Fiber Optic Spectrometer Working Principle and Key Specifications

1. Introduction

Spectral analysis is one of the most fundamental and widely applied analytical methods in modern scientific research and industrial inspection. From substance composition identification, color measurement and thin-film thickness analysis to plasma emission spectroscopy diagnostics, spectral analysis permeates nearly every field involving light-matter interaction. Traditional spectrometers typically use large grating-scanning mechanisms that are bulky, structurally complex and demanding of the working environment, making them difficult to meet the needs of field and online inspection.

The emergence of the fiber optic spectrometer has completely changed this situation. It uses an optical fiber as the light-signal acquisition and transmission medium, integrating core components such as the slit, collimating mirror, diffraction grating and array detector into a compact module to realize fast, portable spectral measurement. Thanks to its outstanding advantages - compact structure, no moving mechanical parts, fast acquisition and easy system embedding - the fiber optic spectrometer has become the preferred spectral measurement tool in industrial online monitoring, environmental testing, food safety, biomedical diagnostics and other fields.

This article systematically explains the working principle of the fiber optic spectrometer's optical system, its core component composition, key technical specifications and their deeper meanings, and on this basis provides practical selection recommendations to help readers deeply understand the technical essence of the fiber optic spectrometer.

2. Working Principle

The core of the fiber optic spectrometer is a crossed optical-path system based on the Czerny-Turner optical configuration. Its workflow can be summarized as follows: the light under test enters the spectrometer through fiber coupling, passes in turn through the entrance slit, collimating mirror, diffraction grating and focusing mirror, and finally forms a wavelength-dispersed spectrum on the detector array. The following sections introduce the function and design points of each optical element in sequence.

2.1 Entrance Slit

The entrance slit is the first gate of the entire optical system; its physical width determines the light throughput entering the spectrometer and the ultimate spectral resolution the instrument can achieve.

As the field stop of the optical system, the slit width is a key trade-off parameter: widening the slit increases throughput so the detector receives more signal, but reduces spectral resolution - because the geometric image of the slit on the detector broadens, causing adjacent wavelengths to be poorly separated; conversely, narrowing the slit improves resolution but reduces throughput and consequently lowers the signal-to-noise ratio. Common slit widths range from $5\mu m$ to $200\mu m$, and users can balance resolution against sensitivity according to their actual measurement needs.

2.2 Collimating Mirror

Light emerging from the entrance slit is diverging; without treatment it cannot form a uniform wavefront on the grating. The role of the collimating mirror is precisely to convert the diverging light from the slit into a parallel beam and project it onto the diffraction grating surface at a uniform incidence angle.

The collimating mirror is generally a concave spherical mirror, and its focal length is a key parameter affecting the overall system size and spot quality. A longer collimating-mirror focal length yields better beam collimation and fuller utilization of the grating's diffraction efficiency, which favors higher resolution; but the physical size of the instrument also increases accordingly. In compact fiber optic spectrometers, the collimating-mirror focal length is typically between $40mm$ and $80mm$.

2.3 Diffraction Grating

The diffraction grating is the heart of the spectrometer; its performance directly determines the instrument's spectral resolution, wavelength range and diffraction efficiency. Through the diffraction effect of the periodic groove structure on its surface, the grating spreads polychromatic light in space by wavelength - that is, it performs dispersion.

The dispersion behavior of the grating follows the grating equation:

$$ m\lambda = d(\sin\alpha + \sin\beta) $$

where $m$ is the diffraction order (first-order diffraction with $m=1$ is typically used), $\lambda$ is the wavelength, $d$ is the grating period (groove spacing), $\alpha$ is the incidence angle, and $\beta$ is the diffraction angle. Within the same diffraction order, light of different wavelengths emerges at different diffraction angles, thereby achieving linear separation of wavelengths in space.

