Redesigned VCSELs take a leap in power

26 Mar 2004

High-power infrared or visible light from a VCSEL-type architecture is what US-based firm Novalux says its product offers. The company's president and CEO Jeff Cannon tells Siân Harris how NECSELs offer an attractive alternative to conventional laser technology.

From Opto & Laser Europe April 2004

When the telecoms industry started to fail a couple of years ago, manufacturers of vertical-cavity surface-emitting lasers (VCSELs) saw the market for their products plummet - and many firms did not survive.

"Many VCSEL manufacturers ended up with a single product specifically designed for this one market, and when this market declined dramatically they did not have an adaptable product platform to address other non-telecoms opportunities," commented Jeff Cannon, president and chief executive officer of California-based Novalux. He speaks from experience - his company avoided disaster by swiftly adapting its platform to deliver visible wavelengths and tackle different markets.

Novalux was able to make this shift because its founder and chief technology officer Aram Mooradian , a former professor at Massachusetts Institute of Technology, had developed some ingenious techniques to extend the performance of standard VCSELs. The resulting technology, known as a Novalux extended-cavity surface-emitting laser (NECSEL), is now starting to offer a real alternative to diode-pumped solid-state lasers and gas lasers in many applications.

Moving on from VCSELs VCSELs first emerged onto the market about 10 years ago as a low-cost, high-yield laser source. They emit in the infrared, which is ideal for the telecoms market. Unlike other semiconductor lasers, VCSELs emit light from the top of the chip so that each device can be tested on the wafer. This cuts out the expensive and time-consuming process of cutting up the wafer, polishing the devices, mounting them on a heat sink and then connecting to the electricity supply (which all side-emitting lasers diodes require before they can even be tested). Testing on the wafer saves time and processing costs and reduces the risk of devices being damaged during production.

However, despite their advantages, Cannon says that standard VCSELs suffer the drawback of a limited upper output power. This is because their epitaxy layer is on top of the gallium arsenide (GaAs) substrate, with the heat sink below that. This structure limits the amount of injection energy that can be put into the device because GaAs is an ineffective thermal conductor.

Another significant limitation of VCSELs is the aspect ratio of the diameter of the active area to the thickness of the epitaxial sandwich. This creates a highly diverging, round beam. They also only lase at the wavelength at which they were grown to operate.

These disadvantages restrict VCSELs to applications that do not require high power and in which the diverging beam can be focused by means of an additional component, such as an optical fibre. But in the heyday of optical telecoms this situation suited many VCSEL manufacturers very well.

Meanwhile, however, Mooradian and his colleagues at Novalux were accumulating intellectual property that appeared to solve the problems of VCSELs and open up other application areas. "Mooradian discovered that the way to increase the power was to put the quantum well region as close to the heat sink as possible - by turning the VCSEL upside-down," explained Cannon. In this new design, the light goes through the GaAs substrate, the poor optical transmission of which is addressed by using the minimum thickness possible with low doping.

In addition, said Cannon, there was a need to open up the active region to distribute a higher injection current more uniformly. While VCSELs typically have an active region of 5-10 µm, Novalux has demonstrated NECSELs with active diameters of between 20 and 300 µm.

Having tackled the output power issue - Novalux's first products have output powers of 5-15 mW - the company began to look at the highly diverging beam problem. It discovered that this can be solved by simply collimating the light with a third mirror that is integrated into the structure.

The presence of the third mirror opens up more possibilities for the device because there can be an intracavity space between the third mirror and the VCSEL chip into which the company can put a nonlinear optical material, making frequency doubling achievable. This creates NECSELs that can yield laser beams in the visible region from 400 to 575 nm. Indeed, added Cannon, this conversion can be highly efficient because the third mirror can be 100% infrared reflective but transmit visible light in order to optimize the cavity parameters for a very high infrared intracavity recirculating power.

Novalux's first product, known as Protera, takes advantage of this frequency doubling. It evolved from a 980 nm source that the company was developing for telecoms applications. When that market started to dry up the company realized that frequency-doubling its source resulted in a 490 nm source - just 2 nm away from the established wavelength of 488 nm, which is widely used in biological applications. And shifting the fundamental lasing wavelength of the epitaxy by 4 nm from the telecoms wavelength yielded 976 nm NECSELs, which could be frequency doubled to give 488 nm.

