1 COUPLED CAVITY LASER SOURCE FOR TELECOMMUNICATIONS
GOVERNMENT RIGHTS
The present invention was made under U.S. Air Force Phillips Laboratory Contract F29601 95 C 0047, U.S Navy Naval Undersea Warfare Center Division, Newport Contract, N66604 97 M 0207, and U.S. Army CECOM Contract Number DAAB07 97 C J228, and the U.S. Government has certain rights therein.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to lasers, and particularly to lasers that provide a light source for telecommunications systems
2. Description of Related Art
Computer networks and advanced communications systems of many types have made vast amounts of information widely available Cable TV the Internet, telephones, and satellite communications are just a few of the communications systems in use today, and in the future it is almost certain that these communications systems will expand and other types of communications systems will become available In order to meet the ever increasing need for expanded communications capacity telecommunications has become a rapidly growing industry, and is likely to continue to grow in the future
In general, any communication system requires a way to communicate a message from a transmitter to a receiver. In many instances, cables are used for transmitting a signal. In the past, cable TV systems typically used coaxial cables, but more and more cable TV systems are now installing fiber optic cables in one or more links, and in some locations, all the way to the consumer. Fiber optic cables generally include a bundle of optical fibers, which are flexible waveguides designed to deliver light from a laser source to a light detector
One large segment of the telecommunications industry-the telephone industry-now almost exclusively uses fiber optic cables to transmit modulated laser signals over long distances For example, prior to 1983, less than 1 % of long distance telephone traffic in the United States was transmitted optically on glass fiber, but by 1990, greater than 90% of long distance telephone traffic traveled over optical fibers
The cost of installing and maintaining cables is a very significant portion of the overall expense of a communication system. In order to minimize this cost, systems have been developed to increase the transmission capacity of installed cables (i.e , these systems increase "bandwidth") However, present technology can support modulation only at rates of about 10 Gbitlsecond per channel due to limitations imposed by external modulators and by opto electronic receivers Therefore, in order to take full advantage of installed fiber optic cables, an increase in data rate can best be realized by increasing the number of optical channels simultaneously transmitting information through a single optical fiber. One such type of system under development is a wavelength division multiplexing (WDM) system or a dense wavelength division multiplexing (DWDM) system, in which each separate channel is provided by a slightly different wavelength. The International Telecommunications Union ("ITU") has agreed upon the standard for DWDM to include the range from 1532.68 nm to 1562.23 nm ± 0 4 nm However, one significant component remains to be developed-a suitable laser source.
A suitable laser source for high bandwidth telecommunications and optical fiber sensor systems would provide (1) high power to overcome losses due to external modulation and to transmit over long distances, (2) wavelength selectabiiity to increase network bandwidth by multiplexing multiple laser sources onto the same optical
2 fiber, and (3) narrow newidth to reduce dispersion over long runs of optical fiber.
The conventional light source for optical fiber systems is the laser diode, and specifically the distnbuted feedback IDFB) laser diode which has the narrow hnewidth necessary for communications. However, DFB lasers are limited in power; commercially available high power DFB lasers provide only 35 mW. Even though higher power DFB lasers are being developed, it is unlikely that commercial units with powers approaching 100 mW will be available in volume, at least for the foreseeable future. Furthermore, like all semiconductor lasers, DFB lasers typically exhibit relative intensity noise (RIN) at frequencies from zero to ten or more gigahertz (GHz), which limits their transmission capacity Also, the temperature of semiconductor lasers must be carefully controlled to avoid temperature dependent frequency changes. The lack of power available from DFB lasers remains a significant problem, particularly for long runs of optical fibers. Optical signals lose power as they travel along an optical fiber; for example, one conventional single mode optical fiber attenuates the signal by about 0.3 dB/km After a long run of optical fiber, a signal may become so attenuated that it no longer can be received intact. In order to amplify an optical signal before it becomes irrecoverably attenuated, erbium doped fiber amplifiers (EDFAs) have been developed. EDFAs can be inserted directly into a cable, and are less expensive than the previous alternative of regenerating the signal using a second receiver and transmitter to receive the optical signal, convert it to an electrical signal, and then re transmit it along an optical fiber. However, EDFAs create noise (and thereby degrade the signal to noise ratio), they are expensive, and it would be a significant advantage if the laser source could provide enough power to reduce or even eliminate the need for EDFAs. One type of laser that holds promise for telecommunications is a diode pumped solid state laser (DPSSL), which includes a solid state gain medium pumped by optical radiation from a laser diode. DPSSLs only exhibit RIN over a range from zero to tens of megahertz (M Hz), making DPSSLs more advantageous than semiconductor lasers from a noise standpoint. DPSSLs have other advantages such as higher power than semiconductor lasers; however, DPSSLs suitable for telecommunications systems have not been available For small to moderate levels of optical power, one particularly useful type of DPSSL is a "microlaser", which comprises a short element (i.e. less than about five mm) of solid state gain medium positioned in a resonant cavity that is defined by two opposing reflective surfaces formed directly on opposing ends of the solid state gain medium. A pump beam supplied by a semiconductor diode laser pumps the solid state gain medium to provide energy to support laser operation.
SUMMARY OF THE INVENTION In order to overcome the limitations of the prior art, a coupled cavity laser is provided that produces a single longitudinal mode TEM00 output with a narrow hnewidth suitable for communications. The coupled cavity laser includes a gain medium, a thin etalon that comprises a polarization selective material, end mirrors on opposing ends of the coupled cavity, and a partially reflective interface situated between the end mirrors that separates the coupled cavity into a first cavity and a second cavity The gain medium is situated within the first cavity, and the etalon is situated within the second cavity. In one embodiment, the coupled cavity laser compnses a monolithic laser assembly including an electro optic modulator element, the gain medium, and the etalon, with the electro optic element situated within the first cavity, and the gain medium situated between the electro optic element and the etalon Advantageously, the electro optic element can be modulated to "chirp" the laser output by shifting the frequency at a predetermined amount and a constant rate, which is useful to reduce losses in optical fibers. The
laser assembly can be formed as a monolithic structure by, for example, optically contacting the modulator element, the gain medium, and the etalon, which advantageously eliminates air from the laser cavity, encloses ail intracavity surfaces, and protects them from contamination by external sources.
