Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Table of Content Zone
locationtop
styledisc

Table of Contents
outlinetrue
indent25px
stylenone

Terminology

Expand

Wet Etching - Substrates are immersed in a reactive solution (etchant). The layer to be etched is removed by chemical reaction or by dissolution. The reaction products must be soluble and are carried away by the etchant solution.  Generally, wet etching is isotropic.

Dry Etching - Substrates are immersed in a reactive gas (plasma). The layer to be etched is removed by chemical reactions and/or physical means (ion bombardment). The reaction products must be volatile and are carried away in the gas stream.

      • Plasma Etching - Typically high pressure, no ion bombardment.  (Substrate is placed on a grounded electrode)
      • Reactive Ion Etching - Typically lower pressures, ion bombardment. (Substrate is placed on a powered electrode)

Anisotropic Etch - Etch rate is not equal in all directions.

Isotropic Etch - Etch rate is equal in all directions.

Etching - The process by which material is removed from a surface.

Mask - Used to protect regions of the wafer surface. Examples include photoresist, Ni, Cr, or oxide layer.

Selectivity - The ratio of etch rate of film to etch rate of substrate or mask.

Aspect Ratio - Ratio of depth to width of an etched feature.

Plasma - Partially ionized gas containing an equal number of positive and negative charges, as well as some other number of none ionized gas particles. 

Glow Discharge - Globally neutral, but contains regions of net positive and negative charge. (Many thin film processes utilize glow discharges, but “plasma” and “glow discharge” are often used interchangeably)

...

Expand


Effects of Cl2 and Fluorine Chemistry on Chamber Walls and Etch Parameters

New chamber walls conditioning and cleaning strategies to improve the stability of plasma processes

Changes in the chamber wall conditions (e.g. chemical composition) are identified as being one important cause of process drifts such as changes in etch rates, etching profiles, etching selectivity or etching uniformity across the wafer. As the reactor wall conditions are modified due to the deposition or removal of etch products, the plasma chemistry changes, which in turn modifies the process performances. Controlling the reactor wall conditions and their stability is of primary concern when trying to create repeatable processes.

In the plot to the right, the first column (a) represents a new, unaltered, Al2O3 chamber wall. In the second column (b), we see the effect of Chlorine etching. There is a clear build up of Silicon Oxychloride layers, as well as, Cl+Br bonded to the wall. The third column (c) represents an 80s fluorine etch following the chlorine etch. We see that the Silicon Oxychloride film and Cl+Br bonds have been removed and we are left with a thin fluorinated aluminum layer. Column (d) represents subsequent fluorine etches totaling 20 minutes of plasma exposure. We see that the fluorinated aluminum layer is building but surface chemistry has not changed.

This will lead to several issues including flake off of AlxFy particles on the wafer and process drifts (due both to the progressive growth of AlF material on the SiO2 windows and to the release of F atoms from the chamber walls during the etching process). 

Plasma chemistry can be used to control the conditioning of the chamber. Transitions from chlorine to fluorine and visa versa require different approaches, see below for details.

Transition from Fluorine to Chlorine etch chemistry

Observed Effects

Klimecky_JVSTA_wfigs.pdf

A thin film of silicon oxychloride grows on the chamber walls during Chlorine plasma etches. Fluorine etch chemistry will quickly strip away the thin oxychloride film from chamber walls resulting in significant transient effects in plasma density and real-time etch rate when transitioning back to chlorine etch chemistry. The root cause of these transients is assumed to be a result of dynamic oxychloride layer build-up processes at the chamber walls.

As shown in the plot to the right, Chlorine etch rate will be lower when transitioning from Fluorine etch back to Chlorine etching.

Chlorine etch rate will also be lower when following a chamber mechanical clean as this can also strip the oxychloride film from the chamber walls.


Black trace represents a well seasoned chamber.

Green trace represents first Chlorine etch after Fluorine etch.

Proposed Mechanism for Effects

The transient effects of oxychloride film buildup persist until it has achieved a steady state creation/loss ratio. At this point the Chlorine etch conditions become constant. In the paper referenced, it is suggested that transient effects are eliminated after approximately 300 seconds of Chlorine etching.

The chamber seasoning alters recombination rates at the walls, which affects both the neutral species density and the plasma density as shown in the plot to the right. The plot shows that SiCl4 is low and increasing at the start of first Chlorine etch after a Fluorine etching. SiCl4 is the primary etch product of Si etched in Cl2 chemistry and is a good indicator of real time etch rate.

 

Black trace represents a well seasoned chamber.

