Radiation Sources for Spectroscopy

Modern UV-Vis spectrophotometers almost universally pair a tungsten-halogen lamp (visible/NIR, 320–2500 nm typical range) with a deuterium arc lamp (UV, 190–370 nm), automatically switching between them near 350–370 nm depending on instrument design — a design decision that flows directly from the Planck-curve and molecular-continuum physics described here, and is foundational to instruments treated more fully under UV-Vis Spectrophotometer Instrumentation. Clinical chemistry analyzers, environmental water-quality monitors, and pharmaceutical dissolution-testing instruments all depend on this dual-source design remaining stable to well under 1% drift over an analytical run, since any source fluctuation propagates directly into absorbance error via Beer's law. Atomic absorption spectrometers used in environmental labs (heavy-metal testing in water, soil, and food), clinical toxicology (blood lead, mercury), and metallurgical quality control still rely on element-specific hollow cathode lamps (or, increasingly, electrodeless discharge lamps for volatile elements like arsenic and selenium, which give higher intensity for those species). Multi-element hollow cathode lamps, containing alloyed cathodes of several metals, let a single lamp serve panels of related analytes, trading off intensity per element for instrument throughput. In the infrared, Fourier-transform infrared (FTIR) spectrometers used across pharmaceutical identity testing, polymer characterization, and forensic trace analysis rely on globars almost universally, their broad, smoothly varying continuum being essential to the multiplexed (Michelson interferometer) measurement scheme central to FTIR — a connection developed further under Infrared Spectroscopy for Quantitation. Portable and process-analytical IR instruments increasingly substitute stabilized diode or quantum-cascade laser sources for specific narrow-band applications, but the thermal globar remains the default for broadband laboratory-grade mid-IR work because of its simplicity, long lifetime, and genuinely continuous output across the whole fingerprint region. Synchrotron infrared beamlines, used in advanced materials and biological microspectroscopy, push this same thermal-continuum idea to its brightness limit, delivering globar-like spectral coverage with orders of magnitude higher spatial brightness for microscale chemical imaging.

Ask the tutor: How does the choice of continuum vs. line source change the shape and linearity of a Beer's-law calibration curve in practice?

Every emission source in this article is either a thermal radiator or an atomic/molecular discharge, and the distinction is the single most useful organizing fact in the whole topic. A thermal radiator is a hot, dense object — a solid filament, a hot gas at high pressure, a ceramic rod — in which so many atoms are colliding and radiating simultaneously that their individual quantum transitions blur into a continuous distribution of photon energies. The emitted spectrum depends almost entirely on temperature (and a material-dependent efficiency factor called emissivity), not on which element is glowing. This is why a tungsten filament, a carbon arc, and the sun all produce recognizably similar-shaped continuous spectra: they are all, to a good approximation, obeying the same universal law — Planck's law — just at different temperatures. Tungsten (2870–3000 K) glows white-hot and is used across the visible and near-IR; deuterium technically breaks this pattern (its UV continuum comes from a molecular dissociation process, not blackbody radiation, though it is grouped with continuum sources for practical purposes); globars and Nernst glowers (1200–1500 K) glow a dull orange-red, radiating mostly in the infrared where Planck's law predicts the bulk of their output to fall at those cooler temperatures. An atomic discharge source is different in kind. Here, individual gas-phase atoms of one specific element are excited — by an electrical discharge, an RF field, or sputtering bombardment — into higher electronic states, and they then relax by emitting photons at their own characteristic, quantized transition energies. The hollow cathode lamp (HCL) is the workhorse example: a cathode made of (or lined with) the element of interest sits in a low-pressure noble gas (Ne or Ar). A few hundred volts ionizes the fill gas; the resulting cations bombard the cathode, physically sputtering off metal atoms (a process called cathodic sputtering); those free atoms are then excited by further collisions in the surrounding plasma "negative glow," and they emit sharp lines at wavelengths intrinsic to that element's electronic structure — the same wavelengths, not coincidentally, that the same element absorbs in atomic absorption spectroscopy. This is why AAS requires a different lamp for essentially every element: the source itself does the wavelength selection that a monochromator does for continuum-source techniques. The width of these emission lines matters enormously. Because the cathode is cooled (relative to a flame or plasma) and operated at low pressure, Doppler broadening and pressure (collisional) broadening are both suppressed, giving hollow cathode lines with widths on the order of 0.001–0.005 nm — narrower than the absorption line profile of the same atoms in a hotter flame or furnace (typically 0.002–0.01 nm). This narrow-line-into-broader-line geometry is precisely what makes Beer's law behave linearly in atomic absorption, a point developed further under measurement techniques for that method.

