Artificial spiking neurons are one of the building blocks of neuromorphic systems. Such type of hardware is often promised to offer energy-efficient platforms to run Deep Learning algorithms on large datasets. These systems require large numbers of neuronal primitives, raising the question of the environmental footprint of their fabrication. In this work, we propose to address this matter by performing a cradle-to-gate Life Cycle Assessment (LCA) of an artificial neuron based on Vanadium Dioxide, fabricated in a lab-scale fab. We focus on the Cumulated Energy Demand (CED), Global Warming Potential (GWP) and Ultra Pure Water consumption (UPW) metrics. We identify the use of gold for the electrical contacts (the dominant practice in the fabrication of such devices) as the critical step, representing respectively 38 % and 49 % of the total CED and GWP associated to the device fabrication. We further explore the validity of this result by varying key parameters in different scenarios. In an eco-design approach, we propose the replacement of gold by aluminium as a way to curb the total process CED and GWP by respectively 25 % and 43 %, and show first evidence that the device electrical properties are preserved in this case. Finally, we provide a first attempt at comparing the direct environmental footprint and performance of VO2 neuron with CMOS-based implementations.