Turning on the glass: Chalcogenides and the future of photonics

Though its name is decidedly difficult to pronounce, a family of glass called chalcogenides (pronounced kal-kə-jə-,nīd)—used for a wide variety of optics and photonics applications—is drawing renewed attention from a panoply of recent studies that point toward an exciting, auntapped potential.

Chalcogenide glass is nothing like window glass. Chalcogenides are often alloys based on germanium-antimony-tellurium, a.k.a. GST, which are so-called phase change materials that can be readily switched from amorphous (opaque, like coal) or crystalline (clear, like diamond) states. That is a singular behavior that has been utilized in photonics devices like optical switches, nonvolatile display, reconfigurable metaoptics, tuneable emitters and absorbers, photonic memory devices, and optical computers.

These amorphous materials contain one or two chalcogens—elements in the oxygen family of the periodic table—namely, sulphur, selenium, or tellurium. Their history dates to at least the 1960s, when the former Soviet Union became interested because, as glass instead of crystals, they would cost less to produce than single crystals of silicon or germanium. Depending on the exact ingredients, the glass can be opaque, shiny, or have a deep red color.

What is more, the mass of chalcogenide glass is tuneable via the choice of chalcogens added. That’s because, moving down the periodic table from oxygen, their atomic weight increases. The effect is that the transparency of the glass can be shifted to longer and longer wavelengths, says Pierre Lucas, a professor of materials science and engineering and of optical sciences at the University of Arizona.

In contrast, the window glass found in our homes and offices is usually soda-lime-silicate, an oxide glass. Oxygen, being a light element, will couple with any wavelength longer than 3-4 µm and be absorbed. If you want to hide from somebody who has an infrared (IR) camera, quips Lucas, you can hide behind a commercial window as all oxide glass is opaque to mid-IR light.

Because of its transparency to IR, one of the earliest applications of chalcogenide glass was for lenses in night-vision goggles for military use, but these were initially rather expensive. Now, the material can be found in thermal IR cameras, or even in the bumpers of high-end cars to avoid collisions in nighttime driving. Chalcogenides are coming into their own in lesser-explored areas such as IR-sensing of biomolecules, nonlinear optics for supercontinuum generation, as well as by providing a nifty toolkit for manipulating new wavelengths of light using nanostructures.

 On the other hand, phase-change materials like  GST are well-known in the electronics industry for their use at the core of optical disks and, more recently, in Intel and Micron’s Xpoint memory technology. In the case of GST, as the material switches from an amorphous to a crystalline state, the refractive index change is more than 2.5, and other properties, like electrical resistivity, change. The material can be preserved in either state at room temperature without the need for any additional energy.

There is no other class of materials with that large a refractive index change, notes Arka Majumdar, an associate professor in the departments of electrical and computer engineering and physics at the University of Washington, who works with silicon photonics circuits.

GST came into prominence for optical disks because it met the criterion of low electrical resistivity to cause sufficient heating and high optical loss which facilitates laser writing of a CD, or DVD. The large contrast of both refractive index and optical loss that exists simultaneously in traditional GST, though, also limits its performance for many applications.

But in 2019, researchers created an alloy of GST with properties tuned for optical systems, suggesting that this coupling of refractive index/optical loss can be broken. They found that by adding the element selenium to GST, creating a new material called GSST, the absorption contrast between the two phases was drastically reduced.

A nonvolatile, electrically programmable photonic switch with GST thin film, a type of chalcogenide glass well known as a phase-change material. Photo credit: Rui Chen

Today, chalcogenide glass is set to transform laser switching in optical systems. With traditional optical materials large mechanical gimbals are often needed to physically move lenses to change the direction of a laser beam. But the phase-changing nature of chalcogenide glass means they can effect this beam steering by being made to be more or less transparent.

With phase-change chalcogenide glass, set-and-forget switches are possible that require zero static power, are miniaturized to micron-size, and are modular. They can be used in photonic integrated circuits for applications such as programmable gate arrays, optical computing, optical memories, and optical neural networks. “You set it, and then you just forget about it,” says Majumdar, who has been working with Intel Corp. “Then, if the circuit fails, you can go back and reprogram it.”

A programmable gate array using phase change materials is the most interesting near-term application that Majumdar sees for chalcogenide glass, particularly in quantum computing.

In the silicon photonics transceiver chips currently in the market, whenever a laser or modulator breaks down, the component must be replaced. With multiple devices on a chip, there is also a higher chance of breakdown.

Programmable photonic circuits based on phase change materials would bring in much higher system reliability and reduce power consumption by orders of magnitude. The increase in reliability would apply to quantum photonics circuits as well, where single photons are generated and converted into a quantum frequency.

“This kind of approach did not really exist before, it is just not possible with existing integrated photonics,” says Majumdar, “The only way you can do that is with phase-change material.”

All the components for such a system have been shown to work separately; now it must be put together. In a magnified image of a silicon photonics circuit, Majumdar points out a ring that is visible with a tiny strip of chalcogenide, that brings about the added functionality.