In one embodiment, the gain medium comprises Er,Yb: glass, which has a broad gain bandwidth that is dependent upon the loss within the laser cavity. The etalon within the second cavity has a thickness that selects a single lasing wavelength within that gain bandwidth.
Some embodiments include a system that controls the pump source to reduce relative intensity noise (RIN), which is advantageous for communications systems. Extremely low relative intensity noise (RIN) ( < 165 dB/Hz at 10 MHz and above) and a narrow hnewidth ( < 100 kHz, typically < 40 kHz) have been observed. Furthermore, high output power can be achieved; for example some embodiments provide greater than 100 mW of linearly polarized laser light coupled into the optical fiber ( > 140 mW uncoupled). It is believed that even higher power levels can be achieved. Advantageously, this high power extends the range that an optical fiber can run without requiπng a fiber amplifier. One significant use for the laser source descnbed herein is in telecommunications systems: cable TV, dense wavelength division multiplexing ("DWDM"), and satellite communications. The laser could also be used in other systems that require a single mode, narrow bandwidth laser source.
The single longitudinal mode TEM00 output can be readily coupled into a single mode optical fiber. The wavelength produced by one embodiment of the laser, around 1550 nm, is optimal for many currently installed fiber optic systems; and furthermore, the laser source is suitable for DWDM systems because a wavelength can be provided within a range of about 1530 nm to 1570 nm. Furthermore, the wavelengths around 1550 nm are "eye- safe" and therefore suitable for free-space communication systems.
BRIEF DESCRIPTION OF THE DRAWINGS
For a more complete understanding of this invention, reference is now made to the following detailed description of the embodiments as illustrated in the accompanying drawing, wherein:
Fig. 1 is a schematic diagram of a coupled cavity solid state laser apparatus for communications; Fig. 2 is a graph of noise intensity vs. frequency, illustrating relative intensity noise (RIN) and the effect of a noise reduction circuit to reduce RIN;
Fig. 3 is a block diagram of one embodiment of the noise reduction circuit for reducing RIN; Fig.4 is a side view of one embodiment of the laser assembly shown in Fig. 1; Fig. 5 is an exploded, perspective view of the laser assembly shown in Fig. 4; Fig. 6 is a flow chart that illustrates steps for making a laser using a silicon base; Fig. 7 is an exploded view of one embodiment of a telecommunications laser, illustrating a silicon base on which the laser components are mounted, a cooler for cooling the silicon base, and a frame for encasing the base and cooler;
Fig. 8 is a cross sectional side view of a telecommunications laser implemented on a silicon base; Fig. 9 is an exploded side view of a telecommunications laser implemented on a silicon base; Fig. 10 is a top view of a telecommunications laser implemented on a silicon base;
Fig. 11 is an exploded view of a mounting structure, a laser assembly, and a welding strip for mounting the laser assembly on the silicon base,
Fig. 12 is a top view of a mounted laser assembly of Fig. 11 ; and
Fig. 13 is a diagram of a wavelength division multiplexing (WDM) communication system that utilizes the
4 communications laser described herein.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
This invention is described in a preferred embodiment in the following descπption with reference to the Figures, in which like numbers represent the same or similar elements
For ease of descnption, the laser may be described herein as a "1550 nm" laser. It should be apparent from the descnption herein that the laser can operate at one wavelength within a range of wavelengths between about 1530 and 1570 nanometers (nm), and that reference to a "1550 nm laser" is not meant to be limiting to a specific wavelength. Some optical components are referred to herein as "etalons", which is meant to refer to components that have two opposing fiat, parallel surfaces polished to etalon tolerances. For example the parallelism between the flat sides of an etalon should be less than λ/4, and if possible less than λ/10. The roughness of the flat surfaces should be less than 10 A rms in order to facilitate optical contacting with other optical components.
Reference is first made to Fig. 1, which is a schematic diagram of a coupled cavity solid state laser apparatus for communications. The laser is designed to produce a single longitudinal mode TEM00 output, which can be easily matched to a single mode fiber. A suitable pump source 1 10 produces a pump beam 115. The pump source 110 can be any suitable source of optical radiation, and preferably comprises a laser diode. The pump source 110 is controlled by a diode dnver 117 A sample 118 of the pump beam 115 is sensed by a photodiode 119. The photodiode may be positioned just off the beam path of the pump beam where it can receive a lower power radiation lobe and thereby sample the pump light. Alternatively, the pump beam may be sampled by sensing backscattered light from the optical elements such as coupling optics 120 or a partial reflector (not shown). The photodiode 119 operates as a light sensor to produce a feedback signal, which is then supplied to the diode dnver 117 to adjust the bias current so that the pump source 110 supplies an approximately constant beam power over the expected lifetime of the laser. One advantage of an approximately constant pump power is that thermal flows within the laser cavity remain approximately constant, which permits the laser components to be designed with substantially predictable temperatures and predictable temperature variations, which in turn provides substantially predictable performance over the lifetime of the laser.
The pump beam 115 is applied through suitable coupling optics 120 to a coupled cavity laser assembly, shown generally at 130. in one embodiment, the coupling optics comprise a lens, such as a ball lens that focuses the pump beam into a narrow spot within the laser assembly in such a way as to provide single mode operation.
The laser assembly 130 defines a coupled cavity, illustrated by an arrow 132, defined between a back end 134 and a front end 136. One embodiment of the laser assembly is descnbed with reference to Figs.4 and 5; generally, the laser assembly includes an electro optic modulator, a gain medium, and a polarizer connected together to define a coupled cavity. A frequency modulation circuit 140 supplies signals to the electro optic element, which in response modulates the laser beam's wavelength in a manner suitable for the intended use; for example, in one embodiment the modulation circuit 140 supplies a "chirp"; specifically the modulation circuit 140 shifts the laser's frequency by a relatively small amount (e.g. a few hundred MHz ), at a constant rate of about 10 kHz. It has been found that a frequency shift of about 200 MHz to 300 MHz at a constant rate of 10 kHz is useful to compensate for problems that may be caused by stimulated Bnllouin scattering (SBS) processes in optical fibers. Other embodiments may require different frequency shifts and/or different rates. In one embodiment, the modulation circuit 140 supplies a user-selected voltage between zero and ten volts to the electro optic material, which allows the laser's frequency
5 to be shifted by up to about 40 MHz. In another embodiment the modulation circuit may be used to enhance the phase stability of the laser.
