Specific Process Knowledge/Thin film deposition/Deposition of Silicon Oxide/Reactively sputtered SiO2 in Sputter-System Metal Oxide (PC1)

From LabAdviser

Feedback to this page: click here


Reactive sputtering of SiO2

This page presents the results of SiO2 deposition using RF reactive sputtering in Sputter-System Metal Oxide(PC1) commonly known as "Cluster Lesker". The deposition target is Si, and the reactive gas is O2. Source #1 (RF) was used.

The fabrication and characterization described below were conducted in 2021 by Evgeniy Shkondin, DTU Nanolab. The attempt to investigate SiO2 wet etch has been done in a collaboration with Narwan Kabir Noori (DTU Photonics Engineering). The prepared samples were investigated by X-ray Photoelectron Spectroscopy (XPS), Spectroscopic Ellipsometry, and cross-sectional SEM inspection. The main focus of the study was the ability of SiO2 RF reactive sputtering to passivate sidewalls of the narrow trenches and the analysis of the refractive index. For this purpose, RF bias on a substrate and substrate temperature values were the main parameters of interest.

For a side-wall passivation study, multiple templates have been fabricated using DUV lithography and DRIE etching in Pegasus-1. The resulted Si-trenches have a height of 500 nm and a pitch of 400 nm with the thickness of each wall (groove) - 200nm


The process recipe in a Sputter-System Metal Oxide(PC1) is following:


  • Deposition mode: Downstream
  • Power: 140 W (Uniformity measurements were performed with 140W, 120W, and 100W).
  • Pressure: 3 mTorr
  • Ar flow: 50 sccm
  • O2 flow: 15 sccm
  • Deposition time: 5000s (Uniformity measurements were performed with 5000s and 3600s).
  • Rotation speed: 10rpm
  • RF bias: between 0 W and 50 W
  • Deposition temperature: between 20°C and 400°C


Two different dark space configuration were used in this investigation:

  • Opened dark space shield was installed during side-wall passivation inspection, where RF bias and temperatures were varied. Deposition rates for this part of the experiments were also measured, while opened dark space shield was installed.
  • Closed dark space shield was installed after the user found the Cu contamination in his film. Uniformity inspections and deposition power influence on deposition rate were performed with a closed dark space shield.

Side-wall passivation

Influence of the RF bias

It was soon realized that the power values below 20W were not optimal due to process instability. Only at 20W and above the reflected power was almost at 0W. For this reason, the RF bias power of 20W, 30W, 40W, and 50W was tested and compared. The results are present in the figure below:

  • Figure 1. Different RF bias on a substarte.
  • Deposition without RF bias.
    Deposition without RF bias.
  • Deposition with 20 W RF bias.
    Deposition with 20 W RF bias.
  • Deposition with 30 W RF bias. Visible erosion of the Si trench.
    Deposition with 30 W RF bias. Visible erosion of the Si trench.
  • Deposition with 40 W RF bias. Promoted damage of the top part.
    Deposition with 40 W RF bias. Promoted damage of the top part.
  • Deposition with 50 W RF bias. Milling turns the trench top into triangular shape.
    Deposition with 50 W RF bias. Milling turns the trench top into triangular shape.

Temperature irregularities

One of the observed issues regarding the deposition with RF bias was uncontrolled fluctuations of the temperature measured on a substrate. Even though the heater was off during the whole process the temperature was moving up and down without any control as it is present in the next figure:


  • Figure 2. Temparature changes without control during the RF substarte bias process.
  • Temperature fluctuations.
    Temperature fluctuations.


This can be problematic for many processes, and great care needs to be taken before considering using it. Temperature-sensitive substrates can be damaged and/or film characteristics can be influenced by elevated temperatures.

Influence of substrate temperature

The next part of the experiment was the investigation of possible temperature influence on high aspect ration passivation. Three temperatures were selected: room temperature, 200°C, and 400°C.

