{
  "id": "fourier-transform-infrared-spectroscopy",
  "title": "Fourier-Transform Infrared Spectroscopy",
  "lang": "en",
  "source_lang": "en",
  "summary": "Fourier-transform infrared spectroscopy measures infrared spectra by mathematically transforming interferograms, enabling chemical identification, structural analysis, and quantitative measurement.",
  "category": "science",
  "subcategory": "chemistry",
  "type": "concept",
  "aliases": [
    "FTIR spectroscopy",
    "FT-IR spectroscopy",
    "Fourier transform infrared spectrometry"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/fourier-transform-infrared-spectroscopy/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/fourier-transform-infrared-spectroscopy/",
      "json": "https://aiwiki.page/zh/fourier-transform-infrared-spectroscopy.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Fourier-transform infrared spectroscopy (FTIR spectroscopy) is a form of infrared spectroscopy in which an instrument records the interference of infrared radiation and uses a Fourier transform to recover its spectral distribution. Rather than measuring successive narrow wavelength intervals, an FTIR spectrometer encodes a broad spectral range in an interferogram. Comparing measurements made with and without a sample reveals how the sample absorbs, transmits, or reflects infrared radiation.\n\nThe method is widely used in analytical chemistry, materials science, environmental measurement, and biological research because infrared bands carry information about chemical bonds, molecular structure, and local environments. A single spectrum may identify a polymer, distinguish crystalline forms of a pharmaceutical compound, reveal changes in protein secondary structure, or quantify gases in a mixture. Its usefulness depends not only on the Fourier-transform measurement principle but also on sample preparation, optical geometry, spectral processing, and appropriate interpretation."
    },
    {
      "heading": "Molecular basis and spectral regions",
      "text": "Infrared absorption occurs when incident radiation couples to a transition between allowed energy states of matter. In molecules, the most commonly measured transitions involve vibrations, often accompanied by rotational structure in gases. The transition energy satisfies \\(\\Delta E=h\\nu\\), where \\(h\\) is the Planck constant and \\(\\nu\\) is frequency. A vibration is infrared-active when its motion changes the molecular electric dipole moment. Thus, the existence of a vibrating chemical bond does not by itself guarantee an absorption band: the symmetry and collective motion of the molecule determine whether the radiation can couple to that vibration.\n\nSpectra are usually plotted against wavenumber, \\(\\tilde{\\nu}=1/\\lambda\\), conventionally expressed in reciprocal centimetres. The mid-infrared region is commonly taken as approximately 4,000–400 cm\\(^{-1}\\), corresponding to wavelengths of 2.5–25 micrometres. It contains many fundamental molecular vibrations. The near-infrared region, approximately 12,800–4,000 cm\\(^{-1}\\), is dominated by overtones and combination bands, while the far-infrared region below about 400 cm\\(^{-1}\\) includes low-frequency molecular motions, lattice vibrations, and some rotational transitions. Exact regional boundaries vary slightly between disciplines and instrument specifications.\n\nA simple diatomic model illustrates why band positions depend on both bonding and atomic mass:\n\n\\[\\tilde{\\nu}=\\frac{1}{2\\pi c}\\sqrt{\\frac{k}{\\mu}},\\]\n\nwhere \\(c\\) is the speed of light, \\(k\\) is an effective force constant, and \\(\\mu\\) is the reduced mass. Stronger bonds generally vibrate at higher wavenumbers, whereas substitution by a heavier isotope usually lowers a vibration’s frequency. Real polyatomic molecules have normal modes involving several atoms. A nonlinear molecule with \\(N\\) atoms has \\(3N-6\\) vibrational normal modes; a linear molecule has \\(3N-5\\). Anharmonicity, coupling, and resonance complicate the simple harmonic picture.\n\nCharacteristic group frequencies provide a practical starting point for interpretation. Carbonyl stretching bands often occur within roughly 1,650–1,800 cm\\(^{-1}\\), while O–H stretching commonly appears between about 3,200 and 3,700 cm\\(^{-1}\\), with substantial shifts and broadening caused by hydrogen bonding. The region below approximately 1,500 cm\\(^{-1}\\) is often called the fingerprint region because its many coupled vibrations can distinguish closely related compounds. Band positions and shapes also depend on phase, temperature, molecular interactions, and crystal structure. Consequently, reliable identification considers the overall pattern rather than treating an isolated band as conclusive evidence for a substance."