The laser output 148 from the laser assembly 130 is transmitted through an isolator 1 0 that allows oneway optical transmission, thereby blocking undesirable optical feedback into the optical cavity. The isolated laser output 152 is then coupled through a fiber optic coupler 155 to an optical fiber 160. A ferrule 161 is used to hold the optical fiber in position. In the illustrated embodiment, the coupler 155 compnses a substantially transmissive partial reflector 162 and a ball lens 164, but it should be apparent that other embodiments may utilize other optics. Backscattered light 165 from the partial reflector 162 is detected by a photodiode 168 that provides a feedback signal to a noise reduction circuit 170, which produces a noise reduction control signal on a line 172 that is applied to the diode dnver 1 17 The noise reduction control signal then operates to reduce low frequency (e.g. less than 2 MHz) noise such as relative intensity noise that may result from relaxation oscillations.
Reference is now made to Fig. 2, which is a graph of noise intensity vs. frequency, to illustrate relative intensity noise (RIN) and the effect of the noise reduction circuit 170 to reduce RIN, which is evidenced as frequency dependent amplitude variations in the laser output at low frequencies (e.g. less than 2 MHz). Relaxation oscillations have been defined as "small-amplitude, quasi sinusoidal, exponentially damped oscillations about the steady state amplitude that occur when a continuously operating laser is lightly disturbed," A.E. Siegman, Lasers, University Science Books, Mill Valley, CA, Chapter 25, 1986 The nature of certain lasers to exhibit relaxation oscillations is believed to be dependent upon factors such as the fluorescence lifetime of the lasant material, the reflectivity of the end mirrors, and cavity losses. The relaxation oscillation phenomena increases RIN Like many noise processes, RIN degrades communication at the frequencies where it appears. Specifically, the effect of RIN is to reduce the maximum signal to noise ratio, increase bit error rates (for digital communications), and reduce the maximum achievable transmission rate. Of course RIN is only one noise process; other noise processes can also adversely affect communications.
Fig. 2 shows a graph 180 typical of free running (i e uncompensated) RIN (measured in dB) as a function of frequency for one embodiment of the diode pumped solid state laser described herein. Apart from the very low frequencies (e.g. below 20 kHz) which are generally not of interest, RIN increases with frequency to a maximum frequency (fm typically between 500 kHz and 1 MHz At higher frequencies, RIN decreases sharply until i mt (several MHz ), where it approaches the shot noise limit At high frequencies (e.g. above several MHz) RIN is essentially shot noise limited ( < 170 dB/Hz). The free running peak RIN frequency and corresponding peak RIN amplitude vary from laser to laser, but these values are generally in the region of 500 kHz to 1 MHz and 85 dB/Hz to 97 dB/Hz. At low frequencies, RIN follows a 1 /f dependency, and at 20 kHz, RIN is typically < 120 dB/Hz.
Semiconductor lasers are also affected by RIN For such semiconductor lasers, the peak RIN is typically much higher— in the GHz range (e.g. one GHz)-and as a result, communication systems that use semiconductor lasers as a light source are adversely impacted by RIN over a very broad range of frequencies. Therefore, semiconductor lasers have a substantial inherent disadvantage when compared with solid state lasers, which in comparison exhibit substantial RIN only over a few MHz.
In order to reduce RIN, the noise reduction circuit 170 produces a noise reduction control signal that is applied to the diode driver 1 17 in order to reduce relative intensity noise (RIN) from relaxation oscillations. Such noise reduction systems are known and disclosed, for example, in a publication by Thomas J. Kane, Intensity Noise in Diode Pumped Single Frequency Nd:YAG Lasers and its Control by Electronic Feedback, IEEE Photonics Technology Letters, Vol. 2, No 4, April 1990, pp. 23 24, and in U S Patents 5, 177,755 and 5,253,267, both of which are
entitled Laser Multiple Feedback Control Circuit and Method, by Keith Johnson. Noise reduction systems are also shown in the following publications: De Geronimo et al., Optoelectronic feedback control for intensity noise suppression in a codoped erbium-ytterbium glass laser, Electron. Lett., 1997, Vol. 33, No. 15, pp. 1336-1337, and Taccheo et al.. Intensity noise reduction in a single-frequency ytterbium codoped erbium laser, Opt. Lett., 1996, Vol. 21 , pp. 1747-1749. In general, these publications disclose a system where the intensity of the laser output is monitored over time and used as feedback to actively control the current to the laser diode, which has the effect of modulating the pump beam to cancel out RIN.
The noise reduction circuit 170 compnses any suitable circuit that functions to phase shift the laser's amplitude fluctuations, and amplifies the signal as appropriate to approximately cancel out RIN from the particular laser system. In general the gain of the circuit 170 at the maximum RIN frequency is set to a value that just compensates for RIN at the maximum RIN frequency (i.e. reduces RIN by the maximum amount). This gain value and the maximum RIN frequency may be determined in the factory on an individual basis for each laser and then set accordingly. Alternatively, an approximate gain value and/or an approximate maximum RIN frequency may be predetermined and built into each laser. Still other alternatives may include active electπcal systems that monitor the RIN frequency and automatically adjust the gain value accordingly. At other frequencies of interest that may be affected by RIN, the circuit is designed so that the gain of the noise reduction circuit vanes with frequency.
