DropleX: Liquid sensing on tablet touchscreens

TL;DR

DropleX repurposes commodity tablet capacitive touchscreens to sense microliter-scale liquids by temporarily disabling the adaptive filter via a priming-drop technique and exploiting spatial heatmaps of mutual capacitance. The approach combines physics-informed modeling of capacitance, an empirical framework to capture fringing fields, and a CNN-based classifier with a frame-wise training regime to detect adulteration and concentration, including through-container sensing. The paper presents extensive empirical characterization, a calibration map, and an end-to-end software stack, achieving high accuracies across adulteration (96-99%), trace chemical detection (93-96%), and through-container sensing (86-96%). This work demonstrates a practical path to turn everyday tablets into portable laboratory tools for safety, education, and accessible sensing.

Abstract

We present DropleX, the first system that enables liquid sensing using the capacitive touchscreen of commodity tablets. DropleX detects microliter-scale liquid samples, and performs non-invasive, through-container measurements to detect whether a drink has been spiked or if a sealed liquid has been contaminated. These capabilities are made possible by a physics-informed mechanism that disables the touchscreen's built-in adaptive filters, originally designed to reject the effects of liquid drops such as rain, without any hardware modifications. We model the touchscreen's sensing capabilities, limits, and non-idealities to inform the design of a signal processing and learning-based pipeline for liquid sensing. Our system achieves 96-99% accuracy in detecting microliter-scale adulteration in soda, wine, and milk, 93-96% accuracy in threshold detection of trace chemical concentrations, and 86-96% accuracy in through-container adulterant detection. Given the predominance of touchscreens, these exploratory results can open new opportunities for liquid sensing on everyday devices.

Figures (24)

• Figure 1: Capacitive touchscreen architecture.(a) Layers of a capacitive touchscreen. (b) Mesh of driving and sensing lines with a dielectric in between form a 2-D grid of capacitive electrodes or cells. The driving lines produce a voltage $V_{drive}$. (c)$V_{drive}$ signal measured from an unmodified touchscreen tablet using an oscilloscope with probe in contact with the display. Figures are drawn for conceptual illustration and are not to scale.
• Figure 2: Capacitive sensing mechanisms under different interactions.(a) The two layers of driving and sensing electrodes form a mutual capacitance electric field $C_{mutual}$. (b) The presence of a conductive grounded object like a finger creates a lower impedance path and electric charge is drawn towards it to $GND_{earth}$, this reduces the mutual capacitance field $C_{mutual}$ via the conductivity effect. (c) When a liquid sample is placed on the surface it acts simultaneously as a conductor and a dielectric: as a conductor, the liquid draws away charge to $GND_{parasitic}$ as it is coupled to the air and other parasitics, while as a dielectric, it's relative permittivity $\epsilon_r$ increases its polarization creating a store of more energy and increases the effective capacitance between the electrodes and the liquid. The net change in capacitance is the sum effects of conductivity and permittivity.
• Figure 3: Effect of liquid rejection filter on measured capacitance values.(a) With the filter enabled by default, the touchscreen controller gradually re-adapts the baseline, causing the measured signal to return to its original level over time scales of approximately 1–20 s. (b) When the filter is disabled, the sample is continuously registered and does not decay away over time. We note that the device units of capacitance have the inverse sign of the actual capacitance change.
• Figure 4: Disabling the touchscreen's liquid rejection filter by mimicking a permanent finger touch event.(a) Depositing a priming drop (water sample) on the screen increases the capacitance as it is dominated by the permittivity effect, and the electric field is confined within the sample. (b) We disable the filter by mimicking a permanent finger touch event. To do this, we draw up the priming drop which leaves a thin film that causes the electric field to be shunted to parasitic ground, similar to a finger. (c) With the adaptive filter disabled, subsequent samples of interest to be measured can be deposited on the screen. (d) Shows the capacitance at the centroid of the priming drop when it is initially deposited then drawn up.
• Figure 5: Equipment-free deposition and drawing up of priming drop.(a) User deposits pendant drop onto screen, and (b) draws up priming drop using tissue paper via capillary wicking.