Optoelectronics, Frequency Changing

Lasers produce monochromatic light; that is, light with a single frequency or wavelength. Many laser applications, such as atomic spectroscopy, depend on the ability of a particular laser to be frequency tunable. Fine tuning (small shifts in output frequency) can be achieved by adjustments of operating characteristics of dye and semiconductor lasers.

In the area of fiber-optic communications, a large number of output wavelengths is desirable for wavelength division multiplexed (WDM) multichannel optical communication systems that utilize many wavelengths to increase data capacity. This can be achieved by nonlinear effects that create new optical frequencies from fundamental frequencies (frequency doubling), or with tunable laser sources.

A nonlinear laser-induced effect, discovered by Peter Franken and co-workers in 1961, occurs when an intense laser beam propagates through a nonlinear optical medium (quartz, for example). Light at double the frequency of the input beam is produced, an effect known as frequency doubling or second harmonic generation. Second harmonic generation can usefully convert the coherent output of a fixed-frequency laser to a different spectral region.

For example, the infrared radiation of a Nd:YAG laser operating at 1064 nanometers (infrared) can be converted into 532-nanometer visible radiation with a conversion efficiency of more than 50 percent. Novel laser sources produced by frequency doubling may offer advantages over existing laser sources that are bulky or inefficient; ultrashort laser pulses produced by frequency doubling or tripling may also have applications in the early twenty-first century for high-density optical storage, increased data transmission rates, and ultrafast spectroscopy of biochemical processes.

As the intensity of the incident light exceeds a certain threshold value, a nonlinear effect known as stimulated Brillouin scattering, which scatters light to different wavelengths, becomes important in fibers. Power is lost and the frequency-shifted wave may ‘‘cross-talk’’ with wavelengths in neighboring channels. Both of these effects degrade the optical signals and are undesirable.

Tunable dye lasers were discovered in 1966 by Peter Sorokin and John Lankard in the U.S. and Fritz Schafer in Germany. Liquid lasers of this type can be made for almost any wavelength from the ultraviolet to the infrared, dependent upon the fluorescence of the dye. Dye lasers may be tuned, through 100-nanometer ranges or more, by either changing the concentration of the dye or by adding and turning a diffraction grating in place of one of the cavity mirrors.

The class of laser used in the majority of today’s telecommunication systems is the semiconductor laser. This class of laser was first made from gallium arsenide (GaAs), but now they are commonly formed by a compound of elements from groups III and V of the periodic table (see Semiconductors, Compound). The end faces of the crystal (0.1 to 1 millimeter thickness) form mirrors that create the necessary cavity to trap the light and sustain stimulated emission.

Semiconductor lasers operate as either ‘‘fixed’’ (a single wavelength, or frequency) or ‘‘tunable’’ (offering coarse or fine tuning of many frequencies across a specific frequency band). The wavelengths of tunable semiconductor lasers depend upon both the properties of the III-V gain medium and the physical structure of the laser cavity surrounding the gain medium. Specifically, the length of the cavity (i.e., the distance between the mirrors) and the speed of light within the gain medium within the cavity determine a laser’s wavelength.

A semiconductor laser can therefore be tuned by mechanically adjusting the cavity length or by changing the refractive index (the speed) of the gain medium. Alternatively, light can be adjusted externally to the laser source, using micro- machined elements such as micromirrors or actuators.

Tunable semiconductor lasers can be grouped into four types:
1. The ‘‘edge emitting’’ triad of distributed feedback (DFB)
2. Distributed Bragg reflector (DBR)
3. External cavity diode lasers (ECDL)
4. The ‘‘surface emitting’’ type known as vertical cavity surface-emitting lasers (VCSEL)

As one might suspect, edge-emitting devices emit light at the substrate edges, whereas VCSELs emit light at the surface of the laser diode chip.

Rather than placing the resonator mirrors at the edges of the device, the mirrors in a VCSEL are located on the top and bottom of the semiconductor material. Somewhat confusingly, these mirrors are typically DBR devices. This arrangement causes light to ‘‘bounce’’ vertically in a laser chip, so that the light emerges through the top of the device, rather than the edge. As a result, VCSELs produce beams of a more circular nature than their cousins and beams that do not diverge as rapidly.

These characteristics enable a more efficient coupling of VCSELs to optical fibers. VCSELs, furthermore, benefit from single-process manufacturing and a relatively straightforward tuning process involving a microelectromechanical-systems (MEMs) cantilever arm placed directly above the optical cavity. Moving the arm a matter of a few micrometers up or down changes the frequency of the device by up to 5 percent.

The edge-emitter family features diffraction gratings etched on a single chip, as in DFB lasers; gratings placed near the active region of the laser cavity, in the case of DBRs; and one or two mirrors combined with a conventional laser chip to reflect light back into the cavity, as found in ECDLs. DFB lasers can be tuned by controlling the temperature of the laser diode cavity.

Because this technique requires large temperature differences, a single DFB has a small tuning range. However, it is possible to link multiple DFBs together to create multiple cavities and therefore wider tuning outputs. DBRs are actually variations of the DFB. In addition to having their grating (or mirror) section in a separate portion of the chip, a DBR has a gain section and a phase section. Tuning occurs when current is injected into the phase and mirror sections to change the carrier density of those two sections and, as a result, the wavelength of light they refract.

As with DFBs, DBRs have a somewhat limited tuning range, but techniques have been developed to expand that capability; for example specialized gratings known as ‘‘sampled gratings’’ and grating and bidirectional coupler combinations. The ECDL achieves tunability by physically moving a wavelength selective element, such as a grating or prism, to tune the laser output. One method involves moving a reflector up and down relative to the surface of a diffraction grating, with the varying distances determining specific wavelengths. This particular tuning method gives it a wide tuning range, but a slow tuning speed compared to the other methods.

Because of their inherent design, specifically characteristics of the gain medium and cavity, VCSELs have lower power than the other three types and as a result are used principally for local or metro (metropolitan area) wavelength applications. ECDLs have the highest power output of the tunables discussed here and are used for long distance networks. The others lie somewhere in the ‘‘middle space’’ between VCSELs and ECDLs and are used for metro and regional applications.

As noted above, tunable lasers have emerged for use in optical communication systems and specifically ‘‘wave division multiplexing’’ (WDM) or dense WDM (DWDM) applications, which entail the process of sending many different wavelengths carrying information down a single fiber optic strand to increase data capacity. WDM emerged as a solution to the bandwidth crunch imposed on telecommunications by the ever-increasing growth of the Internet and other broadband applications and it presents an exciting and new enabling application for tunable lasers in local, regional and long-distance networks.

In future all-optical networks (i.e., those without conversions to electronic signals), fiber optic systems will require tunable lasers that can provide a signal into any WDM channel and that can switch among channels, tuning to a new output wavelength in nanoseconds. In 2001 and 2002, some of the first true tunable semiconductor lasers moved from prototypes to early production.