Miniature fibre-optic spectrometers such as those produced by Ocean Optics have transformed optical spectroscopy. Their compact size, broad wavelength coverage and fibre coupling make them ideal for process monitoring, field measurements and laboratory automation.
I have been working with fibre-coupled spectrometers from Ocean Optics for more than 25 years, applying them across a wide range of optical spectroscopy applications.
A common application is absorbance measurements.
However, unlike conventional scanning spectrophotometers, these instruments rely on CCD or CMOS detector arrays. Consequently, absorbance measurements are influenced by detector characteristics such as dark current, stray light, detector saturation and temperature.
Understanding these effects is essential for obtaining accurate absorbance measurements.
How absorbance is using the Beer-Lambert Law.
For array spectrometers
Where;
Reference= solvent or blank
Sample = absorbing sample
Dark = detector signal with no incident light
Unlike scanning spectrometers, every wavelength is measured simultaneously by an individual detector pixel.
# Measuring Dark Counts
Dark counts arise from
* detector dark current
* electronic bias
* read noise
* thermal electrons
A dark spectrum should always be collected using
* shutter closed, or
* fibre capped, or
* light source switched off
using exactly the same integration time as the measurement.
For long integration times (>100 ms), dark counts become increasingly important
# Temperature Dependence of Dark Counts
One characteristic often overlooked is that dark current approximately doubles every 5–8°C for silicon CCD detectors.
Consequences include
* changing baseline
* increased noise
* poor absorbance stability
* wavelength-dependent baseline offsets
Miniature spectrometers without active cooling are particularly susceptible.
This explains why measurements made immediately after power-up may drift as the detector warms.
Best practice:
* allow thermal equilibrium
* periodically remeasure dark spectra
* avoid direct sunlight on the spectrometer
* use thermoelectrically cooled detectors for demanding measurements
# Detector Dynamic Range
Every detector has a finite number of digital counts.
Typical Ocean Optics spectrometers use a 16-bit ADC:
* maximum ≈ 65,535 counts
* noise floor typically 5–20 counts
* practical usable range approximately 50,000 counts
A good rule is to adjust integration time so the brightest wavelength reaches about 80–90% of full scale
Too little signal wastes dynamic range.
Too much causes saturation.
# Optical Resolution versus Absorbance Accuracy
Higher spectral resolution is often assumed to be better.
This is not always true.
A narrow entrance slit provides
* higher resolution
* lower throughput
* lower signal-to-noise ratio
A wider slit provides
* more photons
* lower noise
* reduced spectral resolution
For absorbance measurements, signal-to-noise ratio is frequently more important than achieving the highest possible spectral resolution.
# Stray Light — The Biggest Limitation
For absorbance measurements, stray light is often the dominant source of error.
Stray light originates from
* diffraction grating scatter
* internal reflections
* fibre reflections
* detector blooming
* imperfect optics
Instead of measuring
Sample transmission = 0
The detector actually receives
Measured signal = Stray light +True transmission
Even a tiny amount of stray light severely limits measurable absorbance.
Example:
Reference = 50,000 counts
True sample = 0 counts
Stray light = 50 counts
Measured transmission
T = 50 / 50000 = 0.001
So the maximum measurable absorbance
A=- log (0.001)= 3
Thus, a spectrometer with 0.1% stray light cannot accurately measure absorbances greater than approximately 3 AU, regardless of integration time.
This is a fundamental instrument limitation.
# Integration Time Selection
A practical workflow:
- Measure dark.
- Measure reference.
- Adjust integration time until the brightest wavelength reaches 80–90% of saturation.
- Reacquire dark.
- Measure sample.
- Repeat dark if temperature changes significantly.
# Common Mistakes
| Mistake | Consequence |
| Reference saturated | Incorrect absorbance |
| Dark measured with different integration time | Baseline error |
| Ignoring detector warming | Drift |
| Measuring >4 AU absorbance | Stray light dominates |
| Using very low counts | Poor SNR |
| Changing fibre alignment between reference and sample | Large absorbance errors |
# Practical Recommendations
For routine absorbance measurements:
* Always collect a fresh dark spectrum using the same integration time.
* Keep the reference signal between 80–90% of detector full scale.
* Allow the spectrometer to reach thermal equilibrium.
* Recollect dark spectra whenever detector temperature changes.
* Recognize that stray light—not detector sensitivity—usually determines the maximum measurable absorbance.
* Choose slit width to balance spectral resolution and signal-to-noise ratio for the application.
# Choosing the Right Light Source
The performance of an absorbance measurement is determined not only by the spectrometer but also by the stability, spectral output, and coupling efficiency of the light source.