The technique that allowed for industrial production of chalcogenide glass with high-quality optical properties in the IR was precision glass molding, specifically a technique developed by a start-up in the 2000s. Before this, the lenses of IR cameras were made of single crystal germanium, delicately grown. This was expensive, but the crystal would also require polishing with a diamond. A single lens could cost upwards of $10,000, says Lucas.

Scanning electron microscope images of chalcogenide glass show that precision molding, rather than polishing, leads to a much smoother surface and may also improve fracture toughness. Polishing a lens can take hours whereas molding takes a minute or so. “In terms of the economy of production, it’s very, very advantageous,” Lucas adds.

Chalcogenide glass remains amorphous while exhibiting optical transparency over the full IR region of 2-20 µm. This family of glass is also composed of polarizable elements that give it high nonlinear properties several orders of magnitude higher than silica.

This characteristic can be used to develop what is called a supercontinuum: intense coherent light in an IR range between 2-15 µm. With such light sources, the sensitivity of applications like hyperspectral imaging becomes superfine, though the high-intensity pump required is not yet readily available.

Nonlinearities affect what happens to light propagation under very high intensities available mainly from laser sources. Particularly, these nonlinearities can be used to manipulate light at wavelengths otherwise not easily accessible.

This trait is commonly used in commercial high-power lasers where near-IR laser light passes through a crystal and produces shorter wavelength green light, the so-called second harmonic generation. What happens inside that crystal is that photons are generated at the higher frequency, though not all light from the pump is converted.

The typical way this frequency conversion is done is by using crystals that have birefringence, where the crystal is cut up in such a way that the pump beam and the generated beam have the same refractive index, otherwise the phase merely oscillates. Chalcogenide glass, being amorphous, is typically not well-suited to the kind of frequency doubling being done by the crystals in a commercial laser. However, it is useful for third harmonic generation.

Engineered materials such as periodically poled lithium niobate where a trick is used to create constructive interference, while not perfectly phase-matched, still work. But if harmonic generation at multiple wavelengths simultaneously is desired, that isn’t possible with the traditional engineered crystal materials with their physical limitations.

Moving forward, a group from Duke University and the US Naval Research Laboratory (NRL) is looking at geometries from what are called quasicrystals, realized in chalcogenide glass, to enhance the nonlinear signal. Quasicrystals are structures with long range order but no periodicity.

“We wanted to satisfy phase-matching at multiple frequencies and this is why we chose to design quasicrystals,” says Natalia Litchinitser, a professor of electrical and computer engineering who led the work at Duke and is an expert on nanostructures.

The patterned quasicrystal in the chalcogenide glass is not visible to the naked eye because they are structures on the scale of the wavelength of light. “You want it to have some exotic appearance because it is such an exotic material,” says Jesse Frantz head of the NRL’s Specialty Waveguides Section. But it doesn’t look like anything special. “It’s not until you put a laser beam into it and generate a different color of light, then you can understand it’s something special.”

Another way to enhance the signal is through phase locking. The Duke/NRL group and collaborators used the versatility of chalcogenide glass to produce a weak source of UV light with single-layer nanostructures. That was thought to be precluded as chalcogenide absorbs UV light. UV light has several applications, among them underwater communications, chemical detection, and underwater imaging.

In a recent experiment by the group, light was pumped at a near-IR wavelength into the chalcogenide glass, in this case as arsenic trisulfide nanowires, with its designed nanostructure. What would normally happen is that third harmonic UV photons would be generated but quickly absorbed, even in such a thin material of only a few hundred nanometers. But by creating a nanostructure which allowed phase locking the third harmonic to the fundamental frequency, ultraviolet light made it through the glass and was not absorbed.

Chalcogenide glass nanostructures. Photo credit: Jiannan Gao

Litchinitser takes the explanation another step. When the laser pump enters the nanowire sample, it generates two components of the third harmonic wave: homogeneous and inhomogeneous.

While the homogeneous component propagates as expected, with high losses due to UV absorption by the chalcogenide glass, the inhomogeneous component is trapped by the pump at the fundamental frequency and propagates together with the pump.

Since the pump frequency is tuned to the near-IR, it experiences no absorption. Therefore, the inhomogeneous component of the third harmonic experiences no losses either. “It’s actually well-known in nonlinear optics that you generate these homogenous and inhomogeneous components,” she says, “But people never thought about looking at this inhomogeneous component and using it for any application.”

This phase-locking mechanism together with strong field enhancement by the nanostructure allowed third harmonic generation from the nanostructure to be enhanced more than five times compared to the original thin film. In subsequent numerical simulations from the same group, chalcogenide meta surface stacks enhanced the third harmonic conversion by two orders of magnitude.

While the group’s work is still nascent, Frantz points to its potential. “There’s a whole variety of applications where if you can generate any desired color of light on an optical chip,” he says, “that opens up all kinds of possibilities.”

Virat Markandeya is a science writer based in Delhi.