Reference is now made to Fig. 3, which is a block diagram of one embodiment of the noise reduction circuit 170. The signal from the photodiode 168 is first amplified in an amplifier 190 that two stages including a first differentiator 192 that increases signal gain (voltage) with increasing frequency by a predictable amount of +6 dB/octave (+6 dB is a doubling of voltage, one octave is a doubling of frequency), which operates to cancel out the naturally declining -6 dB/octave response of the particular laser at higher frequencies. The second stage of the amplifier amplifies the signal by an additional +6 dB/octave around the maximum RIN frequency (e.g. 500 kHz to 1000 kHz). Next, a limiter 194 operates to reduce the amplified signal gain if it exceeds a predetermined limit, which enhances the stability of the noise reduction process. Next, a phase shifter 195 changes the phase of the amplified signal by an amount necessary to cancel the RIN. In one embodiment, the signal from the photodiode 168 has already been shifted by 90°, and therefore the phase shifter reverses it by 270° to provide a net 180° out-of-phase waveform. The phase shifter 195 may be adjustable in order to provide a more precise null (i.e. more precise cancellation of the noise).
Next, a variable gain amplifier 196 provides for fine tuning of the overall circuit gain in order to achieve maximum noise reduction, while maintaining a margin of stability. A resultant feedback signal 197 is supplied to the diode driver 1 17 where it is summed with the bias value for the DC diode current, and then the appropriate drive current is applied to the laser diode to drive and modulate the laser diode output to cancel the unwanted system resonance. The graph line 185 in Fig. 2 shows an observed result of the noise reduction system, in which the peak has been substantially flattened and RIN has been substantially lowered at all frequencies around fm„. In one embodiment, the maximum RIN has been typically reduced by 20 dB or more, and often 25-34 dB reductions are observed.
Reference is now made to Figs. 4 and 5, in conjunction with Fig. 1. Fig. 4 is a side view of one embodiment of the laser assembly 130 shown in Fig. 1 , and Fig. 5 is an exploded, perspective view of the laser assembly. Generally, the laser assembly 130 includes an electro-optic matenal 210 (sometimes called a "Pockels cell"), a gain medium 220, and a polarization-selective etalon ("polaπzer") 230 that has smooth parallel sides. The elements of the laser assembly 130 are coupled together by optical contact or other suitable method to form a monolithic structure
7 that advantageously eliminates air from the laser cavity and protects the intra cavity surfaces from external contamination, resulting in greater stability of laser operation over a wide range of temperatures, pressures, and humidities. The roughness of all flat surfaces on the optical elements should be less than 10 A rms in order to facilitate optical contacting with other optical components In the following discussion, reference may be made to the x , y , and z- axes. The x-y-z axes are defined by the laser axis; the z-axιs is aligned with the center of the laser beam, and the x and y axes are perpendicular to the z axis as illustrated in Fig. 5.
The electro optic mateπal 210 has a spherical curve on the back end 134, and opposite thereto a flat surface 212. The back end 134 is coated for high reflection (e.g. > 99.9%) at 1550 nm and for antireflection at the pump wavelength (e.g. 975 nm). The flat surface 212 may remain uncoated, or it may be coated for anti reflection at 1550 and the pump wavelength. The spherical curve (e.g. about 50 mm) enhances stability of the cavity's laser modes, and also provides focusing power for the pump beam. On its other end, the electro optic mateπal has a rectangular cross section. The electro optic material preferably includes (LιNb03), but other electro- optic materials such as potassium dihydrogen phosphate (KDP) or ammonium dihydrogen phosphate (ADP) could be used as alternatives. The frequency modulation circuit 140 is electrically coupled to the frequency modulation circuit by a pair of electrodes 235 positioned on opposites side of the electro optic mateπal, which allow a voltage to be asserted across the electro optic material. The electrodes 235 may comprise metallized contacts formed on the sides of the electro optic material; alternatively silver or gold epoxy, for example could be used to provide a conductive surface.
The gain medium has the form of a block with two flat opposing surfaces including a back side 222 and front side 224 that are parallel to each other. The gain medium in this embodiment a square cross section, and the front side 224 has a bevel 226 around its sides. In the preferred embodiment the gain medium 220 comprises erbium co-doped with ytterbium in a phosphate glass host mateπal ("Er,Yb: glass") available from Kigre, Inc. of Hilton Head, South Carolina. In operation, the ytterbium ions efficiently absorb energy from the pump radiation at 980 nm, and then transfer the absorbed energy to the erbium ions to create a population inversion, which allows lasing operation to begin. The erbium system has three energy levels relevant to laser operation, with the terminal state of the lasing transition lying near the ground state which makes the terminal state population of erbium sensitive to the temperature of the glass medium. The Er,Yb: glass comprises a suitable concentration of erbium and ytterbium, which may vary between designs. For example Taccheo et al , Widely Tunable Single Frequency Erbium Ytterbium Phosphate Glass Laser, Appl. Phys. Lett. 68 (19), May 6, 1996, pp. 2621 -2623, discloses a Er,Yb: glass laser with an erbium concentration of 1 x 1020 ions/cm3 and a ytterbium concentration of 2 x 102' ions/cm3. U.S. Patent No. 5,225,925 discloses a variety of concentrations and proportions, such as a proportion of ytterbium ions to erbium ions in a range between 4:1 and 20:1 , for example. Alternatively, other co doped materials could be used, such as those descπbed in U.S. Patent No. 4,701,928, entitled "Diode Laser Pumped Co Doped Laser," by Fan et al. In other alternatives, any other suitable solid state materials, glass or crystal, may be used. Such alternatives may be suitable if they have a broad gain bandwidth, which is defined herein as about 5 nm or greater.
The gain medium has a characteristic gain bandwidth, which defines a range of wavelengths within which it can lase. The Yb,Er: glass gain medium has a much broader gain bandwidth than typical crystalline microiaser gain media, which advantageously allows operation over a wide range of wavelengths. Specifically, for Er,Yb: glass, a range between about 1532 nm and 1567 nm (about 35 nm) can be obtained. As disclosed in Taccheo et al., Widely Tunable Single Frequency Erbium Ytterbium Phosphate Glass Laser, Appl. Phys. Lett. 68 (19), May 6, 1996, pp. 2621 2623, the gain bandwidth of Er,Yb: glass varies dependent upon the losses in the cavity, and by varying the
8 reflectivity of the output coupler, several ranges of wavelengths can be obtained.
