Are Electron-blocking Layers Always Beneficial?
BY Z. SIMON LI AND CHANGSHENG XIA FROM CROSSLIGHT
If you peruse the lighting section of a good hardware store, you’ll find plenty of bulbs to choose from. Those with the most impressive performance figures will be based on LEDs, which hold the key to higher efficiency, a lifetime of a decade or more and an output that hits its maximum in the blink of an eye. However, the price you’ll have to pay may put you off – a 60 W-equivalent from a well-known brand can retail for tens of dollars.
Given this state of affairs, it is clear that prices must plummet before solid-state lighting becomes the obvious choice for everyone. This will require a trimming of the cost of many of the components in the bulb, including the biggest contributor to expense, the packaged LED, which accounts for about half the total bill of materials. To drive this down, chipmakers can improve their manufacturing efficiencies, so the cost of making the devices tumbles; or LED bulb makers can drive the devices harder while maintaining their efficiency, so far fewer are needed.
There is much to recommend the latter option, including the potential to slash costs and shrink the dimensions of the light engine. However, this pathway is thwarted by a mysterious malady known as LED droop – a decline in device efficiency as the current through the chip is cranked up (see Figure 1). What causes droop is highly controversial, but two of the primary culprits are believed to be electron leakage and a poor hole injection efficiency.
Figure 1. Simulations of the LED using Crosslight APSYS software were fitted to the performance of a reference LED. Plots show (a) light output power and (b) efficiency, which reveals the level of droop in the device.
To address these transport-related issues, many groups insert a thin AlGaN layer – called the electron-blocking layer (EBL) – between the multiple quantum well region and the top p-contact (see Figure 2). But not everyone does this, because some do not believe that the EBL is beneficial. Their reasoning may be based on some published experimental data, which indicates that LEDs without an EBL perform better than those with it (see for example, the paper from researchers at Gwangju Institute of Science and Technology and Samsung Electro-Mechanics: S. H. Han et. al.Appl. Phys. Lett.94231123 (2009)). In addition, experimental reports of the varied effectiveness of EBL have appeared, adding to the controversy surrounding the employment of the p-type AlGaN EBL in GaN-based LEDs. So it is now critical to clarify what the role of the p-type AlGaN EBL is, and to explicitly answer the question of whether it is useful – and if so, under what conditions.
Figure 2. Crosslight simulations considered a six quantum well GaN LED with an electron blocking layer (EBL) on the p-side.
At Crosslight Software, a leader in compound semiconductor device simulations based in Vancouver and Shanghai, we have sought answers to this question by performing a series of systematic simulations on GaN-based LED structures. This investigation has clarified the physics behind the operation of the EBL in a GaN-based LED.
Figure 3. Light output power for the LEDs with differing EBL polarization compensation factor as a function of aluminium composition at 120 A cm-2.
For this work, we have considered a typical InGaN/GaN LED: a structure with six quantum wells (see Figure 2) and a 20 nm-thick, p-doped EBL. To ensure a realistic simulation study, we use our APSYS models to fit a reference device that had an EBL with an aluminium composition of 0.15 (see Figure 1).
Key features of our effort, which ensure that we simulate a realistic device structure, are the inclusion of a polarization compensation factor and a band-offset ratio for the EBL. The polarization compensation factor offers a mechanism for scaling down the ideal theoretical polarization interface charge, and it can account for partial compensation of the polarization charge by defects and other interface fixed charges. Meanwhile, the band-offset ratio provides a means to tune the ratio between conduction band discontinuity and total band discontinuity.
Note that both quantities are difficult to measure experimentally, and making matters worse, the polarization compensation factor may depend on interface quality and growth conditions. It is believed that a reasonable value for the polarization compensation factor is between 0.2 and 0.7, while that for the band offset ratio could span 0.5 to 0.65.
Fitting real data
When we fitted the experimental data, we obtained a polarization compensation factor of 0.3 for all the interfaces and determined a value of 0.5 for the band offset ratio for the EBL (between AlGaN and GaN). To better focus on the impact of the EBL, we then fixed the polarization compensation factor for the entire device, allowing it to only vary at the EBL.
Using a reference value of 0.3 for the polarization compensation factor and 0.5 the band offset ratio, we simulated LED performance for various AlxGa1-xN compositions to reveal whether the EBL is useful or not. These simulations showed that the output of the LED initially increases with aluminium composition (x) up to a maximum at 0.05, before decreasing continuously thereafter.
