RESUMO
The determination of the longitudinal spin Seebeck effect (LSSE) coefficient is currently plagued by a large uncertainty due to the poor reproducibility of the experimental conditions used in its measurement. In this work we present a detailed analysis of two different methods used for the determination of the LSSE coefficient. We have performed LSSE experiments in different laboratories, by using different setups and employing both the temperature difference method and the heat flux method. We found that the lack of reproducibility can be mainly attributed to the thermal contact resistance between the sample and the thermal baths which generate the temperature gradient. Due to the variation of the thermal resistance, we found that the scaling of the LSSE voltage to the heat flux through the sample rather than to the temperature difference across the sample greatly reduces the uncertainty. The characteristics of a single YIG/Pt LSSE device obtained with two different setups was (1.143 ± 0.007) 10-7 Vm/W and (1.101 ± 0.015) 10-7 Vm/W with the heat flux method and (2.313 ± 0.017) 10-7 V/K and (4.956 ± 0.005) 10-7 V/K with the temperature difference method. This shows that systematic errors can be considerably reduced with the heat flux method.