Q3IC Technology
Question
Q.3. (a) Discuss about the hot wall reactor and cold wall reactor of LPCVD system. [6]
(b) What are the differences between growth mechanism of MOCVD and MBE? [4]
(c) Explain a technique for synthesis of GaAs homoepitaxy layer. [6]
Answer
Low pressure chemical vapor deposition (LPCVD) is carried out at reduced pressure (typically 0.1 to a few torr) compared to atmospheric pressure CVD, which substantially increases the mean free path and diffusivity of gas-phase species, shifting the process into the reaction-rate-limited regime discussed above, which in turn gives LPCVD excellent uniformity and step coverage since deposition rate becomes primarily governed by surface reaction kinetics (dependent mainly on temperature) rather than on local gas flow variations across the wafer or across a large batch of wafers.
Hot Wall Reactor
In a hot wall LPCVD reactor, the entire reactor tube, including its walls, is heated to the process temperature (typically by a resistance-heated furnace surrounding the tube), meaning both the wafers and the reactor walls themselves are at essentially the same elevated temperature. This design allows a large batch of wafers (commonly 50-200 wafers) to be loaded vertically, closely spaced in a quartz boat, and processed simultaneously with excellent wafer-to-wafer and within-wafer uniformity, since the entire tube is at a uniform temperature. However, because the reactor walls are also hot, deposition occurs on the walls just as it does on the wafers, leading to a buildup of deposited film on the tube walls over repeated runs, requiring periodic cleaning or replacement of the reactor tube to prevent particulate contamination from flaking wall deposits.
Cold Wall Reactor
In a cold wall reactor, only the wafer (or the susceptor holding the wafer) is directly heated, typically by radiant lamp heating, induction heating, or a resistively heated susceptor, while the reactor walls remain at a much lower temperature, often close to ambient, through active water cooling or simply by being positioned away from the direct heating source. Because the walls remain cool, little or no deposition occurs on them, substantially reducing particulate contamination from wall flaking and reducing the frequency of required reactor cleaning; however, cold wall reactors are generally single-wafer or small-batch systems, since achieving uniform, controlled heating of only the wafer (while keeping the surrounding chamber cool) becomes progressively more difficult to scale to large multi-wafer batches, making cold wall reactors more common in modern single-wafer processing tools where precise, rapid temperature control of an individual wafer is prioritized over high batch throughput.
MOCVD versus MBE Growth Mechanism
Metal-organic chemical vapor deposition (MOCVD) grows epitaxial layers by flowing metal-organic precursor gases (such as trimethylgallium and arsine for GaAs growth) over a heated substrate, where these precursors undergo pyrolytic decomposition (thermal breakdown) at the hot substrate surface, releasing the group III and group V atoms which then incorporate into the growing crystal lattice while the organic byproducts and excess reactants are carried away by the gas flow; MOCVD therefore operates through a chemical vapor-phase reaction mechanism at relatively higher pressure (from near-atmospheric down to low vacuum) and higher growth temperature, and offers high growth rates and good suitability for high-volume production. Molecular beam epitaxy (MBE), by contrast, is a physical vapor deposition technique carried out in ultra-high vacuum (typically below 10^-10 torr), in which elemental sources (such as solid gallium and arsenic held in separate effusion (Knudsen) cells) are heated to produce thermal beams of atoms or molecules that travel in straight-line, collision-free trajectories (a true molecular beam, given the very long mean free path at ultra-high vacuum) directly onto the heated substrate, where they condense and react to form the epitaxial layer with essentially no gas-phase chemical reactions occurring en route, since the process relies on direct physical deposition and surface reaction rather than a gas-phase-mediated chemical decomposition. Because MBE operates at much lower substrate temperatures than MOCVD and offers precise, real-time control of individual source shutters, it allows exceptionally sharp interfaces and atomic-layer-level control of composition and doping profiles, making it the preferred technique for research and for demanding device structures requiring extremely abrupt heterojunctions (such as quantum well lasers), while MOCVD's higher growth rate and greater compatibility with production-scale reactors make it the preferred technique for high-volume commercial device manufacturing.
Synthesis of GaAs Homoepitaxy Layer
A common technique for synthesizing a GaAs homoepitaxial layer (growing single-crystal GaAs on a GaAs substrate) is vapor phase epitaxy using the chloride transport process, in which arsenic trichloride (AsCl3) vapor is passed over a source boat containing liquid gallium at an elevated source-zone temperature, reacting to form gallium monochloride (GaCl) vapor and releasing arsenic vapor as a byproduct; this GaCl and arsenic vapor mixture is then transported by a carrier gas (typically hydrogen) downstream to a cooler substrate zone containing the heated GaAs substrate wafer, where the reverse reaction occurs, depositing epitaxial GaAs on the substrate while regenerating AsCl3 vapor that is carried away. The growth rate and layer quality in this transport process are controlled by carefully setting the source and substrate zone temperatures, the AsCl3 partial pressure in the input gas stream, and the total gas flow rate, since these parameters together determine both the thermodynamic driving force for the forward (deposition) reaction at the substrate and the mass-transport-limited or reaction-limited deposition kinetics analogous to those discussed for general CVD processes above. Alternative modern techniques for GaAs homoepitaxy include MOCVD (using trimethylgallium and arsine precursors as described above) and MBE (using elemental gallium and arsenic effusion cells), both of which have largely supplanted the older chloride vapor-phase transport method in current commercial GaAs device manufacturing due to their superior interface abruptness and compositional control, particularly for the compound heterostructure devices used in modern optoelectronics and high-frequency electronics.
It is also worth noting that the choice between MOCVD and MBE in practical compound semiconductor manufacturing is very often not an either-or decision, but rather reflects a deliberate split between research and development activity (where MBE's superior interface control and real-time monitoring capability, such as reflection high-energy electron diffraction used to observe the growing surface in-situ, are indispensable for developing new device structures) and high-volume production activity (where MOCVD's higher throughput and greater compatibility with large-batch or continuous-flow production reactors make it the economically preferred technique once a device structure has been fully developed and characterized using MBE-grown material), meaning most compound semiconductor device manufacturers maintain both types of epitaxial growth capability within their overall technology development and production pipeline.