Fabricating ultra-sharp tungsten STM tips with high yield: double-electrolyte etching method and machine learning

Abstract

The double-electrolyte etching method is a simple and effective way to fabricate ultra-sharp scanning tunnel microscopy (STM) tungsten tips. However, it is still challenging for this method to get ultra-sharp tungsten tips with high yield. In this work, significant enhancing of the yield is presented through optimized etching parameters as follows: temperature of 26 °C, applied voltage 7.5 V, electrolyte concentration 4 mol L−1 and length below the liquid lamellae of 2 mm. Under these conditions, the smallest tip radius is around 8 nm and the yield (radius < 10 nm) is 63.5%. These tips are capable of producing high-quality atomic resolution STM images, as demonstrated by testing on Si (111) and highly oriented pyrolytic graphite (HOPG) samples at room temperature. Furthermore, in order to find the relationship between the tip features and experimental etching parameters, an artificial neural network (ANN) model is built by machine learning. Garson’s algorithm is used to analyze the relative importance of each experimental parameter on tip features. The tip features can be estimated by this model with a correlation factor R over 0.85 indicating great predictive performance. Importance analysis indicates that the length of the tungsten wire below the liquid lamellae is the most important parameter to obtain high-quality STM tungsten tips in the double-electrolyte etching method. This result provides a clear direction for rapidly selecting optimized tip fabrication parameters in the future.