Over the last decade the gas sensor field has produced thousands of publications, each typically describing a specific synthesis method of one material or variations of a doped material or heterostructure and reporting the gas sensor properties of each variation over a range of temperatures, gas analytes, and analyte concentrations. Occasionally a review paper will tabulate as much of the relevant data as feasible to aid comparisons by future researchers to established data. However, the high dimensionality of the data (compositions, temperatures, analytes, concentrations, selectivity) restricts published tables to only report the optimum conditions for each paper with abbreviated notes on nanostructure morphology, experimental setup and selectivity. Much like several established databases for phase diagrams, x-ray diffraction and crystallography, we have here established an open-access database to aid the gas sensor researcher in finding relationships in properties and performance across many studies, identifying the most notable deficiencies in need of study, and to aid the commercial manufacturer in selecting the most promising systems for specific applications.

A small segment of the sensor field often carries out carefully crafted experiments that give us new insights on the theory and mechanisms that explain some empirical results. However, the majority of the sensor field focuses on generating empirical data from new materials. It has become increasingly difficult if not impossible to keep track of the data being generated and make meaningful comparisons between studies from different institutions. High variability is often seen in results from seemingly similar tests from different laboratories. Comprehensive yet time-consuming data-mining studies of the literature in other fields have often yielded valuable results that could not otherwise have been achieved by empirical results from one laboratory alone.

The recent technological advancements in data storage, networking, and computer processing power have lead many fields toward so-called “big data” analysis to identify trends and correlations across data points from many different sources. The data generated by the gas sensor field is on a smaller scale than that to which “big data” generally refers, but nevertheless the same principles can be applied for the benefit of all. The first step toward utilizing these principles is assembling the data from the field into a single searchable database. By enabling the discovery or verification of correlations between the structures and resultant properties, this database can help the field toward a “materials-by-design” approach. These correlations should further be verified by computational modeling from first-principles.

Driven by similar needs, the U.S. government launched the Materials Genome Initiative (MGI), to foster the integration of empirical data with computational tools and faster identification of the most promising technologies. Assimilation of data in this field will also help researchers focus on what has not been done, rather than repeating experiments that they may otherwise have never seen in the literature.

This database can be useful in a variety of ways. Below are some examples of questions that the database is designed to help answer:

“What is the best response to 1000 ppm ethanol achieved using plain SnO2 nanowires?”

“What material(s) typically have the best response toward H2S?”

“At what temperature(s) does ZnO have the fastest response time toward H2?”

“What work has been published with NiO doping of SnO2?”

“What morphologies of nanomaterials have had the most publications since 2010?”

“What synthesis technique(s) is/are typically used to create hierarchical nanostructures?”

“Which additive material to SnO2 gives the best selectivity toward ethanol?”

“Do p-type or n-type materials have better response toward xylene?”

Many of these questions do not have clear-cut answers and some researchers may disagree on the answer compared to another group at a different institution. The purpose of the database is then to collect as much of the data as possible and present the information to the user in a scatter plot so that the most common result can be seen. It is up to the user to make the decision that most studies show a given result, but that the user may want to read the specific papers with contradictory results to see if they might be explained by different processing or testing methods.