Formation of Free Hydrogen Gas By Annealing ALD-Al2O3/Si Stacked Structure

Abstract

1. IntroductionAtomic layer depositions (ALD) of Al2O3 films are very essential technique to realize next generation electronic devices thanks to their high-k properties [1]. However, ALD-Al2O3 films usually suffer blistering during thermal treatments, which have caused serious problems for decades in the field of device fabrication [2]. Many groups have studied the cause of this phenomenon, and various models have been proposed. such as, stress at the interface between Al2O3 and Si [3], diffusion of Si atoms at the interface[4], residual H2O in the film after the ALD process [5], and aggregation of hydrogen atoms at the Al2O3/Si interface [6], and so on. However, the actual mechanism has hitherto remained unclear due to a lack of conclusive evidence. Here, in this study, we have tried to reveal the origin of this phenomenon using spectroscopic measurements [7]. As a result, we have succeeded in directly identifying the gas inside the blisters, which can be a concrete evidence to reveal the roots of a phenomenon that has long been obscured.2. Experimental ProcedureIn the experiment, (100) oriented Si substrates with resistivity of 0.1-0.5 W∙cm were carried out. After cleaning the substrates by acetone, sulfuric-acid hydrogen-peroxide mixture solution, and hydrofluoric acid solution, all the substrates were loaded into ALD chamber which was heated up to 300oC. Then, Al2O3 layer with ~108 nm thickness was deposited on the substrates with conventional thermal ALD process using trimethylaluminum (TMA) and H2O as precursor gasses. After deposition, samples were annealed in rapid thermal annealing (RTA) furnace for 5 min with 400 – 1000 oC at N2 ambient. Finally the samples were characterized by laser microscopy, field emission scanning electron microscopy (FE-SEM), spectroscopic ellipsometry, and Raman spectroscopy.3. Results and DiscussionsFigs. 1(a) and 1(b) show typical microscopic image of sample before, and after RTA (500oC, 5 min), respectively. We can clearly see that several blisters appear by annealing the samples, while there was no such sign before annealing. From the microscopic observations, we have summarized density of blisters as a function of RTA temperature in Fig. 2. Here, we can clearly see that by increasing RTA temperature, blister density increases, which is largely consistent with a previous result reported by O. Beldarrain et al, where 400-cycle Al2O3 layers were grown at 200 oC.To discuss the origin of the blister, we have conducted cross-sectional SEM observations of the formed blisters. Fig. 3 shows a typical SEM image of a blister after annealed at 500oC. The cross-sectional SEM observations clearly show that the blisters are completely hollow underneath. The surface of the flat Si substrate can also be seen under the blister, indicating that the upper Al2O3 film has completely delaminated from the interface between Si and Al2O3. From these results, we have speculated that blistering of Al2O3 after annealing is due to outgassing from the Al2O3/Si interface. Here, we came up with an idea that by identifying the gas inside the blisters, we can clarify the cause of this phenomenon.To identify the content inside blisters, Raman spectroscopy was carried out. The 532 nm wavelength laser used for Raman measurements can pass completely through Al2O3, making it possible to measure the gas inside without destroying the blisters. Fig. 4 shows a typical result of the Raman measurement of a blister of a sample after RTA (500oC). Here, as a result, characteristic spectra with peaks at positions ~4126, ~4144, ~4156, and ~4162 cm-1 were successfully observed. These four peaks can be assigned as peaks due to the difference in the rotational-quantum-number (J) of H2 as shown in Fig. 4, which is typical spectra of free H2 gas. These results provide evidence that blister formation is caused by the generation of hydrogen gas from the Al2O3/Si interface, a finding that is of great importance in clarifying the physics of blister formation mechanisms.In the presentation, we will also present details on the physics of hydrogen gas generation and methods to suppress blister formation.[References][1] R. Zhang, et al., IEEE Trans. Electron Devices 59, 335-341, (2012).[2] G Dingemans, et al., J. Vac. Sci. Technol. A, 30, 040802, (2012).[3] J. Thurn., et al., J. Mater. Sci. 39, 4799, (2004)[4] M. Broas, et al., Appl. Phys. Lett. 111, 141606, (2017)[5] O. Beldarrain, et al., J. Vac. Sci. Technol. A, 31, 01A128 (2013)[6] D. G. Xie, et al., Nat. Mater., 14, 899, (2015).[7] R. Matsumura, et al., accepted to ACS Applied Materials & Interfaces, https://doi.org/10.1021/acsami.1c20660 Figure 1