This invention relates to opto-electronic systems using semiconductor lasers driven by feedback control circuits which control the laser's optical phase and frequency. Feedback control provides a means for coherent phased array operation and reduced phase noise.
The optical analogs of electronic components such as amplifiers and filters have undergone significant advances with the development of wavelength division multiplexed optical communications systems. However, several important electronic components, namely the voltage controlled oscillator (VCO) or current controlled oscillator (CCO), do not have high performance optical equivalents. A high performance optical VCO/CCO has the potential to play a key role in future optoelectronic systems, comparable to the role of its radio frequency (RF) counterpart in phased-array radar systems.
The optical CCO functionality can be realized in a primitive fashion by use of a standard semiconductor distributed feedback (DFB) laser. The “FM” or frequency modulation response of the DFB laser has the potential to provide extremely high bandwidths in excess of 20 GHz. However, the frequency of semiconductor lasers depends in a relatively complex way on the level of injection current and these lasers exhibit the potential for frequency mode-hopping, phase inversion and hysteresis. Typically, the FM response or CCO gain is highly frequency dependent and exhibits a 180 degree phase reversal for modulation frequencies in the vicinity of 1 MHz. The phase reversal occurs when the modulation frequency is sufficiently high that the out-of-phase thermal FM response dominant at low frequencies vanishes, leaving only the in-phase electronic contribution. The competition between thermal tuning and electronic plasma tuning is known to be a significant barrier to designing a fundamentally stable, high bandwidth optical phase-locked loop (OPLL).
For the laser LO to precisely track the phase of the reference oscillator (RO), while overcoming the LO's intrinsic phase noise, it is known in the art that the OPLL circuit bandwidth should be designed to provide ten to a hundred times the resulting LO/RO beat note linewidth. To provide this relatively large bandwidth with low phase lag, the physical delay of the OPLL (both optical and electrical) is typically no more than the 1/10 of the inverse bandwidth of the circuit. This is typically a challenging condition to satisfy because of the need for high speed and compact circuitry exhibiting low time delay.
The optical performance of an OPLL system is typically quantified by calculating the residual rms phase error between the local oscillator laser and the reference laser. The rms phase error resulting in 95% coherent power combining is 0.4 rad. For typical laser linewidths of 10 MHz, this requires at least 100 MHz of loop bandwidth. Standard, commercially available DFB lasers do not typically exhibit a well-behaved FM response for frequencies from dc up to 100's of MHz.
A two-section distributed feedback (DFB) laser can be designed to produce an FM response with relatively constant amplitude and phase from dc frequencies up to several GHz. These two section DFB's are typically designed to maximize their tuning coefficient or “CCO gain” to levels in excess of several GHz/mA so that their electronic tuning response overwhelms their thermal response. Alternately, they can be designed to null out the high frequency FM response to produce low parasitic chirp.
The magnitude of the CCO gain directly impacts the OPLL performance. In actual phase-locked loop implementations, it is important to minimize the impact of current noise in the phase-locked loop feedback signal from degrading the laser's spectral purity. Therefore, it is advantageous that a two section laser be designed such that the magnitude of the CCO gain is less than 1 GHz/mA, preferably a few 100's of MHz/mA. Typical two section lasers have significantly larger FM coefficients. In addition, typical two section DFB's provide relatively low optical output powers of a few 10's of mW. For those applications requiring high optical power, new lasers designs are required.
To achieve high optical output power, an array of relatively low power semiconductor laser elements may be used. The practical realization of arrayed semiconductor laser-based OPLLs impose several requirements on the laser: they must be single longitudinal mode/single frequency; the phase of the laser's FM response must be relatively constant over the bandwidth of the feedback control circuit; the laser output power should be greater than or equal to 1 W per emitter; the lasers must be surface emitting; the lasers across the array must be fabricated close in wavelength so they can be tuned to the same optical frequency by changing their bias currents; the laser array layout must allow for compact integration with high speed electronics, and the multi-section laser should be monolithic. At the present time, these multiple and varied laser characteristics have not been realized in a single laser structure, much less an array. In addition, prior art phase-locking approaches have not been compact, integrated nor scaleable, and have not been extended to laser arrays.
In this invention, we disclose phase-locked semiconductor lasers whose optical phase and frequency characteristics are precisely controlled by use of high speed integrated circuits. Potential single frequency semiconductor laser elements include the vertically cavity surface emitting laser (VCSEL) and distributed feedback laser (DFB). For example, an emitter may be comprised of a two section DFB oscillator driven in an asymmetric, push-pull configuration to provide controlled and well-behaved frequency modulation response. The lasers are arranged as individual elements, bars, or two dimensional arrays. In a further example, each DFB laser includes a tapered, electrically pumped optical amplifier section to increase the optical power.
