables, were studied. The objective was to find the best descriptors, which could predict interesting properties of the measurements. Using a forward selection procedure, applying the root mean square error from a multilinear regression model as the selection criterion, it was possible to get good prediction accuracy from a back-propagation neural network (ANN). This procedure was fast, compared to the usual trial and error variable selection, objective, and can be made fully automated. In addition, the use of principal component analysis (PCA) and partial least squares (PLS) score vectors used as descriptors were examined. The ANN models constructed with either the PCA or the PLS score vectors as input gave for the first, rather smooth, data set, errors of the same size or smaller than for the forward selected parameters. For the second data set, containing many noisy variables, the forward selected parameters outperformed the other two data reduction methods.

STM study of the SiC(0001) √3 × √3 surface

Abstract We have used scanning tunnelling microscopy to study the √3 × √3 reconstruction of the Si-terminated 6H-SiC(0001) surface. We find that the images are consistent with a structural model composed of 1 3 layer of Si or C adatoms in threefold-symmetric sites on top of the outermost SiC bilayer, similar to the reconstructions observed for 1 3 monolayer of e.g. group-III metals on the Si(111) surface. Additionally, we discuss the nature of the most frequently occurring defects on the √3 × √3 surfaces as well as the stacking sequence of the atomic layers below the reconstructed surface layer.

Current status of silicon carbide based high-temperature gas sensors

Silicon carbide (SiC) based field effect gas sensors can be operated at very high temperatures. Catalytic metal-insulator-silicon carbide (MISiC) Schottky diodes respond very fast to a change between a reducing and an oxidizing atmosphere, and cylinder-specific combustion engine monitoring has been demonstrated. The sensors have also been suggested for high-temperature electronic nose applications. Car applications and other harsh environments put very strong requirements on the long-term stability of the sensors. Here we review the current status of the field of SiC based Schottky diode gas sensors with emphasis on the work in our group. Basic work on understanding of the detection mechanism and the influence of interfacial layers on the long-term stability of the sensors is reviewed. The direction of future research and device development in our group is also discussed.

Morphology, Atomic and Electronic Structure of 6H‐SiC(0001) Surfaces

Recent findings concerning primarily the √3 x √3 and 6√3 x 6√3 reconstructed surfaces of 6H-SiC(0001) are reviewed. First, the morphology of some different types of 6H-SiC crystals is discussed. The scanning tunneling microscopy (STM) and atomic force microscopy (AFM) results presented show that surfaces with a morphology suitable for surface investigations can be prepared using sublimation- or hydrogen-etching. Then results obtained concerning the atomic and electronic structure for the reconstructed surfaces, prepared using an ex situ method for oxide removal and in situ heating, are presented. For the √3 x √3 reconstruction, recent STM and photoelectron spectroscopy (PES) data are discussed in view of available theoretical results. The STM images presented are shown to be consistent with a structural model of Si or C adatoms in threefold symmetric sites. The theoretical results favor Si adatoms in T4 sites as the optimal configuration for this reconstruction. However, the surface shifted components extracted in studies of the C 1s and Si 2p core levels and the location of a surface state band mapped out in angle resolved experiments cannot be explained using this structural model. At present, there is no structural model that satisfactorily can explain all experimental findings for the √3 x √3 reconstruction. A monocrystalline graphite overlayer on top of bulk-terminated or √3 x √3-reconstructed SiC has previously been proposed to explain the 6√3 x 6√3-reconstructed surface. However, STM and PES results are presented that unambiguously show that there is no graphite on the surface when a well developed 6√3 x 6√3 low-energy electron diffraction (LEED) pattern is observed. The STM images recorded during the gradual development of the 6√3 x 6√3 surface show growing fractions of pseudo-periodic 6 x 6 and 5 x 5 reconstructions. These reconstructed regions dominate on the surface, but small √3 x √3-reconstructed regions are still present when a well developed 6√3 x 6√3 LEED pattern is observed. It is shown that the 6 x 6,5 x 5 LEED pattern can be fully explained by scattering from surfaces with a mixture of 6 x 6, 5 x 5 and √3 x √3 reconstructions. Due to the complexity of the STM data, no structural model is proposed for the 6 x 6 and 5 x 5 reconstructions. STM and PES results are presented showing that graphitization of the surface is obtained only after heating at higher temperatures than that required for observing a well developed 6√3 x 6√3 LEED pattern. The STM images then show that the graphite appears as a monocrystalline overlayer on top of the 6 x 6 reconstruction and not on bulk-terminated or √3 x √3-reconstructed SiC(0001).

Influence of carbon monoxide, water and oxygen on high temperature catalytic metal-oxide-silicon carbide structures

High temperature sensors, Schottky diodes and capacitors, based on catalytic metal-oxide-silicon carbide devices are investigated. Reducing gases like hydrogen and other hydrogen containing gases, ...

Enhanced selectivity of MOSFET gas sensors by systematical analysis of transient parameters

Abstract Transient response curves from Pt-MOSFET gas sensors exposed to mixtures of hydrogen and ethanol were studied. Parameters of two types were extracted from the response curves: simple parameters like pulse heights, derivatives and integrals, and coefficients from different curve fitting algorithms. All parameters were evaluated with respect to their signal to standard deviation ratio, the correlations in a principal component analysis, and the parameters ability to predict the concentration levels using an artificial neural network. Combining these three evaluation methods it was possible to select transient parameters that with good accuracy could simultaneously predict the levels of hydrogen and ethanol in the test gas using information from only one sensor.

The SiC(0001)6√3 × 6√3 reconstruction studied with STM and LEED

Abstract We have used scanning tunneling microscopy (STM) to study the 6√3 × 6√3 reconstruction obtained by heat treatment of 6H-Sic(0001) samples at temperatures above 1100°C. For surfaces showing a well-developed 6√3 × 6√3 low-energy electron diffraction (LEED) pattern, we observe with STM two pseudo-periodic reconstructions with approximate periodicities of 6 × 6 and 5 × 5, respectively, in addition to √3 × √3-reconstructed regions. The fraction of the surface exhibiting the √3 × √3 reconstruction in STM images and the intensity of the √3 × √3 spots in the 6√3 × 6√3 LEED pattern decrease with increasing annealing temperature and time. The 5 × 5 reconstruction is observed on a small fraction of the surface (⪅ 10%) and the 6 × 6 reconstruction becomes dominating upon annealing at temperatures above 1200°C. For the 6 × 6 reconstruction, a Fourier analysis of the STM images reveals an underlying incommensurate 2.1 × 2.1-R30° lattice with long-range order. The features defining the 6 × 6 periodicity have well-defined positions with respect to this lattice. A comparison between the Fourier transforms of the STM images and the 6√3 × 6√3 LEED pattern shows that the LEED pattern can be fully explained by scattering from surfaces with a mixture of the √3 × √3, 5 × 5 and 6 × 6 reconstructions. For surfaces heated above 1250°C, we observe a partial graphitization which results in a modification of the 6 × 6 structure observed in STM.