Providing an alternative "We had our telco product in the final stages of development when that market collapsed," explained Cannon. "In December 2001 we took the decision to move away from the telco platform and by November 2002 we were shipping our first frequency-doubled product, having completely re-engineered our technology platform in 11 months."

Due to the well-established fluorochromes in biological instrumentation, a 488 nm source with a stable output power is important. This area has been served for many years by argon gas lasers, but these consume considerable power and generate a large amount of heat. Optically-pumped solid-state (DPSS) lasers are more efficient and smaller, but they are currently two to three times more expensive than gas lasers.

A NECSEL, claims Cannon, is a good alternative to a gas laser because it is about the same price but its electrical power consumption is considerably lower. For a gas laser to give 5-15 mW of output power an input power of 1.5 kW is typically required, whereas 50-100 W of input power is needed for a diode-pumped solid-state laser. Yet according to Cannon, a NECSEL will yield the same output power of 5-15 mW with just 5 W of input power. In addition, he says that Protera has a considerably longer lifetime than a gas or DPSS laser.

One market of interest is instrumentation. Many biological instruments require multiple wavelengths. In fluorescence measurements, for example, instruments need to incorporate red, green and blue laser sources and the companies that make these devices often need to talk to three different manufacturers and build in three different types of devices of various sizes and power requirements. A 488 nm blue gas laser head, for example, is about the size of a 2 litre Coca-Cola bottle, with a power supply the size of a shoebox. Green light at 532 nm would usually be provided by a diode-pumped solid-state Nd:YAG laser that had been frequency doubled, while red light would come from either a helium-neon gas laser or from a red diode.

The NECSEL chip has approximately the same footprint as an edge-emitting semiconductor laser: around 1 x 1 mm, with a height of 100-200 µm. However, the external cavity is then added on top and the size of that depends on its application. The external cavity for an infrared emitter for telecoms applications would be made of monolithic glass and would add 2-3 mm to the height. If this cavity includes a frequency doubler the device would be around 15 mm long.

Because NECSELs are semiconductor lasers they can be grown at a range of different fundamental wavelengths. This means that Novalux's second product, a 532 nm green source, can be made by simply shifting the parameters of epitaxial growth to create a 1064 nm NECSEL and selecting an appropriate nonlinear material to frequency-double the output. The company's third product is at 460 nm, which is another important wavelength for biology and mandatory for graphic art and digital imaging applications. It uses the same architecture as the Protera 488, which results in the same stable output power and low noise beam characteristics.

Other devices are fixed to certain wavelengths because of gas transmission lines or design constraints. Although NECSELs are not tunable, their wavelengths can be adapted for specific applications. "Today, with epitaxial growth on GaAs substrates you can go from 915 to 1150 nm and we can grow a particular wavelength in this region to within ±1 nm," said Cannon.

Currently frequency doubling only yields visible wavelengths from 458 nm blue light up to 575 nm yellow light, but other wavelength ranges are possible by growing different III-V materials onto the GaAs substrate, says Cannon. The firm is looking at ways to directly generate red light without having to frequency double an infrared NECSEL.

"The high demand we are seeing now is for red in the 635-650 nm region," he commented, adding that there is also interest in 405 nm for bioanalytical instruments and holographic optical storage.

In producing high-power visible source NECSELs Novalux has moved well away from the roots of the technology in VCSELs. "We don't really follow the VCSEL market," said Cannon. "VCSEL manufacturers mostly target the telco market and we don't look at them as competitors."

And in the future the company could move even further away from traditional VCSELs. It is currently developing two other concepts that could evolve into commercial products. The first product, known as Stellar, is much smaller than the Protera and is able to generate visible light. The firm's other project, the Magnus laser, is to make an array or strip of NECSELs. These could be cut from a single wafer and yield higher-power lasers for applications such as flow cytometry and micro-materials processing.

Novalux is already discussing its ideas with customers. And if Cannon's predictions are right, then ideas such as these could pose a significant threat to the manufacturers of traditional lasers.