🔬 applied physics

On-Chip Glucose Sensing at Terahertz Frequencies

This paper demonstrates a label-free, on-chip terahertz glucose sensor utilizing a coplanar stripline on a silicon nitride membrane that successfully detects anhydrous D-glucose by identifying characteristic absorption signatures at 1.42 THz and 2.07 THz, with experimental results aligning closely with simulations and existing literature.

Original authors: Mohsen Haghighat, Thomas Darcie, Levi Smith

Published 2026-03-27

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Mohsen Haghighat, Thomas Darcie, Levi Smith

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a tiny, invisible musical instrument made of gold and silicon, so small it fits on a computer chip. This instrument is designed to "listen" to sugar molecules by making them sing a specific note. That is the core idea behind this research paper: a new, high-tech way to detect glucose (sugar) using invisible light waves called Terahertz (THz) radiation.

Here is a simple breakdown of how it works, using everyday analogies.

1. The Problem: Finding Sugar in a Crowd

Detecting sugar is important for doctors (to check blood sugar) and food manufacturers (to check soda or juice quality).

The Old Way: Traditional methods are like trying to find a specific person in a crowded stadium by shouting their name and hoping they shout back. It's often messy, requires chemicals, or needs big, bulky machines.
The Optical Way: Using light (like lasers) is like trying to see through a foggy window. The light scatters, making it hard to get a clear picture of the sugar.
The New Way (This Paper): Terahertz waves are like a "super-sonic" flashlight that can see through fog better than regular light. But usually, these machines are huge and expensive. This team wanted to shrink it down to the size of a postage stamp.

2. The Solution: The "Silk Road" for Light

The researchers built a Coplanar Stripline (CPS). Think of this as a tiny, golden highway built on a super-thin sheet of Silicon Nitride (imagine a sheet of paper made of rock).

The Highway: The gold lines are the road.
The Bridge: The silicon nitride membrane is a bridge suspended over a gap.
The Passengers: Instead of cars, invisible Terahertz waves travel down this road.

Because the "bridge" is so thin (only 1 micrometer thick—about 1/50th the width of a human hair), the waves don't just stay on the road; they spill over the sides, like water spilling over a very low dam. This allows the waves to "touch" whatever is sitting on the other side of the bridge.

3. The Magic: The Sugar "Fingerprint"

The researchers placed a tiny layer of dry glucose (sugar) on the back of this bridge.

The Analogy: Imagine you are walking down a hallway. If the hallway is empty, your footsteps echo clearly. But if you hang a heavy, fuzzy blanket on the wall, your footsteps sound different; they get muffled at a specific pitch.
The Science: Glucose molecules have a unique "voice." When the Terahertz waves hit the sugar, the sugar absorbs energy at very specific frequencies (notes).
- The sugar "sings" (absorbs energy) loudly at 1.42 THz.
- It sings again at 2.07 THz.

When the waves hit the sugar, they lose energy at these exact frequencies. The machine at the end of the line sees a "notch" or a dip in the signal, like a missing note in a song.

4. Why This is a Big Deal

The researchers proved two main things:

It Works on a Chip: They didn't need a room-sized machine. They built a sensor on a tiny chip that fits in your hand.
It's "Label-Free": You don't need to dye the sugar or mix it with chemicals. You just put the sugar near the chip, and the chip instantly recognizes its unique "fingerprint."

The "Double-Check" Safety Net:
The paper mentions that detecting just one note (1.42 THz) might be tricky because other things (like table sugar/sucrose) might make a similar sound. But because this sensor is smart, it listens for two notes at once (1.42 THz and 2.07 THz). If both notes appear together, the computer knows for sure, "Yes, that is definitely glucose!" This reduces the chance of a false alarm.

5. The Future: From Lab to Life

Currently, the researchers put the sugar on the chip by dropping a liquid mixture and letting it dry, which is a bit messy (like spilling coffee on a table). They admit this isn't perfect yet.

The Goal: In the future, they want to build tiny "micro-channels" (like tiny plumbing pipes) directly onto the chip. This would allow blood or juice to flow right over the sensor automatically.

Summary:
This paper is about building a miniature, high-tech sugar detector that uses invisible light waves to listen to the unique "song" of glucose molecules. It's a major step toward making medical devices and food quality testers smaller, cheaper, and more accurate.

