Quantum Wells and Superlattices Come of Age

Quantum Wells Feature

Quantum wells and superlattices come of age

Zh.l. Alferov A.F. loffe Physico-Technical Institute

I Heterostructure quantum wells and superlattices have a pivotal role in the history of compound semi- conductors. This short review looks at the underlying physics, as well as the applications of quantum wells and superlattices. It starts by looking generally at the major developments in the field, before looking specifically at the advances made at the A.F. Ioffe Physico-Technical Institute in Russia.

t he development of the its use in preparing room tempera- parabolic band would break into double-heterostructure (DH) tureAIGaAs DHS lasers in 1977. mini-bands separated by small for- T laser is a sensible starting Low dimensional semiconduc- bidden gaps and having Brillouin point for looking at the history of tor structures have formed a major zones determined by this period. heterostructure quantum wells. new branch of physics research. Similar ideas were described much Owing to electron confinement in Structures with such a small scale earlier by L.V. Keldysh in 1962 [3] DH, the double-heterostructure in one or two spatial dimensions, when considering the periodic po- laser became an important precur- their electronic properties are sig- tential produced on semiconduc- sor of in the development of quan- nificantly different from the same tor surface by intense ultrasonic tum well structures. This results material in its bulk form. These wave. from the fact that when the mid- properties are changed by quan- At the Physico-Technical dle-layer of these lasers had a thick- tum effects. A quantum well in Institute in St Petersburg in the ness of some hundred angstroms, which chargecarriers are restricted then Soviet Union, R. Kazarinov the electron levels would split to moving in two dimensions is an and R. Suris considered the theory because of the quantum-size example of such a structure. of current flow in superlattice effect. A clear manifestation of the structures in the beginning of the Further, the development of quantum-size effect in the optical 1970s [4].Their work revealed that epitaxial growth techniques for spectra of GaAs-AIGaAs semicon- the current between wells was de- these structures brought with it ductor heterostructures with ultra- termined by tunnelling through the ability to fabricate high-purity thin GaAs layer (quantum wells) the potential barriers separating and ultrathin layers that were to was demonstrated by Raymond the wells. It also predicted very im- become very important for quan- Dingle et al. in 1974 Ill. The au- portant phenomena: tunnelling un- tum wells. Two main methods of thors observed a characteristic der electric field when the ground growth with very precise control step-like behaviour in absorption state of a well coincides with an of thickness, planarity and other spectra and systematic shifts of the excited state of the next well and parameters were developed in characteristic energies with a stimulated emission resulting from 1970s -- molecular beam epitaxy quantum well width decrease. photon-assisted tunnelling be- (MBE) and metal organic chemical Experimental studies of super- tween the ground state of one well vapour deposition (MOCVD). MBE lattices, meanwhile, started after and excited state of a neighbour- was the first method to be of prac- the initial work of IBM scientists ing well, which is lower by the en- tical importance for III-V het- L.Esaki and R.Tsu in 1970 [2]. ergy due to applied electric field. erostructure technology, following These researchers considered elec- At that time L. Esaki and R. Tsu the pioneering work in the begin- tron transport in a superlattice i.e. independently considered reso- ning of 1970s by A. Cho at Bell at an additional periodic potential nant tunnelling in superlattice Laboratories. Not long after created by doping or changing the structures. MOCVD, which had its origins in composition of semiconductor ma- the early work of H. Manasevit, terials with the period bigger, but Superlattice structures found broad application in III-V comparable with, the lattice con- heterostructure research after stant of the crystal. In what Leo Pioneering experimental studies of R.Dupuis and PDapkus reported Esaki called a 'man-made crystal', a the superlattice structures were

IIl-Vs Review• Vol.lO No. 7 1997 26 I I I I 80 lnGaAsP/GaAs W = 100 ~m k = 0.8 ~tm CW L= 1.2mm 70

2 60

g 3 50

1--45-A757 2--3-A799 30

E // ] I I I I 20 0 1 2 3 4 5 1 (A)

Figure 1: The CW light-current characteristics of InGaAsP/GaAs separate-confinement single quantum well DHS laser diodes. In (1), the diode had high and low reflective coatings, while in (2) the diode had a high reflective coating only.

