During operation in the amplification mode, light is input through the front facet and is amplified within the body of the optical gain material. An at least partially optically reflective surface is positioned a first
predetermined distance from one of the facets during operation of the device in the amplification mode. In the signal generation mode, the at least partially optically reflective surface is positioned a second
reason, however, why a 1.5-µm mode-locked VECSEL should not generate quasi-soliton pulses in the sub-500-fs regime.
5. Future prospects
In conclusion, it seems possible that VECSELs will in the future be applied increasingly to the sort of task for which a diode-pumped solid state laser is currently used but where the wavelength, or another specification, is more easily met by a quantum well gain medium. Active mirrors are thinner than laser rods; therefore it is possible to make highly compact VECSEL devices, even with quite complex cavities. Injection pumping is clearly an option for medium power devices, where the application justifies more elaborate wafer design and fabrication. Optical pumping, however, is the most straightforward way in which to achieve a uniform carrier distribution over a large aperture; it is likely to remain the method of choice at the highest power output levels. In particular, it allows the gain structure to be optimized for optical performance, without the compromises imposed by the need for good electrical characteristics. The alignment tolerance for VECSEL optical pumping is less stringent than that required, for example, to launch light into single-mode fibre.
To date, many experimental and theoretical studies have been reported on the FGFP laser   . How- ever, in most of these studies, the temperature effect is not taken into account. In addition, they assumed that the externalcavity diode laser was under strong OFB; i.e. the effect of external OFB was not investigated in weak and moderate levels. Thus, full visualizations of the temperature and the external OFB effects on the output characteristics were not provided. Therefore, an accurate knowledge on the effects of these parameters is very important for avoiding ECSL-FBGs to operate in inoperable regime.
Similar arguments were used in Ref. 13. The en- hancement factors ⌫ cav and ⌫ RPG can be calculated based on our knowledge of the composition of the semiconductor structure used in this paper, and their dependencies as a function of wavelength are shown in Fig. 9. The set of dashed curves are the enhance- ment factors 共⌫ cav and ⌫ RPG ) at normal incidence, while the other set of solid curves are the factors at an incident angle of 27°. The curves shift to shorter wavelengths as the incident angle increases. Due to the antiresonant structure of the wafer, the main peak of the ⌫ RPG curve does not overlap with the peaks of the ⌫ cav function. From the overlapping function 共⌫ RPG 兲 for the case of the 27° incident angle, a wave- length of 847 nm would be optimum. However, it is far off a microcavity resonance and rather close to the laser wavelength, which will lead to strong absorp- tion bleaching; hence it is not suitable for pumping.
From the engineering point of view, it is not only the carrier density which is important but also the pump intensity needed to sustain it. Hence it is crucial to study the loss processes in the semiconductor. Using our many-particle theory, the loss rate due to sponta- neous emission and Auger recombination is evaluated for an (Al 0 . 108 Ga 0 . 785 In 0 . 107 )As / (Al 0 . 26 Ga 0 . 74 )As quantum well (Fig. 5). From the analysis we conclude that Auger processes play a minor role in this particular system com- pared to the radiative losses for the entire carrier density regime relevant for lasing. The recombination via defects is negligible. Since the VECSEL is expected to operate at about 350 K, the calculations are carried out for this tem- perature. Compared to the results obtained at room tem- perature, the total loss rate is reduced by more than one order of magnitude. It is true, that the Auger rate increases with temperature, but more important for this material sys- tem are the radiative losses which become considerably lower because the probability of spontaneous emission is reduced due to the broadened carrier distributions for the higher temperature. However, the higher operating temper- ature also involves higher threshold carrier densities so that the benefits achieved by the reduced total laser losses are suspended.