At first glance, these results are at odds with experimental findings, which show that the LED with the EBL had poorer performance than the one without. However, the LEDs that had an EBL had aluminium compositions of either 0.22 or 0.32 – values that are not close to the optimal alloy formulation. So we can conclude that the EBL is useful, but only at a small range of aluminium compositions, which were unfortunately missed in the experimental report.
To gain further insights into droop and device behaviour, it is imperative to understand the reasons behind an optimal aluminium composition. We have investigated this, beginning by considering energy band diagrams for LEDs with different aluminium compositions under an EBL polarization compensation factor of 0.3 at 120 A cm-2 (see Figure 4).
Figure 4. Energy band diagrams of the LEDs (a) without an EBL, (b), (c) with aluminium composition of 0.05 and 0.3 in the EBL under an EBL polarization compensation factor of 0.3 at 120 A cm-2.
Improving hole injection
For the LED without an EBL, the energy band within the active region is bent due to polarization charges – and this is to blame for the large electron leakage and poor hole injection efficiency. However, when an AlGaN EBL with an aluminium composition of 0.05 is introduced, an electron barrier forms in the conduction band, with the triangle barrier for holes at the last well/barrier interface pushed upwards by the polarization charges. This leads to an increase in the effective barrier height for electrons from 423 meV to 478 meV, and a corresponding cut in hole barrier height from 299 meV to 267 meV. This explains the suppression of electron leakage and the improvement to hole injection efficiency. These results reflect that as aluminium composition of the EBL increases, there is a rapid rise in the energy of the EBL valence band barrier and a strong polarization-induced downward bending effect. Consequently, the valence band of the EBL becomes the dominant barrier impeding the injection of holes.
This scenario is highlighted by simulations considering an aluminium composition of 0.3 (see Figure 4(c)). In this case, the effective barrier height for holes is increased to 413 meV, while that for electrons falls to 365 meV, even though the EBL conduction band barrier is also increased with higher aluminium composition. A poorer performing LED results, due to the combination of increased electron leakage and inferior hole injection.To prove that polarization charges play a central role in the aluminium composition dependence, the polarization compensation factor was varied from zero to 0.5 (see Figure 3). These simulations show that LED behaviour is highly sensitive to the polarization charges: if they were absent, the EBL would be mostly beneficial in a wide range of aluminium compositions; but if the polarization charges are significant (a polarization compensation factor of 0.5 or more), the EBL would be mostly useless, and only aid performance for a very narrow range of aluminium compositions. These findings explain why some reported experimental data failed to uncover the benefits of using an EBL.
Figure 5. Light output power of the LEDs with different EBL polarization compensation factors as a function of aluminium composition at 120 A cm-2 with an EBL band offset ratio of 0.6.
We have also considered the impact of the band-offset ratio on electron leakage, hole injection and the effectiveness of the EBL. Turning to a higher value of band-offset ratio of 0.6, we investigated how variations in the composition of the EBL influence LED performance. At the higher band-offset ratio, the EBL always enhances LED output power (see Figure 5). This is not surprising, since a high electron barrier suppresses electron leakage, while a low hole barrier enhances hole injection
Thanks to its name – the electron blocking layer – the capability of the EBL to block the overflow of electrons is rather well understood. But this moniker may also be partly to blame for a less well known, but equally important consequence: the blocking of hole injection. To illustrate this, we have plotted the hole concentration in the last quantum well as a function of aluminium composition for band offset ratios of 0.5 and 0.6 (Figure 6). It is clear that at a band offset of 0.5, increases in aluminium composition produce a substantial decrease in hole concentration. However, at a higher offset of 0.6, a consistently higher hole concentration is possible for all aluminium compositions.
Our simulations of LED behaviour for a range of devices – different EBL compositions, polarization compensation factors and band offset ratios – explain the apparently conflicting reports on the usefulness of the EBL. We have shown that the polarization compensation factor and the band offset ratio play critical roles in determining the performance of the LED. Learning how to optimise them will help to minimise droop and spur the adoption of solid-state lighting.
Figure 6. Maximum hole concentrations at the last quantum well as a function of aluminium compositions at different band offset ratios. Calculation at 120 A cm-2 with polarization factor of 0.3.
C. S. Xia et. al. Appl. Phys. Lett. 103233505 (2013)