Search

This report describes 40 programmes for use with microcomputers. The progra111Ties have been written in the most 11portable11 Basic language possible so that they will be useable with a wide range of machines. The programmes are designed for the analysis of observational data especially that generated by Clinical Review and Epidemiological research. However, they may prove useful in other situations \'ttlere data from observational studies requires analysis. Alvan Feinstein, whose work has been quoted frequently in this report, has said that we should take advantage of the increasing availability of new ,,., technology to innovate and seek better methods for data analysis rather than ysing it merely to automate a defecti._ve status quo. In this report I have ·tried to heed his advice. Major innovations include the use of the Mantel score test for analysis of i'ndependent samples of interval data, and the provision of a group of programmes to assist in the study of two factor interactions in Clinical Review research. The Jackknife variance estimator for non-null variance calculation has been used in several situations in these programmes when a satisfactory variance estimator has not been available. In addition, several new point estimators in relation to interval data analysis and the stuqy of interactions are proposed which it is hoped will prove of more practical use than those which are currently available.

Thin films (5nm ± 2nm thick) of ZrO2, Al2O3, TiO2, Sm, Gd, Zr, Ni, and Pt were deposited onto TbFeCo films (100nm ± 20nm thick) on silicon and graphite substrates and analyzed with XPS as prospective candidates for TbFeCo diffusion barriers. Metals were chosen primarily according to electronegativity. Samples were typically heated to 272°C in UHV for 20 hours to enhance diffusion. Experiments with the metals were performed in a more consistent manner than with the oxides. The Sm, Gd, and Zr were reactively oxidized during the deposition. The Sm/Sm-oxide and Gd/Gd-oxide appeared to be favorable candidates for TbFeCo diffusion barriers. TbFeCo was not detected near the surface before or after heating the samples to 272°C for 20 hours and depth profiles indicated oxygen contamination decreased steadily as the barrier/TbFeCo interface was approached. For the other materials examined, either the oxides were reduced (at least partially) during heating to 272°C (381°C for Al₂O₃) or diffusion of TbFeCo was detected after heating, indicating that they would not be favorable candidates for TbFeCo diffusion barriers.

Thin films (5nm ± 2nm thick) of ZrO2, Al2O3, TiO2, Sm, Gd, Zr, Ni, and Pt were deposited onto TbFeCo films (100nm ± 20nm thick) on silicon and graphite substrates and analyzed with XPS as prospective candidates for TbFeCo diffusion barriers. Metals were chosen primarily according to electronegativity. Samples were typically heated to 272°C in UHV for 20 hours to enhance diffusion. Experiments with the metals were performed in a more consistent manner than with the oxides. The Sm, Gd, and Zr were reactively oxidized during the deposition. The Sm/Sm-oxide and Gd/Gd-oxide appeared to be favorable candidates for TbFeCo diffusion barriers. TbFeCo was not detected near the surface before or after heating the samples to 272°C for 20 hours and depth profiles indicated oxygen contamination decreased steadily as the barrier/TbFeCo interface was approached. For the other materials examined, either the oxides were reduced (at least partially) during heating to 272°C (381°C for Al₂O₃) or diffusion of TbFeCo was detected after heating, indicating that they would not be favorable candidates for TbFeCo diffusion barriers.