This article reports the results and trends of an optimized process for PEALD deposited TiN. The process was conducted at 200°C, and the metal-organic precursor TDMAT was employed as the titanium source.
By carefully controlling the plasma conditions, a resistivity of 180 µΩcm for 74 nm of film can be obtained. The resistivity shoots to 250 µΩcm for 26 nm due to thin film scattering and other effects.
This is a significant improvement to the resistivity of 361 µΩcm for TiCl4 at 250°C, reported in the past. This makes the TDMAT process suitable when there is a need to avoid chlorine-containing by-products or where there is a need to reduce the thermal budget.
A mixed N2/H2 plasma in the Oxford Instruments FlexAL® remote plasma ALD reactor (Figure 1) and sequential dosing of tetrakis(dimethylamino)titanium (TDMAT) were used to prepare the TiN layers. The TDMAT precursor was stored in a stainless steel blubber at 60°C before it was transported to the reactor through bubbled doses containing 200 sccm of Ar.
Previous observations show that the low pressure plasmas create higher ion energies and enables higher film quality in both oxides and nitrides. The plasma pressure was maintained at a constant 3 mTorr, while other plasma conditions were varied to observe the trends.
Figure 1. Photograph of the FlexAL® reactor
Deposition was made on p-type silicons that have resistivity between the range of 1 and 10 Ωcm. Deposition onto PECVD grown SiO2 on silicon was performed with the previous experiments. The resistivity measurements produced by this deposition were no more accurate.
All the depositions were performed directly onto oxide-coated Si in order to make the ellipsometer modeling easy. A Woollam M2000-V was used to make ex-situ ellipsometer measurements, and a GenOsc model containing three Lorentz oscillators were used to measure the thickness. A Signatone four-point-probe attached to a Keithley 2410 source meter was used to measure the resistivity.
Results and Discussions
As is common for ICP sources, peak ion energy and electron temperature are inversely related to plasma power (Figure 2). This procedure confirmed the suggestion that obtaining low resistivity in TiN is a highly ion driven process (Figure 3), as the lowest resistivity is observed at 100 W. At 50 W the source started functioning in a dominating manner leading to under-saturation.
Figure 2. Mean ion energy (top) and electron temperature (bottom) for FlexAL (<25 mTorr)
Figure 3. Resistivity and thickness versus plasma power
Plasma Exposure Time
As is the case with most conductive nitrides, long exposure is essential to reduce the resistivity. This case is no exception, where the shoulder is seen at ~30 seconds (Figure 4). A ±10% resistivity uniformity can be achieved.
This shows that compared to the previous assumptions regarding the remote ICP source, there is a good uniformity in the ionic species related to relativity. In this case, the distribution of plasma species is probably caused by the low pressure plasma.
Figure 4. Resistivity and Thickness versus Plasma Time
Unlike the other films observed, there is no defined plateau in thickness versus resistivity in places where there are minimal surface scattering effects. For the 200°C and the 350°C depositions, a deposition can be seen at 20 nm in the graph shown in Figure 5.
The fall in resistivity at thicknesses over 20 nm could be due to a slow crystallization effect. There has been a strong slow crystallization effect in hotter TiCl4 processes.
Figure 5. Resistivity versus Thickness for TiN deposited from TDMAT at 350°C (orange) and 200°C (blue). Note: 350°C is acknowledged to be above the decomposition temperature for TDMAT and a large increase in growth rate is observed however the films still appear uniform and comparable to 200°C deposition
Plasma Gas Composition
Resistivity is strongly influenced by gas composition (Figure 6). Certain factors, such as the decreasing nature of H radicals promoting the removal of carbon or the increased presence of H2, resulting in increased the ion energy for a particular power and pressure (Figure 7), are at work.
Figure 6. Resistivity and Thickness versus %N2 in mixed H2/N2 plasma
Figure 7. Ion energy distribution measured by RFEA on homebuilt TU/e system. Comparing to FlexAL, the O2 the peak ion energy is ~4 eV lower but the trend is assumed the same
Surface and Composition
Figure 8 shows AFM analysis with a 65 nm TiN film deposited at 200°C from TDMAT with a 0.71 nm Ra. A significantly lower Ra of 3.3 nm was obtained from a film of comparable thickness deposited at 550°C from TiCl4. This reaffirms the belief that resistivity is connected to crystal grain size.
Figure 8. AFM image of TiN deposited from TDMAT at 200°C (top) and from TiCl4 at 550°C (bottom)
Figure 9 shows the sputtered XPS with the a stoichiometric TiN film with 2% carbon content, close to the detection limit, and 4% oxygen.
Figure 9. Sputtered XPS trace of TiN deposited from TDMAT at 200°C
Table 1. shows the overall parameters and corresponding results of the optimized process providing the lowest resistivity.
Table 1. Results from the optimized process
|Thickness Uniformity (8”)
Compared to process conditions that use TiCl4 as the titanium source, the PEALD of TiN employing TDMAT, the metal organic precursor has shown to provide lower resistivity. In conditions where there is a need to control thermal budget, lower process temperature and lower surface roughness can make optimized process a suitable option.
This information has been sourced, reviewed and adapted from materials provided by Oxford Instruments Plasma Technology.
For more information on this source, please visit Oxford Instruments Plasma Technology.