The broad bandwidth of Er,Yb- glass precludes use of a conventional approach to single-mode cavity design, which would be to design the cavity with an optical path length small enough that the free spectral range (FSR) of the laser cavity will be larger than the hnewidth of the gain medium, which allows no more than one mode at any given time to experience ga above half of the maximum. For many rare-earth transitions in crystalline media, conventional non coupled cavity approaches would dictate an optical path length of at least several hundred microns (e.g. 500 microns), but for Er,Yb: glass, which has a broad bandwidth, conventional non coupled cavity designs would require an optical path length so small (e g. about 35 microns), that sufficient gain would not be available for laser operation. In order to overcome this problem and still provide single mode operation, a coupled cavity configuration is utilized.
A coupled cavity configuration, defined within the main laser cavity 132, includes a first cavity 260 defined between opposite ends of the electro optic material and the gain medium, and a second cavity 270 defined between opposite sides of the polarizer 230 (i.e. the thickness of the polarizer determines the dimensions of the second cavity). The interface between the first and second cavities compnses a partially reflective coating 275 that provides mode selection via coupled cavity effects.
In one embodiment, by using the polarizing etalon to select a single wavelength, and providing the appropriate reflectivity on the output coupler to select a predetermined cavity loss lasing operation can be achieved at any wavelength within the range of about 1530 to about 1570. The etalon may be formed with a predetermined thickness within a range of 80 to 120 microns, and preferably about 100 microns. The ability to utilize a single manufacturing process to produce lasers with different wavelengths is advantageous to produce cost-effective lasers for communications systems such as dense wavelength division multiplexing ("DWDM") systems, in which multiple beams, each having a separate wavelength, are modulated separately, and then multiplexed together and transmitted simultaneously along an optical fiber.
It has been found that a reflectance of about 20% is effective; however, it is believed that other reflectivities (e.g. between about 15% and 25%) may provide suitable performance in some embodiments. The partially reflective coating 275 defines the separation between a first cavity shown at 260 and a second cavity shown at 270, all within the main laser cavity 132.
In order to provide a back mirror on the back end, the electro optic material is coated with a first optical coating 280, which has a high reflectance (e.g. 99.9% or more) at the lasing frequency in order to promote lasing operation. The first coating is also coated for anti reflection at the pump frequency, in order to allow passage of the pump beam. An output mirror is provided on the front end 136 by coating the polarizer with a second coating 282 that is substantially reflective at the lasing frequency. In one embodiment the reflectivity of the second coating 282 is about 97%; however, the reflectivity of the second coating 282 may be varied to obtain different lasing wavelengths. Depending upon the embodiment, the second coating may be reflective or traπsmissive at the pump frequency. The interface between the electro optic mateπal and the gain medium may remain uncoated, or may be coated with a coating 290 for low absorption and/or antireflection.
The polarizer has the form of a thin (e.g. about 100 microns) etalon with two flat opposing parallel polished surfaces including a back side 232 and the front end 136. The parallelism between the fiat sides of an etalon should be less than λ/4, and if possible less than λ/10 In the embodiment shown in Figs 4 and 5 the polarizer has a square cross section (e.g. 2 mm), alternatively other cross sectional shapes, such as circular, can be used. In one embodiment the polarizer 230 is oriented to so that the E-field parallel to the y axis is transmitted. The polaπzer 230
performs multiple functions: in addition to defining the second of two cavities in the coupled cavity configuration, the polaπzer also functions to restπct laser operation to one poianzation. Polaπzed emission is required by the external (i.e. outside the cavity) modulators currently used in communication systems. In one embodiment the polarizer compnses a poianzation selective material such as Polarcor™ optical material, which is type of polarizing glass available from Corning Incorporated, Corning, New York 14831. Polarcor™ is a broad bandwidth, high transmission, dichroic glass polarizing matenal made of borosihcate glass with aligned silver particles, having an index of refraction of n - 1.510 at 1550 nm, and an extinction ratio greater than 10,000:1. Alternatively, other suitable polarizers may be used.
The polarizer 230 has a thickness that is relatively thin (e.g. less than about 200 microns and preferably about 100 microns). In embodiments that utilize Polarcor™, which is presently available only in a 200 micron thickness, the thickness is reduced and both sides are polished. In alternative embodiments, other polarization selective mateπals may be used, such as a thin (e.g. < 100 micron) etalon of silicon where polarization selectivity is achieved by thinly etched diamond-turned parallel lines along one eigen-axis of the silicon crystal. One advantage of use of silicon is that the resulting Fresnel reflection at the interface between the polarizer and the gain mateπal could be sufficient to provide the intra cavity reflection required for the coupled cavity laser, without requiring the partially reflective coating 275. In alternative embodiments that do not require a polarized output, the polaπzer can be replaced with any suitable optical material, such as optical glass or sapphire, which could provide more effective heat transfer.
Generally, the coupled cavity laser described herein may be implemented in a vaπety of ways. One implementation is described in Ser. No. 08/988,947, entitled "Laser Assembly Platform with Silicon Base" filed December 1 1 , 1997, assigned to the same assignee, and hereby incorporated by reference herein. The following specifically describes one such assembly process and platform in which the elements of the telecommunications laser are mounted on a silicon optical bench (or "base"). This process can provide a low cost, reliable method for manufacturing the telecommunications laser described herein; alternatively, conventional processes could be used. Generally, the silicon base begins as a monolithic structure that is precisely formed with features for affixing the various laser components. The same or additional features may also clearly define the position of the laser components to within very close tolerances. Silicon is an advantageous material for the base for any of a number of reasons: it is readily available, it has good thermal conductivity and it has a low coefficient of thermal expansion. Silicon in a single-crystal form is particularly useful, although polycrystalline silicon could be used for some applications, as could other materials with similar characteristics. Use of a single crystal monolithic silicon base for all of the laser components that require alignment has a number of advantages: one advantage is that all the laser components can be easily situated within close tolerances, which is very useful for the final alignment step, and another advantage is that a single TE cooler can be used, which simplifies design and reduces cost. However, the advantages of silicon are complicated by the fact that the materials that are typically used to hold and support optical components, such as copper, generally cannot be directly connected to silicon.