The core parameters of a grating include the groove density (grooves/mm) and the blaze wavelength. The higher the groove density, the greater the angular dispersion and the better the spectral resolution, but the smaller the free spectral range (the wavelength range within which no order overlap occurs in the same diffraction order). The blaze wavelength determines the peak position of the diffraction efficiency; by optimizing the groove profile, as much diffracted energy as possible is concentrated into the target diffraction order.

2.4 Focusing Mirror

After diffraction by the grating, the light is a combination of parallel beams dispersed by wavelength, and the focusing mirror must converge them separately onto different positions of the detector array. The focusing mirror is likewise a concave spherical mirror, and its focal length is usually the same as that of the collimating mirror (the symmetric Czerny-Turner configuration), to balance imaging quality against system compactness.

The plane containing the detector surface is the focal plane. On this plane, the light spots of different wavelengths are arranged according to a linear dispersion relationship: short wavelengths fall at the starting-pixel end of the detector, and long wavelengths at the end-pixel end. The linear dispersion $D$ (in units of $nm/mm$) defines the wavelength width corresponding to unit length on the focal plane; combined with the detector pixel size, it determines the wavelength range covered by a single pixel.

2.5 Array Detector

The wavelength-distributed spectrum imaged by the focusing mirror is finally received by the linear array detector and converted into electrical signals. The detector is the core of the light-to-digital signal conversion, and its performance largely determines the overall signal-to-noise ratio, dynamic range and acquisition speed of the spectrometer.

Each pixel (also called a photosensitive element) of the detector corresponds to an extremely narrow wavelength interval. When photons strike a pixel, charge is generated through the photoelectric effect, and the amount of charge is proportional to the incident light intensity and the integration time. After integration ends, the charge in the pixels is read out and amplified in sequence, and converted into digital signals by an analog-to-digital converter (ADC), forming a complete spectrum curve. Common array detectors include CCD (charge-coupled device), CMOS (complementary metal-oxide-semiconductor) and InGaAs (indium gallium arsenide) linear arrays, each with its own applicable wavelength range and performance characteristics, as detailed in the next section.

3. Core Components

3.1 CCD/CMOS Detectors

Silicon-based CCD and CMOS are the most commonly used array detectors in the visible-to-near-infrared band, with a spectral response range roughly covering $350nm$ to $1100nm$.

CCD detectors transfer and read out signals pixel by pixel through charge coupling, and their advantages lie in low dark current, low noise and good inter-pixel uniformity, making them especially suitable for measurement scenarios with low light signals and long integration times. A cooled CCD (a cooled back-illuminated CCD) can lower the detector operating temperature to tens of degrees Celsius below ambient temperature, significantly suppressing dark-current noise and thereby greatly improving the signal-to-noise ratio. A commonly used detector configuration in high-sensitivity spectral measurements is the $1024$-pixel cooled back-illuminated CCD.

CMOS detectors integrate an independent amplifier and readout circuit on each pixel, enabling parallel high-speed readout with a frame rate far higher than CCD. They also offer advantages in power consumption, integration density and cost. In recent years, the emergence of scientific CMOS (sCMOS) has further narrowed the gap with CCD in noise performance.

For the near-infrared band ($900nm$ to $1700nm$), the quantum efficiency of silicon-based detectors is already extremely low, so an InGaAs (indium gallium arsenide) linear array detector is required. InGaAs has high quantum efficiency and responsivity in this band and is the core detector of near-infrared fiber optic spectrometers.

3.2 Diffraction Grating

The core function of the diffraction grating has been detailed in the principle section above. Here we further emphasize its criticality in engineering selection: the combination of groove density and blaze wavelength determines the spectrometer's wavelength coverage and resolution profile. A high groove density (e.g. $1200\ grooves/mm$) provides high resolution and relatively narrow wavelength coverage; a low groove density (e.g. $300\ grooves/mm$) provides wide-band coverage and lower resolution. Many high-end fiber optic spectrometers support user-replaceable gratings to suit different measurement needs.

