November_EDFA_Digital

edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 22 NO. 4 30 n- [9] andp- [10] typedoping inGaAs, andmoreexoticexamples suchas p-type doping india-mond [11] and silicon (B doping confirmed by CL, unpublished results [12] ). Processing parameters opti- mization. When using processes such as implantation or dry etching, damage is sometimes dealt to the material, impairing its proper function. Here, cath- odoluminescence spectroscopy was used tomeasure dry etching- induced strain in InP, [6] or simply irradiation damage in GaN. [13] The effects of Mg implantation doping on GaN and its interplay with threading dislocations have also been measured. [14-15] Micro-LED. With micron-sized LEDs anticipated as the next gen- erationof display technology, the highspatial resolutionachievable with cathodoluminescence spec- troscopy can be extremely valu- able for both the development of the etching process and themea- surement of the homogeneity of the optoelectronic properties of micro-LEDs. [8] STATISTICAL PROCESS CONTROL Dislocation and point defects density in high electron mobility transistor (HEMT) structures. Rapid dislocation density measurement in GaN HEMT stacks is of notable interest, especially because GaN-on-Si technologies are becoming dominant due to their cost advantage but can contain ten times the dislocation densities observed in GaN-on-sapphire or on-silicon carbide. The impact of dislocations on device performance is expected to be significant. [16] The effect on its reliability remains to be demonstrated, but pipe diffusion of Mg into the active region [14] impacting the 2D electron gas at the AlGaN/GaN interface can be expected. V-shaped pits size, density, and distribution. V-shaped pits have long been known in InGaN-based structures, [17] but their impact on InGaN/GaN multi quantum wells (MQW) has been unclear until recently. Recent studies [18-19] highlight their beneficial influence on the emissions of MQW, which can be attributed to their influence on the composition of semiconductor alloys. Indeed, any varia- tion in composition impacts the near band edge emission of the resulting semiconductor. This has, for example, been performed on AlGaN, [3] AlGaAs, [4] and an example of the type of data needed is shown in Fig. 3. By the same token, the depth and spatial extent of local variations in alloy composition are easy to spot with a great accuracy. [5] The only limitation of this technique is that strain in semiconductor structures is known to affect the results in a fashion similar to alloy variation [6] and must also be considered; however the role of strain is dependent on the exact material composition [7] and epitaxial structure, leading to negligible effects in some instances. [3] Determination of dopant concentration. One crucial component in the fabrication of semiconductor devices is controlling dopant concentrations, both globally and locally. Here, CL spectroscopy can help as well. Indeed, measuring the full widthat halfmaximumof the near band edgeemissionofGaNallows the inferenceof Si dopant con- centration. [8] Similarmethods havebeendesigned for both Fig. 3 Examplesof the twomainacquisitionmodes incathodoluminescence spectroscopy. The panchromatic mode, top, uses a point detector to acquire the signal intensity over a certain span of wavelength. The hyperspectral mode, bottom, acquires a full emission spectrum in each pixel of the data set. The example given is that of a highelectronmobility transistor (HEMT) structure stackwhosegrowthwas stopped at the C-compensated layer. Example spectra drawn from the hyperspectral stack shown on the left are shown on the right, showing the ability to discriminate both alloy compositions (in the case of AlNandAlGaN) anddoping (in this case, C inGaN) by carefully analyzing the spectra acquired as a function of position.