What is Liquid-Phase Epitaxy (LPE)?
This is a continuation from the previous tutorial - temperature dependence of threshold current of semiconductor lasers.
The lattice-matched crystalline growth of one semiconductor over another is called epitaxy.
The development of sophisticated epitaxial techniques has been of major significance in the development of high-quality, reliable semiconductor lasers. The commonly used techniques are liquid-phase epitaxy (LPE), vapor-phase epitaxy (VPE), and molecular-beam epitaxy (MBE).
In LPE a saturated solution of the constituents of the layer to be grown is cooled while in contact with the substrate.
In VPE, the epitaxial layer is grown by the reaction of gaseous elements or compounds at the surface of a heated substrate. The VPE technique has also been called chemical vapor deposition (CVD) depending on the constituents of the reactants.
A variant of the same technique is metal-organic chemical vapor deposition (MOVPE), in which metal alkyls are used as the compound source.
In MBE the epitaxial layer is grown by reacting atomic or molecular beams of the constituent elements (of the layer to be grown) with a crystalline substrate held at high temperature in an ultrahigh vacuum.
In this tutorial we describe the three epitaxial techniques and their use for the fabrication of double-heterostructure semiconductor lasers.
Liquid-Phase Epitaxy
The LPE technique as first demonstrated in 1963. Since then, it has been successfully used to fabricate various types of III-V compound semiconductor devices including injection lasers, light-emitting diodes, photodetectors, solar cells, bipolar transistors, and field-effect transistors.
In LPE a supersaturated solution of the material to be grown is brought into contact with the substrate for a desired period of time. If the substrate is single crystalline and the material to be grown has nearly the same lattice constant as the substrate, some of the material precipitates on the substrate while maintaining the crystalline quality. The precipitated material forms a lattice-matched epitaxial layer on the surface of the substrate.
LPE Apparatus
Three basic types of growth apparatus have been used for LPE. They are
- The tipping furnace, in which the substrate is brought into contact with the solution by tipping the furnace
- The vertical furnace, in which the substrate is dipped into the solution
- The multibin furnace, in which the substrate can be brought into contact with different solutions kept in successive bins. The multibin furnace type of growth apparatus is extensively used for the fabrication of laser structures that require the successive growth of several epitaxial layers.
Figure 4-1 shows the tipping furnace originally used by Nelson. The substrate is held at one end of a graphite boat inside a quartz tube. The solution is at the other end of the graphite boat. A thermocouple connected to the boat is used to control the temperature of the furnace. A flow of hydrogen through the system prevents oxidation.
The temperature of the furnace is reduced slowly; when the desired temperature is reached, the furnace is tipped so that the solution is in contact with the substrate, which allows an epitaxial layer to grow on the substrate, The furnace is then tipped back to the original position after the desired epitaxial layer thickness is obtained.
Figure 4-2 shows a vertical growth apparatus, in which the epitaxial layer is grown by dipping the substrate in a saturated solution.
The solution is kept in a graphite or Al2O3 chamber. The substrate is held by a holder just above the solution. Growth can be started and terminated by dipping and withdrawing the substrate from the solution at a desired temperature.
Figure 4-3 shows the multibin-boat apparatus generally used for growing double-heterostructure lasers.
The graphite boat has a number of reservoirs, each of which contains a saturated solution corresponding to the epitaxial layers to be grown.
The substrate (seed) is placed in a graphite slider that has a groove to hold the substate. The slider is attached to a long rod that allows an operator outside the furnace to position the substrate under different reservoirs.
In this way, several epitaxial layers of different materials and desired thicknesses can be successfully grown on the substrate. Hydrogen or helium is generally used as the ambient gas during the growth process.
Instead of a horizontal furnace and slider as shown in Figure 4-3, a rotary boat with a vertical furnace has also been used to grow several epitaxial layers successively on the substrate.
In this case the graphite boat is in the form of a circular cylinder, and the substrate is moved from underneath one reservoir to another by rotating the disc-shaped slider that holds the substrate.
Growth Methods
Several LPE methods have been used to grow InGaAsP material. They are
- Step cooling
- Equilibrium cooling
- Supercooling
- Two-phase method
The kinetics of the LPE growth process have been studied extensively.
Step Cooling
In the step-cooling technique, the substrate and the growth solution are cooled to a temperature \(\Delta{T}\) below the saturation temperature of the solution.
The substrate is slid under the solution and a constant temperature is maintained during the growth period. The growth is terminated by sliding the substrate (with the epitaxially grown layer) out of the solution.
The growth rate is determined by the diffusion rate of layer constituents from the solution to the substrate surface.
The thickness \(d\) of the grown layer is related to \(\Delta{T}\) and the growth time \(t\) by the relation
\[\tag{4-2-1}d=K\Delta{T}t^{1/2}\]
where \(K\) is a constant that depends on the diffusivity of each solute and on the solute's mole fraction in the solution at the growth temperature.
Equilibrium Cooling
In the equilibrium cooling technique, both substrate and solution are at the saturation temperature of the solution. Growth begins when the substrate is brought into contact with the solution and both are cooled at a uniform rate.
The growth is terminated by sliding the substrate with the grown layer out of the solution. The thickness of the grown layer is given by
\[\tag{4-2-2}d=\frac{2}{3}KRt^{3/2}\]
where \(t\) is the growth time and the cooling rate \(R=\text{d}T/\text{d}t\).
Supercooling
The supercooling technique is a combination of step-cooling and equilibrium cooling.