3.3 Fiber Interface (SMA905)

SMA905 is the most universal fiber connector standard in the field of fiber optic spectrometers. It uses a threaded locking structure that ensures the repeatability and stability of mechanical alignment between the fiber end face and the entrance slit. The inner diameter of the SMA905 interface is $3.17mm$ ($1/8$ inch), and the mechanical tolerance of the connector is strictly controlled at the micrometer level, ensuring that after insertion the fiber core is highly coaxial with the optical axis of the optical system.

The other end of the fiber can be equipped with sampling accessories such as a cosine corrector, collimating lens, integrating sphere or cuvette holder as required by the application, to realize different measurement modes such as irradiance measurement, diffuse reflectance measurement and absorbance measurement.

4. Key Technical Specifications Explained

Understanding a spectrometer's technical specifications is the basis for scientific selection and correct use of the instrument. The following sections analyze each key specification of the fiber optic spectrometer and its physical meaning item by item.

4.1 Wavelength Range

The wavelength range refers to the spectral interval the spectrometer can effectively detect, and is constrained by multiple factors including the grating groove density, the detector's spectral response range and the coatings of the optical elements.

A typical silicon-based CCD array spectrometer has a wavelength range of $350nm$ to $1100nm$, covering the ultraviolet-visible-near-infrared (UV-Vis-NIR) region. For near-infrared applications, a spectrometer using an InGaAs detector can cover the $900nm$ to $1700nm$ range. Note that a wider wavelength range is not always better: with a fixed number of detector pixels, the wider the wavelength range, the larger the wavelength interval covered by a single pixel (i.e. the sampling interval), and the weaker the ability to capture spectral detail. Therefore, the choice of wavelength range should match the specific application requirements.

4.2 Optical Resolution

Optical resolution characterizes the spectrometer's ability to distinguish adjacent wavelength components, and is usually quantified by the full width at half maximum (FWHM). $2.5nm$ (FWHM) is a typical resolution level for compact fiber optic spectrometers.

Optical resolution is not a fixed value; it is affected jointly by the following factors:

  • Entrance slit width: the narrower the slit, the higher the resolution, but the lower the throughput;
  • Grating groove density: the higher the groove density, the greater the angular dispersion and the higher the resolution;
  • Optical system focal length: the longer the focal length, the greater the linear dispersion on the detector and the higher the resolution;
  • Detector pixel size: the smaller the pixels, the finer the sampling of spectral detail.

Note that the actual resolution is the convolution of the instrument function (determined by the slit width and optical aberrations) with the grating diffraction limit. When evaluating resolution, the narrow atomic emission lines of a low-pressure gas discharge lamp (such as a mercury-argon lamp) are usually used as a standard light source for actual measurement.

4.3 Signal-to-Noise Ratio

The signal-to-noise ratio (SNR) is a key indicator of a spectrometer's ability to extract signal from noise, defined as the ratio of signal strength to noise-floor strength. $1200:1$ is a typical signal-to-noise ratio level of current mainstream fiber optic spectrometers.

The signal-to-noise ratio directly affects the detection limit and quantitative accuracy of spectral measurements. A high signal-to-noise ratio means the instrument can reliably detect weaker signal variations, making it suitable for trace analysis and precise measurement of low-reflectance/low-transmittance samples. The main factors affecting the signal-to-noise ratio include:

  • Detector dark-current noise: strongly correlated with temperature, and effectively suppressed by detector cooling;
  • Readout noise: the inherent electronic noise of the charge-readout circuit;
  • Photon shot noise: the statistical fluctuation caused by the temporal randomness of photons arriving at the detector, which follows a Poisson distribution and whose root-mean-square value equals the square root of the mean signal - this is the lower limit of noise and cannot be eliminated through instrument design;
  • Light-source stability: noise introduced by light-source fluctuations.

4.4 Sensitivity

Sensitivity reflects the efficiency with which the spectrometer converts photons into digital counts, and is usually quantified by the number of counts produced per unit optical power and per unit integration time. A typical high-sensitivity fiber optic spectrometer can achieve $445,000\ counts/\mu W/ms$.