The spectral radiance of an ideal thermal emitter (a blackbody) as a function of wavelength and absolute temperature is Planck's law:

where:

• $M_\lambda$ = spectral radiant exitance, the power emitted per unit area per unit wavelength (W·m⁻³, usually reported as W·m⁻²·nm⁻¹)
• $h$ = Planck's constant, $6.626\times10^{-34}\ \text{J}\,\cdot\,\text{s}$
• $c$ = speed of light, $2.998\times10^{8}\ \text{m/s}$
• $k_B$ = Boltzmann's constant, $1.381\times10^{-23}\ \text{J/K}$
• $\lambda$ = wavelength (m)
• $T$ = absolute temperature (K)

Reading: at any fixed temperature, $M_\lambda$ rises steeply from zero at short wavelengths, peaks, and falls off more gently at long wavelengths — the classic asymmetric blackbody hump. Raising $T$ raises the whole curve and pushes the peak to shorter wavelength. A real source (tungsten filament, globar) is not a perfect blackbody; its true output is $\varepsilon(\lambda)\,M_\lambda(\lambda,T)$, where the emissivity $\varepsilon(\lambda) \le 1$ is a material-specific correction (tungsten's $\varepsilon$ is roughly 0.4–0.5 in the visible, itself wavelength-dependent, which is why tungsten lamps deviate somewhat from ideal blackbody shape). Integrating Planck's law over all wavelengths gives the total radiant exitance, the Stefan–Boltzmann law:

where:

• $M$ = total power radiated per unit area (W/m²)
• $\sigma$ = Stefan–Boltzmann constant, $5.670\times10^{-8}\ \text{W}\,\cdot\,\text{m}^{-2}\text{K}^{-4}$
• $\varepsilon$ = emissivity (dimensionless, $0<\varepsilon\le 1$; $\varepsilon = 1$ for an ideal blackbody)
• $T$ = absolute temperature (K)

Reading: total output climbs with the fourth power of temperature — a 10% increase in filament temperature very nearly doubles the total radiated power, which is why lamp lifetime and output are both exquisitely sensitive to the operating voltage. Differentiating Planck's law with respect to $\lambda$ and setting the derivative to zero yields Wien's displacement law, giving the wavelength of peak emission:

where:

• $\lambda_{\max}$ = wavelength of maximum spectral radiance (m or μm)
• $T$ = absolute temperature (K)
• $b$ = Wien's displacement constant

Reading: this is the single most useful "back of the envelope" relation for choosing a source — it converts a filament or plasma temperature directly into the wavelength region where that source's output is concentrated, letting you sanity-check whether a given lamp can plausibly cover a given spectral range.

