The observation of the spin Hall effect^1,2,3 triggered intense research on pure spin current transport⁴. With the spin Hall effect^1,2,5,6, the spin Seebeck effect ^7,8,9 and the spin Peltier effect ^10,11 already observed, our picture of pure spin current transport is almost complete. The only missing piece is the spin Nernst (–Ettingshausen) effect, which so far has been discussed only on theoretical grounds^12,13,14,15. Here, we report the observation of the spin Nernst effect. By applying a longitudinal temperature gradient, we generate a pure transverse spin current in a Pt thin film. For readout, we exploit the magnetization-orientation-dependent spin transfer to an adjacent yttrium iron garnet layer, converting the spin Nernst current in Pt into a controlled change of the longitudinal and transverse thermopower voltage. Our experiments show that the spin Nernst and the spin Hall effect in Pt are of comparable magnitude, but differ in sign, as corroborated by first-principles calculations.

对自旋霍尔效应^1,2,3的观察引发了对纯自旋电流输运^4的深入研究。在已经观察到自旋霍尔效应^1,2,5,6，自旋塞贝克效应^7,8,9和自旋珀尔帖效应^10,11的情况下，我们对纯自旋电流输运的了解几乎是完整的。唯一缺失的部分是自旋能斯特（–Ettingshausen）效应，到目前为止，仅在理论基础上对此进行了讨论^12,13,14,15。在这里，我们报告自旋能斯特效应的观察。通过施加纵向温度梯度，我们在Pt薄膜中产生了纯的横向自旋电流。对于读出，我们利用与磁化方向相关的自旋转移到相邻的钇铁石榴石层，将Pt中的自旋能斯特电流转换为纵向和横向热电势电压的受控变化。我们的实验表明，Pt的自旋能斯特效应和自旋霍尔效应具有可比较的幅度，但是在符号上有所不同，如第一性原理计算所证实的那样。