The higher this metric, the stronger the signal response the instrument produces under the same optical input conditions. High sensitivity is crucial for detecting weak light signals: it effectively shortens the integration time needed to reach the target signal-to-noise ratio and improves measurement throughput. Sensitivity is affected jointly by multiple stages including quantum efficiency, grating diffraction efficiency, optical-system transmittance and ADC gain.

4.5 Integration Time

Integration time refers to the time window during which the detector pixels accumulate photo-generated charge, analogous to the "exposure time" in a camera.

A typical commercial fiber optic spectrometer has an integration time range of $5.2ms$ to $60s$. The shortest integration time ($5.2ms$) is crucial for real-time monitoring of strong light signals or high-speed dynamic processes, avoiding detector saturation; the longest integration time ($60s$) makes it possible to measure extremely weak light signals (such as Raman scattering, fluorescence and bioluminescence). A wide integration-time adjustment range means the instrument can cover extreme dynamic-range scenarios from direct strong-source illumination to near-total darkness.

4.6 ADC Bit Depth and Sampling Rate

The ADC (Analog-to-Digital Converter) converts the analog voltage signal output by the detector into a digital gray value. A $16$-bit ADC is now standard on mainstream fiber optic spectrometers, corresponding to $2^{16} = 65536$ quantization levels, i.e. a dynamic range of $65536:1$ for a single pixel.

A sampling rate of $250kHz$ guarantees high-speed readout of a $1024$-pixel array. The higher the ADC bit depth, the higher the quantization precision of the signal and the finer the light-intensity changes that can be resolved. In actual use, the effective number of bits may be lower than the nominal value due to circuit noise, so the instrument's overall signal-to-noise ratio and dynamic range must be evaluated comprehensively rather than relying on the ADC bit depth alone.

4.7 Stray Light

Stray light refers to the undesired optical signal that reaches the detector at non-target wavelength positions, and is an important source of measurement error caused by light scattering inside the spectrometer.

Sources of stray light include: higher-order stray diffraction of the grating, non-ideal reflection and scattering at the surfaces of optical elements, and diffuse reflection from the inner walls of the mechanical structure. The direct consequence of stray light is reduced spectral linearity and dynamic range: when measuring highly absorbing samples, the absorbance value that should be read at a specific wavelength is seriously underestimated because of stray light - this is one of the main reasons the Lambert-Beer law deviates from linearity.

The main technical means for controlling stray light include: placing stray-light baffles in the optical path, treating the inner walls of the mechanical structure with absorptive coatings, selecting high-quality holographic gratings, optimizing the grating blaze-angle design, and using higher-order cutoff filters to eliminate order overlap.

5. Dark Noise Test Method

Dark noise is a basic test indicator for evaluating a spectrometer's signal-to-noise ratio and detection capability, used to characterize the noise-floor level generated by the detector itself in the complete absence of photon signal input. Dark noise consists of dark-current shot noise and readout noise, and is a key factor limiting the instrument's ability to detect weak signals.

The standard test procedure is as follows:

  1. Block the light path: completely seal the spectrometer's entrance slit or SMA905 interface with a metal light-blocking cap, ensuring no external photons reach the detector;
  2. Set the integration time: set the integration time to $100ms$, a commonly used reference integration time for dark-noise evaluation;
  3. Acquire two dark spectra: under the fully light-blocked conditions above, acquire two consecutive dark spectra $S_1(\lambda)$ and $S_2(\lambda)$;
  4. Calculate the difference spectrum: compute the difference spectrum of the two dark spectra $\Delta S(\lambda) = S_1(\lambda) - S_2(\lambda)$;
  5. Calculate the RMS value: compute the root mean square (RMS) value over all pixels of the difference spectrum:

$$ RMS_{dark} = \sqrt{\frac{1}{N}\sum_{i=1}^{N} \left[\Delta S(\lambda_i)\right]^2} $$

where $N$ is the total number of detector pixels. Because the difference spectrum of the two dark spectra removes the fixed dark-current bias and fixed pattern noise, the RMS value of the difference spectrum reflects purely random noise. Dividing by $\sqrt{2}$ yields the dark-noise RMS value of a single measurement.

