Specific Process Knowledge/Thin film deposition/ALD Picosun R200/TiO2 deposition using ALD

From LabAdviser

Feedback to this page: click here


This page is written by Evgeniy Shkondin @DTU Nanolab if nothing else is stated.
All images and photos on this page belongs to DTU Nanolab.
The fabrication and characterization described below were conducted in 2022 by Evgeniy Shkondin, DTU Nanolab.


Deposition of TiO2 is a standard, and well explored ALD process. Titanium(IV) isopropoxide (TTIP), Tetrakis(dimethylamido)titanium (TDMAT), and TiCl4 are the most popular. The last one is what is used at DTU Nanolab, because of the low cost and abundance. It can tolerate high temperatures (at least up to 500-550C) and can be employed for the deposition of TiO2 as well as TiN. The only disadvantage is the minor chlorine contamination, especially pronounced at the lower temperature range (below 200C). The TiO2 film is very popular in many applications due to the ease of deposition, excellent uniformity, strict thickness control, and reliability. At DTU Naolab you can either use ALD-1 or ALD-2 for the deposition of TiO2 with thermal (ALD-1 or ALD-2) or plasma (ALD-2) methods.

Unlike Al2O3, the TiO2 deposition process is quite complex. While increasing the temperature, the film will change the phase from amorphous to a polycrystalline (anatase) state. Typically users select an amorph layer if they need very smooth continuous films with high refractive index (many different photonic applications within waveguides and cavities). The refractive index for TiO2 is higher for polycrystalline anatase, but so also roughness. Anatase films are dominating in photovoltaics and solar cells relevant research. Another crystallographic phase known in TiO2 is rutile. This has an even higher refractive index but is unreachable with ALD as temperatures above 600 °C are usually required. It is possible to achieve a rutile form with plasma-enhanced ALD (PEAD) at a lower temperature but requires a biased stage which we do not have. If the rutile form is needed the users should consider reactive sputtering.

TiO2 ALD deposition using TMA and H2O

TiO2 can be deposited in the range between 100 °C to 350 °C. The recommendation is to use 150 °C for amorphous TiO2 and 300-350 °C for anatase.

TiO2 recipes

Recipe: TiO2

Maximum deposition thickness: 100 nm

Temperature: 100 oC - 350 oC

TiCl4 H2O
Nitrogen flow 150 sccm 200 sccm
Pulse time 0.1 s 0.1 s
Purge time 4.0 s 5.0 s

This is a standard recipe of TiO2. It is suitable for coating the flat samples and structures with an aspect ratio of around up to 1:10.

Recipe: TiO2T

Maximum deposition thickness: 100 nm

Temperature: 100 oC - 350 oC

TiCl4 TiCl4 H2O H2O
Nitrogen flow 150 sccm 150 sccm 200 sccm 200 sccm
Pulse time 0.1 s 0.1 s 0.1 s 0.1 s
Purge time 0.5 s 20.0 s 0.5 s 20.0 s


This is a more complicated recipe that allows coating the sample with very high aspect ratio profiles. The recipe is good for deposition on very high aspect ratio structures (with aspect ratio more than 1:20) such as deep trenches, pillars, pores etc. TiO2T recipe can also be used for deposition on polymers at low temperatures (80-120 oC).


Deposition characteristics

The deposition rate for TiO2 depends on the temperature, see the ALD-window graph below. The uniformity, thickness, refractive index has been obtained using Ellipsometer VASE. Besides the temperature dependencies, the process exhibits a certain growth delay especially pronounced in the transition zone where the growth characteristics change significantly. This occurs in the temperature range where the phase shift from amorphous to polycrystalline (anatase) happens. Very interesting behavior observes between 200 °C and 250 °C. Here is a very abrupt change in the deposition rate present. To understand better this phenomenon, films with more dense temperature steps need to be deposited and analyzed.

Deposition rate

Uniformity across 100 mm wafer

Recipe "TiO2" Samples: 100mm ssp-Si

Results have been obtained for <100> 100 mm Si wafers with native oxide, based on ellipsometry study. Note, that for TiO2 a growth delay will occur, which can be quite pronounced depending on the selected temperature and substrate.

Temperature (°C) / Number of cycles Deposition rate (nm/s) Average thickness (nm) Standard deviation Uniformity (%)
100°C / 1000cycles 0.0548 57.64 0.53 1.62
150°C / 1000cycles 0.0462 45.02 0.46 1.87
200°C / 1000cycles 0.0392 37.45 0.39 1.68
250°C / 1000cycles 0.0600 48.36 0.28 0.94
300°C / 1000cycles 0.0519 45.43 0.60 2.22
350°C / 1000cycles 0.0455 41.57 0.28 0.95




Optical functions

TiO2 has been deposited on natively oxidized Si wafers (1000 cycles of TiCl4 and H2O) in the temperature range between 100°C and 350°C. The ellipsometry analysis is based on the Tauc-Lorentz model.

