Flux measurements

In application of RHEED oscillations to us, growing lattice-matched InAs on InAs substrate and GaAs on GaAs substrate are separately applied on calibrating fluxes of indium (In-fluxes) and fluxes of Gallium (Ga-fluxes). The relation of molecules flux vs. RHEED oscillation frequency can be expressed as

F Group III f Hz (3.1)

where F is the molecules flux, a_s is the substrate lattice constant, and f is RHEED oscillation frequency; it means the flux for one monolayer effectively corresponds to 2 atoms in square unit cell of zincblende structure. Therefore, the growth rate (GR) can be directly calculated by

GR f Hz, 1/sec , : the thickness of one monolayer (3.2)

Substitute (3.2) to (3.1), we get

Flux GR (3.3)

Due to the lattice constant of InAs (aAlAs = 5.6611Å) being almost same as that of GaAs (a_GaAs = 5.65325Å), therefore, Al-flux can be indirectly estimated from RHEED oscillations of GaAlAs growth on GaAs substrate with a linear interpolation of FAl+Ga = FAl +FGa. In other words, the Al-flux contributed RHEED oscillations frequency can be approximately obtained by fAlAs = fAlGaAs - fGaAs.

With this method, the measured In¹-BEP (i.e. BEP of cell In-1) by flux gauge which will be regarded as a long-term base and In¹-flux (i.e. flux of cell In-1) from RHEED oscillations measurement are indicated by Arrhenius plot in Fig. 3.3(a) and (b), respectively; the Ga-BEP and Ga-flux of Ga¹ and Ga² are separately shown in Fig. 3.4(a) and (b); also, the Al-BEP and Al-flux of Al¹ and Al² are separately shown in Fig. 3.5(a) and (b).

Fig. 3.3. The Arrhenius plot of measured BEP (a) and flux (b) of In¹-cell at different cell temperatures.

Fig. 3.4. The Arrhenius plot of measured BEP (a) and fluxes (b) of Ga¹- and Ga²-cells at different cell temperatures.

Fig. 3.5. The Arrhenius plot of measured BEP (a) and fluxes (b) of Al¹- and Al²-cells at different cell temperatures.

Once the GaAs growth rate is calibrated, the doping cell fluxes of Si and Be can be calibrated at different cell temperatures by a basic structure as the schematic diagram shown in Fig. 3.6. In order to measure their carrier concentrations (CC) from Hall effect [3.9], each sample is prepared in size of 5×5 mm² for soldering indium metal contact near to four corners of samples, which ohmic contact is finally formed in condition of 1 minute RTA treatment at 330^oC. Considering doping concentration in alloys is directly relative to growth rate; in other words, a fixed doping flux with a higher growth rate results a lower doping concentration, vice versa. Therefore, people prefer to know a combined capability of growth rate to carrier concentration in an Arrhenius plot. The Si-cell ability has been calibrated by Hall measurement as indicated in Fig. 3.7; the results show the flux slope of Si-cell at Riber C21T MBE system is almost same as our older MBE (2300RD). Also, the trend is consistent with other early reports [3.10], [3.11]. The Be-flux related to growth rate is also shown in Fig. 3.8; although its flux slop is a little different from 2300RD and other groups [3.10], [3.12], the trend still make sense for application. However, in case of achieving same carrier concentration, the necessary heat/temperature of C21T doping cells is higher than that of 2300RD and others. In Fig. 3.10, we show the measurement data of Hall mobility vs.

carrier concentration on Si-doped GaAs; the results also correspond to a predict trend according to Hilsum [3.13].

Fig. 3.6. The basic structure for determining the doping cell fluxes of Be and Si.

Fig. 3.7. Si-cell: the Arrhenius plot for growth-rate-related doping capacity.

Fig. 3.8. Be-cell: the Arrhenius plot for growth-rate-related doping capacity.

Fig. 3.9. The relation of carrier concentration vs. Hall mobility on Si-doped GaAs.

Once the characteristics of cells are obtained, the first emergency is to catch the accurate condition for achieving three basic lattice-matched bulk alloys, In¹0.523Al¹0.477As, In¹_0.532Ga¹_0.468As, and 1eV bandgap In¹_0.528Ga²_0.26Al²_0.212As. The procedure is started on predicting a temperature for In¹-cell (TIn1

) when the growth rate of In¹0.532Ga¹0.468As is set to 0.75 μm/hr; in other words, the effective InAs growth rate on InP (GRInAs@InP) is equal to 0.4 μm/hr (0.75×0.532). According to (3.3) and as = a_InP = 5.8687 Å, we firstly obtain the relation of flux vs. GR on InP substrate

F molecules cm⁄ sec 5.497 10 GR µm/hr (3.6) In case of GRInAs@InP = 0.4 μm/hr (0.75×0.532), hence, the necessary In¹-flux is 2.193 10 molecules cm⁄ sec ; simultaneously, substitute this flux into the fitting formula

for In¹-flux [Log F_In 23.6 9826 T K⁄ ], TIn1

= 788^oC is therefore determined as the beginning temperature point. For run-by-run, in fact, the In¹-cell is also very stable without clearly flux variation for long term growing. Based on T_In¹ = 788^oC, T_Ga¹, T_Al¹, T_Ga² and TAl2

are following estimated by the same approach. In the beginning of test run, we are used to grow three 300-nm thick ternary alloys but using different T_Ga¹ or T_Al¹ for realizing how much flux error from Ga¹ and Al¹ should be corrected.

On the other hand, in order to precisely analyze the group-III mole-fraction in quaternary alloy, single layer growth is needed to keep away the confusion from multi-peak signals presenting in DCXRD and PL results. In usual, a couple of growth runs with assistant of DCXRD and PL can determine the optimum conditions for lattice-matched alloys. After all cells fluxes are fine tuned, three basic lattice-matched materials were achieved and demonstrated by DCXRD measurements, in comparison with InP substrate as shown in Fig. 3.10. As for InAlAs in Fig. 3.10, another two diffraction peaks presented in the left side is in reason of three same thick layers but different TAl1

involved in growing InAlAs; among those three layers, one layer is lattice-matched to InP and its signal is embedded into substrate one. The results also show epitaxial quality is well controlled to suppress FWHM within 50-arcsec. According to the room-temperature (RT) PL measurement in Fig. 3.11, the PL peak of Be-doped lattice-matched InGaAlAs is just consistent with the desire one at wavelength = 1.24 μm (1eV).

Fig. 3.10. The X-ray diffraction in rocking angle for three basic lattice-matched materials, InGaAs, InAlAs, and 1eV InGaAlAs, in comparison with InP substrate.

Fig. 3.11. The room-temperature PL result from lattice-matched InGaAlAs.