Interpretation points:

  • The dark-noise RMS value increases with integration time, because dark-current accumulation is proportional to the integration time;
  • Cooling the detector (for example, lowering the CCD temperature to $-10^\circ C$ or below) can reduce dark noise to a few tenths of its value at room temperature;
  • The smaller the dark noise, the stronger the instrument's ability to detect extremely weak signals at long integration times, and the more suitable it is for low-light application scenarios such as Raman spectroscopy and fluorescence spectroscopy.

6. Communication Interfaces

Modern fiber optic spectrometers support high-speed data transmission to match their fast acquisition capability. Mainstream communication interfaces include:

  • USB 3.0: provides a theoretical bandwidth of up to $5Gbps$, meeting the real-time data-transmission needs of high-frame-rate continuous spectral acquisition, while powering through the USB bus simplifies system wiring;
  • Gigabit Ethernet (GigE): supports networked deployment over long distances (within $100m$), enabling remote control and data acquisition via the TCP/IP protocol, suitable for distributed multi-point monitoring scenarios at industrial sites.

The two interfaces complement each other, providing flexible options for precision laboratory measurements and industrial online integration.

7. How to Choose a Fiber Optic Spectrometer

Based on the in-depth analysis of the principles and specifications above, the following are the key considerations for selecting a fiber optic spectrometer:

1. Determine the wavelength range and resolution according to the application:

  • UV-Vis absorbance measurement (e.g. water-quality analysis) → choose $350nm$-$1100nm$; a resolution of about $2.5nm$ is usually sufficient;
  • Near-infrared composition analysis (e.g. grains, oils) → choose an InGaAs detector, $900nm$-$1700nm$;
  • Laser wavelength characterization and atomic emission line analysis → prefer a high-resolution configuration (e.g. a $1200\ grooves/mm$ grating paired with a narrow slit);
  • Raman spectroscopy → requires high sensitivity, low dark noise (a cooled detector) and cutoff filtering near the narrow excitation wavelength;

2. Evaluate detector performance:

  • For weak-signal detection → prefer a cooled back-illuminated CCD, paying attention to the dark-noise and sensitivity specifications;
  • For high-speed dynamic processes → pay attention to the minimum integration time and frame rate; a CMOS detector may be more advantageous;

3. Match communication and system-integration needs:

  • Laboratory benchtop use → the USB 3.0 interface is plug-and-play and the most convenient;
  • Industrial online integration → Gigabit Ethernet supports long-distance deployment and reliable industrial communication;

4. Pay attention to the stray-light level:

  • When measuring highly absorbing samples or over a wide dynamic range, stray-light suppression capability is crucial; pay attention to the stray-light test data provided by the manufacturer.

5. Consider the flexibility of the grating configuration:

  • If different measurement tasks may be involved in the future, it is recommended to choose a spectrometer model that supports grating replacement to protect the initial investment.

8. Summary

As a miniaturized spectral-analysis platform that highly integrates the diffractive dispersion optical path with an array detector, the fiber optic spectrometer - with its significant advantages of speed, portability and easy integration - has established an irreplaceable position in scientific research and industrial inspection.

Mastering its working principle - the complete optical chain from the slit, collimation and grating dispersion to detector signal conversion - and understanding the physical nature and mutual constraints of its key technical specifications are the basic prerequisites for scientific selection and correct use of a fiber optic spectrometer. Whether facing the high signal-to-noise ratio demands of precision laboratory measurements or the fast online-inspection demands of industrial sites, one should conduct a systematic evaluation across multiple dimensions including wavelength range, resolution, signal-to-noise ratio, sensitivity, detector type and communication interface, in order to select the instrument configuration that best fits the actual application scenario.

This article was compiled by Pynect.