    },
    {
      "heading": "Interferometric measurement and Fourier reconstruction",
      "text": "The defining optical component of a conventional FTIR instrument is a Michelson interferometer. A beamsplitter divides incoming radiation between two paths, commonly ending at a fixed mirror and a moving mirror. After reflection, the beams recombine. Their relative phase depends on the optical path difference, so their combined intensity varies as the moving mirror changes position. For a simple arrangement in air, moving one mirror by a distance \\(d\\) changes the optical path difference by approximately \\(2d\\). The resulting intensity record is the interferogram.\n\nFor monochromatic radiation, the modulated signal varies sinusoidally with optical path difference. Broadband radiation produces a superposition of many such oscillations. After removing the unmodulated component, an ideal interferogram may be represented as\n\n\\[I(\\delta)=\\int_{0}^{\\infty}S(\\tilde{\\nu}) \\cos(2\\pi\\tilde{\\nu}\\delta)\\,d\\tilde{\\nu},\\]\n\nwhere \\(\\delta\\) is optical path difference and \\(S(\\tilde{\\nu})\\) is the spectral signal reaching the detector, including the source spectrum and the instrument’s wavelength-dependent response. Fourier transformation separates the encoded oscillations into their constituent wavenumbers. Practical instruments use complex-valued transformations and phase correction because measured interferograms are not perfectly symmetric or ideally sampled.\n\nNear zero optical path difference, contributions from a broad range of wavenumbers reinforce one another, producing the strong central feature called the centerburst. Farther from zero, broadband contributions increasingly cancel, although narrow spectral features can maintain oscillations over longer distances. This relationship explains why resolving closely spaced bands requires measurement to a larger maximum optical path difference. Spectral resolution scales approximately as the reciprocal of that maximum difference, although the precise numerical relationship depends on the resolution convention and processing. A routine setting of 4 cm\\(^{-1}\\) is suitable for many condensed-phase identification tasks, whereas resolving gas-phase rotational structure can require much finer resolution.\n\nThe Fourier-transform approach has several classical advantages over scanning dispersive spectrometers. The Fellgett, or multiplex, advantage arises because many spectral elements are recorded together; it is most beneficial when detector noise dominates. It is reduced or absent when photon noise dominates, and radiation from one strong spectral region can contribute noise elsewhere. The Jacquinot, or throughput, advantage reflects the absence of the narrow entrance and exit slits used in many dispersive instruments, although apertures still constrain performance. The Connes advantage concerns accurate wavenumber referencing through interferometer metrology. None of these advantages eliminates detector limitations, sample effects, or the need for independent performance verification."
    },
    {
      "heading": "Instrument architecture and optical components",
      "text": "A laboratory FTIR spectrometer typically contains a broadband source, interferometer, sample compartment, detector, reference-laser system, and electronics for acquisition and processing. Mid-infrared instruments commonly use a heated silicon carbide source, often called a Globar, which emits a broad thermal spectrum. Tungsten-halogen sources are common in the near-infrared, and mercury discharge sources have been used in the far-infrared. Source brightness, stability, and spectral distribution affect achievable signal-to-noise ratios. Specialized instruments may instead use synchrotron radiation, particularly when high brightness is needed for small sampling areas.\n\nThe beamsplitter must divide and recombine radiation efficiently over the intended spectral range. Potassium bromide substrates with suitable coatings are widely used in the mid-infrared, while calcium fluoride, quartz, or other materials serve different ranges. Thin polymer-film beamsplitters are used in parts of the far-infrared. No single combination of source, beamsplitter, windows, and detector provides optimum performance across the entire infrared spectrum. Changing the spectral region may therefore require exchanging several components. Hygroscopic salts also require protection from moisture, since surface deterioration reduces transmission and increases unwanted scattering.\n\nDetector choice strongly affects speed and sensitivity. Deuterated triglycine sulfate and related pyroelectric detectors operate near room temperature and are common in general-purpose instruments. They respond to changes in absorbed radiant power and therefore suit the modulated radiation produced by a scanning interferometer. Mercury cadmium telluride detectors provide high sensitivity and faster response over composition-dependent wavelength ranges; many laboratory versions require cooling. InGaAs detectors are commonly used for near-infrared measurements. Detector linearity and dynamic range are important because the centerburst can be much stronger than the interferogram wings. Saturation near zero path difference can corrupt a large part of the reconstructed spectrum.\n\nA reference laser, traditionally a helium–neon laser, provides an accurate measure of mirror motion. Its interference fringes support sampling or resampling at known optical path intervals. This stabilizes the wavenumber scale but does not make every measurement automatically traceable: beam alignment, apertures, optical dispersion, and geometry can introduce residual errors. Instruments also require control of atmospheric absorption. Internal purging with dry air or nitrogen reduces interference from water vapour and carbon dioxide, which have strong infrared bands. External accessories, long optical paths, and imperfectly sealed compartments can remain sensitive to laboratory humidity and changes between background and sample measurements."
