HgCdTe detector (MCT)

Mercury cadmium telluride, Hg₁₋ₓCdₓTe, often abbreviated MCT, is a semiconductor alloy between HgTe, a semimetal with a negative gap, and CdTe, with a bandgap of about 1.6 eV at 77 K. Choosing the cadmium fraction $x$ places the absorption edge anywhere from the near infrared to the very-long-wave infrared, and the lattice constant changes by only about 0.3% across the whole range, so layers of different composition can be grown on nearly lattice-matched CdZnTe substrates. At 77 K, $x \approx 0.3$ gives a cutoff near 5 µm for the mid-wave band and $x \approx 0.21$ a cutoff near 12.5 µm for the long-wave band. It is the most widely used material for high-performance cooled infrared detectors, in single elements for FTIR spectrometers and in large focal-plane arrays for imaging and astronomy.

Composition and cutoff wavelength

The gap as a function of composition and temperature is well described by the empirical relation of Hansen, Schmit and Casselman (1982), with $E_g$ in eV and $T$ in kelvin:

$$\begin{aligned} E_g &= -0.302 + 1.93x - 0.81x^2 \\ &\quad + 0.832x^3 + 5.35\times10^{-4}\,T(1-2x). \end{aligned}$$

The cutoff wavelength is $\lambda_c = 1.23984/E_g$ µm. At $T = 77$ K this gives:

The steep dependence at low $x$ is a practical difficulty: a change of 0.01 in composition moves a long-wave cutoff by about 2.4 µm (from 12.5 to 14.9 µm between x = 0.21 and 0.20), so uniformity across a wafer directly limits array uniformity. The temperature term has the opposite sign to most semiconductors for $x < 0.5$: the gap widens as the crystal warms, so at 300 K the $x = 0.3$ alloy cuts off near 4.3 µm instead of 5.1 µm.

Device types and cooling

Early MCT detectors were photoconductors; modern arrays are photodiodes, usually p-on-n or n-on-p junctions hybridized with indium bumps to a silicon readout circuit, as in other infrared focal planes. The direct gap gives a high absorption coefficient, so a layer roughly one cutoff wavelength thick absorbs most of the light, and quantum efficiencies of 70% or more are common with an anti-reflection coating.

Cooling is required because thermally generated dark current scales roughly as $\exp(-E_g/k_BT)$. At 77 K, $k_BT$ is 6.6 meV, and a 0.1 eV gap is about 15 $k_BT$; at room temperature the same gap is only about 4 $k_BT$, and dark current would swamp any signal. Long-wave arrays are therefore mounted on a cold finger in a vacuum dewar and cooled by liquid nitrogen or a Stirling cryocooler to about 60–80 K. Mid-wave arrays tolerate higher operating temperatures, and short-wave MCT can run on thermoelectric coolers. Tennant's empirical "Rule 07" summarizes the best reported dark-current density versus cutoff wavelength and temperature and is used as a benchmark when comparing MCT against competing materials.

MCT avalanche photodiodes are unusual in that, for suitable compositions, only electrons multiply, which gives an excess noise factor near 1 and allows linear-mode photon counting in the infrared. This distinguishes them from InGaAs and silicon avalanche photodiodes, where both carriers contribute to multiplication and the excess noise is higher.

Where it is used

The main applications are cooled thermal imaging in the 3–5 µm and 8–12 µm atmospheric windows, FTIR spectroscopy (the "MCT detector" option on most benchtop instruments), infrared astronomy, where large short-wave and mid-wave arrays operate at still lower temperatures for minimal dark current, and laser-based gas sensing. For the 0.9–1.7 µm band, InGaAs is cheaper and needs less cooling; extended-cutoff MCT covers 1.7–2.6 µm where InGaAs becomes noisy, including in some SWIR cameras.

Pitfalls

The Hg–Te bond is weak, so mercury diffuses and the material is sensitive to processing temperature; defects give cluster and blinking pixels in arrays. Photoconductive MCT elements in FTIR instruments saturate and become nonlinear at high flux, which distorts the measured spectrum unless corrected. Cold-shield and window emission contribute background photons that can exceed the scene signal at long wavelengths. Figures of merit are best compared through specific detectivity at a stated temperature and field of view.

Common questions

Why do MCT detectors need liquid nitrogen?

For long-wave cutoffs the bandgap is only about 0.1 eV, comparable to a few $k_BT$ at room temperature, so thermal generation of carriers produces dark current far above the photocurrent. Cooling to about 77 K reduces it by many orders of magnitude.

What is the difference between an MCT detector and a DTGS detector in FTIR?

DTGS is a pyroelectric thermal detector that works at room temperature with a flat spectral response but lower sensitivity and speed. MCT is a cooled photon detector, much more sensitive and faster, with a response that falls off beyond its cutoff wavelength.

Is MCT the same as HgCdTe?

Yes. MCT stands for mercury cadmium telluride, the common name of Hg₁₋ₓCdₓTe; the composition, often written as the value of $x$, determines the cutoff wavelength.

References: G. L. Hansen, J. L. Schmit, T. N. Casselman, J. Appl. Phys. 53, 7099 (1982); W. E. Tennant et al., J. Electron. Mater. 37, 1406 (2008); A. Rogalski, Infrared and Terahertz Detectors, 3rd ed. (CRC Press, 2019).