General Electric The Diode Laser The First 30 Days, 40 Years Ago Russell D. Dupuis In the space of only a month, history was made four decades ago when four groups of researchers independently developed and demonstrated their own versions of the injection laser. The author describes how the experiments laid the foundation for the vast array of materials and technologies used in the fabrication of modern compound semiconductor alloy devices. little over 40 years ago, on Sept. versions of the injection laser. The efforts lasers were fabricated by zinc (Zn) diffu- 16, 1962, Gunther Fenner, a of the other three groups that also suc- sion into conventional (and commer- member of the team headed by ceeded in making a semiconductor laser cially available) n-type gallium-arsenide A 11 Robert N. Hall at the General Electric in late 1962 were led by: Nick Holonyak (GaAs) “bulk” crystals, one of the “first” Research Development Center in Jr.4-6 at General Electric’s Syracuse, N.Y., laser diodes (Holonyak’s) was created Schenectady, N.Y., operated the first facility; Marshall I. Nathan7, 8 at IBM from a small piece of single-crystal semiconductor diode laser.1-3 Within Research Laboratory, Yorktown Heights, ternary gallium-arsenide (GaAsP) alloy about 30 days, workers in three other N.Y.; and Robert Rediker 9, 10 at MIT material grown by vapor-phase trans- laboratories in the United States had Lincoln Laboratory, Lexington Mass. port, making it the first alloy compound independently demonstrated their own While three of these early p-n junction semiconductor device to assume © 2003 IEEE. Reprinted with permission from the Feb. 2003 issue of IEEE LEOS Newsletter. 30 Optics & Photonics News ■ April 2004 1047-6938/04/04/0030/6-$0015.00 © Optical Society of America THE DIODE LASER commercial importance. thermodynamics!10 What From this humble beginning, made the announcements the vast array of materials electrifying was that it sud- and technologies used in the denlybecame clear that a fabrication of modern com- semiconductor diode could pound semiconductor alloy be a very efficient generator devices has emerged. of photons, perhaps the most efficient “converter” of elec- What are GaAs p-n trical energy into optical junctions good for? energy ever demonstrated.16 While germanium (Ge) and This revelation was an silicon (Si) were well-known important motivation for semiconductors by the late some working in the field to 1950s, the III-V compound pursue their dreams of mak- semiconductors (sometimes ing a semiconductor laser. referred to at that time as General Electric The race for the semicon- “intermetallic compounds”), ductor injection laser was on. were not well studied and had no obvious (Above) The first semiconductor laser was Of course, Lincoln Labs had a sizable unique application. By 1960, Si had made of GaAs and had a diffused p-n junc- lead in some respects since it already had become the dominant semiconductor tion. It had polished facets and operated a relatively large research group experi- only under pulsed conditions at cryogenic because of its large bandgap relative to temperatures. (Facing page) The GE enced in the study of GaAs diodes. Earlier Ge and because of the beneficial features Corporate Research Lab semiconductor laser in 1961 and even before, various other of its native oxide, SiO2, discovered by team, led by Bob Hall, examines the cryostat groups, including IBM Research Labs, Carl Frosch in 1955. It was not clear used to test the first diode laser. (Left to right) had considered the concept of a semi- exactly what benefits could be obtained Jack Kingsley, Dick Carlson, Gunther Fenner, conductor laser. The IBM group even by replacing Si with GaAs, since GaAs Ted Soltys and Bob Hall. had a U.S. Army-sponsored research had no stable native oxide and was much contract to make such a device.8,17 more difficult to make in high-purity Research groups in the United Kingdom form. Consequently, as late as 1958, few been under study for some time and their had also joined the chase for a semicon- device researchers considered GaAs optical properties were beginning to be ductor laser; there were well-organized worth much effort. Only in 1952, in fact, explored as well. In fact, it was the amaz- GaAs p-n junction research activities had GaAs been identified as a semicon- ing electroluminescence efficiencies of at the Royal Signals and Radar Estab- ductor by Heinrich Welker of Siemens in such diodes that were reported at the lishment under Cyril Hilsum. In the West Germany.12 Solid State Device Research Conference Soviet Union, Nikolay G. Basov and The search for a higher voltage tunnel (SSDRC) on July 9, 1962, by R. J. Keyes co-workers at the Lebedev Institute in diode was one of the motivations behind and T. M. Quist13 of the MIT Lincoln Moscow and