To study wavelength-scale cavities, we no longer rely on achieving phase matching, but rather just use the taper as a convenient means to produce a micron-scale evanescent field for sourcing and out-coupling the micron-scale cavity field. The taper effectively serves to bridges the disparate length scales from conventional fiber and free-space optics to chip-based microoptics, and does so entirely off the chip, so that on-chip structures do not require any additional complexity. Although the coupling we observe might not be as efficient as phase matched coupling, the power transfer is more than adequate enough to probe many of the important properties of the cavity. By using an external waveguide as the coupling element, this method is inherently non-invasive, can be used to rapidly characterize multiple devices on a chip, and the ability to vary the position of the taper with respect to the cavity (not an option for microfabricated on-chip waveguides) allows for quantitative investigation of not only the Q factor but also V eff . Furthermore, the resonant coupling from the ex- ternal waveguide is polarization selective, providing additional information about the cavity modes that is not easily obtainable through techniques such as NSOM. Knowledge of a mode’s spectral position, polarization, Q, and V eff will in many cases be enough to unambiguously determine the identity of the mode in comparison to simulation or theoretical results. Thus, in some respects, the versatility of the fiber-based approach that we have described in this chapter makes the technique an optical analog to electrical probes used to study microelectronic devices.
Finally, Section 5 will summarize our conclusions.
2. Iterative Traveling-Wave Model
Consider the configuration of Figure 1. A single-longi- tudinal-mode laser diode is in resonance with an external Fabry-Perot cavity. We assume that r 1 , r 2 and r 3 are all real and dispersionless. For this three-mirror system, the dominant resonator is defined by the mirrors with reflec- tion coefficients r 1 and r 3 , and multiple round trips inside the externalcavity should be in general taken into ac- count for an arbitrary feedback level.
5.2.4 A M Mode-Locking Experim ent
The composite cavity configuration used in th is experim ent is sho'svn in Fig. 5.5.b. The same laser was used as in section 5.2.3. The m odulator was designed as a tra v e llin g wave Y -ju n ctio n m odulator. E ffe ctive ly, w ith in the laser ca vity i t acted as a M ach-Zehnder in te rfe ro m e te r structure. The waveguides were formed using a long, d ry diffusion on X- cut, Y -propagating lith iu m niobate. The Y -junction was form ed using a b ra n c h in g angle o f 0.5® in order to provide low-loss s p littin g and recom bination. The distance between the bran ch in g waveguides was 35|im, large enough to prevent any optical coupling. A 200nm Si 0 2 buffer layer was deposited on the top o f the waveguides in order to reduce the electrode loading loss. The electrodes were designed as a coplanar w aveguide s tru c tu re a llo w in g p u s h -p u ll o p eration o f the in te n s ity m odulator. A n in te ra ctio n length o f 1cm was chosen as a compromise between high-frequency capability and efficient m odulation. The w idths of the central electrode and the gap were 20|Lim and 15pm leading to the impedance o f 35Q. The electrodes were fabricated using chrome and gold evaporation and electroplated to the thickness of 1 . 1 pm. The measured DC resistance of the central electrode was IIQ . This m odulator was fixed to an a lu m in iu m holder and electrical connection was made w ith SMA connectors. The electrical re tu rn loss of the m odulator was o f the order of -lOdB in the frequency range from 3-18GHz. Two resonances o f the order o f -6 to -7dB have been detected below 3GHz and were probably caused by the m odulator package.
The output spectra of the ECL system collected from the above sensors, shown in Figure 2.14, clearly demonstrate the porous PhCs’ wavelength selec- tion function. The raised background represents the broadband SOA spon- taneous emission, where the porous PhC resonance registers a transmission dip at the center of the SOA gain spectrum near λ = 855 nm. The dynam- ics of establishing lasing behavior is illustrated by three curves, representing the output below (red), just above (blue), and well above (green) the lasing threshold, respectively. Above the threshold, a laser emission occurs at the transmission dip, overlapping with the porous PhC resonance. The relative broad PhC resonant reflection peak translates into a narrow laser emission spike via the process of stimulated emission. The laser output surpasses the spontaneous emission and gradually increases intensity with increasing injection current. The laser output power is approximately 1 mW, which is considerably lower than passive WGM biosensors or other active pulse pumped optical sensors . A precise calibration of the ECL laser emission using a scanning Fabry-Perot interferometer demonstrated single-mode las- ing with a FWHM of 0.03 pm . The narrow linewidth enables resonant wavelength shifts to be resolved with sub-picometer accuracy. Therefore, the porous ECL establishes a record high FOM of 1.05 × 10 7 .
