Optical modulators get ready for a high-speed network future
EO modulators rely on the application of an electric field to change the refractive index of the material through which the light is passing. They are generally deployed in a Mach-Zehnder (MZ) interferometer format. The modulator has two arms and the incoming light is divided and passed into these (figure 1). A field is applied to one or both of the arms (single or dual drive), changing the refractive index so that the light emerging from one arm will be out of phase with that emerging from the other. When the light is recombined, it interferes destructively, effectively switching the light off. Without an applied field the light is in phase and remains on.
In an EA modulator an applied electric field shifts the absorption band edge in such a way that the material becomes opaque to the light at the wavelength of interest. Modulating the field turns the light on and off.
Lithium niobate Lithium niobate MZ modulator technology is the most mature of the various options. It is well entrenched in optical networks and because of this has an edge on the newer technologies, the cost and performance of which are not as well understood as those of lithium niobate. The material is inherently very broadband, meaning that a device optimized to work in a particular spectral region can also perform in another region with some adjustment to its drive voltage. This is important in DWDM systems where ideally a single device is required to handle a variety of wavelengths.
Lithium niobate is being challenged in both size and functionality. As bit rates go up, the modulator size also increases. At 40 Gbit/s, lithium niobate modulators are about 5 inches long. Those working with lithium niobate are constantly looking at ways to reduce package sizes. "A lot of people want to offer 4 inch transponders and our challenge is to fit the modulator in that package," says Dan Wilt, components product line director in Agere s optical networks division. "The Holy Grail would be a 2 x 3 inch case that some lower-end transponders are in."
Lithium niobate is very much a single-function material, unlike its semiconductor rivals. As Ian Croston, a product development manager with JDS Uniphase, concedes, "GaAs and InP are streets ahead in getting more functionality in a smaller volume than lithium niobate. But there are ways of integrating lithium niobate with other functions that will give GaAs and InP a run for their money." For example, increasing the functionality of the modulator box by putting the control electronics into the modulator package without impacting its size is an area that the modulator suppliers are currently concentrating on.
Drive voltage One of the areas where the different modulator materials compete is drive voltage. This is an important consideration, because a large voltage swing - the change in voltage between the on and off states - gets harder to produce as bit rates get higher. Those working with lithium niobate s competitors are quick to point out its high drive voltages, quoting anything from 6 to 8 Vp-p. Fujitsu has recently announced a 40 Gbit/s product that requires a drive of about 5 Vp-p Drive voltage is dependent on variables such as waveguide width and degree of optical confinement, and those involved with lithium niobate seem confident about their ability to reduce it further.
Agere is now sampling a 40 Gbit/s product that has benefited from new etching processes and electrode structures that have reduced drive voltage to 5 Vp-p compared with 7 Vp-p for its older technology. Both Agere and JDSU are looking at getting this nearer to 4 Vp-p.
Apart from the difficulties involved in producing big voltage swings at high frequencies, another reason for lowering the drive voltage is to enable the use of cheaper semiconductor materials in the drive electronics. "We want to drive our devices with ICs that are made using the cheapest possible technology. Our goal is to use SiGe. At the moment, 4 Vp-p is a bit of a stretch, but it is certainly not unthinkable," says Wilt.
Indium phosphide The electrical and optical properties of InP enable a large number of functions to be realized in one material system. The most obvious advantage of using InP as an EA modulator is that the device can be made relatively small. The bandgap energy of InP-based materials is close to the photon energies at telecoms wavelengths, so electro-optic and electroabsorption effects can be brought into play at low voltages and over short distances. In an InP-based EA modulator, light absorption can be achieved in about 300 µm with about 3 V.
As a material for EA modulators, InP is at present finding most widespread use in electroabsorption modulated lasers (EMLs), where a DFB laser is monolithically integrated with an EA modulator. Modulating a laser externally as opposed to direct electronic modulation of the laser itself reduces laser chirp. Chirp is particularly troublesome over longer distances because a chirped pulse is badly affected by chromatic dispersion.
At present, lithium niobate wins in the longer-reach (more than 100 km) high-bandwidth applications. In the shorter-reach applications it is being challenged by the use of EMLs. The development of EMLs is not stopping lithium niobate from continued penetration into shorter-reach applications. Indeed, JDSU is part of a multisource agreement for short-haul transceivers into which lithium niobate parts are being supplied.
The availability of cheap lasers is also being seen as a helpful nudge for lithium niobate modulators in cost-wary markets. "With the price of CW lasers falling rapidly, a lithium niobate modulator does not look all that unattractive now compared with EAs or EMLs," says Agere s Wilt. It s a close trade-off, with the decision becoming less one of price and more to do with application.