Example 1 (Foundational): Where does a tungsten lamp's peak fall, and is it useful for UV-Vis work? A tungsten filament lamp used in a UV-Vis spectrophotometer's visible channel operates at $T = 2870\ \text{K}$ (a standard rated operating temperature for long-life tungsten-halogen lamps). Find $\lambda_{\max}$ and assess its suitability for measurements across 350–800 nm. Using Wien's law,

So the peak of tungsten emission falls at 1010 nm — solidly in the near infrared, well outside the visible range. This is not a flaw: Planck's curve is broad, and at 2870 K there is still substantial output tailing back through the visible (400–700 nm) and even into the near-UV edge (down to about 350 nm), just at intensities well below the peak. This is exactly why tungsten lamps are rated for use from 350–2500 nm despite peaking near 1000 nm — the usable range is much broader than the peak location alone would suggest, but it also explains why tungsten output falls off sharply below about 350 nm: shorter wavelengths lie increasingly far down the steep short-wavelength flank of the Planck curve, where $M_\lambda$ drops approximately exponentially. A quick check: at 350 nm, $x = hc/(\lambda k_BT)$ evaluates to

and since $M_\lambda \propto \lambda^{-5}(e^x-1)^{-1}$ with $e^{14.3}\approx 1.6\times10^6$, output at 350 nm is suppressed by roughly six orders of magnitude relative to the exponential factor alone (partially offset by the $\lambda^{-5}$ term), confirming that tungsten is a poor deep-UV source — which is exactly why instruments pair a tungsten lamp for the visible/NIR with a deuterium lamp for the UV below about 360 nm.

Example 2 (Applied): Comparing hollow cathode line width to atomic absorption line width in a flame — does Beer's law hold? An AAS instrument measures chromium at 357.9 nm using a Cr hollow cathode lamp. The HCL emission line has a Doppler-limited width (full width at half maximum, FWHM) given approximately by

where $m$ is the mass of the emitting atom and $T$ is the effective temperature of atoms in the cathode's negative glow discharge, typically $T_{\text{HCL}} \approx 400\ \text{K}$ (cooler than a flame because the cathode is often water- or air-cooled and operated at low current). The same formula applies to the absorbing Cr atoms in an air–acetylene flame at $T_{\text{flame}} \approx 2500\ \text{K}$. For chromium, $m = 52.0\ \text{u} = 8.633\times10^{-26}\ \text{kg}$, $\lambda = 357.9\ \text{nm}$, $c = 2.998\times10^8\ \text{m/s}$, $k_B = 1.381\times10^{-23}\ \text{J/K}$. HCL linewidth at 400 K: First compute the term inside the square root:

Numerator: $8 \times 0.6931 \times 1.381\times10^{-23} \times 400 = 3.062\times10^{-20}$. Denominator: $8.633\times10^{-26} \times 8.988\times10^{16} = 7.760\times10^{-9}$. Ratio: $3.062\times10^{-20}/7.760\times10^{-9} = 3.947\times10^{-12}$. Square root: $1.987\times10^{-6}$.

Flame linewidth at 2500 K: since $\Delta\lambda_D \propto \sqrt{T}$, scale by $\sqrt{2500/400} = \sqrt{6.25} = 2.500$:

(Real flame linewidths are somewhat larger still, roughly 0.002–0.005 nm, because pressure/collisional broadening adds to the Doppler contribution — the calculation above isolates the Doppler piece alone to show the scaling.) The result: the HCL emission line (≈0.0007 nm) is roughly 2.5 times narrower than the flame absorption line it probes (≈0.0018 nm, Doppler-only estimate; wider still with collisional broadening included). This is the physical basis of atomic absorption spectroscopy — the narrow source line sits well inside the broader, flatter-topped absorption profile, so the effective absorptivity the instrument sees is nearly constant across the narrow source linewidth, and Beer's law $A = \varepsilon b c$ behaves linearly over a useful concentration range. If the source line were instead broader than the absorption line (as would happen if a continuum source were used directly with a low-resolution monochromator), the absorbance would be underestimated and calibration curves would bend downward at higher concentrations — a classic instrumental deviation from Beer's law.