Reference is first made to Fig. 6, which is a flow chart that generally illustrates the process for making a laser assembly using a silicon base. A first step 310 is providing a silicon blank, which can be a monolithic, single- crystal piece, and forming the desired features thereon. Generally, one surface of the silicon blank can be taken as the reference plane, and the other surfaces are formed in a predetermined relationship with the reference plane, such as parallel or perpendicular. The well defined crystal axes of single crystal silicon permit etching as a means for forming very precise features that can be used to simplify alignment of the laser elements during subsequent
10 assembly. Also, etching techniques, such as chemical etching, are well known in the semiconductor industry and provide a very precise and reproducible method for forming the desired features on the silicon base. Advantageously, etching techniques provide a way to produce large quantities of precise, uniformly formed silicon bases at low cost. Alternatively, the sihcon base can be precision machined using diamond turning techniques to form the desired features.
At step 320, one or more of the predetermined silicon surfaces on which attachment of the laser elements is to occur are first plated with a suitable material such as nickel followed by gold ("nickel-gold"). The plated surfaces are then pre-tmned at the locations on the silicon where soldering is desired.
Next, at step 330, welding stπps are soldered to the plated, pre tinned surfaces of the silicon platform at predetermined locations, using any suitable solder such as, for example, Sn/Pb/Cd solder having ratios of
51.2/30.6/18.2, which has a melting point of about 145°C. The welding strips are positioned appropπately, and then soldered to the silicon platform in order to provide a suitable surface for welding components thereto. The welding strips may comprise a material such as kovar that is suitable for soldenng to a plated and pre tinned silicon surface. Advantageously, kovar has a small coefficient of thermal expansion (about 5 x 10 δ m/°C.) similar to that of sihcon (about 4.6 x 10 ° m/°C ). Also, kovar has good thermal conductivity (about 14.2 W/m/°C). Additionally, kovar has good absorption characteristics at a wavelength of 1 06 microns-the typical wavelength of commercially available Nd:YAG welding lasers-which means that kovar is well suited for laser spot welding. For example, about 7.0 Joules is needed to spot weld a lap joint of 0.3 mm kovar, and about 4.0 Joules is needed for a filet joint. Furthermore, kovar is relatively easy to machine in comparison with invar. However, other mateπals may be used as the welding stπps.
Next, at step 340, a laser component is placed in its exact desired position, which may require optical alignment using a small HeNe laser, for example. Once in position, the component may be clamped to hold it duπng the welding steps. Next, at step 350, the component is spot welded to the welding strip using a bracket or other suitable connecting device. Spot welding is performed using a laser with a wavelength, power, and beam size suitable for welding the component and welding strip together, such as a high power pulsed Nd:YAG laser. For each of the laser components, the positioning and welding steps are repeated. Advantageously, such a welding technique provides a quick and permanent connection. Laser welding is a particularly useful manufacturing technique for all components that require alignment.
Sometimes it may be preferable to connect components directly to the unplated silicon. For example, a thermistor may connected directly to the base in order to continuously monitor its temperature. To attach components directly to the silicon, a one-part thermal epoxy, such as available from Epoxy Technology of Billenca, MA, can be used. Such an epoxy will not outgas until 300° C and 107 torr; however, curing the epoxy requires heating to high levels (e.g. 85° C. for twelve hours). Alternatively, a soldenng process may be used for some mateπals; for example, aluminum nitride can be soldered directly to silicon that has been plated and pre-tmned. Reference is now made to Fig. 7, which is an exploded view of one embodiment of an implemented telecommunications laser A silicon base 400 is shown having features formed thereon for positioning and connecting laser components, these features including a flat diode platform 410, a flat optics platform 420 formed below the diode platform, a vertical abutment 430, and a flat output coupler platform 440. In preparation for the assembly process, certain surfaces have been plated and pre tinned, including a pair of surfaces 412 on the diode platform 410, a surface 422 on the optics platform 422, a surface 442 on the output coupler platform 440, and the underside of the silicon base. The silicon base 400 is connected on its underside to a conventional thermoelectric
1 1
("TE") cooler 450, which is utilized to control the temperature of the sihcon base by maintaining it at a constant temperature. The "cool" side 452 of the TE cooler is attached to the plated and pre tinned underside of the silicon base, by for example soldering The opposite (i.e. "hot") side of the TE cooler 92 is connected onto a suitable frame 460 by any suitable means such as soldenng. The frame 460 acts as a heat sink for the TE cooler, and also can provide support for electπcal connections to and from the laser components through, for example, a hole 462 formed on one side of the frame. A second hole 464 may allow the optical fiber to pass through. After providing for passage of appropπate cables, the holes 462 and 464 can by sealed by any appropriate means, and the frame 460 can sealed with a cover 466 and held in place by screws 468 to prevent contamination of the optical components by external sources such as dust Referring now to Figs 8, 9, and 10, a more detailed description of the laser assembly and their manufacture will be provided Figs. 8, 9, and 10 respectively show a cross sectional side view, an exploded side view and a top view of the telecommunications laser implemented on the silicon base 400. As discussed for example with reference to Fig. 1 , the components assembled on the base include the pump source 1 10, the photodiode 119, the coupling optics 120, the laser assembly 130, the optical isolator 150, the fiber optic coupler 155 and the photodiode 168.
Alternatively, one or more additional optical components, such as an optical modulator shown in Fig. 13 may be mounted on the base For example, an optical modulator may be situated between the isolator 150 and the coupler 155.
