SWIR-LED Technology: Food Inspection & Sorting

1. Introduction

In today’s fast-moving tech landscape, the food & agriculture sector is undergoing a digital revolution. Short-wave infrared (SWIR) light-emitting diodes (LEDs), operating between 900 nm and 1700 nm, are rapidly becoming the technology of choice for non-destructive food testing and high-speed sorting thanks to their unique optical fingerprint and ever-shrinking size and cost. This paper explains how SWIR LEDs work, why they outperform conventional methods, and where they are already adding value—from measuring sugar in fruit to spotting bruises inside meat.

2. How SWIR LEDs Work – and Why They Win

- Deep penetration: photons travel millimetres into fruit, cheese or poultry, bypassing skin, fat or plastic wrap

- Low scatter: wavelength is ~2× longer than NIR-850 nm, so images stay sharp even inside turbid media

- High contrast: water (1 450 nm, 1 950 nm), sugar (1 050–1 500 nm), protein (940 nm, 1 200 nm) and lipids (1 700 nm) show strong, distinct absorption peaks that translate into chemical maps

- Solid-state reliability: LED sources are instantly on, mercury-free, shock-resistant and rated >50 000 h; spectral drift <1 nm over temperature

3. Food-Quality Applications

3.1 Moisture & Dry-Matter

Principle: quantify 1 450 nm or 1 940 nm absorption relative to a reference band (e.g. 1 300 nm)

Example: Tianhui 1 W 1 450 nm module embedded in a conveyor belt photometer measures whole-kernel corn moisture at 2 t h⁻¹; repeatability ±0.3 %, no grinding required

3.2 Sugar / Brix

Principle: fructose & sucrose exhibit combination bands at 1 580 nm and 1 650 nm; ratio to water peak gives °Brix

Example: on an apple sorting line, four Tianhui LEDs (1 060, 1 300, 1 450, 1 650 nm) + InGaAs array predict °Brix with RMSEP 0.4 °Brix; throughput +40 % versus spindle penetrometer

3.3 Protein & Fat

Principle: –NH and –CH overtones at 1 200 nm and 1 730 nm correlate with protein and lipid concentration

Example: dairy plant uses 1 200 nm/1 700 nm trans-reflectance probe to monitor skim-milk powder protein in real time; calibration R² = 0.92 versus Kjeldahl

3.4 Pesticide Residue Screening

Principle: many agro-chemicals show SWIR fingerprints (e.g. 1 550 nm for azole fungicides); chemo-metric models flag outliers

Example: portable 1 550 nm LED reflectance gun inspects 300 apples h⁻¹; 95 % correct rejection of samples above 0.01 ppm limit

4. Sorting & Foreign-Body Detection

4.1 Fruit & Vegetable Grading

SWIR hyperspectral cubes sort on internal browning, watercore or freeze damage while still on the line; good product routed to premium channels

4.2 Internal Defects in Meat

1 310 nm LED arrays detect PSE (pale-soft-exudative) pork and haemorrhages inside chicken breast; automatic air-blast removal cuts customer complaints by 60 %

4.3 Package Integrity

SWIR penetrates opaque plastic trays and paper cartons to verify fill level, missing inserts or leaky seals; 100 % inspection at 120 m min⁻¹