carried out by L.Esaki and R.Tsu very low concentration of defects. with the superlattices grown by Many years later in 1983, after vapour phase epitaxy (VPE) in a G.Osbourn's theoretical study at GaPxAsl x/GaAs system. At the Sandia lab and the first successful same time, in our own laboratory, preparation of a high quality the first multi-chamber apparatus strained-layer superlattice (GaAs/ for VPE was developed. As men- In0.2Ga0.sAs by M.Ludowise at tioned before, this was used to pre- Varian), N. Holonyak at the pare the superlattice structure University of Illinois achieved the GaP0.3As0.7/GaAs with the thick- first continuous wave (CW) room ness of each layer 100 A and total temperature laser action using number of the layers of 200. these structures. From research Observed peculiarities of the volt- such as this it became clear that in your age-current characteristics, their strained-layer superlattice the lat- temperature dependence and tice strain becomes an additional photoconductivity were explained degree of freedom.This meant that by the splitting of the conduc- by varying the layer thicknesses tion band due to the one-dimen- and compositions it was possible sional periodic potential of the to vary, continuously and indepen- superlattice. dently, the forbidden gap, lattice These first superlattices were constant and other parameters of also the first strained-layer super- the overall superlattice. lattices. E. Blakeslee and J. At the beginning of 1970s Matthews, who were working with L.Esaki and his co-workers started L. Esaki and R.Tsu at IBM, succeed- using MBE to study the AIGaAs sys- ed in the mid-1970s in growing tem resulting in the submission of strained-layer superlattices with a a paper in March 1974 to Physical Fax +46 46 t68g 81 WWW httpd/www.epigreSS.se Quantum Wells Feature

Review on resonant tunnelling -- up to 420 GHz were reported in a application of the high-mobility, the First experimental demonstra- GaAs resonant tunnel diode at two-dimensional electron gas for tion of quantum well heterostruc- room temperature. microwave amplification. In 1980 ture physics. In the study, they The restriction of the motion of this led to new types of transistors measured the tunnelling current the electrons to two dimensions in based on single n-AIGaAs/n-GaAs and conductance as a function of field effect transistors has long modulation-doped heterostructure applied voltage in GaAs-Ga&lAs been recognized and was First veri- being developed.The development double barriers and found current fied, for the trapped electrons in occurred almost simultaneously in maxima associated with this reso- inversion layers, in a magneto-con- France, where they were labelled nant tunnelling. Later in the same ductance experiment conducted TEGFETs (two dimensional elec- year L.Esaki and L.L.Chang ob- by A.B.Fowler et al.. in 1,966. tron gas FETs), and in Japan, where served resonant tunnelling in a su- Spectral effects due to spatial they became known HEMTs (high perlattice. This strong interest in quantization were observed in thin electron mobility transistor). resonant tunnelling was obviously bismuth Films in 1968 by V.N. The first quantum well laser op- connected with its potential appli- Lutskii and L.A. Kulik at the eration was demonstrated by J.P. cations in high-speed electronics. Moscow Radio-Electronics Institute. van der Ziel et al.. in 1975, al- The success of these endeavours is Pioneering work on modula- though the parameters of the las- seen in the fact that by the end of tion-doped superlattices [5] ing were much worse than for the 1980s picosecond operation demonstrated a mobility enhance- average DHS lasers. By 1978, how- had been achieved in a double res- ment with respect to the bulk crys- ever, R. Dupuis and P.Dapkus, in onant tunnel diode and oscillations tal. This stimulated research on collaboration with N. Holonyak, were reporting quantum well lasers with parameters comparable (a) with their conventional DHS coun- terparts. Their paper was also no- table for its specific use of the • :."',/,i,~,~" ~ x",~: ,,, " ,,, term 'quantum well'. Even with this advance, it was not until 1982 that the real advantages of quan- -~ ,,~ tum well lasers were demonstrated. At that time, W.TTsang at Bell Telephone Labs used the many im- provements of MBE growth tech- t=O nology, as well as an optimized (b) structure (GRIN SCH), to achieve threshold currents as low as 160 A. cm 2. The idea of stimulated emission in superlattices that had been pub-