Green trace represents first Chlorine etch after Fluorine etch.

Recommended Chlorine Etch Chamber Preparation

In our shared facility, it is often difficult to know exactly what has happened in the chamber before you arrived to use the tool. In addition, most users are more concerned with final depth of their etch rather than efficiency of time usage. For these reasons, it is best to return the chamber to a known condition before you begin your processing. Since the transient effects discussed above are consistent and predictable, we can generate a recipe that will bring the chamber to a stable condition before Chlorine etching.

Before beginning your Chlorine etch, it is best to strip the oxychloride film from the chamber walls and bring the chamber walls back to a known starting point. Do this by running the clean and coat recipe available at the tool.

Once returned to the known starting point, you may proceed with your Chlorine etch as usual.

Example Recipe for Panasonic E620 (other tool recipes may vary from this but purpose is the same)

Recipe Step

TCP Coil Power (Watt)

Platen Power (Watt)

Gas 1 (sccm)

Gas 2 (sccm)

Pressure (mTorr)

Time (seconds)

1. Strip and Clean

8000CF4 > 50O2 > 4022.5900
2. Chamber Purge00N2 > 100NA22.530
3. Grow Oxychloride Film on chamber walls4000Cl2 > 50NA22.5300


Chlorine etch depth achieved for first three etches after Fluorine etching. Samples etched for 30 seconds each, significant etch rate increase for each sample. This is known as the "First Wafer Effect" in industry.

The etch rate is increasing for each run as the wall seasons with
oxychloride buildup and the recombination rate goes down.

Transition from Chlorine to Fluorine etch chemistry

Observed Effects

During Chlorine plasma etches a thin film of silicon oxychloride grows on the chamber walls. Fluorine etch chemistry will strip away the thin oxychloride film from chamber walls resulting in significant transient effects in plasma density and real-time etch rate when transitioning back to fluorine etch chemistry. The root cause of these transients is assumed to be a result of dynamic oxychloride layer stripping processes at the chamber walls and subsequent build up of Al-F layers.

The oxychloride stripping effects of fluorine etch plasma are short lived and typically last less than a minute. The subsequent build up of Al-F layers can take a few minutes to stabilize.

Black trace represents a Cl2 seasoned chamber.

Green trace represents first Chlorine etch after Fluorine etching, indicating plasma losses due to replacing oxychloride film that was stripped in a 30 second fluorine etch.

Proposed Mechanism for Effects

The transient effects of oxychloride film stripping persist until the film has been removed. At this point the fluorine etch conditions become constant. In the paper referenced, it is suggested that transient effects are eliminated after approximately 30 seconds of fluorine etching.

The chamber seasoning alters recombination rates at the walls, which affects both the neutral species density and the plasma density as shown in the plot to the right. The plot shows that SiCl4 is low and increasing at the start of first Chlorine etch after a Fluorine etching. SiCl4 is the primary etch product of Si etched in Cl2 chemistry and indicates low etch rate due to plasma losses.

 

Black trace represents a Cl2 seasoned chamber.

Green trace represents first Chlorine etch after Fluorine etching, indicating low etch product due to replacing oxychloride film that was stripped in a 30 second fluorine etch.


Recommended Fluorine Etch Chamber Preparation

In our shared facility, it is often difficult to know exactly what has happened in the chamber before you arrived to use the tool. In addition, most users are more concerned with final depth of their etch rather than efficiency of time usage. For these reasons, it is best to return the chamber to a known condition before you begin your processing. Since the transient effects discussed above are consistent and predictable, we can generate a recipe that will bring the chamber to a stable condition before fluorine etching.

Before beginning your fluorine etch, it is best to strip the oxychloride film from the chamber walls and bring the chamber walls back to a known starting point. Do this by running the chamber clean recipe available at the tool.

Once returned to the known starting point, you may proceed with your fluorine etch as usual.

Example Recipe for Panasonic E620 (other tool recipes may vary from this but purpose is the same)

Recipe Step

TCP Coil Power (Watt)

Platen Power (Watt)

Gas 1 (sccm)

Gas 2 (sccm)

Pressure (mTorr)

Time (seconds)

1. Initial Step

7000CF4 > 50O2 > 4037.55
2. Strip and Clean7000CF4 > 50O2 > 4037.530
3. N2 Purge1000N2 > 100NA18.810


Chlorine etch depth achieved for first three etches after Fluorine etching. The etch rate is increasing for each run as the wall seasons with oxychloride buildup and the recombination rate goes down, indicating previous fluorine etch stripped oxychloride film.


...