Laser designs which satisfy the unique requirements of phase-locking advantageously provide for high optical power, high electrical efficiency, high beam quality, single temporal and spatial mode output, and constant phase FM response over a bandwith in excess of 100 MHz are disclosed. These design features enable the optical fields of large numbers of lasers to be coherently combined to produce a high brightness semiconductor laser source. In addition, the phase of each laser within an array can be locked to be exactly in-phase with the reference laser or with programmable phase offsets. Electronic frequency and phase-locking is achieved by high-speed integrated electronics that provide both a large electrical bandwidth as well as the control and functionality necessary for stable coherent beam combination. Alternate opto-electronic implementations provide a low noise laser source or a mode-locked pulse train. Implementations to provide beam steering and beam shaping features are also disclosed.
In this invention we disclose techniques for coherent optical beam combining of one or two dimensional semiconductor laser arrays driven by optical phase-locked loops (OPLLs).
The laser outputs 11 are collimated by a lens array 22 which produces a composite, collimated output field 15 with an effective aperture given by the dimensions of the laser array 24. Each lens of the lithographically patterned GaP micro-lens array is in precise alignment with the corresponding emitter element. The curved surfaces of lens array can be either on the local emitter side or the reference laser side of the optical system, depending on considerations of optical abberation and backreflection management.
Individual OPLL circuits inject current into the laser to modulate the emission frequency to synchronously drive the lasers. A pickoff mirror 38 reflects a small fraction (˜0.1%) of each laser's output 11′ back onto its photodetector 12. A reference laser 40 with output 10 is directed by a beamsplitter 36 to be colinear with the reflected outputs 11′. The lens array 22 then focuses the colinear reference output 10′ into individual spots which overlap with beams 11′ and optically mix in each photodetector 12. This electrical mixing signal serves as the input to the electronic feedback control circuit.
Each OPLL circuit receives a phase control input produced by controller 51. The individual phase control inputs set the relative phases of each laser emitter 14. The phase control can be programmed to give a target waveform based on real time measurements from a wavefront measurement apparatus 34, for example. In one implementation, this waveform can be set to provide a diffraction limited output by maximizing the optical power passing through a diffraction limited aperture. Additionally, the relative phase of each laser element may be updated at a high rate to provide adaptive wavefront control.
More specifically, the phase control unit 51 utilizes one or more detector arrays, such as a charge coupled detector (CCD) or CMOS detector, to measure the intensity profile at one or more locations along the beam. For example, unit 51 may include a shearing interferometer or a Shack-Hartmann type interferometer, which uses a lens array to transform phase variations to position variations of focused wavefront elements on the two dimensional detector array. Alternately, an aperture followed by a photodetector can be utilized to provide a measure of “times diffraction limited” by determining the power-in-the-bucket. The phase control unit 51 includes electronic signal processing and digital logic to translate these measurements into an optimal set of phase control outputs for each laser element. For example, the phase of each laser element is dithered at a particular frequency and its effect on the composite wavefront identified by extracting that frequency component from the wavefront measurement.
Alternately, in
The output of the mixer 57 is a baseband signal which is input into loop filter 58, for example, a passive lead/lag type with a pole and zero to provide a second order PLL response. A phase/frequency detector may be used in lieu of the mixer. Phase locked loops are characterized as first order, second order, or third order, based on the number of integrators in the loop. It is further advantageous for this loop filter to include an electronic integrator which holds the laser bias current necessary to maintain locking under thermal drift, for example. In this case, the PLL circuit is third order.
The OPLL circuit advantageously includes an acquisition function 53 which sweeps the LO laser frequency until a beam note within the bandwidth of the photodiode 12 is detected. The acquisition circuit tunes the bias current at 52-1, using a search algorithm based on stepping through or ramping the current, for example, until the baseband beat note is detected within the loop bandwidth of the OPLL.
The output of the loop filter 58 is summed with the dc bias current and input to the gain section of a local oscillator. In a particular example, the output of the loop filter 58 is input into a current amplifier 56 which is summed with bias currents 52-1 and 52-2 and injected into the two oscillator sections 16-1 and 16-2 of laser element 14. The feedback current signals are split into two paths, one of which is summed with the section 2 bias current and injected into section 2, and the other which is input into an inverting gain stage 54 to provide the proper ratio of modulation currents, summed with the section 1 bias current and injected into section 1. A constant current supplied by source 52-3 drives the MOPA laser amplifier section 18. The physical size of the actual circuit and the resulting time delay through the feedback loop is preferably kept as small as possible (i.e., below 1 ns) to enable a feedback loop bandwidth of about 100 MHz.
