Applications for photonics beyond the communications sector

Slowly, over the last decade, money has returned to the photonics industry, but it has been spread far more widely than before as applications spread out from the communications sector into areas such as medicine and defence. Terror attacks have stirred interest in technologies that will provide more sensitive and accurate gas detection – to look for evidence of toxins and explosives. Not only have the applications spread, so have the frequencies: from the visible spectrum down through infrared and into the terahertz regime. And a wider range of materials is under consideration, thanks to a better understanding of quantum mechanical effects. For example, transparent conductive oxides may provide cheaper options for the waveguides that are used to steer light beams around an optical switch. One obvious candidate is indium tin oxide, but its appeal is limited by its cost – a combination of indium's rarity and the popularity of the material, which is used in laptop and mobile phone displays to connect millions of transistors to driver circuitry. The EU funded project Novel Advanced Transparent Conductive Oxides (NATCO) used quantum mechanics to predict the optical and electronic properties of alternative materials. For example, it calculated that, by doping strontium cuprate – on its own, not a good material – with barium, the required properties could be obtained. Even optical interconnect between chips and boards – promised 20 years ago when it looked as though FR4 fibreglass was hitting the end of the line for high speed signals – is threatening to make a comeback. In March, FPGA maker Altera said it planned to demonstrate devices with built in optical interfaces by the end of the year. Optical provides a route to get to 12Gbit/s and higher; potentially to more than 100Gbit/s per channel between chips and across backplanes in communications switches. The trade off between materials costs and the expense of integrating optical interfaces into chip packages is now more finely balanced, making a transition to the optical domain more realistic, thanks to advances in multichip packaging. Optical interfaces do not integrate well with silicon. The material is good for control of optical signals, so is used in switches and waveguides. But the lack of a direct bandgap makes it a poor choice for building lasers. As a result, it seems likely that FPGAs with optical interfaces will combine a regular digital logic part, with laser diodes and photosensors built on a III-V process assembled into the same package and connected through a silicon substrate or interposer. For its work on interchip communications, Intel has focused on integrating III-V materials on a silicon substrate – at the wafer level to reduce cost – and proposed using wavelength division multiplexing (WDM) to increase the datarates achievable without massively increasing the cost of the optics. Last year, the chip giant demonstrated a 50Gbit/s interchip link as a spin off from work that has turned into the Thunderbolt I/O scheme used on Apple's latest family of portable computers. In communications overall, III-V elements remain the elements of choice. Thanks to the development of dense WDM telecom systems, where many gigabit rate laser pulses are combined on one fibre, there is no pressing need to upgrade to materials that can provide faster switching times. Regular III-V materials can easily cope with the speeds required from individual laser diodes, although work is progressing to put the materials onto silicon to reduce manufacturing cost and to cut power consumption. The problems in telecom start when trying to recover the information, splitting the signal so that it can be processed at the other end or switched to different destinations within the network. Electronic switching is not, yet, a realistic prospect at terabit speeds. The absolute maximum oscillation rate of CMOS is only 400GHz, so researchers are looking at materials, such as lithium niobate, that exhibit optical properties that allow photons to be controlled in useful ways. One way around the problem is to use material properties. Chalcogenide glasses – the materials used in writeable DVDs – offer one possibility. A team from Denmark, Australia and China exploited the Kerr Effect, where an electric field changes the refractive index of a material. Chalcogenides such as arsenic sulphide have a non linear response to the Kerr effect, making it possible to split the individual beams making up a multiplexed signal in different directions. Outside chip to chip communication and telecom, venture capitalists expect a boom in areas such as biophotonics and other sensor applications. Many medical diagnostic and research techniques use light. A popular tool for DNA research is a fluorescent gene found in a species of jellyfish and this is now routinely used in lab on a chip systems. More recently, researchers have turned to the plasmons generated by lasers on the surface of living cells to diagnose diseases and other problems – potentially avoiding the need to perform invasive biopsies. Lasers can also be used to manipulate living cells, as Professor Ming Wu of the University of California at Berkeley has demonstrated. With optoelectronic tweezers, optical energy is converted to electrical energy using a photoconductive surface made from amorphous silicon – the same material as that used in solar cells. The particles themselves move on the nitrided surface of the photoconductive layer, with a layer of glass coated with indium tin oxide over the top. Wherever the light hits the photoconductive layer, it acts as a conducting electrode. Everywhere else, it behaves as an insulator. Using techniques such as dielectrophoresis, particles can be scooted around the surface. First found 50 years ago by Herbert Pohl in experiments at Oklahoma State University, dielectrophoresis provides a way of controlling the movement of uncharged particles in a liquid. The technique relies on the presence of a changing electric field across the liquid and its effect on the liquid around the particles. Under a field, the interfaces between the liquid molecules and the particles can become polarised. A particle with interfaces that are easily polarised will move to regions where the electric field is higher; less polarisable particles move the other way. The frequency at which the field changes has an effect on the polarisability, so different types of particles will move depending on the rate at which the field oscillates. That makes it possible to target microscopic particles precisely, even though they are insulators. For detection of hazardous materials, photons have one big benefit: you can bounce them off the target without necessarily getting too close. For bomb material detection, the mid infrared spectrum is the most popular candidate. The mid infrared region is used by chemists to identify molecules. Many of the common bonds found in organic chemicals vibrate in the mid infrared, absorbing