Answers

Gatan Answers 2

Defect distribution and compositional variations in a light-emitting material revealed by CL imaging and spectrum imaging

David Stowe and Simon Galloway, Gatan UK

The increasing use of opto-electronic materials and devices is core technology to our modern world. Emitting, amplifying and sensing devices enable high-speed telecommunications whether through fibre optic, mobile or wireless. There are also growing fields like optical data storage, energy efficient solid-state lighting and photovoltaics. In order to develop and manufacture devices with novel performance and suitable lifetimes, it is important to understand the material science governing the opto-electrical behaviour. Included in this, is an understanding of problems associated with scaling from laboratory prototypes to high yield, large-scale manufacturing. Characterisation plays a pivotal role at all stages in this process, especially when it comes to identifying the role defects play in limiting the performance or lifetime. Thus, a technique which enables spectroscopic characterisation of opto-electronic properties at high spatial resolution is extremely powerful. Cathodoluminescence (CL) is such a technique and Gatan’s MonoCL3 system remains at the forefront of this discipline.

CL is the emission of photons from a material when excited by an electron beam. A wide range of materials, outside the obvious example of phosphors, give rise to CL emissions. Indeed those materials used for opto-electronic devices are ideal. The number and energy of photons emitted provide a wealth of information about the material and the underlying physics. In compound semiconductors, the CL will be a function of the band gap (associated with alloy composition and crystal symmetry), dopants, impurities, line and point defects, stress, internal electric fields, temperature, quantum confinement, and more! Such sensitivity to material and experimental parameters (e.g. injection energy and density), when coupled with the analytical environment of a scanning electron microscope (SEM) or a transmission electron microscope (TEM), makes CL a potent and highly flexible tool. Although a scanned image showing CL intensity can be informative, especially with suitable resolution, the real wealth of information is often only extracted by employing spectroscopic techniques at cryogenic temperatures.

Read on for a detailed look at how Gatan’s MonoCL3+™ and XiCLone™ system and CF302 liquid-helium cooled SEM stage were applied to understand defect distributions and alloy inhomogeneities in the commercially important and scientifically interesting AlGaN system.

Case study: cathodoluminescence investigation of defects in crack free (Al,Ga)N layers with low Al content.

All results shown are courtesy of Dr Uwe Jahn of the Paul Drude Institute for Solid State Electronics, Berlin.

Nitride semiconductors have attracted widespread interest for more than a decade for their extensive application potential. Aside from the extra-ordinary radiative efficiency that can be achieved with novel growth techniques, much of their potential, and indeed functionality in devices, comes from band-gap engineering and doping. Alloying GaN with aluminium (AlGaN) or indium (InGaN) theoretically allows the bandgap to be fine-tuned all the way from the UV to IR wavelengths. On the quantum scale this enables confinement effects necessary, for example, for room temperature laser operations, as used in blue laser digital storage technology. Application areas for wide band gap AlGaInN semiconductors are diverse and include solid-state lighting, radiation detectors / solar cells and as high power / high-frequency electronic materials (e.g. 11 W/mm and 10 GHz respectively already demonstrated).

Higher Al concentration pushes the emission wavelength to shorter wavelengths. However, growing high quality epitaxial AlGaN, with good control over the electrical conductivity and radiative efficiency, remains a significant challenge. Surface cracking is a serious obstacle for AlGaN growth on traditional substrates, as higher Al concentration leads to higher strain levels. Sophisticated buffer layer engineering techniques are required to achieve bulk, crack-free, AlGaN but the higher the Al content the more taxing the problem. Two important factors in determining the quality of a semiconductor are the defect distribution and compositional homogeneity. This requires the characterisation of a material’s optical properties on the sub-micron scale.