A laser diode assembly 500, including a laser diode chip 502 and a metal block 504 that may be formed of copper, is mounted on the diode platform 410. The laser diode chip 502 comprises any suitable laser diode as long as it provides the desired output characteristics such as wavelength and power. In one embodiment, a conventional broad area laser diode for optical pumping at an appropriate wavelength is used, for example at 980 nm. Alternatively, a laser diode bar or an array of laser diodes may be used in place of a broad area laser diode. To affix the laser diode assembly 500 in position, a pair of welding bars 510 are first soldered in predetermined positions to the plated, pre tinned diode platform 410. The laser diode assembly 500 is then situated between the two welding bars 510, which aid in roughly positioning the laser diode assembly, in order to more effectively conduct heat from the laser diode into the base, a suitable foil such as an indium foil (not shown) can be placed between the laser diode assembly and the base.
An alignment step may be utilized to align the emitter of the laser diode in its desired position on the base, using any appropπate technique, and once the laser diode is properly aligned, the laser diode assembly is clamped in position for welding. To couple the iaser diode assembly to the welding bars, a pair of L shaped brackets 512 are positioned so that one side of the "L" contacts the welding bar and the other side contacts the Iaser diode assembly and then spot-welded using a laser welding process in which a high energy beam of Iaser radiation is directed at the spot to be welded. The welding spots are indicated by dark spots in Fig 10 In one embodiment the L shaped brackets 512 are formed of kovar which can be readily welded to the kovar welding bars 510.
The coupling optics 120 include a ball lens 520 to focus optical radiation from the Iaser diode 502 into a focal point within the gain medium of the Iaser assembly 130, thereby providing high intensity optical radiation to pump the gain medium The ball lens may be antireflection coated. Alternatively, other types of known focusing optics can by used or the coupling lens may be omitted. For example the Iaser diode may be situated sufficiently close to directly pump the gain medium without the need to focus the pump radiation, a configuration commonly termed "close-coupled"
12
The ball lens 520 is situated in a lens holder 522 which may be a metal cylinder. The lens holder 522 is placed on the silicon base in a predetermined location, and may be aligned using a HeNe alignment beam. Alternatively, silicon etching techniques could be used to provide precisely positioned alignment marks and/or πdges on silicon base to align the lens holder reliably and quickly A single welding pad 524 formed of kovar is connected to the optics platform 420. A U shaped clip 526 or other suitable structure, also formed of kovar, includes two legs that straddle the lens holder 522 and hold it in position The two legs of the U shaped clip 526 are then spot welded to the welding pad 524.
Referring now to Figs. 1 1 and 12 in conjunction with Figs 8, 9, and 10, a further descnption of the Iaser unit 130 and its assembly will be provided. Fig 1 1 is an exploded view of components for mounting the Iaser assembly 130 on the silicon base, and Fig. 12 is a top view of a mounted Iaser assembly. In the embodiment descπbed elsewhere with reference to Figs. 4 and 5, the Iaser assembly 130 includes the electro optic modulator 210, the gain medium 220, and the polarizer 230, all coupled together by optical contact, for example. A Iaser cavity is defined by optical coatings formed on opposing ends 134 and 136 of the Iaser unit 130. Such a Iaser cavity may define a flat flat configuration, a curved flat configuration, or any other suitable Iaser cavity configuration. Other embodiments could include one or more optical elements affixed to the solid state ga medium, such as a heat spreader
A mounting structure, shown generally at 529, includes two mounting arms 530 and 531 on which a pair of electrodes 533 have been formed. The mounting arms 530 and 531 define a center opening 536, and the electro optic material 210 has a shape that fits within this opening As illustrated in Fig. 12, which is a top view of the Iaser assembly assembled to the mounting arms, the back side of the gain medium 220 includes surfaces 538 extending outwardly from the boundaries of the electro optic unit These surfaces 538 directly abut the mounting arms with the electro optic material situated between the mounting arms. Advantageously, the abutting relationship between the gain material and the mounting arms provides a thermal path for heat flow from the gain material, through the mounting structure 529, and to the silicon base 400 The Iaser assembly 130 may be soldered to the mounting arms by, for example a gold solder. One advantage of using a metal solder is that it provides a conductive path from the electrodes 533 to the electro optic material. Alternatively any suitable method, such as thermal epoxy, may be used.
The mounting structure 529 is affixed to the silicon base 400 in a position so that the Iaser unit can be optically pumped by pump radiation from the Iaser diode 502 The sihcon base has a vertical face 532 formed in the abutment 430 at a predetermined distance from the ball lens 520. The mounting structure 529 has a vertical surface 534 formed below the mounting arms, which abuts against the face 532 on the abutment. The mounting arms may also rest partly on the optics platform 420. As a result, the Iaser unit 130 can be positioned approximately within the proper z plane (i.e. along the lasing axis) Also, the interface between the face 532 and the surface 534 may have a close, thermal contact, in order to conduct heat from the Iaser unit 130 into the silicon base 400, where it can be dissipated. Thermal conductivity between adjacent surfaces may be enhanced by use of heat conductive materials such as indium foil or thermal epoxy
To affix the mounting arms to the base, a welding strip 540 is first soldered across the bottom of both of the mounting arms A pair of L shaped brackets 542 (Figs 5 and 6) are then used to permanently affix the mounting arms by spot welding. Before welding, the iaser unit 130 may be aligned to the proper x y coordinates by any suitable method such as using a HeNe Iaser. In some circumstances, it is desirable to electrically isolate the Iaser unit 130 from the conductive silicon base 400 while still providing a path for heat to flow from the solid state Iaser unit to the base In one such
13 embodiment the mounting arms that connects the Iaser unit with the silicon base may compπse aluminum mtπde ("AIN"). The electrodes 533 formed on the mounting arms may then be used to deliver the required dπve voltage/current to the electro optic modulator of the Iaser unit with the necessary bandwidth. Alternatively, the modulator may be electrically coupled via a wire or any other suitable method. Advantageously, AIN can be spot- welded to kovar using a high power Iaser by first metallizing predetermined locations to provide soldenng pads, which can then be soldered to kovar brackets, which in turn can be Iaser welded to kovar welding stπps on the base. Furthermore, AIN has a coefficient of thermal expansion very close to that of silicon, it has high thermal conductivity (170 W/m/°C), and it has a low thermal expansion coefficient (4.6 x 106 m/°C). However, alternative materials can be used, and it would be advantageous if the alternative material is thermally conductive and has a thermal expansion coefficient approximately equal to the thermal expansion coefficient of silicon.