while the optical pulses reported in  are fairly irregu- lar and have a relatively small space to mark ratio (the
“pulse” duration is a substantial fraction of the repeti- tion period), the output of the shorter Fabry–Perot laser is of a much more regular self-mode locking type, produc- ing short pulses (more than an order of magnitude smaller than the repetition period/round trip time) and we believe would have been completely periodic were it not for the gain bandwidth limitation. The likely explanation is that in the relatively short, very fast tuned Fabry–Perot res- onators, each longitudinal mode is only present in the laser output for a few round-trip times. During such short time, output instabilities have no time to develop and the regime remains stable. We note also that in the more conven- tional semiconductorlaser constructions with a saturable absorber, stable mode locking is also difficult to achieve at repetition rates comparable to, or below, the carrier lifetime in the gain section . We note indeed that in simulations with lasercavity lengths 2–5 times the value reported above, we found that for the same parameters studied here, the stable long pulse structures of Figure 7b (and reminiscent to those seen in ) are simulated for a fairly broad range of tuning speeds; however the short pulse, self-ML type dynamics of the type of Figure 7c is not seen for any operating parameters.
First of all, one needs to put a camera after the cavity on the transmission path of the beam trying to get a signal on which to start the optimization of the beam adjustment. Then one needs to scan the laser frequency more than one FSR by scanning the PZT in the lasercavity either using an external signal generator or the RAMP function on the laser servo. The purpose is to find the resonance line of ULE cavity. Thirdly, start to adjust the beam. The light path before the cavity is shown in Figure 4.2. In order to decouple the adjustment of the beam position and its tilt angle, it is better to have one mirror rather close to the cavity, in this case is M3 and put one mirror far away from the cavity, in this case is M1. Adjusting the mirrors in both horizontal and vertical directions until one can see some patterns on the camera. Figure 4.3 shows some example transmission patterns one can see on the camera.
transmission are, to our knowledge, the highest reported to date without using electrical circuit such as equalization to enhance the device’s data-transmitting capability.
To further improve the high speed performance, photon lifetime tuning techniques can be incorporated. The device presented in Section 2.3 does not incorporate post- fabrication process for photon lifetime tuning. Post-fabrication top DBR shallow etching has been demonstrated on 850 nm VCSELs to reduce photon lifetime and improve the modulation bandwidth and the optical output intensity [29, 30]. However, an etching technique may have uniformity issues across a sample or wafer as well as raise concerns for reliability. Another approach that can be used is depositing dielectric material such as SiN x to tune the photon lifetime. In both cases, careful theoretical calculation and process calibration should be done to accurately tune the photon lifetime.
single mode oscillation could be possible. This single- mode oscillation was observed in AlGaAs lasers when the injection current is well-above the threshold level . In lasers made from the quaternary compound InGaAsP, the gain is rather shallow and spectrally asym- metric, which supports oscillation of several modes on the long-wavelength side of the central mode [23–26]. It was observed in experiments that these modes exhibit hopping, which was attributed to the large value of the linewidth enhancement factor that violates the asymmet- ric gain suppression (AGS) and competition among these long-wavelength modes . It was shown that this multimode hopping can be released and the laser supports single-mode oscillation by subjecting the laser to very strong optical feedback and avoiding the chaotic state [28, 29]. They pointed out that the range of feed- back that corresponds to single-mode oscillation shifts to weaker values with shortening the externalcavity . It is then motivating to investigate the possibility of achieving stable single-mode oscillation under the pre- dicted stable operation of semiconductorlaser coupled with a very-short externalcavity. Introducing selective feedback to lock or tune the predicted single-mode oscil- lation could add to the advantages of this coupled laser as a promising device in wavelength-division multiplex- ing transmission systems , and optical pumping . The selective feedback can be achieved using diffraction gratings  or tunable wavelength selective grating structures, such as grating waveguide structures  and resonant grating mirrors .