, ..: , •. lished by R.Kazarinov and R.Suris [4] back in 1971, was Fmally real- ized nearly a quarter of century later after a proposal by Federico t = 15 min t = 6 min Capasso. The proposed structure was strongly improved and a cas- ~ .~ , ,: • ... ,!:.~,: :.:::. ;.': ,~: ::. -.... cade laser developed by E Capasso gave rise to the new generation of unipolar lasers operating in the ~,. ; '..i.-. middle-infrared range. In considering the many mile- • "~r:. A.~: i :. stones in the study of quantum wells, perhaps the crowning achievement was the discovery of t = 25 min the quantum Hall effect. This discovery, and its comprehensive Figure 2: The evolution of active region luminescence patterns under high level photoexcita- tion for: (a) AIGaAs/GaAs SCH SQW (excitation level, 104 W.cm-2); and, (b) InGaAsP/ GaAs study in AIGaAs-GaAs heterostruc- SCH SQW (excitation level, 105 W.cm2). The diameter of the Kr+ laser excitation beam was tures shortly led to the discovery 40 IJm. of the fractional quantum Hall ef-

I II-Vs Review • Vol. 10 No. 7 1997 28 The d

• / e: E :=L r'.l : '-2. 0

E "-tE 0,0 -'1 0 <5 0 o E io

Figure 3: A SPSL QW SCH laser structure grown by MBE.

tion - the: fect discovery, which has had a solid state physics. Observation of principal effect on the whole of the effect, which deals only with

10 5 and diverse. Silicon ca~; as the basis of semicone~tor

104 4.3 k A/cm 2 (1968) Impact of Double Heterostructui'e 103 "~ . flmpact of < 900 iA/cm 2~ /Quahtum Wells e- ~:(1970) ~- 10 2 il 160 A/cm 2 (198il) v . 40 A/cm 2 'l (1988) 10 .., ...... : ...... Impact Of SPSL Q,

0 1960 65 70 75 80 85 90 95 2000 Park Years Figure 4: Evolution of the current threshold for semiconductor lasers. www http:/t www.epigress.se Quantum Wells Feature

Table 1. The contributions and importance of heterostructure quantum wells and superlattices.

Fundamental Physical Phenomena * IR quantum cascade laser * 2D electron gas * SPSL QW laser * Step type density-of-state function * Optimization of electron and light confinement * Quantum Hall effect and waveguiding for semiconductor lasers * Fractional Quantum Hall effect * 2D electron gas transitors (HEMT) * Excitons at room temperature * Resonant-tunneling diodes * Resonant tunelling in double-barrier structure * Precise resistance standards and superlattices * SEEDs and electrooptical modulators * In superlattices energy spectrum determined by * IR Photodetectors based on quantum size levels choice of potential and strain absorption * Stimulated emission at resonant tunelling in superlattices Important Technological Peculiarities * Pseudomorphic growth of strained structures * Lattice-match unnecessary * Principally require low growth-rate technology Important Consequences for Applications (MBE, MOCVD) * Shortened the emission wavelength, reduced * Submonolayer growth technique threshold current, increased differential gain * Blockading mismatch dislocations during and reduced temperature dependence of the epitaxial growth threshold current for semiconductor lasers

fundamental quantities and does with NTO AN, a scientific instru- using the rotating boat system. In not rely on peculiarities of the ments company attached to the our laboratory we developed a band structure, carrier mobility or •Academy of Sciences in Leningrad, new LPE method with the usual densities in semiconductor, has during the mid-1980s. Both types translational motion in a standard shown that heterostructures can of MBE systems are still working at horizontal system for InGaAsP het- be used to model some very basic the Ioffe Institute, as well as in oth- erostructures and a low-tempera- physical effects. Recently, many er laboratories. ture LPE method for AIGaAs studies in this area have concen- Initially, MOCVD systems were heterostructures. These methods trated on the understanding of the developed wholly within the enabled us to prepare a wide vari- condensation of electrons and the Institute. Later, in the 1980s, the ety of good quality quantum well search for Wigner crystallization. Swedish company Epyequip, with heterostructures having active re- our participation, designed a cou- gions with thicknesses down to Work at the A.F. Ioffe ple of systems which are still used 20 A and with the size of their in- in our research. terface regions comparable tO one Physico-Technical The strong interest in the ex- lattice constants. Institute perimental study of low-dimen- In relation to InGaAsP laser het- We started to develop MBE and sional structures, coupled with the erostructures, a development of MOCVD methods for growing III- lack of equipment for MBE and great practical importance was the V heterostructures towards the MOCVD, stimulated our research creation of a record threshold end of the 1970s.The first step was on the development of liquid current density for InGaAsP/InP the design and construction of the phase epitaxy CLPE) for quantum 0~=1.3 and 1.55 ~tm) and for first Soviet MBE machines. In the well heterostructures. Until the InGaAsP/GaAs 0~=0.65-0.9 lam) course of a few years, we helped to end of the 1970s, however, it single quantum well separate con- develop three generations of MBE seemed impossible to grow III-V fmement lasers. machines, with the final version heterostructures with an active-re- For high power InGaAsP/GaAs being fully capable of meeting our gion thickness of less than 500A by 0~=0.8 pm) lasers (Figure 1), a total needs. This unit is named "Cna' af- LPE because of the existence near efficiency of 66% with CW power ter the river not very far from the the heterojunction of extended in- of 5W for a 100 larn width stripe- city of Ryazan which is home to terface regions with varying chem- geometry laser was achieved. NITI, the industrial laboratory that ical compositions. This situation These lasers incorporated, for the carried out development of the changed in 1977 following the first time, the effective cooling of a machine. publication of work by N. semiconductor power device by In addition to the NITI system, Holonyak and colleagues, for su- recombination radiation -- a phe- we also developed a MBE system perlattice like InCmAsP structures nomenon that had been predicted

