I recently successfully defended my PhD thesis, entitled “Atomic Layer Deposition of Metal Oxide Thin Films for Si Heterojunction Solar Cells”. As the title suggests, I have worked on ALD of various metal oxides for the application in Si heterojunction solar cells. It was however one material in particular, hydrogen-doped indium oxide In2O3:H, that dominated my thesis and research. As it turns out, this material holds a lot of promise for application in a range of optoelectronic applications, as it is both highly electrically conductive and optically transparent at the same time. In this mini-review, I want to share briefly how we got involved in this work, why In2O3:H is so promising and explain the physics behind its success, and finally give you a glimpse of what it might bring us in the future.
Introductory talk during my Phd defense
What is In2O3:H?
In2O3:H is an upcoming transparent conductive oxide (TCO), originally developed by sputtering.1 It is much more transparent and conductive than conventional Sn-doped In2O3, ITO. The enhanced conductivity and transparency make this new material extremely interesting to replace ITO in a range of optoelectronic devices, and in particular solar cells. Indeed, substantial efficiency gains in a range of different solar cell types have already been demonstrated using this material. 2–7
In a nutshell, the main reason for the enhanced conductivity and transparency of this material is as follows: electrons are able to move much more easily in In2O3:H as compared to ITO, i.e. the electron mobility is much higher (~140 vs. ~40 cm2/Vs). Therefore, the same level of conductivity is achieved at a much lower electron density. This has the benefit that the absorption of (mostly) IR light which is induced by free carriers is negligible in In2O3:H as compared to ITO (see figure below).
Comparison of the absorption coefficient of sputtered ITO and ALD In2O3:H. The solar spectrum is added for reference.
How In2O3:H came into my project
Given the development of In2O3:H, one might think that I deliberately pursued the development of this material by ALD. However, this was not entirely the case, there was a bit more serendipity at play. The original plan was to develop ITO by ALD for silicon heterojunction solar cells. This was mainly motivated by the fact that we could already grow doped ZnO in our group, but it was thought that ITO would be better for our solar cells. Looking back in my logbook, it is now almost 3 years ago that I started to work on ALD of indium oxide with two other colleagues, with the intention to combine this at a later stage with Sn-doping to make ALD ITO. We selected the existing ALD In2O3 process as developed at Argonne National Lab based on indium cyclopentadienyl (InCp) and both H2O and O2 as oxidants.8
Schematic of the surface reactions during ALD of In2O3:H using InCp, H2O and O2. The H2O serves for the release of Cp ligands, whereas O2 further oxidizes the In. Note that the order of steps B and C can be interchanged, or combined in one step.8
It was however around that time that I became familiar with this novel material. Maybe with a bit of luck, I found that the ALD In2O3 films grown with InCp, H2O and O2 actually contain a lot of hydrogen, about 4 at. %. In other words, I had inadvertently grown In2O3:H!
How to make high-quality In2O3:H by ALD
Although I had grown In2O3:H by ALD, the properties of the material were mediocre at most, and comparable to ITO. This was because the films were amorphous, whereas high-quality sputtered In2O3:H films are polycrystalline. Therefore, I thought: Why not post-crystallize the films by thermal annealing? Indeed, this turned out to work wonderfully on the first try: After only annealing the amorphous In2O3:H for ten minutes at 200 oC record-high mobility values of 138 cm2/Vs and a low resistivity of 0.27 mΩcm were obtained. These promising initial results were quickly communicated in Rapid Research Letters9 in December 2014, and the manuscript was covered on the cover of that issue. In addition, Oxford Instruments, the supplier of the OpALTM ALD system that was used in this work, saw the great potential of this ALD process and featured it on their website
The In2O3:H ALD process featured on the cover of the Rapid Research Letters issue.
At that point, it became clear that the ALD In2O3:H material had a lot of potential. Besides the reproducible record optoelectronic properties, the main merits over conventional sputtering are the absence of plasma damage10,11, and the ability to conformally deposit on non-planar substrates such as nanowires.12
TEM picture of a silicon nanowire conformally coated with ALD In2O3:H.
Why does the ALD process work so well?
Clearly the process was working beautifully, but we were not yet so sure why. Therefore, a few more fundamental questions arose, that I would like to address in this mini-review:
- How does the post-crystallization work, and what are the critical parameters?
