To propel widespread uptake of solid-state lighting, LEDs must be cheaper and more efficient. One way to do that is to switch the material used to make these devices from nitrides to cuprous halides, which have incredibly high exciton binding energies and can be grown on silicon substrates, argue Doyeol Ahn from the University of Seoul, Korea, and Seoung-Hwan Park from Catholic University of Daegu, Korea.

Although sales of LED lamps are rising fast and grabbing market share from incandescent and fluorescent bulbs, they are too pricy to dominate the market today. And that's not their only issue: Their efficiency is not that much higher than a fluorescent and can fall fast when the drive current through the LED is cranked up "“ a problem known as droop.

Look more closely and you'll find that the performance of LEDs, which feature a stack of InGaN quantum wells where electrons and holes recombine to generate light, are held back by inherent weaknesses. It is conventional to grow them on the crystal orientation (0001), and this spawns strain in the light-emitting active layers that creates strong built-in polarization fields that ultimately drive down device performance. These piezoelectric and spontaneous polarization fields, which can be as high as megavolts-per-centimetre, pull apart electrons and holes in the quantum well, impairing recombination efficiency.

Making matters worse, devices are typically grown via epitaxial processes on lattice-mismatched substrates, such as sapphire and SiC. This mismatch leads to the generation of many misfit dislocations, which hamper light-generation within the LED and shorten the device lifespan.

To address these issues, many researchers are considering alternative devices.

One popular option "“ which our theoretical team at the University of Seoul, Korea, and the Catholic University of Daegu, Korea, is looking at "“ is to switch to a different growth plane for the nitride LED. This can either reduce or eliminate the internal electric fields. In addition, we are investigating the potential of an even more promising, novel device: An LED built from the alloy CuBr, CuCl and CuBrCl. Such a devices could combine an incredibly high degree of optical gain with lattice-matched growth on a silicon substrate.

Slashing field strengths

Various approaches can be used to reduce the impact of the electric field in the active region of a nitride LED. These include: introducing an ultrathin, indium-rich InGaN quantum-well; inserting a very thin AlGaN layer into a thick InGaN well; employing the quaternary AlInGaN; and using non-square quantum-well structures. On top of all of this, there is also the highly popular method of today "“ growth on a new nitride plane.


The latter approach can be traced back to the pioneering work of Tetsuya Takeuchi, Hiroshi Amano and Isamu Akasaki from Meijo University, Japan. In 1996, they reported the significant reduction in heavy hole effective masses resulting from a switch from aligning the quantum wells on the (0001) plane to (1010) and (1012) orientations. One key consequence of reducing heavy hole effective mass is to increase recombination efficiency.

Following on from this work, researchers throughout the world have looked to trim the piezoelectric and spontaneous polarizations by growing the epitaxial nitride stack on a semi-polar plane "“ a plane titled with respect to the (0001) direction. Initially, those layers that were grown on non-polar and semi-polar substrates were plagued with numerous non-radiative recombination centres, because it is difficult to achieve a high crystal quality on non-polar and semi-polar planes.� However, this is far less of an issue today, and now researchers are reporting brighter devices. Other recent highlights in this area include the finding of a polarization crossover in a single InGaN/GaN quantum well grown on a semi-polar (1011) direction and a high compositional homogeneity in an InGaN quantum well grown on a semi-polar {2021} substrate and non-polar (1010) m-plane. These efforts show that there is the potential for commercial devices grown on non-polar and semi-polar substrates.

Our contribution to this field is to consider the optical gain of the LED. This is a measure of the luminous efficiency of this device. We have performed calculations for a 3 nm-thick In0.2Ga0.8N quantum well with a carrier density of 2 x 1013 cm-2 that is sandwiched between GaN barriers (see Figure 2, which shows the xï‚¢ and yï‚¢-polarized transverse electric (TE) optical gain spectra for several crystal orientations and optical anisotropy as a function of crystal orientation).



