Optimization of the growth process

The growth of Fe₃Si films was performed on well-ordered As-rich GaAs(113)A templates.

First, a 70-nm-thick GaAs buffer layer was prepared in a conventional III-V compound semiconductor growth chamber at a temperature of 610^◦C. Similar to the case of Fe films, we chose the As-rich surface of GaAs for the growth of Fe₃Si films, which was obtained by cooling down the substrate with the As shutter open until 400^◦C. The As-rich surface was chosen to avoid the formation of macroscopic defects on the surface similar to the case of Fe on GaAs(001) and GaAs(113)A substrates [63]. The growth of Fe₃Si was then performed in the separate As-free metal deposition chamber. Fe and Si were co-deposited from high-temperature effusion cells at a base pressure of 1×10⁻¹⁰ Torr. The following systematic approach has been adopted to optimize the growth of Fe₃Si on GaAs(113)A substrates.

First we kept the Fe to Si flux ratio constant and varied the growth temperature from 100 to 500^◦C. With the obtained optimum growth temperature, we then adjusted the growth rate to obtain a smooth surface morphology. Finally, to tune the Fe-Si composition, we

Figure 4.3: AFM images of Fe3Si films as a function of the growth temperature. Note that the scan area of the sample grown atT_G = 400^◦C is magnified to show the nanoscale

“ripples-like" structures.

varied the Fe to Si flux ratio at these optimized growth conditions.

Substrate temperature dependence

To obtain an optimum growth temperature, we chose an off-stoichiometric composition of the Fe_3+xSi1−x films and varied the substrate temperature of the growth. Herex denotes the deviation from exact stoichiometry. For an off-stoichiometric composition the layer peak is well separated from that of the GaAs and this makes it easier for HRXRD studies.

Figure 4.2(a) shows normalizedskew-symmetricω−2θscans near the GaAs(004) reflection for Fe3Si films grown at different temperatures from 100 to 500^◦C. The measurements were performed with an analyzer crystal in the diffracted beam optics with the sample tilted by 25.24^◦, the inclination angle of the (004) planes with the film plane. The skew-geometry for the measurement of asymmetric reflections can be found in Fig. 2.2 of chapter 2.2.

The measurement in theskew-geometry provides a better comparison with (001)-samples.

As can be seen, the sample grown at 100 ^◦C did not show any layer peak in the ω−2θ scans nor any RHEED pattern during growth, implying that the layer is amorphous.

Though the samples grown at T_G = 200, 250, 300, and 400 ^◦C show a layer peak due to the Fe₃Si(004) reflection, only the sample grown at 250 ^◦C shows distinct interference fringes, indicating a high structural ordering and an abrupt interface. However, the temperature range where these fringes are observed is much narrower compared to that on GaAs(001) [161], indicating a narrower optimized growth temperature window on the GaAs(113)A substrates.

For the sample grown atT_G = 400^◦C, we found additional peaks in wide-range skew-symmetric ω−2θ scans [as shown in Fig. 4.2(b)] at 2θ= 34.9^◦ and2θ = 73.9^◦, which are very close to the Fe2As(110) and (220) reflections, respectively [184]. Though the exact chemical composition for this layer at 400 ^◦C is not known, the presence of the these additional peaks indicates the formation of interfacial compounds. However, no additional

Figure 4.4: AFM images of Fe₃Si films grown at 250 ^◦C with growth rates of (a) 0.26 nm/min and (b) 0.13 nm/min, yielding RMS roughness of 5 and 1.6 Å, respectively.

peaks were observed for T_G ≤300 ^◦C [see Fig. 4.2(b)] indicating that the growth of Fe₃Si films on GaAs(113)A can be performed at a much higher temperature compared to Fe on GaAs. Noteworthy, the optimum growth of Fe₃Si on GaAs(113)A occurs at the same T_G (though the range is much narrower) as that for Fe₃Si on GaAs(001), whereas for the growth of Fe films on GaAs(113)A a lower T_G was required [63] as discussed in Sec. 3.3.

For the sample grown at T_G = 500 ^◦C, neither a layer peak nor any additional peak was observed in theskew-symmetric ω−2θscans, which may be due to the three-dimensional growth mode and/or formation of other crystalline phases.

Surface roughness and the effect of growth rate

Figure 4.3 presents AFM images of a set of samples for which T_G was varied. The RMS roughness of the films measured from these scans is plotted vs T_G in the inset of Fig. 4.2(a). Fe₃Si films with T_G ≤ 250 ^◦C exhibit minimal RMS roughness values of about5−6Å (measured over a 5×5µm² area). A significant increase in RMS roughnessσ occurs forT_G >250^◦C in agreement with the structural degradation of the films observed in HRXRD. Interestingly the AFM images of the samples grown at T_G = 200and 400 ^◦C show the formation of a “nanoscale ripples-like" structures, which were not observed in the case of [001]-oriented films. These “nanoscale ripples" were found to be parallel to h332iand h110i for T_G = 200and 400 ^◦C, respectively. This seems to be some indication of the anisotropic growth of Fe3Si layers on GaAs(113)A substrates, especially regarding the sample grown at T_G = 400 ^◦C. In principle, this could technologically promising.

However, as discussed in the previous paragraph, the sample grown at T_G = 400 ^◦C also shows the formation of interfacial compounds. The AFM image of the sample grown at even higher temperature of T_G = 500 ^◦C shows the formation of a large number of pyramidal-shaped nanocrystals indicating a three-dimensional growth mode.

The RMS roughness of the films can be reduced even further by lowering the growth rate (determined from thickness calibration) of the Fe₃Si layer. This is demonstrated in Fig. 4.4, which shows AFM images of two samples grown at 250 ^◦C with a growth rate of (a) 0.26 and (b) 0.13 nm/min. It should be noted that for the experiments reported before in Fig. 4.2, the growth rate was maintained at 0.26 nm/min. For the lower growth rate [Fig. 4.4(b)], the RMS roughness is reduced from 5 to 1.6 Å (measured over a 5×5 µm² area). Moreover, the growth rate reduction also improves the magnetic properties as

Figure 4.5: Normalized skew-symmetric ω −2θ scans for Fe_3+xSi1−x/GaAs(113)A films grown at 250 ^◦C with different Si cell temperatures. The curves are normalized to the GaAs(004) reflection and are shifted with respect to each other for clarity. The dotted line shows a simulation for a sample with (∆a/a)⊥ = 1.2%. See text and Ref [7] for simulation details.

discussed in Sec. 4.2.