- Where does the H dopant come from?
- Why does hydrogen doping so well? Why is the electron mobility so much higher in In2O3:H?
- Once we know this, can we devise strategies to make it even better?
How does the crystallization process work?
In order to answer the first question, I studied the crystallization process in detail by both scanning and transmission electron microscopy. The main finding is that you only obtain the high-mobility In2O3:H if you deposit the In2O3:H at the lowest possible deposition temperature of 100 oC, and then post-crystallize it. To demonstrate why this is, I have made a small sketch with SEM pictures below.
The crystallization process monitored by top-view SEM. Left: as-deposited sample contains embedded crystallites (bright) in amorphous matrix (dark). Middle: Embedded crystallites grow upon thermal annealing. Right: Fully crystallized film with a lateral grain size of ~400 nm.
As can be seen, the sample deposited at 100 oC is actually not purely amorphous, but contains a few small crystallites that appear in bright. These crystallites act as seeds for further crystallization: Upon annealing, these crystallites grow in size and no new crystallites form in the amorphous material. Because of the low density of embedded crystallites, you end up with an extremely large grain size of ~400 nm, which is one of the crucial factors for the high electron mobility. This is also precisely the reason why you should deposit the In2O3:H at the lowest temperature: Increasing the deposition temperature strongly increases the density of embedded crystallites, and thereby reduces the final grain size. Also other aspects of the crystallization have been studied, which the interested reader can find in a publication in Journal of Applied Physics.12
Where does the H dopant come from?
Since the H dopant plays a crucial role in determining the properties of this material, the natural question arose as to where the H dopant actually comes from. When you think about it, the H atoms could come from the InCp precursor, the H2O reactant or even residual water in the non-perfect vacuum of the ALD reactor. Since our suspicion was that the H comes from the H2O reactant, we decided to do an isotope experiment: We replaced the water (H2O) reactant with deuterated water (D2O). This way, we could distinguish between hydrogen coming from the water reactant and other possible sources. As it turned out, the hydrogen in the film, typically ~4-5 at. %, indeed mainly originates from the water reactant! But there’s more: using Atom Probe Tomography we made a 3-dimensional map of the elemental distribution in our In2O3:H film. From this, we could clearly see that during ALD growth, less H (or D) is incorporated in the embedded crystallites than in the amorphous regions! More info can be found in our recent publication in ACS Applied Materials & Interfaces.13
Atom Probe Tomography image of In2O3:H grown at 100 oC. (left) A fully amorphous region with a homogeneous D distribution. (right) A region with an embedded crystallite, showing less D incorporation in the crystal.
Why does hydrogen doping work so well?
In order to answer the third question, I studied the mechanisms behind hydrogen doping (which determines the carrier density) and electron scattering (which limits the electron mobility) in the crystallized ALD In2O3:H. I actually found that in principle hydrogen is a very bad dopant in In2O3, but at the same time also excellent. Let me try to briefly explain this. I found that only about four percent of the embedded H actually acts as a donor. In other words, 96 percent of the H is inactive in In2O3 and does not donate electrons to the material. But is this a bad thing? Potentially: Sn atoms that are inactive in ITO are known to lead to electron scattering. I have however shown that this is not the case for H as the inactive H atoms do not lead to scattering of electrons!
OK, so the inactive H does not lead to scattering, nor do the grain boundaries due to the large grain size. But what kind of scattering is there in the crystallized In2O3:H? In other words, what is limiting the mobility to 138 cm2/Vs? It turns out that there are only two remaining scattering mechanisms of significance, which are schematically shown in a movie I rendered below.
Video taken from my PhD defense presentation in layman’s terms. Left: Vibrations of the crystal lattice (phonons) lead to scattering. Right: Positively charged H+ dopants (dark blue) attract electrons, also known as ionized impurity scattering.
Firstly, on the left there is phonon scattering: Electrons interact with lattice vibrations. By performing Hall measurements at liquid nitrogen temperatures, I could “freeze in” the phonons and determine their contribution to the total scattering: 34%. The other 66% stems from so-called ionized impurity scattering, on the right: electrons interact with the positively-charged blue H+ dopants. For further reading, see our publication in ACS Applied Materials & Interfaces.14
Can we devise strategies to make the material even better?