D. N. Nasledov and co- the study of GaAs diodes. By the early Labs group and by a group led by workers at the Ioffe Physico-Technical 1960s, heavily doped n-type GaAs sub- J. Pankove at RCA Laboratories. These Institute in Leningrad were also consid- strates—produced by both Czochralski workers reported that their GaAs p-n ering how to achieve population inver- and horizontal Bridgman growth tech- junctions had extremely high internal sion in a semiconductor.18, 19 In France, nologies—were commercially available; quantum efficiencies—as high as Pierre Aigrain at École Normale fabrication of the junction of these 85-100 percent!10,14 The Lincoln Labs Supérieure had proposed that laser devices was in some respects relatively group also reported using a GaAs diode operation of a semiconductor could straightforward since the means of form- to demonstrate the optical transmission occur; in 1961, he was reportedly plan- ing p-n junctions by use of diffusion and of television signals from Mount ning to visit the United States with a alloying techniques were well known. The Wachusett to the roof of the Lincoln Labs working semiconductor laser in his form of the p-n junction in the first GaAs facility, a distance of about 50 kilometers pocket!20 In 1961, Maurice Bernard laser diodes was relatively conventional, as the crow flies—quite possibly the first and Guillaume Duraffourg, working at consisting of an n-type GaAs “host” crys- demonstration of the optical transmis- CNET in France, had also published a tal into which Zn atoms were diffused to sion of an electronic signal.10,15 paper analyzing the possibility of laser create a heavily doped p+ region. One While the basic elements of the light- operation in semiconductors.21 application of contemporary interest for emitting properties of GaAs p-n junc- these diffused and alloy p+-n diodes was tions had been known several years prior How do you make the study of tunneling phenomena in to the1962 report, some of those in atten- a semiconductor laser? heavily doped (degenerately doped) dance at the SSDRC complained that The concept and demonstration of light p+ -n+ diodes. The electric characteris- these high-efficiency electroluminescence amplification by the stimulated emission tics of GaAs diffused-junction diodes had results violated the second law of of radiation, i.e., LASER operation, had Tell us what you think: http://www.osa-opn.org/survey.cfm April 2004 ■ Optics & Photonics News 31 THE DIODE LASER been under active discussion in semiconductor lasers had pro- the late 1950s. The discussion posed to use a macroscopic +V culminated in the demonstra- “external cavity” into which a tion of the solid-state ruby laser M GaAs diode was placed, Hall on May 16, 1960, by Theodore decided to polish parallel faces Maiman, who was working P onto his GaAs diodes so that S1 S2 at Hughes Research Labs. By the Fabry-Pérot optical cavity 1962, it was well known that geometry was built into the the light from a ruby laser—a device. This approach was not visible laser operating in the L universally applied and, in fact, red spectral region—exhibited N the importance of optical feed- several properties unique to back into the diode’s “active coherent radiation, including region” was not fully appreci- the characteristic presence of M ated by many workers. Hall’s “laser speckle” which could be -V team operated their first suc- observed by the human eye cessful GaAs laser diodes when the laser was operated under pulsed conditions at above threshold.22, 23 Since the 77 K on Sept. 16, 1962.1 A first laser demonstration in 1960 (and Figure 1. Schematic diagram of initial con- schematic diagram of Hall’s early concept even before), some researchers exploring cepts for an injection laser developed at for an injection laser is shown in Fig. 1. semiconductor diodes had wondered if a General Electric Research Laboratories by The first verification of laser operation semiconductor had the necessary qualities Robert Hall in 1962. [From R. N. Hall, IEEE J. was made through the observation of Quant. Electron. QE-23, 674 (1987), Fig. 1.] to support stimulated emission and laser near- and far-field interference patterns operation. Others believed that the possi- using an infrared (IR) up-converting bility might indeed exist.24 John von “snooper scope” which was being used in Neumann had considered the essential A practical problem faced by everyone Hall’s lab to study the emission from such elements of a semiconductor laser theo- interested in making a diode laser was infrared light emitters. retically in 1953.25 how to determine that it was in fact las- As noted above, IBM had also ing. While it may seem somewhat diffi- mounted an early effort to develop a Few parallels with the MASER cult to understand with the benefit of diode laser.