    },
    {
      "heading": "Sampling geometries and specimen preparation",
      "text": "Transmission is the most direct sampling geometry. Radiation passes through a specimen, and the transmitted intensity is compared with a background measurement. Liquids may be held between infrared-transparent windows with a spacer defining the path length. Strongly absorbing liquids often require paths of only a few to tens of micrometres in portions of the mid-infrared. Polymer films can sometimes be measured directly, while powders may be dispersed in potassium bromide and pressed into pellets. Preparation can alter hydration, orientation, or crystal form, and particles that are too large can cause substantial scattering. Window compatibility with the sample is therefore as important as optical transparency.\n\nAttenuated total reflectance (ATR) is particularly common because it permits direct measurement of many solids, pastes, and liquids with little preparation. Radiation undergoes total internal reflection inside a high-index crystal, such as diamond, zinc selenide, or germanium, while an evanescent field interacts with material touching its surface. For a weakly absorbing sample, a commonly used penetration-depth expression is\n\n\\[d_p=\\frac{\\lambda} {2\\pi n_1\\sqrt{\\sin^2\\theta-(n_2/n_1)^2}},\\]\n\nwhere \\(n_1\\) and \\(n_2\\) are the refractive indices of crystal and sample, and \\(\\theta\\) is the incidence angle inside the crystal. This depth describes exponential field-amplitude decay rather than a sharply bounded sampled layer.\n\nATR penetration depths are often of the order of a fraction of a micrometre to several micrometres, depending on wavelength, crystal, angle, and sample refractive index. Longer wavelengths generally penetrate farther, so relative band intensities differ from those of an equivalent transmission spectrum. Good physical contact is crucial for rigid or rough solids, and applied force can influence reproducibility. ATR correction algorithms can make spectra more comparable with transmission data, but they depend on assumptions about optical constants and contact. A thick specimen measured by ATR is still sampled principally near its surface, not uniformly through its bulk.\n\nReflection methods address other specimen types. Specular reflection can characterize smooth surfaces and thin films, but measured features depend on both absorption and refractive-index dispersion. Diffuse-reflectance FTIR collects radiation scattered by powders and rough materials; interpretation may use the Kubelka–Munk model when its assumptions are sufficiently satisfied. Grazing-incidence reflection–absorption measurements enhance sensitivity to thin layers on suitable reflective substrates. Gas analysis instead commonly uses sealed or flowing cells, including multipass cells that extend the optical path to metres or more. Photoacoustic detection provides another option by measuring pressure changes generated when absorbed, modulated radiation heats a sample."