A central tenant to Gatan’s MonoCL3 technology is achieving high spatial and spectral resolution results. With SEM specimens, the resolution is primarily dictated by the generation volume, with the spot size only becoming an important factor once the kV is sufficiently low. Modern microscopes are capable of focusing to a small spot size at low kV, but with a near surface generation volume, the ratio of radiative emissions to surface losses becomes small. Excessive injection density is often inappropriate as this perturbs the equilibrium conditions being measured, and may cause beam induced changes to the specimen. In addition increasing the beam current at the desired low kV, often leads to unwanted degradation in the spot size, or charging problems. Hence, high spatial resolution injection conditions are associated with a limited photon flux. All these factors emphasize the necessity of optimum collection, dispersion and detection efficiency in a CL system. MonoCL3 features direct optical coupling to detectors for serial and parallel acquisition modes, and this enhances efficiency, whilst avoiding chromatic losses. In this study, a short working distance CL collection mirror was employed to optimise resolution at low kV.

Cryogenic cooling enhances CL at the expense of non-radiative recombination losses and this can therefore enhance spatial resolution. Furthermore it can minimize the injection levels required to stimulate certain processes, increase spectral discrimination and allow unambiguous interpretation of the underlying opto-electronic physics displayed locally by the specimen. However, high spatial resolution work requires that the cooling stage does not introduce drift or image vibration. Gatan’s CF302 continuous flow liquid helium cold stage is optimised for these purposes and is ideally compatible with most SEMs as well as with MonoCL3 for semiconductor applications.

The sample used in this study was a 2.5 microns thick Al_xGa_x-1N(x = 0.10) layer, grown by metal organic chemical vapor deposition on a sapphire substrate with a thin GaN nucleation layer. The sample was crack free and investigated in a field emission scanning electron microscope (FE SEM). A Gatan liquid-helium cold stage and airlock (CF302 + C1010) was used and all results shown in this report were obtained at a base temperature of 6K.

Figure 1 shows the plan view investigation of the sample. The secondary electron image revealed a relatively smooth surface with some regions showing dark contrast, these were assumed to be defect related. A spectrum recorded at 6 K displayed two strong peaks centred at 3.665 and 3.520 eV. Several smaller peaks were also observed between 3.6 and 3.64 eV. Monochromatic imaging at 3.553 and 3.674 eV clearly showed a direct correlation between the defect distributions observed in the secondary electron image and the lower energy peak. Several of these regions are indicated in Figure 1. The defects are thought to be inversion domain related, with lower Al concentrations. The monochromatic image at the higher energy clearly showed regions of relatively uniform emission between the defects. The energy of this luminescence closely corresponded to the bandgap of Al_xGa_x-1N (x = 0.1) at 6 K.

Figure 1 a) Plan view secondary electron image of the AlGaN film, b) CL spectrum of AlGaN sample, c) monochromatic CL image at 3.533 eV and d) monochromatic CL image at 3.674 eV. All results were recorded at 6 K.

The sample was cleaved and investigated in cross-section using panchromatic imaging, monochromatic imaging and spectrum imaging. Figure 2 shows the secondary electron image of the cross-section - the sapphire substrate, cleaved face and the surface are clearly seen - and a panchromatic image acquired at 6 K. The CL was observed to be non-uniform; some areas appeared bright, whilst dark, vertical lines and dark regions in close proximity to the surface and to the AlGaN/sapphire interface were also observed. Electron-hole pairs generated by the electron beam are free to recombine via a number of mechanisms. CL is a competitive process and any region which has a faster non-radiative recombination rate than surrounding region appears dark. The darker areas are interpreted as regions of higher dislocation density, near to interfaces where the increased non radiative recombination reduces the CL intensity.

Figure 2 a) Secondary electron image and b) panchromatic CL image of the cleaved AlGaN layer at 6 K. The arrows simply link the 2 images at 2 defined points, whilst the dashed line indicates the position of the interface between the substrate and the AlGaN layer.

Figure 3 shows a secondary electron image (a) and four monochromatic CL images (b-e), recorded at 3.524, 3.542, 3.664 and 3.658 eV. The corresponding energies are indicated on the spectrum in Figure 3f. The 4 monochromatic images showed clear differences. The two images formed using the lower energy band-passes showed vertical stripes. These largely ran from the substrate/AlGaN interface to the surface, though some were located in the lower portion of the layer only and terminated within the AlGaN layer itself. The two higher energy images showed bright, horizontally striped regions which were periodically modulated.