Optical functions of TiO2 deposited on different substrates

The physical properties of TiO2 are sensitive to the chemical and crystal nature of the substrate material it is deposited on. After all, ALD is a chemical reaction with the surface. Here the optical properties are shown for TiO2 deposited on Si with and without native oxide, as well as on Si with the 35 nm Al2O3. The nucleation and deposition rate are quite different on all three especially if the selected temperature is chosen to be in the transition zone of 150-250°C, where the phase change from amorphous to anatase polycrystals appear.

Deposition on flat samples (SEM inspection)

Surface morphology was inspected using SEM images. TiO2 layers were deposited on different substrates: Si without native oxide (HF-treated), Si with native oxide, and Si with 35 nm Al2O3 (300°C, 350 cycles). From the images, it is clearly visible that TiO2 growth and nucleation depend on the nature of the substrate surface used in the process. The difference is mainly pronounced in the transition zone region (200°C +/- 50°C), where the phase change appears.


Deposition on high aspect ratio structures (SEM inspection)

Titanium oxide has been deposited on Si trench structures to see the uniformity of the layer. The standard recipe “TiO2” was implemented. The trenches were inspected using SEM in cross-sectional mode. All samples were cut across the groves and along them, and all sides (across and along) were inspected.

X-ray photoelectron spectroscopy

XPS measurements of all samples has been performed using the XPS NEXSA equipment. The purpose of the investigation is to get an idea about the film stoichiometry, composition and possible contamination. Since the XPS NEXSA also offers REELS and ISS analysis these measurements were also performed.

The XPS Nexsa is equipped with the MAGCIS or Mono Atomic and Gas Cluster Ion Source that allows to sputter off material from the sample either using single argon atoms with energies up to 4 keV or by using ionized clusters of argon atoms (between 75 and 2000 atoms) with energies up to 8 keV. For this analysis, the native carbon contamination has been removed by Ar clusters at 4keV and a size of 300 atoms. The moisture level has been sputtered off by implementing 25 levels at 5 s each.

After the surface has been cleaned, the film has been measured, so the survey, valence region, and high-resolution elements (Ti, O, and Cl) scan have been acquired.

Stoichiometry results


Stoichiometry results for TiO2 deposited with TiCl4 and H2O.
Temperature °C 100 150 200 250 300 350
Number of cycles 1000 1000 1000 1000 1000 1000
Ti (At. %) 32.36 32.42 32.50 32.65 32.60 32.85
O (At. %) 65.57 66.68 66.98 67.35 67.40 67.15
Cl (At. %) 2.05 0.9 0.52 0 0 0


  • Small amount of chlorine comes from unreacted TiCl4 processed at low temperatures.


REELS and ISS

Reflected Electron Energy Loss Spectroscopy (REELS) involves the exposure of the sample to a focused beam of monoenergetic electrons at 1keV coming from the flood gun. As these electrons interact with the material some of them will undergo inelastic scattering whereby they lose some kinetic energy. In reflection electron energy loss (REELS) the electrons that are both elastically and inelastically scattered are collected using the electron energy analyzer. The low-energy electron energy loss spectrum (up to ~50eV in energy loss) contains the elastic, or zero-loss, peak as well as the plasmon peaks. This spectrum can be used to determine information about the band structure and dielectric properties of the sample.

Calculated sampling depth for a 1 keV REELS experiment is ca. 15 nm, although it is noted that this will be dependent on the material being analyzed. Information from the REEL spectrum is complementary to other surface sensitive techniques, including XPS.

Ion scattering spectroscopy (ISS) is a technique in which a beam of He+ ions is scattered by a surface. The kinetic energy of scattered ions is measured; peaks are observed corresponding to elastic scattering of ions from atoms at the surface of the sample. Each element at the sample surface produces a peak at a different measured kinetic energy, caused by the momentum transfer between the incident ion and atom. The scattered ion and the scattering atom are normally of different masses, but the total momentum of the atom and ion is conserved. Therefore, as the initially "stationary" atom recoils, some kinetic energy is lost from the scattered ion, and the quantity of lost energy depends on the relative masses of the atom and ion. An ion scattering spectrum typically contains one peak for each element in the sample, with their separation in energy relating to the relative atomic masses of those elements. In some cases, different isotopes of the same element can be resolved, although this usually requires a primary ion heavier than He+. Strong scattering signals are limited to the topmost layer of atoms, so ISS is extremely surface-sensitive. Therefore, the sample must be clean, and even a small quantity of surface contaminant can significantly influence the ion scattering spectrum. Find more information here: https://www.thermofisher.com/dk/en/home/materials-science/learning-center/surface-analysis.html

ISS

REELS

X-ray Reflectivity and Diffraction

X-ray methods have also been involved in TiO2 characterization. The reflectivity method will allow measuring the film density and possible fluctuation of the density through the film thickness. The thickness and roughness can also be calculated. The grazing incidence diffraction can be applied to evaluate crystallinity in films prepared at different temperatures.