No significant influence has been observed. However, the temperature does not remain constant, it oscillated back and forth and is controlled by PID parameters.

  • Figure 3. Different substarte temperatures. No RF bias.
  • Deposition at room temperature. RF bias is not applied.
    Deposition at room temperature. RF bias is not applied.
  • Deposition at 200°C. RF bias is not applied.
    Deposition at 200°C. RF bias is not applied.
  • Deposition at 400°C. RF bias is not applied.
    Deposition at 400°C. RF bias is not applied.
  • Measured temperature fluctuations during the deposition.
    Measured temperature fluctuations during the deposition.

Deposition rate as a function of RF bias and temperature

It shows up that both RF bias value and temperature setpoint affect the deposition rate.


  • Figure 4. Deposition rate as a function of different RF bias and the temperature. Opened dark space shield is used.
  • Deposition rate of SiO2 as a function of the tempearture and RF-bias. Opened dark space shield is used.
    Deposition rate of SiO2 as a function of the tempearture and RF-bias. Opened dark space shield is used.

Dark space shield configuration

The process can run with two different anode enclosure configurations around the cathode. These enclosures are called dark space shields. The "opened" dark space shield is suitable for low-pressure processes where the highest possible area of the target should be exposed. In general, it also has a positive influence on deposition rate, since a higher number of sputtered adatoms leaves the target. The disadvantage of opened dark space shield is the possible Cu contamination from the cathode materials. This contamination becomes pronounced with the aging of the target.

The installation of a "closed" dark space shield will completely prevent cross-contamination but will lower the deposition rate. Additionally, the lower pressure (below 3 mTorr) processes can be problematic to run due to the plasma instability.

  • Figure 5. Two different dark space shield configurations.
  • Opened dark space shield is installed.
    Opened dark space shield is installed.
  • Closed dark space shield is installed.
    Closed dark space shield is installed.

Uniformity

The uniformity measurements were performed with 6-inch wafers and installed closed darks space shield, with three power settings (100W, 120W, and 140W). The pressure and gas flow remain the same. The measurements was conducted using VASE Ellipsometer.

  • Figure 6. Thickness uniformity across 150mm Si wafer.
  • 100W Power. Closed dark space shield is installed. Standard deviation std=0.307. Average thickness 65.26 nm.
    100W Power. Closed dark space shield is installed. Standard deviation std=0.307. Average thickness 65.26 nm.
  • 120W Power. Closed dark space shield is installed. Standard deviation std=0.314. Average thickness 59.95 nm.
    120W Power. Closed dark space shield is installed. Standard deviation std=0.314. Average thickness 59.95 nm.
  • 140W Power. Closed dark space shield is installed. Standard deviation std=0.372. Average thickness 74.01 nm.
    140W Power. Closed dark space shield is installed. Standard deviation std=0.372. Average thickness 74.01 nm.

Deposition rate as a function of power

The uniformity measurements were performed with 6-inch wafers and installed closed darks space shield, with three power settings (100W, 120W, and 140W). The pressure and gas flow remain the same. The measurements was conducted using VASE Ellipsometer.


  • Figure 7. Deposition rate as a function of power.
  • Deposition rate measured with closed dark space shield installed.
    Deposition rate measured with closed dark space shield installed.

Characterization

Optical properties

The SiO2 films have been measured with a VASE Ellipsometer and optical constants were extracted by the Cauchy-Urbach model. No confidant correlation between different RF biases, temperatures, or powers was found.


  • Figure 8. Optical properties.
  • Refractive index of 120 nm SiO2 prepared with 14 0W, 3 mTorr, 50 sccm Ar, 15 sccm O2 and no substrate RF bias. Deposition time 5000s.
    Refractive index of 120 nm SiO2 prepared with 14 0W, 3 mTorr, 50 sccm Ar, 15 sccm O2 and no substrate RF bias. Deposition time 5000s.
  • Absorption coefficient of 120 nm SiO2 prepared with 140 W, 3 mTorr, 50 sccm Ar, 15 sccm O2 and no substrate RF bias. Deposition time 5000s.
    Absorption coefficient of 120 nm SiO2 prepared with 140 W, 3 mTorr, 50 sccm Ar, 15 sccm O2 and no substrate RF bias. Deposition time 5000s.