Example: Hybrid Integration of PLL Circuitry with Laser Array
The emitters are densely packed with adjacent rows of emitters offset from one another, with the ratio of their x and y spacings nominally equal to the laser beam x and y divergence angles. By use of a collimating lens array and Fourier filtering optics above this laser plane, the MOPA outputs form a single, coherent beam of high spectral and spatial purity. The outputs 11 of the in-plane lasers are directed out-of-plane by use of well known fabrication processes that selectively etch deflection mirrors 19 at precise and consistent angles along a crystallographic plane. This produces identical beam deflection angles for all emitters in the array. Alternately, a diffraction grating based output coupler 19′ may be used to direct the laser output 11 out of the plane of the substrate and also to potentially focus the beam for collimation and/or to correct for beam astigmatism.
The laser array requires an external optical system to achieve coherent aperture filling and distribution of a portion of the reference laser beam onto each photodetector. Such an integrated laser system is illustrated in
To collimate the arrayed emitter outputs 11-1, . . . 11-N into a single output beam, a diffraction-limited lens array 22 fabricated of GaP or an equivalent high index of refraction and low optical absorption material is utilized. Dead zones between lenses resulting from fabrication limits are typically 100 microns or less and result in an over 90% effective fill factor. Lenslet array 22 focusing elements are preferably interleaved in an A-B-A-B-A-B pattern to maximize the packing density. Microlens arrays may use toric surfaces to simultaneously collimate both axes simultaneously (as illustrated in
The use of quarter waveplate pairs 37-1, 37-2 rotates the polarization of reflected local oscillator output 11′ to prevent extraneous optical feedback from coupling back into the local oscillator 14. The baffle array 39 further prevents optical crosstalk between adjacent emitters 11-i and 11-j, where j is not equal to i. A polarization beam splitter 36 is placed behind the lens array 22, to couple the reference laser 40 back through the optical system and direct a fraction of it onto each OPLL's photodiode 12. The angle of the RO beam 10′ is selected such that it is aligned with each detector 12 and LO optical beam 11′-j to produce a mixing signal with high contrast.
The reference laser 40 is directed in a counter-propagating sense through the common optical system and its output beam 10 is polarized orthogonal to the local oscillator outputs 11, preventing optical injection locking of the reference laser into the local oscillator lasers. The reference laser 40 is distributed onto each OPLL detector 12 by first passing through a beam expander 41 to increase the reference laser output beam diameter such that the entire laser array 24 aperture will be filled. The reference laser output 10 is polarized orthogonal (p) to the laser array output 11 (s) and a polarization beam combiner 36 allows the reference beam 10′ to propagate back through the lens array system and also to efficiently out-couple the laser array combined output 15, without experiencing significant insertion loss. To prevent undesired optical interactions between the laser array and the reference laser, an optical isolator 44 is placed immediately in front of the reference laser.
This coherent laser array system has several design features to promote stable, phase-locked operation: (1) the polarizer 35 in front of detectors suppresses mixing noise arising from stray reflections and scattered light; (2) tilting of optics so their surface normal is not coincident with beam propagation directions prevents back reflections from coupling back into lasers; (3) the micro baffle array 39 blocks-out optical crosstalk between adjacent emitters; (4) electronic filtering rejects unwanted beat signals arising from adjacent emitters; (5) the use of quarter wave plates 37-1, 37-2 and polarizer 35 allows a well controlled LO signal to be directed back to each OPLL, and (6) the isolator 44 in front of the reference laser 40 prevents LO outputs from being coupled back into the RO.
In a further example, the total reference laser power is 1 W uniform across the laser array aperture. For an array of five thousand 2.5 W lasers, approximately 0.1 mW of reference power falls onto each local detector. In addition, a fraction of each laser emitter's output power is simultaneously reflected back onto each local detector.
By operating the laser array with the acquisition loop activated, the optical outputs of an array of vertically emitting, high-power single-mode DFB-MOPA lasers are independently tuned until their frequencies lies within the locking range of the circuit (typically 100's of MHz to 10's of GHz), after which the phase lock control is activated and the laser frequencies are rapidly pulled-in and locked to the common reference laser. All laser elements 14 are driven electronically such that they are forced into phase synchronism with one another and are mutually phase coherent.
This laser array approach is extendable to systems producing diffraction limited optical output powers of 10-100 kW for large numbers (e.g. thousands) of lasers. In a particular example, the array of single mode, high power (1-5 W), vertically coupled, two section DFB lasers are phase and frequency locked to a single reference laser in the wavelength range of 700 nm to 1600 nm by the use of an array of OPLLs, including integrated GaAs optical detectors and high speed SiGe BiCMOS integrated circuits with critical feature sizes of 90 nm to 250 nm. In a particular implementation, each of the local oscillators are locked to the same rf offset from the reference laser 40, the offset typically in the range of 0.5 to 5 GHz. In this case, an additional rf clock signal (0.5 to 5 GHz) is distributed across the surface of array 24 (not shown in
Example: PLL Electronics Physically Separate from Laser Array
In an alternate embodiment, the electronically phase-locked laser array is comprised of stacked, one dimensional arrays of single mode edge emitters.