OK, so now that we know that there are only two scattering mechanisms left in the crystallized In2O3:H, can we come up with strategies to mitigate these remaining mechanisms, and thereby improve the material further? Unfortunately, or fortunately, depending on your point of view, these last two scattering mechanisms are in a sense unavoidable. One could in principle reduce phonon scattering by reducing the temperature, but that is probably not very practical for most device applications. Scattering by ionized impurities could be reduced by lowering the doping level. Although this can lead to higher mobility values, you also quickly lose in conductivity. To conclude, this basically means that the ALD In2O3:H is of the highest achievable quality.
So, what is next?
Firstly, now that we have this ALD process available, it is time to move to device applications! Having a crystalline silicon solar cell background, the first application that comes to mind is using this process in Si Heterojunction solar cells. As you can see from the cover photo of this article, we have already successfully made such cells! Since the manuscript detailing this process is now under review I cannot give any details at the moment. Besides that, we have succeeded in our group to do area-selective ALD with this process. Also here I cannot tell you too much about it yet, but you might expect a new blog about it soon!
Cover of my PhD thesis.
- Koida, T. et al. Hydrogen-doped In2O3 transparent conducting oxide films prepared by solid-phase crystallization method. J. Appl. Phys. 107, 33514 (2010).
- Barraud, L. et al. Hydrogen-doped indium oxide/indium tin oxide bilayers for high-efficiency silicon heterojunction solar cells. Sol. Energy Mater. Sol. Cells 115, 151–156 (2013).
- Jäger, T. et al. Hydrogenated indium oxide window layers for high-efficiency Cu(In,Ga)Se2 solar cells. J. Appl. Phys. 117, 205301 (2015).
- Steigert, A. et al. Sputtered Zn(O,S)/In2O3:H window layers for enhanced blue response of chalcopyrite solar cells. Phys. status solidi – Rapid Res. Lett. 9, 627–630 (2015).
- Keller, J., Lindahl, J., Edoff, M., Stolt, L. & Törndahl, T. Potential gain in photocurrent generation for Cu(In,Ga)Se2 solar cells by using In2O3 as a transparent conductive oxide layer. Prog. Photovoltaics Res. Appl. 24, 102–107 (2016).
- Fu, F. et al. Low-temperature-processed efficient semi-transparent planar perovskite solar cells for bifacial and tandem applications. Nat. Commun. 6, 8932 (2015).
- Yin, G., Steigert, A., Manley, P., Klenk, R. & Schmid, M. Enhanced absorption in tandem solar cells by applying hydrogenated In2O3 as electrode. Appl. Phys. Lett. 107, 211901 (2015).
- Libera, J. A., Hryn, J. N. & Elam, J. W. Indium Oxide Atomic Layer Deposition Facilitated by the Synergy between Oxygen and Water. Chem. Mater. 23, 2150–2158 (2011).
- Macco, B., Wu, Y., Vanhemel, D. & Kessels, W. M. M. High mobility In2O3:H transparent conductive oxides prepared by atomic layer deposition and solid phase crystallization. Phys. status solidi – Rapid Res. Lett. 8, 987–990 (2014).
- Macco, B. et al. Influence of transparent conductive oxides on passivation of a-Si:H/c-Si heterojunctions as studied by atomic layer deposited Al-doped ZnO. Semicond. Sci. Technol. 29, 122001 (2014).
- Demaurex, B. et al. Atomic-Layer-Deposited Transparent Electrodes for Silicon Heterojunction Solar Cells. IEEE J. Photovoltaics 4, 1387–1396 (2014).
- Macco, B. et al. On the solid phase crystallization of In2O3:H transparent conductive oxide films prepared by atomic layer deposition. J. Appl. Phys. 120, 85314 (2016).
- Wu, Y., Macco, B., Vanhemel, D., Kölling, S., Verheijen, M.A., Koenraad, P.M., Kessels, W.M.M. & Roozeboom, F. Atomic Layer Deposition of In2O3:H from InCp and H2O/O2: Microstructure and Isotope Labeling Studies. ACS Appl. Mater. Interfaces (2016).
- Macco, B., Knoops, H. C. M. & Kessels, W. M. M. Electron Scattering and Doping Mechanisms in Solid-Phase-Crystallized In2O3:H Prepared by Atomic Layer Deposition. ACS Appl. Mater. Interfaces 7, 16723–16729 (2015).