Page 412 - Academic Press Encyclopedia of Physical Science and Technology 3rd Chemical Engineering

P1: FMX Final Pages
Encyclopedia of Physical Science and Technology EN009J-427 July 6, 2001 20:25

Metalorganic Chemical Vapor Deposition 497

However, in Manasevit’s first paper on MOCVD, concern- halides (e.g., AsCl 3 ) and Column III metals; (2) VPE using
ing the epitaxial growth of GaAs on insulators, the actual Column V hydride sources (e.g., AsH 3 ) and Column III
epitaxial process was not even mentioned! trichlorides, e.g., GaCl 3 ; (3) LPE usingColumn III metal
It is interesting to note that besides MOCVD, many solutions (e.g., Ga melts with GaAs source material); (4)
important innovations in III–V compound semiconduc- MBE using pure elemental sources (e.g., Ga and As). Al-
tor epitaxial growth technologies were first developed in ready in 1968, when Manasevit’s paper first appeared, the
the 1966–1967 time frame. For example, J. J. Tietjen VPE and LPE technologies were proven for the growth of
and J. A. Amick (RCA Laboratory, USA) first reported a variety of III–V “high-performance” devices. By 1973,
“open-tube” hydride VPE growth of III–Vs in 1966 and hydride VPE dominated the production of GaAsP light-
H. Rupprecht et al. (IBM Laboratory, United States) first emitting diodes (LEDs) and halide VPE dominated the
reported liquid-phase epitaxial (LPE) growth of ternary production of high-purity GaAs for electronic devices.
alloys of Al x Ga 1−x As in 1967. Also in 1967, J. A. Arthur LPE was the dominant technology for many III–V com-
(Bell Telephone Laboratories, USA) reported the first pounds, especially Al-containing devices, including Al-
studies of the properties of Ga and As molecular beams GaAs LEDs, lasers, solar cells, and other heterojunction
in ultrahigh vacuum—which ultimately lead to growth of devices. By 1975, MBE was being actively researched by a
GaAsbymolecularbeamepitaxy(MBE)reportedbyA. Y. few groups, particularly at Bell Laboratories and IBM Re-
Cho (Bell Telephone Laboratories, USA) in 1970. search Laboratory. Consequently, there was not much in-
Over the next few years after 1968, Manasevit and terest in MOCVD—it was viewed as just “another” III–V
coworkers explored the growth of various III–V, II–VI, materials technology—and the materials results seemed
and IV–VI compounds by MOCVD. Manasevit concen- to be much worse than those achieved by the other III–V
tratedonthegrowthofthinsemiconductorfilmsonvarious epitaxial growth technologies.
insulating oxide substrates including sapphire, spinel, and In 1977 R. D. Dupuis et al. reported high-performance
beryllium oxides. Much of Manasevit’s work was “proof- AlGaAs/GaAs solar cells and injection lasers grown by
of-concept” growth studies on insulators. The impurity MOCVD, showing that this technology could perform
content of these films was relatively high for several rea- at levels equal the other III–V materials technologies. In
sons. First, the purity of the metal alkyl sources was still 1978, they reported the first quantum-well semiconductor
very far behind that of other precursors used for III–V injection lasers operating continuously at 300 K, clearly
epitaxial growth, especially compared to the pure met- showing that the performance of MOCVD-grown devices
als and the metal halides. Contributions to the impurity could, in fact, exceed that of alternate materials technolo-
concentrations were also made by the hydrides. Further- gies. These results caused many groups to reconsider the
more, the MOCVD process is very sensitive to oxygen exploration of MOCVD materials technology, resulting
(more so than LPE and VPE) and the quality of the films in a rapid increase in the rate of publication of research
is degraded when small oxygen leaks exist in the reactor papers on this topic and its development as a production
system. Oxygen incorporation also contributes to exces- process for III–V epitaxial films.
sive C incorporation. Given the state-of-the-art in reactor
system design and construction in the late 1960s and early
C. General Description of the MOCVD Process
1970s, this oxygen sensitivity created serious problems,
especially for the growth of Al-containing alloys. These The MOCVD process (as applied to the growth of III–V
combined effects led to low carrier mobilities, high back- compound semiconductors) generally employs vapor-
ground impurity concentrations, poor surface morpholo- phase mixtures of Column III metalorganic and Column V
gies, and generally low photoluminescence efficiencies hydride sources (precursors) in a carrier gas and is carried
compared to those achieved by other competing and more out in an open-tube process chamber. In some cases, one or
well-developed III–V materials technologies, e.g., LPE more of the Column V precursors may also be a metalor-
and VPE. While Manasevit and other workers studied the ganic source. The carrier gases are typically purified H 2
growth of III–Vs by this process in the early and middle or N 2 . The input gas mixtures are heated above ∼ 350 C
◦
1970s, they were unable to demonstrate materials quality using a heating system, which provides thermal energy
comparable to that of other III–V epitaxial technologies to the growth surface of the substrate. The thermal en-
such as liquid-phase epitaxy and halogen- and hydride- ergy source is most often radio-frequency (RF) induction,
based vapor-phase epitaxy. electrical resistance, or optical infrared (IR) heating sys-
Sincetheearly1960sandintothemiddleandlate1970s, tems. The process chamber total pressure during growth
various other III–V materials technologies had been devel- is typically in the range 20–760 Torr (26–1000 mbar, 2.6–
oped, and had come to dominate the research and produc- 100 kPa). The process or “reactor” chamber is usually
tioneffortsworldwide,including(1)VPEusingColumnV composed of quartz or stainless steel.