The paper in Nature Communications is here: https://go.nature.com/2HvnRGD

Illicit drugs, especially the synthetic ones, have posed serious threat to human health, family harmony, and social stability. As far as drug abuse is concerned, more than 29.5 million people are drug users on the world, which account for ~ 0.4% of the total population, according to World Drug Report published June last year. Therefore, development of sensitive, selective, rapid, and inexpensive methods for in-situ detection of illicit drugs is of great significance for preventing and reducing drug-related crimes. However, no methods can be compared to specially trained animals up to now. Even worse, the reported techniques or instruments are hard to be used in-situ since pre-treatment and/or pre-concentration of the samples need to be conducted before any valuable detection, which explains why on-site screening of illicit drugs still relies on drug-sniffing dogs and drug-suppressing police officers. 
As well-known, sniffing dogs are hard to please, people have been tried to replace them with modern optical or electrical technology-based vapor detection techniques. As an example, hidden explosive detection via vapor phase sensing has been realized by using ion-mobility and fluorescence techniques. Compared with other known methods, film-based fluorescence sensing possesses several unique advantages, such as great designability, outstanding sensitivity, re-usability and low cost. Because of such reasons, research in the field has achieved great progress during the last few decades. However, for drug detection there are other challenges. For example, incompatibility between universality for sensing different illicit drugs and selectivity to avoid interferences, and moreover, most of the drugs exist in salt forms which greatly limit their vapor pressures, resulting in additional difficulty for sensing.
To cope with the challenges, our strategy is to develop individual high-performance illicit drug sensors first, and then combine them into a sensor array. To be more compact, simply-structured and have less power consumption, the component sensors are designed to work in a similar way. Consequently, we developed three film-based fluorescent sensors showing unprecedented sensitivity, selectivity and response speed to the existence of six widely abused illicit drugs, including methamphetamine (MAPA), ecstasy, magu, caffeine, phenobarbital (PB), and ketamine in vapor phase. Importantly, vapor sensing of the drugs can be performed after 5.0×10^5, 4.0×10^5, 2.0×10^5, 1.0×10^5, 4.0×10^4 and 2.0×10^2 times respective dilution of their equilibrated vapor with air at room temperature. More importantly, presence of water, toiletries, fruits, cloths and other most common interferents shows little effect upon the sensing. Particularly, the drugs in their hydrochloride forms with extremely low vapor pressure can be directly detected, a result never reported before. 
During the submission process, the reviewers suggested us to construct a sensor array to confirm the strategy as proposed in a direct way. Accordingly, the film-based sensors were integrated and a conceptual illicit drug detector was developed, which works well as expected.
In summary, we have demonstrated the viability and value of film-based fluorescent sensing in non-contact detection of hidden illicit drugs, which could find important real-life applications.

The paper in Nature Communications is here: https://www.nature.com/articles/s41467-018-04119-6