XRR analysis

X-ray reflectivity analysis (XRR) of samples (deposited at 150°C and 350°C) have been performed to investigate the thicknesses, roughness, and density profiles.

Three samples have been analyzed. 100 nm TiO2 deposited at 350°C, 100 nm TiO2 deposited at 150°C and 1000 cycles (45 nm) TiO2 deposited at 150°C on an amorph Al2O3 interlayer.

The scans have been obtained using Rigaku XRD SmartLab equipment. The voltage and current settings for the Cu X-ray tube were standard 40kV and 30mA. The incident optics contained an IPS (incident parallel slit) adaptor with 5 ° Soller slit. Other slits: IS=0.03mm RS1=0.03mm and RS2=0.075mm. Step size: 0.01 and measurement time - 5s for each point. The fitting process was performed using commercial GlobalFit software assuming the model based on Si or SiO2 substrates with native oxide/interlayer followed by the deposited TiO2 film and moisture surfaces. The results are summarized in the tables below.



XRR results for TiO2 layers
Anatase TiO2 deposited at 350°C 2174 cycles
Layer name Thickness (nm) Density (g/cm3) Rougness (nm) Depth distribution Bottom density (g/cm3) Delta Beta
Moisture 0.45 1.04 0.19 No distribution 3.3876e-6 7.8666e-8
TiO2

104.14

3.99

6.88 Linear 2.27 1.3010e-5 3.0212e-7
SiO2 native oxide 3.84 2.29 0.05 No distribution 7.4756e-6 1.7359e-7
Si wafer 2.33 0.08 No distribution 7.5860e-6 1.7616e-7


Amorphous TiO2 deposited at 150°C 2161 cycles
Layer name Thickness (nm) Density (g/cm3) Rougness (nm) Depth distribution Bottom density (g/cm3) Delta Beta
Moisture 0.66 0.12 0.18 No distribution 4.2943e-7 9.9721e-9
TiO2

91.22

3.81

0.36 Linear 3.00 1.2422e-5 2.8846e-7
SiO2 native oxide 0.66 2.11 0.46 No distribution 6.8850e-6 1.5988e-7
Si wafer 2.33 0.21 No distribution 7.5860e-6 1.7616e-7


Amorphous TiO2 deposited at 150°C 1000 cycles on top of Al2O3 (350°C 350 cycles)
Layer name Thickness (nm) Density (g/cm3) Rougness (nm) Depth distribution Bottom density (g/cm3) Delta Beta
Moisture 0.61 0.91 0.18 No distribution 2.9195e-6 6.7794e-8
TiO2

44.75

3.76

0.39 Linear 3.38 1.2250e-5 2.8447e-7
Al2O3

34.69

2.82

0.28 No distribution 9.1898e-6 2.1340e-7
SiO2 native oxide 1.43 2.30 0.37 No distribution 7.4884e-6 1.7389e-7
Si wafer 2.33 0.04 No distribution 7.5860e-6 1.7616e-7


Fitting parameters
Sample Fitting parameters
Intensity Background Fitting area () R
Anatase TiO2 deposited at 350°C 2174 cycles 1.03302 1.25415E-5 0.4304 - 2.0000 0.00954 0.00116
Amorphous TiO2 deposited at 150°C 2161 cycles 1.15715 4.34885E-8 0.4304 - 6.0000 0.00793 0.00127
Amorphous TiO2 deposited at 150°C 1000 cycles on top of Al2O3 (350°C 350 cycles) 1.42211 5.50839E-8 0.4672 - 6.0000 0.01334 0.00392





Grazing incidence X-ray diffraction (GiXRD)

The grazing incidence X-ray Diffraction (GiXRD) technique was applied to measure 100 nm TiO2 film deposited at 150°C and 350°C. The scans have been obtained using Rigaku XRD SmartLab equipment. The voltage and current settings for the Cu X-ray tube were standard 40kV and 30mA. The incident optics contained an incident parallel slit adaptor with incident 5 ° Soller slit. Receiving optics also included 5 ° Soller slit and parallel slit analyzer of 0.5 °. Other slits: IS=0.5mm RS1=open and RS2=open. Step size: 0.05 and measurement speed - 1° /min for each point. The measurement range was from 20 ° to 110 ° of 2θ. The incident angle was fixed at 2°.