Infrared imaging gets a second colour channel, in eyeglass form

A team at the Beijing Institute of Technology has built colloidal quantum dot upconverters that turn infrared light into visible light while preserving spectral distinctions, and has put them into lightweight, semi-transparent eyeglasses ^[2][3][11]. That matters because infrared sensing has generally handed operators a single-channel image: everything above 700 nm arrives as brightness, not as a distinguishable band ^[4].

The physical constraint is in the eye, not the optics. Mammalian photoreceptors absorb light using opsins and their linked retinals, and light beyond about 700 nm is not absorbed at all, so no electrical signal reaches the brain ^[4]. Human colour perception runs roughly 400 to 700 nm ^[1]. Upconversion works around this by absorbing low-energy photons and re-emitting higher-energy ones, an anti-Stokes shift ^[5]. Earlier versions of the idea have been demonstrated: upconversion nanoparticles bound to photoreceptors and injected into mouse eyes, and wearable lenses for humans that project infrared-derived visible light onto the cornea ^[6]. According to Physics World, those approaches were confined to a narrow near-infrared window ^[6].

The device from the group led by Ge Mu and Xin Tang is a stack: mercury telluride colloidal quantum dots sitting on an organic LED with two emissive layers, one red and one cyan ^[7]. Because the dots are small enough to confine electrons in all three directions, the carriers occupy discrete, atomic-like subbands rather than a continuous band ^[8]. Absorbed photons from the near-infrared through the short-wave infrared generate carriers that move into the OLED, whose electronic bands are aligned with those subbands ^[9]. Tang's explanation of the colour behaviour is that hole-trapping barriers engineered into the emissive layers steer recombination to one layer or the other as the incident intensity or wavelength changes ^[10].

Read plainly, that is a two-channel encoding, since the OLED carries two emissive layers ^[7][15]. It is not colour vision in the sense a human uses the phrase; it is a device that maps parts of the infrared onto two visible outputs instead of one. The reported performance figure is narrow but concrete: a wearer can detect infrared beyond 2 um at a luminance above 700 cd/m2, and the glasses project multispectral infrared onto the retina without disturbing normal vision ^[11][12]. Detection past 2 um is roughly 2.9 times the longest visible wavelength ^[16].

Tang describes the work as the end of a sequence rather than a single result, starting from an efficient single-colour upconverter, then colour-tuneable OLEDs, then upconverters integrated with silicon, germanium and CMOS ^[14]. The retinal application is further out. Tang says the upconverter could be bound to light-sensitive proteins in the retina, stimulating retinal neurons directly and bypassing damaged photoreceptors ^[13].

What to watch: whether the two-layer scheme can be extended to more emissive channels without losing the band alignment that makes it work ^[7][9][10], and whether the 700 cd/m2 figure holds at the panel sizes and duty cycles that head-worn hardware requires ^[12]. The CMOS-integrated line of the same work is the more likely near-term product path, since it does not require a human retina in the loop ^[14]. Mercury telluride in a face-worn consumer device is its own regulatory conversation, and the source material does not address it.