Example: Semiconductor Laser Phased Array with External Optical Amplifier
In an alternate embodiment, relatively low power, single frequency VCSEL laser emitters 14′ or DFB lasers 14 are coherently combined by use of electronic feedback.
The phase of each emitter 14 or 14′ is controlled by phase control unit 51 to produce an optical phased array source in which the phase of each beam segment corresponding to a particular OPLL element can be programmed arbitrarily and with high speed.
Example: Single Temporal Mode Semiconductor Laser Elements
Electronic phase locking of semiconductor lasers places two fundamental requirements on individual emitter elements. First, the laser should emit at a single frequency or single temporal mode. Typically, the optical power at other frequencies, for example, in spectral sidebands, should be less than 1% of the power in the central peak. Typicaly, semiconductor lasers require a frequency selective element such as a grating to filter out unwanted Fabry-Perot modes. This level of sideband suppression further requires minimization of backreflections to prevent coupling back into the laser cavity, which can produce external cavity effects. A second requirement is that the laser's FM response, or frequency change produced by a given injection current change, exhibits a response with relatively constant phase within the bandwidth of the feedback loop. For example, a laser with a 10 MHz linewidth requires a feedback loop bandwidth of 100 MHz. Over this range, the phase of the FM response should vary by less than 90 degrees. Larger phase variation (greater than 90 degrees) can lead to instability of the feedback control loop in the absence of a suitable electronic phase compensation approach.
Semiconductor laser devices which achieve these dual requirements include distributed feedback lasers (DFB's) and vertically cavity surface emitting lasers (VCSELS). Active phase locking can be accomplished at all potential emission wavelengths by use of a fast photodetector with appropriate responsivity. Typical semiconductor laser wavelengths extend from the visible (400 nm) to the near infrared (1700 nm); however, the approaches disclosed herein are not limited to these wavelengths. Typical semiconductor laser materials are comprised of the class including GaAs, InGaAs, InGaP, GaN, and AlGaAs.
Example: Two Section DFB-MOPAs Emitters with Adaptive FM Response
A laser emitter 14 exhibits a well-behaved “CCO” characteristic if the phase of its FM response is relatively constant within the feedback circuit bandwidth required for stable locking. If the FM response has a strong spatial hole burning component, for example, which is of the same phase as the thermal FM response, then it is possible for DFB emitters 14 with a single section to have a sufficiently constant phase FM response. This may be produced by proper selection of the effective phase and reflectivity of the front and rear reflectors of the DFB emitters. The desired constant phase FM response may be achieved by suppressing the front reflection to a value of less than 10%, for example.
For adaptive electronic control of the FM response, we further disclose herein a laser emitter 14 comprised of a two-section DFB oscillator 16 with an additional, monolithically integrated, tapered optical amplifier section 18. This emitter is utilized as an individual element, as a bar or as a two dimensional array. The resulting two-section DFB-MOPA laser produces both high optical power and electronically programmable FM response with well-behaved optical CCO characteristics, making it suitable for the electronic locking approach disclosed herein.
The design of a DFB oscillator with two independently driven sections adds an additional degree of freedom enabling the FM response of a given device to be electronically varied in magnitude and sign by changing the bias and modulation current ratios across the two sections. The FM response of each emitter is optimized adaptively, for example, by electronic control means. The two oscillator sections are driven in an asymmetric push-pull relationship while the amplifier section is un-modulated. In the asymmetric push-pull approach disclosed herein, the bias or “dc” current densities in the two sections 16-1, 16-2 are made dissimilar. This is achieved by injected the same bias current into the two sections of unequal length, injecting different bias currents into two sections 16-1, 16-2 of equal length, or by injecting different bias currents into different length sections. For example, if the lengths of the two sections are made equal, then the ratio of bias currents adjust the magnitude of the FM coefficient. In this example, the relative amplitudes of the modulation currents applied to each section are determined simply by the ratio of bias currents.
The electronic plasma response of the multi-section DFB laser is estimated by solving the semiconductor laser rate equations. Based on the analysis of Yariv [Opt. Letters Vol. 30 No. 17 (2005) pp. 2191-2193], the induced frequency shift of a semiconductor laser arising from the electronic plasma mechanism is:
For a two-section laser, the equivalent expression is:
where the photon density is Pj(t) for section j. Note that χ1+χ2=1, where χ1, χ2 are the length fractions of sections 1 and 2, respectively. τp is the photon lifetime at transparency, e is the electron charge, αj is the linewidth enhancement factor and εj is the gain suppression factor. Typical laser parameter values are listed in Table 1.