Referring again to Figs. 1 , 8, 9, and 10, an output assembly 600 includes a cylindrical metal housing 605 for the optical isolator 150, a partial reflector 162, a fiber coupling lens 164, a fiber ferrule 161, and an optical fiber 160. The housing 605 can be machined from stainless steel with an appropriate shape to accommodate the elements that fit within it A hole 607 is provided in the housing in a position to allow light reflected from the partial reflector 162 to be detected by the photodetector 168.
To connect the Iaser output assembly 600 to the base 400, a kovar welding pad 610 is soldered to the output coupler platform 440 The Iaser output assembly 600 is then placed in its preliminary position and aligned. Next, a U-shaped kovar bracket 612 is put in position over the Iaser output assembly and the bracket 612 is spot welded to the kovar welding pad 610 In this embodiment, it may be desirable to maintain the Iaser diode assembly 500 and the Iaser unit 130 at different, optimal temperatures, even though both components are connected to the same base. We have found that the single base can still allow for thermal differences between optical components by proper selection of the temperature of the base 400 (the "base temperature") and proper selection and control of the drive current applied to the Iaser diode 502. Because the heat generated in the Iaser diode is a function of the dπve current; therefore its temperature is approximately a function of the drive current and the temperature of the silicon base 400. Also, the temperature of the Iaser assembly 130 is approximately a function of the pump power received from the Iaser diode and the temperature of the silicon base. Thus, the pump power and the base temperature can be considered as the two vaπables in two simultaneous equations that determine the temperatures of the Iaser diode and the Iaser element. The thermal resistance between the Iaser diode assembly 500, the Iaser assembly 130 and the base 400 can be considered to be approximately constant. Therefore, by monitoring the output of the iaser assembly while varying the base temperature and the pump power, the base temperature and the pump power for optimizing the desired Iaser output of the Iaser assembly can be determined. Then, the optimized Iaser output can be maintained by using feedback and control techniques for the temperature of the silicon base and the optical pump power. Referring again to Fig. 7, the base temperature can be maintained and controlled by, for example, controlling the single TE cooler 450 by any suitable means. In order to monitor the temperature of the silicon base, a temperature sensor such as a thermistor 620 may be mounted in a hole 622 in the sihcon base,, which provides a signal on a connection 624 to a controller 626, such as a microprocessor, which controls the TE cooler over a connection 630. The controller can thereby monitor and control the temperature of the base 400. To monitor the optical pump power, the photodetector 1 19 (Fig 1 ) is situated to detect a portion of the optical pump radiation, and supply a signal to the diode driver 1 17 Alternatively, the output from the photodetector can be provided to the
14 controller 626, which can control the dπve current responsive thereto to insure that the desired optical pump power is supplied from the Iaser diode.
Reference is now made to Fig. 13, which is a system diagram of a communications iaser installed in a wavelength division multiplexing (WDM) communication system A Iaser source 701 including a coupled cavity, optically pumped solid state laser as described herein (e.g. with reference to Fig. 1 ) supplies optical radiation along the optical fiber 160 at an approximately constant predetermined power and wavelength. The optical radiation is modulated in an external modulator 711 with any suitable information signal, such as a cable TV signal, or a digital signal. The external modulator comprises any suitable form such as a Mach-Zender interferometer; an electro absorption modulator, a waveguide, or a bulk modulator. External modulators are commercially available from Lucent Technologies of Breinigsville, Pennsylvania or Uniphase Telecommunications Products of San Jose, California. Using these external modulators, the Iaser output can be amplitude modulated, phase modulated, and/or frequency modulated for coherent communications, using techniques such as frequency shift keyed or phase shift keyed formats at ulti-Gbit/second data rates.
Additional Iaser sources, such as Iaser sources 702 and 703 may also supply optical radiation at a constant, predetermined power and wavelength, although the wavelength may differ between the Iaser sources. These Iaser sources are respectively modulated in external modulators 712 and 713. Depending upon the embodiment, many additional Iaser sources may be utilized, each having a slightly different wavelength.
After being modulated by the external modulator 71 1, 712, and 713, the modulated optical signals are supplied to a multiplexer 720, such as a wavelength division multiplexer, which multiplexes the modulated optical signals into a single optical signal, so that they are transmitted along a single optical fiber 730. In an alternative embodiment without the multiplexer 720, a single modulated signal, such as the modulated optical signal from the first Iaser source, may be transmitted by itself along the optical fiber 730.
Generally, relatively short runs of optical fiber will not require amplification. For example, if a 100 mW Iaser source is modulated by an external modulator that attenuates the signal by a factor of four, and a typical single mode fiber is used that attenuates the signal by 0.22 dB per kilometer, amplification of the modulated signal will not be required for optical fiber lengths less than 60 km, thereby eliminating the need for light amplification systems. However, for long runs of optical fiber, or if other transmission losses are present, it may become necessary to amplify the combined modulated optical signals in an optical amplifier 735 such as an erbium doped amplifier commercially available from Lucent Technologies of Breinigsville, Pennsylvania or Uniphase Telecommunications Products of San Jose, California.
When the multiplexed optical signals arrive at the receiving end, they are de multiplexed into their respective wavelengths in a de multiplexer 740, and then each of the individual modulated optical signals is individually demodulated in receiver/demodulators 741 , 742 and 743 to provide the respective signals. After demodulation, the demodulated signals are then supplied by any suitable means to their respective end users for their intended uses such as watching television.
It will be appreciated by those skilled in the art that alternative embodiments may be implemented without deviating from the spint or scope of the invention For example, one or more additional components such as a light modulator and/or a polarizer can be situated on the optical base. Various suitable materials and methods can be utilized to hold the Iaser components to the base, including various adhesives and mechanical fasteners. This invention is to be limited only by the following claims, which include all such embodiments and modifications when viewed in conjunction with the above specification and accompanying drawings