Photonics West '96 Summary

Nakamura Discusses Laser Diode

The highlight of the SPIE's conference on Physics and Simulation of Optoelectronic Devices at the Photonics West Meeting was Shuji Nakamura's talk. This was the first opportunity for the technical community to hear the details of the recently announced nitride-based blue laser diode. Dr. Nakamura not only presented the work reported in his recent paper in Japanese Journal of Applied Physics, but went further, presenting new results on subsequent work.

After a brief summary of the Nichia work on blue and green LED's, Dr. Nakamura pronounced the research into LED's "almost finished", and went on to talk about the laser diode structures. All of the laser structures were grown on c-plane sapphire substrates with a multi-quantum well active layer. In addition to the 26-well structure reported in the JJAP paper, the Nichia group has now made lasers with as few as 10 quantum wells. Other features of the structure include an n-InGaN layer on the thick n-GaN bottom layer which is included to "prevent cracking", and a p-AlGaN layer on top of the MQW stack to prevent InGaN decomposition during growth of the p-GaN top electrode.

The lasers are RIE etched stripes 20µ wide and 300µm to 1500µm long. The top contact stripe is 10µm wide. An SEM photo of the structures showed visible vertical striations and layer contrast. The roughness of these "facets" was of the order 500Å. The operating voltage was about 22V for an optical power of 55mW, and threshold currents ranged from 4 to 10 kA/cm². Threshold current for one of the smaller stripes was only 300mA. The lasers are run pulsed at room temperature, with a 0.1% duty cycle and a rep rate of 500/s.

Above threshold, the emission of the laser was almost all TE polarized. Spectra from these lasers showed the clear signatures of lasing, with sharp peaks rising out of a broad background of spontaneous emission. The lasing modes were distributed irregularly in one of the lasers, with 0.02nm linewidths, but in a longer stripe laser, 4 "sub-bands" of lasing modes dominated the emission pattern at high bias. The far-field pattern was the truncated ellipse shown in the JJAP paper, but subsequently, part of the substrate was removed from one of the lasers so that the full ellipical pattern could be seen, as expected. A sin(phi) pattern of spontaneous emission surrounded the elliptical laser beam. The laser beam had a 5 degree horizontral divergence and a 17 degree vertical divergence, with multiple peaks. In a further demonstration the laser light was collumated with a lens to form a photogenic laser "beam".

These photographs of the laser were shown. (Thanks to Shuji Nakamura and Nichia Chemical Industries for permission to publish them).

Here is a laser running below threshold. Note the copious spontaneous emission.

Here is the same laser running above threshold. Note the laser beam, and the redused spontaneous emission. The color of the laser is actually bluer (almost violet) than can be conveyed in a photograph.

Dr. Nakamura concluded by saying that CW was the next step. We don't have to worry about things like dark-line defects, and that the real difference between nitride LED's and II-VI LED's was that nitride LED's were commercially available.

After the talk, Dr. Nakamura entertained a number of questions.

Q. How long do the lasers last?: A. We're not interested in life testing at this time. We ran one under pulsed conditions for 24 hrs, and saw no degradation.
Q. What happens when you try to run CW?: A. The voltage was very high, so the device burned up.
Q. What about the lifetime of LED's?: A. 100,000 hours, we hope. But we don't know yet.
Q. Have you tried a single or double quantum well laser?: A. It doesn't work.
Q. When you run CW, does the resistance drop?: A. I don't know.
Q. Why did you reduce the number of quantum wells from 26?: A. It was easier to grow.
Q. What about other substrates?: A. We've tried other substrates, but I can only talk about c-plane sapphire.
Q. Have you tried smoothing the etched surfaces?: A. Yes, but it doesn't change much.
Q. Are the facets coated?: A. Yes, sorry I forgot to say. The reflectivity is 60%.
Q. Where does all the voltage drop?: A. We think the contact is the problem.
Q. Why is the voltage so high?: A. Contact resistance.
Q. What's the shortest cavity length that worked?: A. 300µm.
Q. What's the difference between the laser in this photo compared to the others?: A. No difference, it's just the illumination that makes it look more violet.
Q. Why did you start with 26 quantum well, and then work down? I would have started with 1, and then worked up.: A. With GaN, everything is backward.
Q. When did you first have the laser?: A. In October.
Q. How long had you been working on the laser?: A. Previously, we only worked on LED's, because the market is bigger. Then, last summer, when the LED research was pretty much finished, we got permission to start working on the laser. We took maybe 100 runs to get it right.