Email: Jnewman@NorComSystemsInc.com Abstract
A production leak test system using digital holography, Optical Leak Testing (OLT), has been developed for simultaneous gross and fine leak testing of hermetic semiconductor, MEMS and Optoelectronic devices. The technique has also shown the unique capability to leak test ceramic SMC’s on PCBs, even if conformal coated. Many devices are manufactured using welded, brazed or soldered metal lids with metal or ceramic packages. The most common conventional leak test methods used in the semiconductor industry include gross leak testing by the bubble leak method and fine leak testing with a helium mass spectrometer. The application of these techniques is highly problematic for many OLED display, Optoelectronic and MEMS devices. For maximum sensitivity, bubble leak testing requires the package be immersed in a perfluorocarbon liquid at a temperature of 125°C, exceeding the 90°C limit for most of these devices. In addition, helium absorption by the fiber optic pigtail causes fine leak testing with mass spectroscopy to be highly inaccurate when the helium degasses during testing. Further, neither method can be applied to SMC mounted to PCBs to locate leaking devices cracked during soldering. The Optical Leak Test method overcomes these concerns, and other problems with conventional leak test methods and reports the leak rate for all devices tested at one time. The hermetic devices are placed in the test chamber and exposed to a pressurized low molecular weight gas such as helium. If the package is leaking, the lid responds to changes in pressure differences as the device cavity pressure and test chamber pressure come to equilibrium. Precision chamber pressure measurements combined with lid stiffness and velocity data, obtained with digital holography are used to determine package leak rates in helium cc-atm/sec. Leaks in the range from the “no lid condition” to 1x10E-8 cc-atm/sec. have been measured. The method has demonstrated a very high level of accuracy and repeatability. Throughput is determined by the number of devices that can be placed on the boat for simultaneous testing. Cycle times vary from 2 to 20 minutes depending on package size and internal volume. For hermetic optoelectronic devices the problems of high temperature exposure, contamination and helium absorption/release experienced with conventional leak test methods are overcome. Finally, automated Optical Leak Testers provide near real-time leak test data for process control of metal lid seam sealing operations, minimizing lost production time, rework and scrap. The Optical Leak Test Method has been included in MIL STD 883E since 1995 for conditions C4 and C5.
PACS number ~ s ! : 42.55.Px
Vertical-cavity surface-emitting lasers ~ VCSEL’s ! are a very interesting breed of semiconductor lasers, both funda- mentally and from an applications point of view. This is particularly true for the polarization properties which are di- rectly linked with the inherent transverse geometry. In a VCSEL the quantum wells ~ QW’s ! are oriented perpendicu- lar to the propagation direction of the light, so the field is always polarized in the plane of the QW. This situation is totally different from that in edge-emitting semiconductor lasers, where there is generally a large difference in gain between TE- and TM-polarized light, i.e., light that is lin- early polarized in the plane of the QW or perpendicular to it, and where the device geometry determines the optical polar- ization. In most practical VCSEL’s the optical polarization is determined by the birefringence, which originates from the electro-optic effect and from stress and strain induced by the electrical contacting of the devices @ 1,2 # . This birefringence in VCSEL’s creates a frequency difference between the two orthogonally polarized modes. Apart from birefringence there is also some dichroism present, which causes a gain or loss difference between two orthogonally polarized modes
Integrated plasmonic sources and detectors are imperative in the practical development of plasmonic circuitry for bio- and chemical sensing, nanoscale optical information processing, as well as transducers for high-density optical data storage. Here we show that vertical-cavity surface-emitting lasers (VCSELs) can be employed as an on-chip, electrically pumped source or detector of plasmonic signals, when operated in forward or reverse bias, respectively. To this end, we experimentally demonstrate surface plasmon polariton excitation, waveguiding, frequency conversion and detection on a VCSEL-based plasmonic platform. The coupling efﬁciency of the VCSEL emission to waveguided surface plasmon polariton modes has been optimized using asymmetric plasmonic nanostructures. The plasmonic VCSEL platform validated here is a viable solution for practical realizations of plasmonic functionalities for various applications, such as those requiring sub-wavelength ﬁeld conﬁnement, refractive index sensitivity or optical near-ﬁeld transduction with electrically driven sources, thus enabling the realization of on-chip optical communication and lab-on-a-chip devices.