Example 3 (Exam/competition level): Designing an FTIR source budget — how much globar power reaches the detector, and what temperature is needed to hit a target radiance at 6 μm? A globar (silicon carbide rod) is operated at $T = 1300\ \text{K}$ with emissivity $\varepsilon \approx 0.85$ (a typical value for oxidized SiC in the mid-IR). An FTIR method requires a minimum spectral radiance of $M_\lambda \ge 1.2\times10^{3}\ \text{W}\,\cdot\,\text{m}^{-2}\mu\text{m}^{-1}$ at $\lambda = 6\ \mu\text{m}$ (a wavelength important for carbonyl C=O stretches). Determine whether the 1300 K globar meets this requirement, and if not, find the minimum temperature that does. Step 1 — evaluate Planck's law at 1300 K, 6 μm. Compute $x = hc/(\lambda k_BT)$:

Then $e^x = e^{1.844} = 6.323$, so $e^x - 1 = 5.323$. Step 2 — compute the prefactor $2\pi hc^2/\lambda^5$. Using $\lambda = 6\times10^{-6}\ \text{m}$:

Step 3 — combine.

Converting to per-μm units (1 m⁻¹ = $10^{-6}$ μm⁻¹):

Applying emissivity $\varepsilon = 0.85$:

This vastly exceeds the required $1.2\times10^{3}\ \text{W}\,\cdot\,\text{m}^{-2}\mu\text{m}^{-1}$ — the 1300 K globar meets the requirement with enormous margin (by a factor of roughly 640). This makes physical sense: 6 μm is close to the globar's Wien peak ($\lambda_{\max} = 2898/1300 = 2.23\ \mu\text{m}$... actually somewhat past the peak but still on the strongly-emitting flank), so the source is comfortably radiant there without needing to be driven any hotter. In practice, the FTIR throughput bottleneck is almost never the raw source radiance at globar operating temperatures; it is interferometer/beamsplitter transmission and detector responsivity that limit the usable signal, which is why globars are run at modest, long-lived temperatures (1200–1500 K) rather than pushed toward tungsten-like temperatures that would shorten rod lifetime for no real analytical benefit. Sanity check via the limiting case: if the requirement had instead been pushed to a wavelength deep in the source's weak tail — say 25 μm, far past the peak — the same calculation would show $x$ becoming small (since $x\propto 1/\lambda$), pushing $M_\lambda$ down substantially, and a real design might then need either a hotter source or a more sensitive detector; this is exactly why far-IR spectroscopy (beyond about 25–40 μm) is harder instrumentally than mid-IR, and increasingly relies on mercury-arc or synchrotron continuum sources rather than globars.

The physics underlying continuum sources traces to Max Planck's 1900 resolution of the "ultraviolet catastrophe" — the failure of classical electromagnetism (the Rayleigh–Jeans law) to correctly predict blackbody spectra at short wavelengths. Planck introduced the ad hoc but ultimately foundational postulate that oscillators in the cavity walls could only exchange energy in discrete quanta $E = h\nu$, a move he considered mathematically necessary rather than physically obvious at the time — a striking historical case of a formula preceding full acceptance of its physical meaning. Wilhelm Wien had already, in 1893, derived the displacement law that bears his name from thermodynamic scaling arguments alone, without needing the full quantum machinery Planck later supplied; Wien's law is in that sense a more classical result that Planck's complete theory later subsumed and explained. The applied engineering side belongs to a different lineage: Walther Nernst, better known for the third law of thermodynamics and the Nernst equation of electrochemistry, also invented the Nernst glower (1897) — a ceramic rod of mixed rare-earth oxides that becomes conductive and glows brightly only once preheated, exploiting the material's negative temperature coefficient of resistance. The globar (a trademarked name for a silicon carbide resistor rod) was developed commercially in the 1910s–1920s as a more convenient, self-starting alternative that dominates infrared spectrometers to this day. Hollow cathode lamps were developed and refined through the mid-20th century, closely tied to the rise of atomic absorption spectroscopy pioneered by Alan Walsh in Australia in the 1950s, who recognized that narrow-line atomic sources — rather than continuum sources with high-resolution monochromators — were the practical route to sensitive, element-selective absorption measurements.