Nakayama Presents the case for II-VI LED's

Just prior to Nakamura's talk, Sony's Dr. Nakayama presented the case for II-VI LED's. LED's in the ZnSe-CdSe system have been made with better than 8% external quantum efficiency in the Green, and 3% in the blue. (This compares to as high as 20% for AlGaAs red LED's. The structures presented by Dr. Nakayama had ZnCdSe triple quantum wells in a ZnMgSSe cladding, all grown bon GaAs in a dual chamber MBE system. Dr. Nakayama emphasized the excellent color purity of the green and blue LED's.

Much of the talk was devoted to studies which used the spectral shift to determine the heating of the active layer during operation at high current. The blue LED's were estimated to heat by 24 degrees for a 20mA bias, compared to only 3.7 degrees for the blue LED's. This was attributed to an "improved structure and growth" resulting in better contacts. Aging studies demonstrated 1000 hour lifetime at 15 mA. The degradation mechanisms were attributed to carrier recombination.

The word in the hallways was that even Sony is now working on nitrides.

Cree reports progress in laser structures

Dr. Bulman represented Cree in this session, presenting results on p-doping and laser structures. (i.e. undoped structures with photopumping). Hall effect measurements on p-doped GaN on SiC indicate a doping density of 1.4x10²⁰ with a 5% compensation ratio, and nearly 100% activation (albeit with a 150meV activation energy. Contacts made on these films had a contact resistivity of 0.046 Ohm-cm².

The photo-pumped lasers were made with a cleaved facet, which looked very nice, and by varying the length, they determined a gain of 1000cm^-1 at 373 nm. These structures had stimulated emission thresholds of 72 kW/cm².

Dr. Bulman also showed results from an edge-pumped separate-confinement-heterostructure with a 100Å quantum well and a 4.8% optical confinement factor. The P_th was 63 kW/cm². The large mode spacing suggested that the optical cavity was being formed by microcracks.

Questions

Q. Which face do you cleave?: A. We can't say.
Q. Is your buffer layer AlN, and if so, how does it effect your back contact?: A. Yes, it is. These results are for undoped structures, so there isn't a back contact.

UNM/Sandia team Reports on LED Reliablility

D. L. Barton reported the results of reliability tests and failure analysis on the Nichia LED's. The results are very encouraging. Of 20 devices tested, only 1 failed a 20mA, 1000Hr stress. This device developed some ohmic leakage apparently related to a discontinuity related to some subtle mechanical damage. 18 of the LED's were from a newer lot, and none of these devices failed the 1000 hour test. More results are expected soon. (Lifetests take a while.)

Calculated thresholds not so far off

Marek Osiński (who did a great job organizing this conference) reported on gain calculations for GaN based laser diodes. The effective masses are not known well enough to do an accurate calculation, but the estimated minimum thresholds of 2 kA/cm² for a double heterostructure device are not so far off from the Nichia lasers. Estimates for VCSEL's and for SCH lasers were 20 kA/cm² and 400 A/cm².

GaN/SiC HBT's unexplained

H. Z. Fardi of University of Colorado at Denver presented modeling results for the GaN/SiC HBT's previously reported. The theory predicts beta of 10⁵; experiments have reported 10⁷. The cutoff frequency is calculated to be only 8MHz.

If anyone wants to summarize the two GaN talks I slept through on Tuesday morning, send me e-mail.

Summary written by E. Hellman.

Coming soon: A Discussion Page on the Nichia Laser.