the radiation strongly at specific frequencies. The 'fingerprint region' – covering wavelengths between 5µm and 20µm – is extremely useful in pinpointing specific chemicals because each has a specific absorption profile. Some mid infrared laser based systems are close to deployment. Northrop-Grumman started test flights for its ASTAMIDS mine detector in the summer of 2008, ten years after almost killing the project because of a lack of funding. The system scans for landmines and improvised explosive devices (IEDs), using the difference in reflections from man made objects compared to the ground around them. Making a laser suitable for material detection is not easy, however. The monochromatic laser is not a good fit for spectroscopy, which relies on the ability to sweep through the spectrum. Tuning is possible in lasers, but is generally performed using gratings or mirrors to select from a number of wavelengths the core laser can generate. This is the technique that chemists have used to perform Raman spectroscopy for close to 30 years, albeit with room sized equipment. Sweeping through a frequency range directly within a solid state laser is trickier, but is possible with new generations of semiconductor lasers, such as vertical cavity surface emitting lasers (VCSELs) and quantum cascade lasers. The big advantage of the quantum cascade laser is that it removes the traditional limitation of laser design: the energy levels within a material determine its lasing frequencies. A breakthrough paper written in 1971 by Rudolf Kazarinov and Robert Suris of the Ioffe Institute in St Petersburg predicted that electron states could be quantised within energy levels by confining the particles within nanostructures such as quantum wells. By creating sandwiches of different materials, it is possible to build 'superlattice' materials that restrict the quantum states of electrons within them to very narrow bands. Working at Bell Labs, Federico Capasso and colleagues turned this concept into the quantum cascade laser. As the name suggests, the only way that electrons can move through the superlattice is by descending a 'staircase' of quantised energy levels, losing defined amounts of energy at each step. The thickness of each well controls the frequency of the light emitted by electrons when they lose energy. As it descends the staircase of energy states, one electron can produce many photons before it moves out of the laser. Typically, manufacturers use molecular beam epitaxy to put down the alloys needed to build the superlattice in very thin layers. The longer the wavelength, the thinner the layers need to be and the trickier the laser is to make. Despite that, the quantum cascade laser has turned out to be a good candidate for work in the mid infrared and even down into the terahertz region. Alongside well thickness, temperature controls the output wavelength of a lasing step. By applying high current pulses to the laser, it is possible to trigger a short sweep of wavelengths as the body of the device heats up and cools down. With some current lasers, it is possible to shift the wavelength by up to ten wavenumbers, a range of about 0.1µm in the mid infrared. Although this shift is not large, it is enough to capture a minimal fingerprint of a gas that can be matched by software. One system, developed by Cascade Technologies working with Sagem, has been used to pick up the signature of hydrogen peroxide gas, which is emitted by some homemade explosives. In September last year, Morpho installed a system based on the laser at Glasgow Airport to test its ability to 'sniff out' the gases given off by homemade explosives. As in telecom, the favoured materials for tunable lasers come from Groups III and V of the Periodic Table. Although silicon is a less than ideal material for electro optical conversion, that has not stopped work on trying to make it work as one. One property it does exhibit strongly is the Raman effect and this may prove useful in tunable lasers. Normally, when photons are scattered after striking the atoms of a material, they keep their original wavelength. But that is not always the case. In the late 1920s, Raman and Krishnan found that, sometimes, collisions result in photons transferring energy to or from the atoms in the material, changing their wavelength and producing light of a different colour. It is possible to concentrate and amplify this light, resulting in a type of laser that does not rely on population inversion. In the mid infrared, silicon is 10,000 times better than glass at demonstrating the Raman effect. Researchers from the University of California at Los Angeles (UCLA) demonstrated the first silicon laser based on the Raman effect in 2004. Intel then followed with a continuous wave design in 2005 that its developers thought would suit sensing applications. The problem is the basic Raman laser needs an external laser to act as a pump: but the quest for the all silicon laser has not stopped. Instead of using the vibrational energy levels of a crystal lattice to determine the output wavelength of a Raman laser, you can use the electronic energy levels of the quantum wells in a quantum cascade laser. The laser can be tuned by changing the electric field applied to the device, which in turn changes the relative energy levels within the superlattice, a technique demonstrated by researchers, including Capasso, from Texas A&M, Rice, Princeton and Harvard. Not only does the combination of Raman and quantum cascade laser promise a wider tuning range, it could yield a silicon based laser with no need for an external pump laser – using layers of silicon germanium, silicon dioxide and other materials to create the necessary heterostructures. Although it is not the only candidate, silicon also turns out to be promising for the region of the spectrum between the infrared and radio, if doped with electron donors such as antimony, arsenic or phosphorus. The terahertz spectrum provides the opportunity to scan solid and liquid materials and can provide information about the weak intermolecular bonds between them. By doing so, it is possible to identify fake pharmaceuticals as the changes in binders and composition from the normal will show up as different fingerprints. In principle, this kind of scanning can find plastic explosives by looking at fingerprints of the core chemical amid binding agents. Terahertz imaging is, however, difficult as there is no way of generating the waves directly, except at the very low end of the spectrum that overlaps with microwaves. Most techniques use optical lasers to fire photons at a crystal that then reemits photons in the target region of the spectrum. However, a new generation of quantum cascade laser based on silicon or other materials may prove successful at generating terahertz waves. As the photonics industry broadens its base in terms of application and frequency, the dominant position of III-V materials is under threat in the quest to yield cheaper, more flexible light processing devices.