Semiconductor lasers also exhibit a thermal FM response arising from Joule heating and the thermo-optic effect. This effect is significant at low frequencies, from dc to typically several MHz. The thermal frequency shift is equal to:
Δωthermal(t)=χ1h1[i1(t)]2+χ2h2[i2(t)]2. (3)
where hi represents the thermo-optic response in units of rad/s-Amp2. Typically, each laser section 16 will have the same nominal thermo-optic response. Substituting in the expansion for the laser current, expressed in terms of the bias Ibj and small modulation Δi j, into equation (3) and linearly expanding for small Δij's, the first order expression for the thermal frequency shift becomes:
Δωthermal(ω)=2χ1h1Ib1Δi1+2χ2h2Ib2Δi2. (4)
Typically, the thermal tuning response is relatively large (˜0.5 to 1 GHz/mA) and, by substitution of physical constants and realistic operating conditions, is found to be 180 degrees out-of-phase in comparison to the electronic tuning response of equation 2. There is potentially also a contribution to the FM response from a spatial hole burning effect; however, this mechanism typically exhibits a high degree of variability and in many laser devices may be smaller in magnitude than thermal and plasma effects. The use of a DFB laser exhibiting an FM response with large phase variations leads to a general instability of the feedback control system. Operating in the asymmetric, push-pull configuration disclosed next significantly reduces the thermal contribution to the FM response.
In the typical push-pull configuration of the prior art, equal and opposite modulation currents are applied to two identical laser oscillator sections in a fashion which nulls-out the electronic FM response. However, in the laser system disclosed herein, a well controlled, non-zero FM response is required. By proper selection of bias current asymmetry, a corresponding asymmetry in the modulation currents introduces a non-zero electronic FM while nulling out the thermal FM. The relation between modulation currents in sections 1 and 2, Δi 1 and Δi2, respectively, which cancel out the thermal FM response is given by:
Equation (5) is the general solution for the modulation current ratio which gives an electronic-only FM response, dependent on the lengths, bias currents and thermo-optic responses of the two sections 16-1, 16-2. The actual value of the FM response of DFB lasers under operating conditions satisfying the above equation is determined by solving the semiconductor rate equations. Table 2 summarizes the calculation results for various configurations, neglecting spatial hole burning effects which can be made small. By varying the ratio of modulation currents (as well as bias currents), the magnitude and sign of the FM response can be adjusted continuously within the target range of a few hundred MHz/mA. The later four examples correspond to the asymetrical push-pull configuration, which nulls out the thermal response while extending the “constant phase” bandwidth.
In practice, even under the condition set forth in equation 5, the thermal FM response may have a small residual component due to spatial non-uniformities in the temperature and thermo-optic coefficient across the laser oscillator sections 16-1, 16-2. To achieve a relatively constant phase for the net (thermal plus electronic) FM response (<10 degrees variation), the effective two section thermal coefficient must be reduced to a value less than 25% of the electronic value. The variation in phase and amplitude of the net FM response for various relative electronic and thermal contributions when passing through the thermal crossover frequency are summarized in Table 3. In general, the OPLL is quite sensitive to FM coefficient phase variations, but relatively insensitive to amplitude variations. A variation in phase as large as 30 degrees still provides adequate phase margin to ensure effective phase locking.
Typical DFB lasers exhibit a Lorentzian linewidth of about 10 MHz. A phase-locking bandwidth in excess of 100 MHz is then required to provide reasonably efficient coherent combining. For these characteristics, the performance has been simulated using the two-section DFB-MOPA emitters disclosed herein. The results are summarized below in Table 4. The RMS phase error is calculated in the case of a “perfect” RO with zero linewidth and also for an RO linewidth equal to that of the LO (10 MHz). The corresponding rms phase errors are 0.04 wave (0.25 rad) and 0.089 wave (0.56 rad), respectively. This level of phase error enables two lasers to be coherently combined with greater than 95% optical efficiency. By extending this technique to thousands of lasers in an array format, a high power and high brightness semiconductor laser is produced.
Example: Beam Combining Optics
Each diode laser element 14 in the array produces a nearly diffraction limited, single spatial mode output 11 which is typically characterized by slight beam asymmetry and astigmatism. When these outputs 11-j are combined by a lens array 22, there remains a significant amplitude ripple 71 at the near field location in the back focal plane 70 of the lens array 22, as illustrated in
where f is the fill factor after the lens array using phase plate 74, amplitude ripple (78 in
Example: Adaptive Wavefront Control
For many of these applications, the ability to arbitrarily set the phase of each emitter at rapid rates eliminates the need for auxiliary adaptive optical systems (e.g., deformable mirrors and micromirror arrays) and, in fact, dramatically improves the performance of existing adaptive optical systems.