An ideal absorbance light source should have:
* High spectral intensity over the wavelength range of interest
* Stable output over time
* Low flicker
* Good coupling into optical fibres
* Minimal thermal drift
The choice depends on the application.
| Light Source
| Spectral Range | Advantages | Limitations |
| Tungsten-Halogen | 360–2500 nm | Smooth spectrum, excellent VIS-NIR
| Weak below 400 nm |
| Deuterium | 190–400 nm | Strong UV output | Limited visible intensity |
| Combined D₂/Halogen | 200–2500 nm | Continuous UV-VIS-NIR coverage | Requires source switching or combining optics |
| Xenon Flash | 220–1000 nm | High brightness, pulsed operation
| Pulse-to-pulse intensity variation |
| White LED | 420–750 nm | Long lifetime, low power | Uneven spectrum, limited UV
|
| Laser-driven light source | 190–2500 nm | Extremely bright and stable | Expensive |
# Lamp Stability
Absorbance assumes the reference and sample measurements are made under identical illumination conditions.
Lamp instability introduces apparent absorbance changes:
* LED output changes with temperature.
* Halogen lamps drift during warm-up.
* Xenon flash lamps exhibit pulse-to-pulse variation.
* Fibre movement changes coupling efficiency.
A reference measurement should therefore be repeated whenever:
* the lamp has been switched off,
* integration time changes,
* fibres are moved,
* optical components are realigned.
# Fibre Coupling Effects
One advantage of fibre-optic spectrometers is measurement flexibility, but fibres introduce their own errors.
Common issues include:
* Connector repeatability
* Fibre bending losses
* Numerical aperture mismatch
* Fresnel reflections
* Modal noise (particularly with coherent sources)
Even reconnecting an SMA fibre can change coupling efficiency by a few percent, affecting absorbance if the reference is not reacquired.
# Scattering vs. Absorption
Another common misunderstanding is interpreting scattering as absorption.
Examples include:
* Cell suspensions
* Milk
* Polymer emulsions
* Nanoparticle dispersions
* Turbid environmental samples
The detector records reduced transmitted light regardless of whether photons were:
* absorbed, or
* scattered out of the collection optics.
Thus, the measured “absorbance” is often more accurately described as optical attenuation.
This distinction is particularly important when using fibre probes, where the collection geometry differs from that of a standard benchtop spectrophotometer.
# Fluorescence – An often overlooked Source of Error
One limitation of absorbance measurements is that the detector cannot distinguish between transmitted light and light emitted by the sample.
If the sample fluoresces, the detected signal becomes:
Detected Intensity = Transmitted Intensity + Fluorescence Intensity
The additional fluorescence increases the measured intensity, making the sample appear to absorb less light than it actually does.
This results in an underestimation of absorbance.
For example
Suppose
Reference = 50,000 counts
True transmitted intensity = 5,000 counts
Fluorescence = 2,000 counts
The detector measures
7,000 counts
instead of
5,000 counts.
Calculated absorbance becomes
True absorbance A= -log (5000/50000) = 1.00
Measured A = -log (7000/50000) = 0.85
The fluorescence causes an error of approximately 15% in absorbance.
# How to Minimise Fluorescence Effects
Several strategies can reduce fluorescence interference:
* Select wavelengths where fluorescence is weak.
* Measure fluorescence separately to assess its contribution.
For highly fluorescent samples, conventional absorbance measurements may not be appropriate without additional optical filtering.
## Conclusion
Fibre-optic spectrometers offer exceptional flexibility for absorbance measurements, but achieving laboratory-grade performance requires careful attention to detector physics and optical limitations. Proper dark correction, management of detector dynamic range, awareness of temperature-dependent dark current, and an understanding of stray light limitations are essential for producing accurate, reproducible absorbance spectra.
Maximum absorbance is usually limited by stray light, not by detector bit depth.
Reference intensity matters — operating near (but below) saturation maximises signal-to-noise ratio.
Temperature stability matters — especially for uncooled CCD/CMOS array spectrometers.
Fluorescence can produce deceptively low absorbance values because emitted photons are interpreted as transmitted light.
Resolution and throughput are a compromise — the narrowest slit is not always the best choice for quantitative absorbance.
Fibre-Optic Absorbance Measurement Checklist
☐ Allow lamp and spectrometer to warm up
☐ Measure dark spectrum at the same integration time
☐ Adjust reference intensity to ~80–90% full scale
☐ Avoid detector saturation
☐ Reacquire reference after changing fibre alignment
☐ Check for fluorescence before interpreting absorbance
☐ Consider stray-light limits for high absorbance samples
☐ Control temperature for long-term measurements