Figure 3 a) Cross-sectional secondary electron image of the cleaved face of the AlGaN on sapphire film. b) - e) monochromatic CL images at 3.524, 3.542, 3.664 and 3.658 eV respectively and f) CL spectrum of AlGaN sample, the 4 monochromatic energies are indicated by the arrows. All results were recorded at 6 K.

The color mix facility included within Gatan’s DigitalMicrograph™ software was used to combine the 4 monochromatic images and the secondary electron image. The colors red, green, blue and yellow were assigned to the monochromatic energy images in ascending energy order and the secondary electron image was kept as a greyscale image, see Figure 4. In the color mixed image, the complex distribution of the CL emission is shown. From this image, it was clear that those regions with red and green coloring corresponded to defects which ran vertically though the layer, whilst those regions with blue and yellow coloring were associated with AlGaN luminescence from regions between defects. The fact that the luminescence was observed to vary both laterally and vertically indicated a corresponding fluctuation in the aluminium fraction of this compound semiconductor.

Figure 4 A color mixed image of monochromatic CL images at 3.524, 3.542, 3.664 and 3.658 eV and the secondary electron image of Figure 2a. The defects are clearly observed as vertical stripes (green and red), whilst, in other areas defect free AlGaN luminescence is observed (blue and yellow).

A more detailed and quantitative investigation of the AlGaN compositional fluctuation was made using spectrum imaging linescan profiles, obtained using the XiClone facility. In this mode, a line is selected by the user along which a number of data points are acquired. At each data point along the line a full spectrum is acquired, thus creating a multi-dimensional data set (wavelength, intensity and position). For this sample, a total acquisition time of only a few seconds was required to obtain the data presented below. The camera employed was LN2 cooled with 1340 x 100 pixels of 20 micron size, read at 16 bit depth and 2 MHz. This sensitivity and configuration helps with low injection conditions, optimises the spectral resolution of the results, and increases the speed of the experiment. (With XiCLone, spectrum imaging can be performed on areas as well as line-scans, in which case dynamic drift correction can be employed to enhance the integrity of results over extended acquisition periods.)

Spectrum imaging linescans can be acquired in just a few seconds and are a powerful means of quickly quantifying the spatial variation in the spectral signal at different depths in the cross section. In this example, 40 spectra were recorded at 53 nm intervals in two positions as shown in Figure 5. In the upper line profile (labelled 1), the high energy, bandgap related CL line dominated. As the peak energy position reflects the Al content in the alloy, Gaussian fitting tools can be used to map the Al concentration of the alloy along the linescan. The results shown in Figure 6 indicate little fluctuation in the Al fraction along the upper part of the cross section (0.120 – 0.124 = ~3%). In contrast, the linescan closer to the interface, shown as linescan 2 in figure 5, shows a completely different story. In place of relatively well defined peaks with little variation in the peak energy positions, there are often multiple peaks of different energy and varying intensity. The local distribution of Al content in the alloy is very much more perturbed due to the higher defect concentration. It is not possible to perform a similar fit of the peak energy position from the linescan spectrum image because of the multiple nature of the CL peaks from a given beam position. However, the complex nature of each spectrum suggests variations in the Al concentration occur on a scale smaller than the resolution of the probe.

Figure 5 Cross sectional secondary electron image of the cleaved sample, with the regions investigated by spectrum imaging indicated by the lines labelled 1 and 2. The raw data from the 2 spectrum images is shown in the lower figure, and is plotted as intensity (a.u.) as a function of energy (eV).

Figure 6 Aluminium fraction as a function of position along the spectrum image linescan from the upper part of AlGaN layer, linescan 1.

Summary

The MonoCL3 family of products is ideally suited to providing “the full picture” when investigating new opto-electronic materials and devices. This note illustrates the wealth of information which can be achieved quickly and with confidence, at the necessary spatial resolution when studying the important AlGaInN system.

Defect distribution and compositional variations in a light-emitting material revealed by CL imaging and spectrum imaging

Case study: cathodoluminescence investigation of defects in crack free (Al,Ga)N layers with low Al content.

Summary

Further Reading