1. Problem Statement

Glucose detection is critical for medical monitoring (e.g., diabetes management) and industrial quality control (e.g., food and beverage manufacturing). While existing sensing technologies exist, they face significant limitations:

Optical Methods: Suffer from high scattering in biological tissues, weak absorption peaks (especially in Near-Infrared), and interference from ambient light.
Terahertz (THz) Free-Space Methods: Although THz waves offer low scattering and strong glucose absorption signatures, current free-space sensing setups are bulky, require complex optical alignment, and suffer from path loss. They are difficult to miniaturize for on-chip integration.
Guided-Wave Limitations: Previous attempts at guided-wave THz sensing (using Coplanar Waveguides or Microstrip Lines) have struggled with high signal attenuation at frequencies above 1 THz, often causing signals to decay below the noise floor.

The core challenge addressed by this paper is the development of a compact, on-chip, label-free glucose sensor that operates effectively in the THz regime without the bulk and alignment issues of free-space systems, while overcoming the high-loss limitations of standard waveguide designs.

2. Methodology

The authors propose a novel sensor architecture utilizing guided waves rather than free-space radiation.

Sensor Architecture:
- Substrate: A thin (1 µm) Silicon Nitride ( $Si_3N_4$ ) membrane suspended over an etched silicon cavity. This thin membrane minimizes dielectric loss and allows the electromagnetic field to interact strongly with materials placed beneath it.
- Transmission Line: A Coplanar Stripline (CPS) with a gap ( $S$ ) of 130 µm and strip width ( $W$ ) of 30 µm is fabricated on the membrane.
- Sample Placement: Anhydrous D-glucose is placed directly on the backside of the membrane (opposite the CPS). The glucose is prepared as a suspension in isopropanol, sonicated to reduce grain size, and drop-coated to form a layer approximately 10 µm thick.
Measurement Setup:
- A modified Terahertz Time-Domain Spectroscopy (THz-TDS) system is used.
- A femtosecond laser (780 nm) drives photo-conductive switches (LT-GaAs) acting as the transmitter and receiver on either side of the CPS.
- The system measures the transmission response of the broadband THz pulse as it propagates through the CPS.
Modeling and Simulation:
- Simulations were performed using ANSYS HFSS.
- The dielectric properties of glucose were modeled using a modified Lorentz model to account for frequency-dependent permittivity, specifically targeting the resonance at ~1.435 THz.
- Simulations predicted attenuation coefficients for various glucose thicknesses (5, 10, and 15 µm).

3. Key Contributions

On-Chip Guided-Wave Sensing: The paper demonstrates the first successful on-chip glucose sensor using a CPS on a thin membrane, proving that guided-wave THz sensing can achieve high sensitivity without the bulk of free-space optics.
Label-Free Multi-Signature Detection: Unlike many metasurface sensors that rely on a single resonance shift, this sensor detects two distinct absorption signatures of glucose simultaneously (at 1.42 THz and 2.07 THz). This dual-frequency detection significantly reduces the probability of false positives compared to single-frequency sensors.
Mitigation of Waveguide Loss: By utilizing a suspended 1 µm $Si_3N_4$ membrane, the design minimizes substrate loss, enabling effective signal propagation up to 2.2 THz, a range where other guided-wave structures typically fail.
Broadband Operation: The sensor operates over a broad frequency range (0.1 – 2.2 THz), facilitating the observation of multiple spectral features rather than just a narrow resonance.

4. Results

Absorption Signatures: Experimental measurements confirmed two distinct absorption dips in the transmission spectrum corresponding to glucose resonances:
- 1.42 THz: The primary absorption peak.
- 2.07 THz: The secondary absorption peak.
Quantitative Agreement:
- The measured attenuation depth at 1.42 THz was approximately 6.2 dB (for a ~~3 mm interaction length), which aligns closely with simulation predictions (~~13.5 dB total attenuation for the specific geometry, accounting for the 3 mm length).
- The attenuation at 2.07 THz was measured at ~6.2 dB, consistent with expected values derived from literature data.
Validation: The results align with existing literature regarding the resonance frequencies of D-glucose (typically cited between 1.40–1.44 THz and 2.05–2.1 THz).
Limitations Observed: The current drop-coating method resulted in non-uniform glucose distribution, which can lead to false negatives if the film does not cover the CPS effectively. The authors acknowledge this as a limitation to be addressed by future microfluidic integration.

5. Significance

This work represents a significant step forward in miniaturized THz sensing technology.

Practical Application: It validates a pathway for integrating THz sensors onto chips for real-world applications in healthcare (continuous non-invasive glucose monitoring), food industry (quality control of sugar levels), and security (material identification).
Reliability: The ability to detect multiple spectral signatures simultaneously enhances the reliability of the sensor, distinguishing glucose from other sugars (like sucrose) that may have overlapping peaks at single frequencies.
Future Outlook: The study establishes a foundation for future on-chip sensors that can be integrated with microfluidics for automated, continuous, and highly sensitive analysis of various biological and chemical materials at THz frequencies.