X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) has been performed using XPS K-Alpha. The scans look similar in all prepared films without any noticeable differences. The stoichiometry ratio between Si and O was around 1.74 in all room setpoint temperature prepared films. The ratio for 200C prepared film was 2.48 and for 400C was 1.67. No clear explanation for these differences.

The presented spectra have been taken from the film that was clean-sputtered with Ar ions at 1000 eV, high-current, and 10s time exposure.

  • Figure 9. X-ray photoelectron spectroscopy.
  • SiO2 survey scan.
    SiO2 survey scan.
  • Si 2p High-resolution scan.
    Si 2p High-resolution scan.
  • O 1s High-resolution scan.
    O 1s High-resolution scan.

Cu contamination in Si films

As it was mentioned earlier, the use of the Si target with the oped dark space shield leads to Cu contamination, once the target gets old and heavily used. This is most likely due to the shrinkage of the target area. The XPS Cu 2p High-resolution scans reveal the difference in Cu content in pure Si nonreactive sputtered film when an open dark space shield or closed dark space shield is installed.

  • Figure 10. X-ray photoelectron spectroscopy.
  • Cu 2p High-resolution scan in sputtered Si thin films using the open dark space shield and the seasoned (old and heavily used) Si target.
    Cu 2p High-resolution scan in sputtered Si thin films using the open dark space shield and the seasoned (old and heavily used) Si target.
  • Cu 2p High-resolution scan in sputtered Si thin films using the closed dark space shield and the seasoned (old and heavily used) Si target.
    Cu 2p High-resolution scan in sputtered Si thin films using the closed dark space shield and the seasoned (old and heavily used) Si target.

Wet etch attempt

It is important to find the procedure to etch the deposited materials. In the case of reactively sputtered SiO2, it showed up to be challenging to apply wet chemistry methods. Normally, one would prepare BHF or HF solutions to do that, and indeed they are quite good if the final result is the complete dissolution of the deposited film, but if the user would like to etch to a certain thickness it is not possible. To investigate the etch rate (or in general the SiO2 film behavior in HF-based solutions) the mask was designed in CleWin5 software. The AZ 5214E was spin-coated to the thickness of 2.2 µm in Spin Coater: Gamma UV and exposed in MLA-2 system, followed by 60s. development in Developer TMAH UV-lithography.

  • Figure 11. The designed mask for wet etch.
  • Mask designed in CleWin5 for wet etch experiment.
    Mask designed in CleWin5 for wet etch experiment.

The prepared wafer with the photoresist mask has been cleaved into many small pieces and tested in different HF solutions with different concentrations. Afterward, the photoresist was stripped in Plasma Asher-1.

It was realized that the longer etch time removes the film completely, but the short (a few seconds exposure) change the film properties.

The ellipsometric measurement did not reveal anything meaningful, but it was realized that the wet etched area heavily scatters light (the beam spot was clearly visible). This indicates a pronounced roughness, so it was decided to imply SEM analysis.

SEM reveals that even very short etch time (1s.) is enough to damage SiO2 reactively sputtered films. The result is presented in the figure. HF attacks the film as ” a whole” leaving the random structure.

  • Figure 12. SEM inspection.
  • SEM images that shows SiO2 films after HF exposure.
    SEM images that shows SiO2 films after HF exposure.

If the user wishes to etch SiO2 films gently and controllable, it is recommended to select dry etch methods.

Retrieved from "https://labadviser.nanolab.dtu.dk//index.php?title=Specific_Process_Knowledge/Thin_film_deposition/Deposition_of_Silicon_Oxide/Reactively_sputtered_SiO2_in_Sputter-System_Metal_Oxide_(PC1)&oldid=50417"