Shaping of the combined wavefront 15 is particularly relevant for several applications, including high power semiconductor sodium laser guide stars at 589 nm (by frequency doubling a 1178 nm diode array, for example), the management and reduction of orbital debris, lidar, and “wireless” power transfer and distribution.
In a particular example, this invention provides a new approach to sodium guide star lasers using an electrically locked laser array. The coherently locked, frequency doubled, vertically emitting high power semiconductor laser diode array provides high optical power at 589.159 nm. The semiconductor laser-based guide star offers several advantages over the prior art. First, these arrays are reliable, light-weight, compact and potentially low cost compared to present day laser guide star approaches. In addition, the high wall plug efficiency of laser diodes (60-70%) and the high doubling efficiency into the visible can produce an efficient laser source with 100's of watts of diffraction limited and single mode output power at the sodium absorption line. Furthermore, the use of coherent beam combining allows for the relative phases of the individual emitter elements 14 to be adaptively controlled at high speeds (GHz) by controller 51 to enable fast beam steering, focal shifting and adaptive wavefront compensation. This high power semiconductor laser array approach can be extended to any wavelength within the semiconductor gain region, such as the atmospheric windows of 1040 nm and 865 nm, and to powers in excess of 10's of kW.
Example: Mode Locking
In a further embodiment of this invention, semiconductor diode laser and laser arrays 24 are electronically mode-locked by configuring each laser emitter 14 as a local oscillator in an OPLL, wherein each local oscillator 14 is frequency locked to the reference laser 40 such that the difference frequency is a unique integer multiple of the pulse repetition frequency. The phases of each laser 14 are locked to be exactly in-phase, or arbitrary phase offsets can be provided. Electronic frequency and phase-locking is achieved by high-speed electronics 20 which provide both the large electrical bandwidth as well as the control and functionality necessary for stand-alone and stable mode-locked laser operation. Since the center frequency of each local oscillator 14 differs from that of the reference oscillator 40 by an integer multiple of the rf oscillator 50 frequency offset, the composite laser array output 15 has a spectrum which is a frequency comb with precise comb spacing and stable relative phase difference between each spectral component.
The electronic mode-locking of array 24 can potentially achieve in excess of 100 kW average power and 1 GW peak power from a diffraction-limited semiconductor laser diode array. The laser array is electrically and optically interfaced to an arrray of PLL circuits 20 with integrated optical detectors 12 and a reference rf oscillator 50 operating at the mode-locking pulse repetition frequency. The optical outputs of the array are transformed by beam combining optics 43 into a single near-diffraction limited spot at the output 15. The output in the locked state produces a single, high-power, mode-locked output, with a peak power given approximately by N2 (where N is the number of lasers) times the average power per emitter 14.
In a particular example, an array of N=5000 single mode, high power (2.5 W) single mode diode lasers 14 are phase and frequency-locked to a single reference laser 40 at frequency offsets equal to integer multiples of, for illustration purposes, 20 MHz by use of an array of OPLLs with integrated optical detectors 12, loop filters 58, rf mixers 57 and multipliers 59. Each OPLL operates by optically mixing the local oscillator 14 with the reference laser 40 in an integrated photodetector 12. The optical mixing process produces a current signal containing high frequency beat components arising from a mismatch between the frequencies of the local oscillator and the reference oscillator. This beat signal is subsequently mixed at rf mixer 57 with the multiplied output of a 20 MHz rf oscillator 50. Each rf multiplier stage 59 provides a different integer multiple of the rf oscillator frequency to each mixer associated with each OPLL element. The output of the rf mixer 57 is passed through a loop filter 58 to produce an error signal suitable for driving the local laser oscillator 14. Each laser 14 functions as a current controlled oscillator (CCO) with a tuning characteristic on the order of 0.1 GHz/mA. By utilizing high bandwidth electronics, the frequency of the local oscillator can track the sum of the reference oscillator frequency and offset frequency, so that the OPLL circuit can phase and offset-frequency lock the current controlled laser to the single reference laser. This same process is applied to every laser element of the array, thereby locking all lasers to a fixed frequency comb with a given free spectral range. A coherent, pulsed output (10 ps pulse width) of high average power (10 kW), high peak power (50 MW), high beam quality (diffraction limited) and high spectral purity (<20 MHz linewidth) is thereby produced at the output of the beam combining optics 15.
These mode locking approaches require wavelength combining optics to combine the multiple, spatially separate optical modes into a single overlapping output 15.
Example: Coherent Laser Power Combining
In
Beam combiner 92 is preferably a 50/50 fused coupler or 50/50 beam splitter which combine like polarizations. Beam splitter 92′ is a 50/50 splitter or alternately, an asymmetric tap coupler (e.g., 1/99%) which directs the majority of optical pwer to output beam 15 while tapping a small amount for detector 12. The outputs of lasers 14-1 and 14-2 are driven to be precisely phase and frequency locked, in addition to having a controllable relative phase relationship. The controllable relative phase relationship enables the maximum optical power to be produced in combined output beam 15.