    },
    {
      "heading": "From interferograms to reliable spectra",
      "text": "A normal absorption measurement requires a background spectrum and a sample spectrum. If their spectral intensities are \\(I_0(\\tilde{\\nu})\\) and \\(I(\\tilde{\\nu})\\), respectively, the transmittance is\n\n\\[T(\\tilde{\\nu})=\\frac{I(\\tilde{\\nu})}{I_0(\\tilde{\\nu})},\\]\n\nand the decadic absorbance is\n\n\\[A(\\tilde{\\nu})=-\\log_{10}T(\\tilde{\\nu}).\\]\n\nThe ratio removes much of the source distribution and instrument response, provided these remain stable between measurements. It does not automatically remove changing atmospheric absorption, contamination, or differences in accessory alignment. A background appropriate for an empty transmission holder may not be appropriate for an ATR accessory or a reflection experiment.\n\nBefore transformation, the interferogram is usually subjected to several processing steps. Finite mirror travel truncates the signal, which produces an instrumental line shape with sidelobes. Apodization multiplies the interferogram by a weighting function that reduces these sidelobes at the expense of broadening spectral features. Phase correction compensates for optical and electronic phase shifts. Zero filling adds numerical points before the fast Fourier transform, giving a smoother sampled representation of the spectrum. It does not recover information beyond the measured path difference and therefore does not improve true resolving power. The selected processing settings matter when comparing narrow lines or spectra acquired on different instruments.\n\nSignal averaging improves precision when successive scans contain independent random noise. Under those conditions, signal-to-noise ratio increases approximately as the square root of the number of scans: quadrupling the scans yields roughly a twofold improvement. Drift, sample changes, and atmospheric variation do not follow this rule and may become more important during long acquisitions. Higher resolution also commonly requires longer scans and distributes information among narrower spectral intervals. The practical acquisition choice is therefore a trade-off among resolution, sensitivity, measurement time, and the stability of the specimen.\n\nInstrument qualification and calibration address different aspects of measurement reliability. Polystyrene films with characterized absorption features are widely used to check wavenumber performance, while other tests assess resolution, repeatability, photometric response, and noise. Saturated bands should not be treated as reliable absorbance measurements, because transmitted intensity near the noise floor produces unstable logarithms. Baseline subtraction, atmospheric compensation, smoothing, and derivative processing can aid analysis but can also introduce or exaggerate features. Documenting such operations, along with sampling geometry and acquisition settings, is essential for assessing measurement uncertainty and for reproducing a result."
    },
    {
      "heading": "Chemical identification and quantitative analysis",
      "text": "Qualitative interpretation combines functional-group analysis with comparison against reference spectra. A carbonyl band may establish the presence of a particular type of bonding environment, but its frequency alone rarely identifies an entire compound. The fingerprint region, relative intensities, band shapes, and absence of expected features all contribute. Search libraries are valuable because they permit rapid comparison against thousands of reference materials. Their matches remain conditional on sampling mode, spectral range, resolution, and specimen state. An ATR spectrum of a filled commercial polymer, for example, need not closely resemble the transmission spectrum of the pure polymer even when both contain the same principal material.\n\nMixtures present a more difficult inverse problem. Absorption bands may overlap, and the most abundant component can mask minor constituents. Difference spectra can reveal changes between related specimens, but small errors in path length, contact, or baseline can create derivative-like residuals that resemble chemical differences. Assignments become stronger when supported by complementary techniques such as Raman spectroscopy, mass spectrometry, or nuclear magnetic resonance. Raman and infrared spectra are especially complementary because their selection rules differ: a vibration’s Raman activity depends on changes in polarizability rather than changes in dipole moment.\n\nQuantitative transmission analysis commonly starts from the Beer–Lambert law:\n\n\\[A(\\tilde{\\nu})=\\varepsilon(\\tilde{\\nu})bc,\\]\n\nwhere \\(\\varepsilon\\) is the molar decadic absorption coefficient, \\(b\\) is path length, and \\(c\\) is concentration. A consistent set of units is required. For noninteracting components, their absorbances are approximately additive, enabling multicomponent fitting. Deviations arise from chemical equilibria, concentration-dependent interactions, scattering, stray radiation, insufficient resolution, and detector nonlinearity. Gas measurements must also account for temperature and pressure, which affect number density, line strength, and broadening. ATR quantification generally requires an appropriate calibration model rather than substitution of a fixed geometric path length into the transmission equation.\n\nChemometrics extends analysis beyond individual peaks. Principal-component analysis can reveal dominant patterns and outliers, while partial least-squares regression can relate many spectral variables to independently measured concentrations or properties. These methods are useful for overlapping bands, complex formulations, and process streams. Their performance depends on calibration samples that represent the intended range of compositions and physical conditions. Independent validation is necessary because a model may learn correlations with moisture, particle size, or batch history rather than the target constituent. A low fitting error on training data does not establish chemical specificity or reliable prediction outside the calibration domain."