III-Vs Review • Vol.10 No. 7 1997 3O much earlier. Another important Acknowledgments conclusion we reached concern- ing the InGaAsP heterostructure, I am deeply indebted to P.S. Kop'ev was its unusual resistance to and N.N. Ledentsov for very fruitful the multiplication of dislocations discussions and to A.V. Gordeeva and defects (Figure 2). It was and N.E. Sergeeva for technical as- this research that initiated the sistance in the preparation of this broad application of M-free overview paper. heterostructures. The most complicated quantum References: well laser structure designed in our laboratory has been for the [1] R.Dingle, W.Wiegmann and creation of GRIN SCH structures C.H.Henry, "Quantized states of (the most favourable for the lowest confined carriers in very thin AIGaAs-GaAs-AIGaAs heterostruc- Jth). First demonstrated in 1988 [6], tures", Phys.Rev.Lett. 33,827, 1974. it combines a single quantum well [2] L. Esaki and R.Tsu,"Superlattice with short period superlattices and negative differential conduc- (SPSL).With the use of SPSL we not tivity ", IBM J Res. Dev. 14, 61. only achieved the desirable profile 1970. of a graded wave-guide region (thus creating a barrier for disloca- [3] L.V.Keldysh, "Effect of ultrason- tion movement to the active layer), ics on the electron spectrum of but were also able to grow differ- crystals", Fiz.Tv.Tela 4, 265, 1962 ent parts of the structure at large (Sov.Phys. :SoLState, 4, 1658, 1963). differences of temperature. In this [4] R.EKazarinov and R.A.Suris, way, we have obtained both an ex- "Possibility of amplification of elec- cellent surface morphology and a tromagnetic waves in a semicon- high internal quantum efficiency ductor superlattice ", Fiz. Tekh.Polupr. on a planar GaAs (100) surface. 5,707, 1971. The lowest Jth we achieved, initial- (Sov.Phys..'Semicond. 5,707, 1971). ly 52 A.cm 2 and shortly after opti- [5] R.Dingle, H.L.Stormer, H.L. mized to 40 A.cm2, is still a world Gossard and W. Wiegmann, record for semiconductor injection "Electron mobilities in modulation- lasers and remains a good demon- doped semiconductor heterojunc- stration of the application of quan- tion superlattices", Appl.Phys.Lett. tum wells and superlattices in 33,665, 1978. electronic devices. [6] Zh.I.Alferov et aL., "Reducing Future challenges of the threshold current in GaAs- AIGaAs DHS SCH quantum well The history of semiconductor lasers (Jth=52 A.cm-2,T=3OOK) with lasers is, from a certain point of quantum well restriction by short view, the history of evolution of period superlattice of variable peri- semiconductor laser current od", Pisma Zh.Tehn.Fiz. 14, 1803, threshold. This is shown in Figure 1988. 4. The most dramatic changes hap- (Sov. Phys.: Technical Physics Letters pened just after the introduction of 14,782, 1988). the DHS concept. The evolution of SPSL QW has tak- Contact: en researchers practically to the Zh. L Alferov, theoretical limit of this most im- A. E Ioffe Physico-Technical Institute, portant parameter, so now interest Russian Academy of Sciences, is turning to the application of the 26 Politechnicheskaya st. new quantum wire and quantum St. Petersburg, 194021, Russia. dot structures. E-Mail: [email protected]