Example: Linewidth Narrowing of Laser Emitter
In
Beam combiner 92 is preferably a 50/50 fused coupler or 50/50 beam splitter which combine like polarizations. Beam splitter 92′ is a 50/50 splitter or alternately, an asymmetric tap coupler (e.g., 1%/99%) which directs the majority of optical power to output beam 15 while tapping a small amount for control purposes at detector 12.
The purpose of locking a high power local emitter to a low power, low noise reference laser is to transfer the low phase noise characteristics onto the high power emitter. The output beam 15 then exhibits the superior optical power characteristics of laser 14 and the superior spectral linwidth characteristics of laser 40. Typical optical power is >1 W and typical spectral linewidth is <10 KHz. The emission wavelength is typically within, but not limited to, the range of 600 nm to 2000 nm. This spectral narrowing approach is of value in applications requiring low phase noise, such as spectroscopy, sensing and coherent communications.
Example: Power Combining Based on Heterodyne Optical Phase Locking
Greater design flexibility and optimized locking performance are possible by frequency and phase locking two lasers with a fixed offset frequency. However, the output power of lasers locked to within an offset frequency can not be coherently combined. To provide both the optimal performance characteristics of offset locking and provide efficient beam combining, two or more local oscillators 14-1, 14-2 are locked to within the same rf frequency offset, to a third, common reference laser 40, thereby locking the local oscillators 14-1, 14-2 to the same optical frequency (typically 100-400 THz) (
This laser system is implemented using fused fiber components, planar lightwave circuits, or bulk beam splitters to achieve the beam splitter 92′ and beam combiner 92 functionality.
Beam splitter 98′ is typically a fused fiber or planar lightwave circuit having 16 or 32 outputs, for example. Beam combiner 98 may in addition take the form of a coherent fiber bundle 100 (
Example: Linewidth Narrowing of High Power Laser Emitters
As illustrated in
In this particular example, laser emitters 14 are high power DFB lasers having an integrated tapered amplifier section which increases the output of the oscillator section(s) from 100 mW to >1 W. Note that the high speed frequency noise characteristics of the DFB laser with tapered amplifier 18 are dictated primarily by the oscillator section 16 in which the frequency selective grating resides. In addition, the oscillator section 16 can generally be FM modulated with high speeds (<1 GHz) by direct current injection into the oscillator gain section(s). Therefore, the feedback control provided by circuit 20 is applied to this oscillator section 16. The amplifier section 18 is driven with a relatively constant current independent of the feedback loop. The FM response of the amplifier section is typically restricted to relatively low frequencies (<10 KHz) for which thermal coupling between the amplifier and oscillator section enable Joule heating in the amplifier to affect the thermal distribution in the oscillator section(s).
Example: Acquisition, Phase Locking and Wavefront Control Process
Robust and efficient phase locking of a semiconductor laser array is accomplished by performing a series of steps including frequency acquisition, phase locking and composite wavefront control steps.
The frequency acquisition process begins with the search for an electronic beat note present at the output of the transimpedance amplifier in step 105-j, where j denotes each of the emitters. All emitters undergo independent and simultaneous search processes to reduce the time to lock the entire array. The photodetector 12/TIA 55 combination typically have a bandwidth in the range of 5-10 GHz. If the initial frequency of the local laser 14-j and the reference laser 40 differ by more than this bandwidth, the beat note will lie outside of the circuit bandwidth and is not detected. In this situation, the acquisition process branches to step 106-j, wherein the bias current injected into the oscillator section(s) 16 of emitter 14-j is stepped or scanned in a search procedure until a beat note within the circuit bandwidth is detected. One such beat note detection process utilizes an rf frequency counter which counts the number of signal transitions between two threshold values in a given time period, for example. In the subsequent step 107-j, the oscillator bias current is varied to shift the nominal beat note frequency to equal that of the rf offset frequency, at which point this value of bias current is held in Step 108-j. Next, the feedback control circuit is activated to phase lock the local laser 14-j to the reference laser 40. This step 109-j is independently repeated for all local lasers in the array 24 in a parallel fashion, until all local lasers are locked to the common reference laser and made mutually coherent.