    },
    {
      "heading": "Chemical, industrial, and environmental applications",
      "text": "In polymer analysis, FTIR identifies resin families, additives, oxidation products, and contamination. Polyethylene and polypropylene show distinctive hydrocarbon patterns, while polyesters and polyamides introduce prominent carbonyl-related bands. Changes in carbonyl absorption are often used to follow oxidative ageing, although a reported carbonyl index depends on the integration limits, reference band, and sampling method. Surface-sensitive ATR measurements can reveal weathered layers that differ from the interior of a specimen. In manufacturing, similar measurements support incoming-material identification, batch comparison, coating assessment, and investigation of unexpected deposits or product failures.\n\nPharmaceutical and fine-chemical applications include identity testing, monitoring of synthesis, and characterization of solid forms. Different crystalline arrangements of the same molecule can alter band positions and splitting because intermolecular interactions and symmetry change. Interpretation may therefore be combined with X-ray diffraction and thermal methods. In situ infrared probes can follow disappearance of reactants and formation of products during a chemical reaction. Measurements of reaction rates require distinguishing true concentration changes from shifts caused by temperature, solvent composition, or changing optical contact. Coupling FTIR with thermogravimetric analysis identifies gases released as a material is heated.\n\nGas-phase FTIR is widely used because many atmospheric and industrial gases have diagnostic vibrational spectra. Examples include carbon monoxide, carbon dioxide, methane, ammonia, and numerous volatile organic compounds. Multipass cells increase sensitivity by extending absorption path length, while heated sampling lines can reduce condensation losses for suitable applications. Open-path arrangements measure radiation over an outdoor path and can detect several gases simultaneously. Their results generally represent path-integrated or path-averaged quantities rather than a uniquely located point concentration. Strong water-vapour absorption, overlapping bands, changing temperatures, and background conditions require explicit treatment.\n\nEnvironmental particle analysis increasingly uses infrared microscopy or imaging to identify microplastics on filters and other substrates. Polymer assignment depends on particle size, weathering, pigments, substrate absorption, and the reference library. The smallest reliably identified particle is determined by optical resolution and signal quality, not merely by detector pixel spacing. In cultural heritage research, FTIR can characterize binding media, varnishes, mineral pigments, and degradation products, sometimes using microscopic samples or reflection measurements. In geology, infrared bands provide information about carbonates, silicates, clays, and hydroxyl-bearing phases. Across these applications, the method measures chemical and structural signatures; provenance or historical interpretation requires additional evidence."
    },
    {
      "heading": "Biological samples, microscopy, and spatially resolved analysis",
      "text": "Biological infrared spectra contain contributions from proteins, lipids, carbohydrates, nucleic acids, and water. The amide I region, approximately 1,600–1,700 cm\\(^{-1}\\), is dominated by peptide carbonyl stretching and is widely used to investigate protein secondary structure. Amide II, commonly near 1,550 cm\\(^{-1}\\), includes substantial N–H bending and C–N stretching character. Changes in these envelopes can accompany protein folding, aggregation, or interactions with other molecules. Quantitative decomposition into structural components is model-dependent because bands overlap and their positions depend on environment, hydration, and coupling between peptide groups.\n\nWater creates a central experimental difficulty: it is biologically essential but strongly absorbs infrared radiation, including in regions relevant to protein analysis. Thin liquid layers, ATR geometries, or substitution with heavy water can reduce particular interferences, although isotopic substitution also changes vibrational frequencies and may affect the system under study. Drying specimens simplifies some optical measurements but changes hydration and can alter structure. Tissue preparation introduces further variables, including fixation, embedding media, section thickness, and residual paraffin. These effects must be separated from the biochemical differences being investigated.\n\nAn FTIR microscope combines spectroscopic measurement with spatial selection or imaging. A single-element detector can collect spectra point by point, while a focal-plane array records signals from many detector elements during each interferometer scan. Transformation produces a spectrum at each spatial position, yielding a hyperspectral data cube. Chemical maps may display an integrated band, a band ratio, or a multivariate classification. Because spatial resolution is limited by diffraction, ordinary far-field infrared microscopy cannot resolve arbitrary submicrometre structures. At a wavenumber of 1,000 cm\\(^{-1}\\), the wavelength is 10 micrometres, so optical resolution is generally on a micrometre scale governed by wavelength and numerical aperture.\n\nBiological FTIR imaging has been used to study tissue composition, microorganisms, and cellular responses. Spectral classification can detect reproducible differences among experimental groups, but such differences do not automatically establish a unique molecular mechanism or validated clinical diagnosis. Cohort composition, specimen handling, instrument variation, and statistical validation affect interpretation. Near-field infrared and photothermal methods can achieve much finer spatial resolution than conventional far-field FTIR microscopy, but they use different detection principles and do not all constitute Fourier-transform spectroscopy. Distinguishing these techniques prevents spatial-resolution claims from being transferred incorrectly between methods."
