Hard X-ray microprobe highlights the structure of multi-quantum wells in electroabsorption-modulated lasers for optoelectronics

Nowadays Multi-Quantum Well (MQW) structures based on quaternary III-V semiconductor alloys are widely used in optical communication systems. Advanced optoelectronic devices often require the integration of two different functions in the same chip: excellent results in the development of monolithic integration have been reached with the Selective Area Growth (SAG) technique [1]. SAG exploits the perturbation of the growth fluxes induced by a dielectric mask: when the metallorganic precursors collide with the dielectric mask, they are deflected and they migrate through the unmasked semiconductor where the growth starts. In this way the reactive species coming from the gas phase are enriched by those deflected by the mask: the result is a variation in composition and thickness of semiconductors grown near (SAG region) and far (field region) from the mask (Figure 1a). Electroabsorption modulated laser (EML), obtained by monolithic integration of an electroabsorption modulator (EAM) with a distributed feedback laser (DFB), is one of the most promising applications of SAG. A voltage modulation applied to the EAM switches it between an opaque and a transparent state by means of the Stark effect and ensures the modulation of the DFB laser emission, allowing long-distance communications (up to 80 km) at high frequency (10 Gb/s). The SAG EML device investigated in the present work is an AlxwGaywIn1xwywAs/AlxbGaybIn1xbybAs (compressive-strained well/tensile-strained barrier) MQW structure grown on InP by metallorganic vapour phase epitaxy. The SAG mask used for the growth (Figure 1a) featured 20 μm wide SiO2 stripes with a 30 μm opening width between them. The X-ray fluorescence (XRF) maps (Figure 1b) reveal that Ga Kα and As Kβ counts are higher in the SAG region owing to material enrichment caused by the SiO2 stripes. The effectiveness of the SAG technique in modulating the chemical composition of the quaternary alloy is proven by the map reporting the ratio between Ga and As counts in which a gradient in the average well/barrier chemical composition is clearly visible. Since the Ga/As ratio is lower in the SAG region than in the field, we can assert that the average Ga content of the MQW structure progressively increases by moving from the SAG to the field in the middle of a couple of stripes (along the Y-line showed in Figure 1a). The structural parameters of the sample were investigated by -X-ray diffraction (XRD): 35 different spatial points were sampled along the Y-line (Figure 2a). With such data, it is possible to obtain the widths (wb, ww, Figure 2c) and the mismatches (mb, mw, Figure 2d) of the barrier and of the well by fitting the observed patterns (Figure 2b) [2]. Both wb and ww undergo a modulated increase moving from field to SAG regions: this is the direct measure of the material enrichment in the SAG region. Moreover, both mb and mw values increase almost monotonically moving from the field to the SAG reflecting the expected modulation of the AlxGayIn1-x-yAs composition of barrier and well layers. This key information, coupled with the μm-determination of the energy gap by photoluminescence, allowed us to understand the structural gradient of the MQW structure along the Y-line from the field to the SAG region. Summarizing, in this study we determined the composition and the structure of the SAG EML device with a spatial resolution of 2 m. This unprecedented characterization allowed us to fully understand what was grown and thus it gives the appropriate feedback needed to improve the growth process, previously based only on a trial/error approach. A new generation of devices may rise by extending this method to other SAG growths with different stripes and opening sizes [3]. References [1] M.E. Coltrin, C.C. Mitchell, J. Crystal Growth 254, 35–45 (2003). [2] C. Ferrari, C. Bocchi, in: Characterization of Semiconductor Heterostructures and Nanostructures; C. Lamberti, Ed.; Elsevier: Amsterdam, 93–132 (2008). [3] L. Mino, A. Agostino, S. Codato and C. Lamberti, J. Anal. At. Spectrom. 25, 831-836 (2010)