Coherent combining with a single diffraction limited composite output beam requires, in addition to phase locking, that the phases of each emitter circuit be adjusted to produce a composite beam with constant phase front. The solution wherein each emitter is locked to the same phase does not necessary lead to a constant phase front because of various imperfections in the optical path, such as optical aberrations and misalignment. Measurement step 110 is thereby incorporated to determine phase set points which accomplish the target phase front. The composite wavefront is measured, for example, by focusing the beam power through an aperture at the back focal plane of a lens while dithering the phase of each emitter independently, in a serial fashion, according to steps 111-1 thru 111-N. The phase set points which maximize the power through the aperture necessarily produce a diffraction limited output. Note that the wavefront measurement may be performed at the exit of the laser, can be remotely located, or can be performed on the light reflected from a distant target, for example. In this latter case, the phase set points may be programmed to correct for atmospheric aberrations or thermal distortions, for example. The phase set points can potentially be updated at high refresh rates to correct for dynamic aberrations or to accomplish beam stearing and/or focusing.
Slight temperature or acoustic variations, for example, can potentially cause the emitter circuits to lose phase lock. As a consequence, each emitter circuit continuously monitors the presence of a beat note in steps 114-1 through 114-N during normal operation. Should the beat note shift outside the bandwidth of the detection circuitry, a step/scan process (steps 115-j through 117-j) to re-acquire is automatically initiated for the particular emitter(s) out-of-lock. This is followed by the reactivation of feedback control 118-j to phase lock the jth emitter. Once locking is restored, the associated phase offset may need to be recomputed based on the composite wavefront measurement. In step 119-j, the wavefront is measured while varying the phase of emittter j. The phase control unit 51 processes this data to calculate and update the emitter with its new phase setpoint 120-j.
The determination of phase offsets can potentially be performed in parallel by associating the optical phase of each emitter with a unique dither frequency in step 119-j. This has the benefit that frequency acquisition can be performed more quickly, since rf spectral analysis decomposes the composite wavefront measurement into independent contributions from each emitter. This enables real-time adaptation of the composite wavefront's phase and amplitude distributions.
Once the local laser's frequency is within the PLL locking range, the main loop with fast response will take over from the acquisition loop and acquire the lock. The acquisition loop is disengaged under locked conditions. The frequency of the beat noise signal produced by the TIA 55 may potentially be divided by frequency divider circuit element 152. Loop filter 58 is preferably of the charge pump type. A portion of loop filter 58 may be optionally input into a drift tracking circuit 58′, which includes, for example, an electronic integrator preceeded by an offset circuit. This low frequency circuit can be implemented by standard op amp/transister circuits. The drift tracking circuit is output to the laser driver 56, such that the loop tracks slow thermal drifts. Thermal drifts play a significant factor because the frequency of typical semiconductor laser emitters drift by 1 MHz per mK. Appropriate phase offsets are provided by phase control unit 51 and summed by element 155 with the output of phase/frequency detector 154.
An acquisition lock detector 53′ and ramp generator 53″ are used for the initial frequency acquisition process. The use of an offset locking approach facilitates the re-acquisition process if the frequency offset (e.g., 1 GHz) is larger than the typical frequency jump event that unlocks the loop (e.g., 100 MHz) because the frequency change of the beat note provides, unambigously, the frequency of the local oscillator. In homodyne locking, the beat note does unambigously determine whether the local oscillator is higher or lower in frequency than the beat note.
The acquisition loop is critical in this laser array system since the initial frequencies of the multiplicity of laser pairs potentially differ by more than the bandwidth of the optical detector 12 and/or the transimpedance amplifier 55. This requires that the bias current of the local laser 14 be scannned until a beat note is detected. The ramp generator 53-2 output produces a bias current ramp at the local laser 14 through the laser driver 5, which tunes or chirps the laser frequency until the beat note frequency is detected and equal to the rf offset frequency. The acquisition loop is engaged in the start-up phase, and is reactivated if a laser loses lock due to temperature changes, laser mode hopping or other perturbations.
To summarize this invention, phase locked laser arrays and various laser designs and OPLL circuit implementations are disclosed. The extension of this OPLL approach to the array format leads to numerous applications in the area of high power lasers and optical phased array lasers. Examples of the use of this technique to linewidth narrowed semiconductor lasers has been disclosed. Those skilled in the art will readily observe that numerous modifications and alterations of the device may be made while retaining the teachings of the invention. Accordingly, the above disclosure should be construed as limited only by the metes and bounds of the appended claims.
This application relies for priority on provisional application 60/674,093 of Yariv et al., filed on Mar. 23, 2005 and entitled “Optical phase-locked loops,” on provisional application 60/692,853 of Kewitsch et al., filed on Jun. 22, 2005 and entitled “Mode-locked semiconductor laser array,” and on provisional application 60/776,773 of Kewitsch et al., filed on Feb. 24, 2006 and entitled “Arrayed semiconductor lasers in optical phase-locked loops.”
Number | Date | Country | |
---|---|---|---|
60674093 | Apr 2005 | US | |
60692853 | Jun 2005 | US | |
60776773 | Feb 2006 | US |