Figure 1. (a) Configuration of the coordinate systems (x, y, z) in (hkil) -oriented crystals. The growth axis, or z-axis, is normal to the substrate surface (hkil), and the coordinate system (x, y, z) denotes the primary crystallographic axes. The Euler angles and� are the polar and azimuthal angles of the direction zin terms of the� coordinates. (b) Nonpolar a- and m- planes with the growth direction parallel to the c-axis.� =2 with� = 6 corresponds to the z= [1120] growth direction and�  =2 with� =0 corresponds to the z= [1010] growth direction

The most striking feature of these graphs is that optical gain peaks have different strengths in different directions. As crystal angle increases, the optical gain for the yï‚¢- polarization shifts to a longer wavelength, while the optical gain for the xï‚¢-polarization shifts to a shorter wavelength. This high degree of anisotropy is not ideal for making an LED, and the optical gain is not as high as it can be in other material systems.

Cuprous halide LEDs

We believe that one material system that could take the performance of an LED to an entirely new level is that of the copper compounds CuBr and CuCl. One of the most attractive attributes of the I-VII cuprous halides is their incredibly high optical gain: It is more than an order of magnitude higher than that of AlInGaN, thanks to a combination of inherent strong excitonic effects and negligible piezoelectric internal fields. What's more, the lattice spacing in CuBr/CuCl quantum wells is close to that found in silicon substrates. That means cheap, widely available silicon substrates could be used as the foundation for producing devices that are free from misfit dislocations.

The I-VII cuprous halides, which include CuBr, CuCl and CuI, are direct bandgap semiconductors with a zincblend crystalline structure. They have piqued the interest of the research community with their very high exciton binding energies: For CuCl and CuBr, binding energies are 190 meV and 108 meV, compared with just 20 meV for GaN and 63 meV for ZnO. A high exciton binding energy is indicative of a strong attractive electron-hole Coulomb interaction, and ultimately enhanced optical transitions, even at temperatures well above room temperature.

We are not the first group to study these I-VII materials from a theoretical perspective. Efforts in this direction date back to the 1970s, led by Manual Cardona's group at the Max Plank Institute for Solid-State Research. However, this initial study and those that have followed have been restricted to bulk materials. Our breakthrough is to consider quantum well heterostructures, the region found in real LEDs.



Figure 2. (a)� x-and� y-polarized optical gain spectra for several crystal orientations and (b) the in-plane optical anisotropy as a function of the crystal orientation of the wurtzite In0.2Ga0.8N quantum well with a width of 3 nm. The decrease of the optical peak is attributed to the reduction of the optical matrix element for the x-polarization. For the case of the y-polarization, the optical gain peak increases significantly with the crystal angle. Note that the optical gain of the (0001)-oriented quantum well is calculated self-consistently, taking into account the piezoelectric and spontaneous polarizations

Our calculations, which include many-body effects such as bandgap renormalization, enhancement of optical gain due to excitonic effects and plasma screening, have determined the optical gain spectra for a CuBr/CuCl quantum well. Gain is 30 times that produced by an In0.2Ga0.8N/Al0.2In0.005G0.7995N quantum well, and significantly higher than that produced by a ZnO/Mg0.3Zn0.7O quantum well.



Figure 3. Optical gain spectra for CuBr/CuCl QW (green), ZnO/Mg0.3Zn0.7O QW (blue), and In0.2Ga0.8N/Al0.2In0.005G0.7995N QW (red) versus photon energy for carrier density of 5 x 1019 cm-2

Although this modelling effort shows that cuprous halide LEDs have tremendous promise, there is obviously a great deal of work still to do before they can make any commercial impact. The first step towards this is to establish a growth technology for forming high quality epitaxial films. Fortunately, some groundwork has already been carried out for this "“ in 2005 a partnership between scientists in Ireland reported the growth of a thin film of polycrystalline CuCl on a silicon (111) surface (this material produced strong room-temperature photoluminescence related to excitonic recombination).

In addition, the doping techniques need to be established, which could involve zinc and magnesium as n-type dopants and oxygen, sulphur, and selenium as p-type dopants. And once this has been accomplished, devices will have to be designed, shown to be robust and manufactured in high volumes. Only if all of this happens will cuprous halides be in with a chance of displacing nitrides in solid-state lighting.

Further reading

S. H. Park and D. Ahn, Proc. SPIE 8625 862511-1, 201
D. Ahn and S. L. Chuang, Appl. Phys. Lett. 102 121114 (2013)