Skip to content
Ahsan’s Lab logoAhsan Suny.

Research · 2025

PhotoScale: an automated continuous-flow photoreactor

ICIQ · Universitat Rovira i Virgili

A continuous-flow photoreactor with embedded sensing, control and logging, built as a platform for data-driven, autonomous photocatalysis. Presented as a poster at the ICIQ Summer School 2026.

Flow chemistryPhotocatalysisRaspberry PiLab automationEdge AI
PhotoScale: an automated continuous-flow photoreactor

Overview#

PhotoScale is an automated continuous-flow photoreactor built at ICIQ as a platform for data-driven, autonomous photocatalysis. Optimising a new photochemical transformation means searching a high-dimensional space (catalyst loading, reagents, solvents, photon flux, residence time, temperature), and manual screening limits both throughput and reproducibility.

PhotoScale embeds sensing, control and logging in the hardware, and defines a layered upgrade path: structured logging with quality-control gating, machine-vision kinetics, edge-AI closed-loop optimisation, and an LLM pipeline that distils catalyst knowledge from the literature.

Architecture#

  • Recirculating flow path: reagent reservoir → peristaltic pump → in-line ultrasonic flow meter → glass tubular reactor.
  • Illumination: eight 450 nm LUXEON high-power LED matrix plates (48 LEDs each) surround the reactor, water-cooled by an external chiller.
  • Control: a Raspberry Pi runs a full-screen touchscreen app, started at boot as a system service, with remote operation and configuration over VNC.
  • Light and logging: four PWM channels set LED intensity from 0 to 100 % (two plates per channel) with independent on/off; temperatures and flow are logged for every run.
Hardware at a glance
Light source 8 × 450 nm LUXEON LED matrix plates (48 LEDs each)
LED power Phoenix Contact 48 V / 20 A supply
Dimming 4 PWM channels, 0–100 %, independent on/off
Flow sensing In-line ultrasonic flow meter
Thermal Up to 8 temperature sensors + external chiller
Control Raspberry Pi · 7″ touchscreen · VNC · per-run logs

Control software#

The control unit runs on a Raspberry Pi 4B with a touchscreen app I wrote in Python with PyGame:

  • Touchscreen control of light intensity for each LED channel, plus manual control through LED dimmers.
  • Real-time temperature monitoring with DS18B20 one-wire sensors on every reactor plate.
  • Wireless control through a 2.4 GHz RF module.
  • Data logging of temperature and light intensity for every run.
  • Fail-safe: if the system fails, the dimmers fall back to a safe default state.

The code is split into an application core, GPIO/PWM control, sensor reading and reusable UI components. The app is open source on GitHub as DisplayUI.

Status and results#

  • The platform is built and validated in routine use: touchscreen UI, PWM LED control, eight-channel temperature logging, flow meter, hotspot and VNC all work and start automatically at boot.
  • Standard recirculation runs are logged as a reference, and the software is tracked under Git.
  • Every module ships with a simulation mode, so APIs, dashboards and logging can progress regardless of hardware availability.
  • Next steps: current and light sensors, the machine-vision pipeline and the edge-AI control loop, which feed into my self-driving lab research.

Poster#

Presented at the ICIQ Summer School 2026 (20–22 July 2026).

MD. Shahnauze Ahsan, Damian Giziński, Julio Lloret-Fillol. PhotoScale: An Automated Continuous-Flow Photoreactor for Data-Driven Photocatalysis. Institute of Chemical Research of Catalonia (ICIQ); Universitat Rovira i Virgili (URV); ICREA.

Download the poster (PDF)

Prof. Julio Lloret-Fillol Research Group, ICIQ, Tarragona, Spain.

Figures