( th ) ( eff1 eff2 ) ( 1 2 )
2π ln 1 ln 1
l τ ν ν l r α r r r r
∆∅ = − + ∅ − − (5) where τ l is the round trip time delay of the optical beam inside the primary lasercavity, ∅ r represents the total phase of effective reflectivity due to two cavities, ν is the new oscillation frequency and ν th is the oscillation frequency before using an external cavities and α is the linewidth enhancement factor in the semiconductor material. From this relation, we can find a new formula for the frequency shift of the laser emission that result due to using dual cavities as follows:
Diode lasers are compact, robust optical sources with a broad range of applications in physics research. However, a major limitation is the incomplete wavelength coverage of commercial laser diodes. The output wavelength is determined by properties of the semiconductor material inside the lasercavity, and for any given atomic or molecular transition, there may not be a corresponding diode available. For example, while many commercial diode products based on AlGaInP semiconductors exist for red wavelengths above 630 nm, few exist for the wavelengths below. Applications for lasers of these wavelengths (often in combination with frequency doubling) include slowing, cooling, trapping, and quantum- state-preparation of several di fferent species of diatomic molecules 1–4 as well as laser cooling of beryllium ions for quantum information processing. 5,6 Techniques that have been previously employed to construct lasers in the 620 nm range include frequency doubling, 7,8 custom-fabrication of semiconductor materials, 9 and cryogenic cooling, 10 but these methods are typically costly and highly dependent on the specific final lasing wavelength. Alternatives to diode lasers include dye lasers, which operate at high power for many red wavelengths 11 but are maintenance-intensive, and optical parametric oscillators, which are broadly tunable across the visible spectrum 12 but are expensive to manufacture.
Diode lasers are compact, robust optical sources with a broad range of applications in physics research. However, a major limitation is the incomplete wavelength coverage of commercial laser diodes. The output wavelength is determined by properties of the semiconductor material inside the lasercavity, and for any given atomic or molec- ular transition there may not be a corresponding diode available. For example, while many commercial diode products based on AlGaInP semiconductors exist for red wavelengths above 630 nm, few exist for the wavelengths below. Applications for lasers of these wavelengths (often in combination with frequency doubling) include slowing, cooling, trapping, and quantum-state-preparation of sev- eral different species of diatomic molecules 1–4 as well as laser cooling of beryllium ions for quantum information processing 5,6 . Techniques that have been previously em- ployed to construct lasers in the 620 nm range include fre- quency doubling 7,8 , custom-fabrication of semiconductor materials 9 , and cryogenic cooling 10 , but these methods are typically costly and highly dependent on the specific final lasing wavelength. Alternatives to diode lasers in- clude dye lasers, which operate at high power for many red wavelengths 11 but are maintenance-intensive, and op- tical parametric oscillators, which are broadly tunable across the visible spectrum 12 but are expensive to man- ufacture.
We note, however, that LFI has been demonstrated previously in a THz QCL over an external path length of
>10 m 40 . As such the spectral resolution of our experiment could, in principle, be increased beyond 15 MHz.
Figure 2(c) and (d) show the normalised FFTs of the complete interferograms corresponding to data presented in Fig. 2(a) and (b), revealing the expected single- and multiple-longitudinal mode emission, respectively. From this data, a longitudinal mode spacing of ∆v FP = ~17 GHz is obtained, in agreement with that expected for a lasercavity length L c = 2.3 mm and active region efective refractive index n = 3.8 using the relation ∆ ν FP = c /2 nL c . Figure 3 shows emission spectra recorded in this way for a range of laser driving currents, and plotted on a logarithmic scale to illustrate the dynamic range of the laser feedback-interferometer. he noise loor in our system is dominated by voltage noise at the input of the digital acquisition board, which is measured to be ~10 µV/√Hz. It is also worth noting that the response of lasers under weak OF is typically greatest just above threshold and rolls of with increas- ing driving current 25 ; in our system, voltage signal amplitudes up to ~40 mV were recorded ater ampliication. As can be seen in Fig. 3, switching of the dominant mode from 2.258 THz to 2.241 THz is observed at low drive cur- rents, with multiple-mode emission dominating at larger drive currents, as is typical behaviour in QCLs.