The substrate is brought into contact with the solution when both are at temperature \(\Delta{T}\) below the saturation temperature of the solution. The growth solution and the substate are further cooled during growth at a rate \(R\).
The thickness of the grown layer is given by a sum of Equations (4-2-1) and (4-2-2), i.e.,
\[\tag{4-2-3}d=K(\Delta{T}t^{1/2}+\frac{2}{3}Rt^{3/2})\]
The value of \(\Delta{T}\) used in the supercooling technique is generally smaller than that for the step-cooling technique.
Two-Phase Method
In the two-phase technique, the cooling procedure is the same as in the equilibrium cooling technique except that a piece of solid InP is added on top of the solution.
The solid InP is in equilibrium with the solution during growth. The two-phase technique can in principle be used to grow very thin layers because the presence of the solid InP piece reduces the growth rate.
The experimental results on the material quality obtained by the above methods or their variations are available in the literature. All of these methods produce good-quality wafers. Run-to-run variations, observed in an LPE process are influenced by wafer quality, surface preparation, and thermal decomposition prior to growth.
When an InP substrate is exposed to high temperature (\(\sim650^\circ\text{C}\)) occurring in an LPE reactor, the phosphorus evaporate from the surface, leaving behind indium rich regions. Epitaxy on this thermally decomposed surface has poor morphology and results in a low photoluminescence efficiency.
Several methods have been used to protect the substrate surface prior to growth by creating a phosphorus overpressure. These include
- Using a cover wafer in close proximity to the substrate to prevent large phosphorus loss
- Providing excess phosphorus in the vicinity of the InP substrate using phosphorus powder or phosphine gas
- Using an external chamber with phosphorus overpressure.
For the last scheme, one of the common methods is the use of a graphite or quartz chamber with an Sn-InP solution.
Figure 4-4 shows a schematic diagram of a multibin LPE boat with a quartz chamber. The substrate is kept inside this chamber prior to growth. The excess phosphorus overpressure created by the Sn-InP solution prevents any loss of phosphorus from the substrate surface.
LPE of InGaAsP
In order to grow lattice-matched layers of InGaAsP on InP, it is necessary to determine the phase diagram of In-Ga-As-P. The liquidus and solidus isotherms of this material system have been extensively studied.
Nakajima et al. determined the liquidus isotherm using the seed-dissolution technique. An undersaturated solution with known amounts of In, Ga, and As obtained from In, InAs and GaAs was saturated with P at \(650^\circ\text{C}\) using an InP seed.
The solution and the seed were kept in contact for \(\sim1\text{ h}\). The amount of P in the equilibrated solution was calculated from the loss of weight of the seed after removal from the solution.
Figure 4-5 shows the liquidus isotherms at \(650^\circ\text{C}\) for several atomic fractions of various elements in the liquid.
In order to grow high-quality epitaxial layers, one needs the correct solution composition for lattice matching. The composition for lattice-matched growth is determined by growing In1-xGaxAsyP1-y layers on InP from In-Ga-As-P solutions whose compositions are determined by liquidus isotherms.
The lattice constant of the grown layer is determined by the x-ray double-crystal diffraction technique. Precise lattice matching is verified using the substrate as a standard.
The melt compositions needed for lattice-matched growth depend strongly on the orientation of the substrate.
Nakajima et al. and Sankaran et al. have determined the melt composition for the lattice-matched growth on (111)B-oriented (P-rich planes) InP, while Nagai and Noguchi and Feng et al. have determined the melt compositions for (100)-oriented InP.
Their results as compiled by Nakajima are shown in Figure 4-6. Pollack et al. have reported similar measurements of the lattice-matched melt compositions for growth at \(620^\circ\text{C}\) on (100) InP.
The composition of the quaternary solid In1-xGaxAsyP1-y is determined using electron-microprobe analysis and wavelength-dispersive x-ray detection which uses a comparison of the intensities of the \(\text{Ga}-\text{K}_\alpha\), \(\text{As}-\text{K}_\alpha\) and \(\text{P}-\text{K}_\alpha\) lines from the grown sample and from GaP and InAs standards.
Figure 4-7 shows the measured composition of the lattice-matched quaternary solid. The solid line is the calculated result on the basis of Vegard's law, according to which the lattice constant \(a(x,y)\) of In1-xGaxAsyP1-y is given by
\[\tag{4-2-4}\begin{align}a(x,y)=x&ya(\text{GaAs})+x(1-y)a(\text{GaP})\\&+(1-x)ya(\text{InAs})+(1-x)(1-y)a(\text{InP})\end{align}\]
Table 4-1 lists the room-temperature lattice constants of GaP, GaAs, InP, and InAs. Using \(a(x,y)=a(\text{InP})\), Equation (4-2-4) yields
\[\tag{4-2-5}x=\frac{0.4526y}{1-0.031y}\]
as the relation between the mole fractions \(x\) and \(y\) for the lattice-matched layer.
Figure 4-7 shows that InGaAsP quaternary alloys obey Vegard's law.
Figure 4-8 shows the scanning-electron-microscope (SEM) cross section of a typical 1.3-μm InGaAsP double-heterostructure wafer grown by the equilibrium cooling technique on (100)-oriented n-type InP.
The four layers grown are n-type InP (buffer layer), undoped InGaAsP (\(\lambda=1.3\) μm, active layer), p-type InP (cladding layer), and p-type InGaAsP (\(\lambda=1.3\) μm, contact layer).
The next tutorial introduces what are vapor-phase epitaxy (VPE) and metal-organic vapor-phase epitaxy (MOVPE).