    },
    {
      "heading": "Historical development and advanced measurement modes",
      "text": "FTIR combines nineteenth-century interferometry with twentieth-century detectors, electronics, and computing. Albert A. Michelson developed interferometric methods capable of resolving spectral structure, establishing an important optical foundation. Practical Fourier-transform spectroscopy expanded in the mid-twentieth century through work associated with researchers including Peter Fellgett, Pierre Jacquinot, and James Connes. Their names remain linked to the multiplex, throughput, and wavenumber-accuracy advantages. Early applications were particularly attractive where weak signals and high resolution made conventional dispersive measurements difficult.\n\nComputational cost initially constrained routine use because every interferogram had to be transformed numerically. The efficient fast Fourier transform algorithm published by James Cooley and John Tukey in 1965 helped make such processing more practical, alongside broader advances in digital electronics and computers. Commercial instruments became increasingly accessible during the following decades, and FTIR largely displaced scanning dispersive infrared instruments in many laboratories. Later developments included compact interferometers, more stable optical assemblies, improved detectors, automated accessories, and extensive spectral libraries. The resulting instruments range from high-resolution research systems to portable devices for field identification.\n\nConventional rapid-scan FTIR records an interferogram while a mirror moves continuously. Because a complete interferogram takes a finite time, a sample that changes during the scan may violate the assumption that all optical path differences describe the same state. Faster scanning can follow sufficiently slow processes, but temporal resolution remains linked to mirror travel, detector response, and spectral resolution. Step-scan FTIR instead holds the mirror at successive path differences and records a time-dependent response at each position. Reconstructing a spectrum at each delay requires a repeatable, synchronized process, making the method suitable for many periodically excited or reproducibly triggered experiments rather than arbitrary one-time events.\n\nOther advanced arrangements adapt Fourier-transform measurement to demanding environments. Polarization-dependent measurements reveal molecular orientation in stretched polymers, films, and ordered assemblies. Infrared reflection measurements under controlled gas flow can follow adsorbed species on catalysts, although intense gas-phase bands and changing backgrounds complicate interpretation. Emission FTIR measures radiation generated by a specimen rather than relying solely on an external transmitted beam; quantitative work must then account for temperature, emissivity, and instrument response. High-resolution atmospheric and astronomical instruments likewise exploit Fourier-transform methods, but require careful control of line shapes, radiometric calibration, and environmental conditions specific to their observations."
    }
  ],
  "links": [
    "infrared-spectroscopy",
    "fourier-transform",
    "analytical-chemistry",
    "planck-constant",
    "electric-dipole",
    "chemical-bond",
    "speed-of-light",
    "isotope",
    "hydrogen-bond",
    "temperature",
    "michelson-interferometer",
    "synchrotron-radiation",
    "water",
    "carbon-dioxide",
    "scattering",
    "total-internal-reflection",
    "refractive-index",
    "fast-fourier-transform",
    "calibration",
    "measurement-uncertainty",
    "raman-spectroscopy",
    "mass-spectrometry",
    "nuclear-magnetic-resonance",
    "beer-lambert-law",
    "pressure",
    "chemical-reaction",
    "methane",
    "cultural-heritage",
    "geology",
    "protein",
    "protein-folding",
    "diffraction",
    "numerical-aperture",
    "diagnosis"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}