• PDF / 290,138 Bytes
• 5 Pages / 420.48 x 639 pts Page_size
• 8 Downloads / 222 Views

DOWNLOAD

REPORT

---

CALCULATION OF THE THERMOPOWER OF NI-P AMORPHOUS ALLOYS JAMES C. SWIHART*, HYUNGJIN YANG*, AND DONALD M.C. NICHOLSON** *Physics Department, Indiana University, Bloomington, IN 47405 *Oak Ridge National Laboratory, Oak Ridge, TN 37831 ABSTRACT We have modeled the Ni-P amorphous alloy system by a set of supercells on which we have carried out calculations of electronic and transport properties. The thermopower is determined by calculating the energy dependence of the dc electrical conductivity from the Kubo formalism. At 15% P we find -0.015 pV/K 2 for the thermopower divided by the temperature, while for 25% P the calculated value is +0.008 /V/K 2 , in reasonable agreement with experiment, particularly with regard to the change of sign with concentration. INTRODUCTION The Ni-P alloy system is of interest in part because it can be prepared as a met-glass over a rather wide range of concentrations, from less than 15 atomic percent phosphorus to more than 25 atomic percent phosphorus.' This system also has interesting transport properties as functions of concentration, with a fairly large change in the electrical 2 2 in sign in the temperature coefficient of the resistivity resistivity ,3. and with changes 3 thermopower. the in and We have developed supercell models for treating the electronic and transport properties of the Ni-P amorphous alloy system and have reported on results for the electronic density of states and for the frequency dependent and dc resistivities with 80 and 160 atoms in each supercell. 4 In order to calculate the thermopower, we found that the 160 atom supercell is not large enough. In the present work, we have concentrated on calculating the contribution to the thermopower from disorder scattering, and we have gone to 300 atom supercells. Our method of calculation consists of calculating the dc resistivity due to disorder scattering at various energies at and near the Fermi energy. The thermopower is then determined from the logarithmic derivative of the resistivity with respect to the energy. Because we need the energy derivative of the resistivity to obtain the thermopower, we can understand why the 160 atom supercell could be sufficiently large for calculating the resistivity but not large enough for the thermopower. METHOD OF CALCULATION In the first step of the calculation, we construct supercells containing 300 atoms in which we employ periodic boundary conditions. We have constructed four such cells at each of the concentrations of Ni 85 P 1 5 and Ni 7 5 P 25 . The Ni and P atoms axe placed in the cell at random and then the positions are relaxed to a local minimum of the potential energy using the phenomenological pair potentials of Weber and Stillinger. 5 In carrying this out we take pains to ensure that no phosphorus atom is near other phosphorus atoms as is found for the real systems. 1 With this procedure we obtain cells which have partial pair distribution functions that are in good agreement with experiment. 6 In calculating the electronic properties, we use the one-electron model