1. Field of the Invention
The present invention generally relates to the plasma processing of a substrate. More particularly, the present invention relates to an improved method for etching vias in the backside of a gallium arsenide substrate within a plasma processing chamber.
2. Description of the Background Art
GaAs devices are used extensively in the wireless telecommunications industry, where the high electron mobility of GaAs makes it well suited for high frequency, low noise, high gain applications. Although it has excellent electrical properties, GaAs is a relatively poor thermal conductor, making it difficult to remove heat efficiently from power devices. A commonly used solution to this issue is the formation of vias from the wafer backside to the frontside circuitry. Such vias provide a good thermal path for heat removal as well as a low impedance ground connection for RF (radio frequency) devices.
After completion of the frontside processing, the wafer is mounted face down on a carrier wafer and mechanically thinned to a thickness of approximately 100 microns. The back of the wafer is then patterned using photo-resist and the vias are plasma etched through the thinned substrate, stopping on the frontside metal. After resist removal the vias are metallized, typically by sputtering a gold seed layer followed by a gold plating step, to act as the heat sink/ground connection.
Ideally the backside etch process results in a via that has smooth sloped walls. However, in practice, pillar (also known as grass) formation often occurs in a number of vias.
Nam et al. addressed pillar formations in a recent journal article. This group concluded that pillar formation depends on the surface condition of the wafer after the grinding process. In addition, the group discovered that pillars form more easily when the physical conditions set forth below are satisfied.
Based upon their observations, the Nam et al. group recommended the following process conditions.
In addition, the Nam et al. group recommended an Ar sputter pre-etch to physically remove any residuals from the grind process. Nevertheless, experimentation by the present inventors has shown that pillar formations may occur even when there is compliance with the suggested process conditions. Additionally, the process conditions suggested by Nam et al. result in a relatively low etch rate for the GaAs (<3 micron/minute) which results in a process time of >30 minutes for a typical 100 micron deep via. This problem is expected to be more severe for 150 mm substrates, compared to the 3 inch substrates used by Nam et al.
In view of the above discussed problems with the prior art, what is needed is a high rate, GaAs via etch process that results in sloped via profiles and eliminates pillar formation. Furthermore, the operating parameters for the process should be specified such that the process can be implemented as a production worthy process.
A preferred embodiment of the present invention is directed toward an improved method for forming via holes in a substrate during an ion assisted etching process. The method commences with the forming of a mask layer on the substrate. The substrate is then placed in an inductively coupled plasma processing system that has an inductively coupled plasma confinement ring. High Cl2 flows are introduced into the processing system and a plasma is ignited using a low RF bias etch. The power level of the inductively coupled plasma processing system is increased to more fully dissociate the high Cl2 flows. The RF plasma is ramped in steps to etch at a high rate and to eliminate pillar formation during the etch process. The ramping of the RF plasma may be accomplished by placing an RC circuit in-line with an analog RF bias setpoint. Alternatively, the ramping of the RF plasma may be incorporated into the operational software used to control the etching process. When the etching process is complete, the substrate is removed from the plasma processing system and the mask layer is stripped from the substrate. The mask may be eroded to produce a via having a sloped profile.
The above described embodiment of the present invention provides a number of benefits over the prior art. For example, the provision of a confinement ring forces the reactive species generated over the surface of the substrate. This insures that enough of the reactive species is present at the surface of the substrate to give a high etch rate and to prevent the formation of pillars. In addition, the increased flow rate of the reactive gas and the increased power level of the inductively coupled plasma, which insures that the gas is more fully dissociated, provides for an excess of etch gas at the surface of the substrate. As a result, the present invention substantially decreases the formation of pillars during an etch process.
Another embodiment of the present invention is directed toward an improved method for forming via holes in a GaAs substrate using an inductively coupled plasma source. The method is centered around decreasing a residence time of an etch gas used during an etching process to prevent the formation of pillars and increasing the power of the inductively coupled plasma source to more fully dissociate the etch gas into reactive species. The etching process is initiated with a low RF bias voltage. The RF bias voltage is ramped between different RF bias voltage levels used during different steps in the etching process. The residence time of a gas is the average time a gas molecule remains in the reactor before being pumped away. Residence time can be calculated by the following formula:
t=pV/Q
Where t is the residence time in seconds, p is the pressure in torr, V is the chamber volume in liters, and Q is the gas flow rate measured in torr·liters/second. For a reactor with a fixed volume, the residence time can be decreased by decreasing the process pressure or increasing the gas flow rate. The residence time of the etch gas is decreased (preferably to less than 1 second) by introducing high Cl2 flows at lower pressures during the etching process to provide an excess of reactive Cl. An aperture is placed between the inductively coupled plasma source and the substrate surface to force reactive Cl generated in the inductively coupled plasma source over the surface of the substrate.
As discussed above, decreasing the residence time of the etch gas and increasing the power of the inductively coupled plasma source substantially limits pillar formation. In addition, ramping, rather than stepping, the RF bias voltage between different levels also eliminates pillar formation due to voltage spikes caused by rapid changes in the bias voltage. Therefore, the present invention represents a substantial improvement upon the prior art.
FIGS. 3(a) and 3(b) graphically illustrate pillar density vs. ICP powers for a given set of operating parameters;
A preferred embodiment of the present invention is directed toward a method of etching GaAs vias utilizing at least one of high Cl2 flows (short residence times), high inductively coupled plasma (ICP) powers, an ICP confinement ring, low RF bias etch initiation and RF ramping. As discussed in more detail below, these process modifications result in a high rate (>6.0 μm/min) via etch process that eliminates pillars. In addition, a sloped via profile is achievable through the erosion of a sloped photoresist mask.
Dry etching of GaAs vias requires etching relatively deep (˜100 um) features into a thinned GaAs substrate ending on a metal etch stop (typically gold). The dry etch process is typically chlorine based (i.e., BCl3/Cl2, Ar/Cl2, SiCl4/Cl2, etc.) due to the relatively high volatility of the etch products as shown in Table 1. While fluorine plasmas will not chemically etch GaAs, small amounts of fluorine containing precursors (CF4, etc.) may also be added to the process to assist in profile control.
Anisotropic GaAs etching in chlorine based chemistries follows an ion assisted chemical etch mechanism. In a purely chemical etch process, the reaction etch products have sufficiently volatility to spontaneously desorb from the surface. In an ion assisted process, desorption of the etch products is assisted by ion bombardment. Ion assisted chemical etch processes may involve the formation and removal of nonvolatile etch products which may roughen the surface morphology as the etch proceeds.
While a Cl2 plasma will readily etch GaAs, a Cl2 only plasma often has trouble etching any native oxides, resulting in a roughened surface. Oxygen scavengers (such as BCl3 or SiCl4) are often added to the process gas mixture to remove any native oxides.
A proposed mechanism responsible for etching GaAs in a Cl-based plasma preferably consists of a number of steps as set forth below.
Ionization
e+Cl2=Cl2++2e
Etchant Formation
e+Cl2=2Cl+e
Adsorption of etchant on substrate
Cl=GaAssurf−nCl
Formation of the etch product
GaAs−nCl=GaClx(ads)+AsCly(ads)
Product Desorption by evaporation
GaClx(ads)=GaClx(gas)
AsCly(ads)=AsCly(gas)
Product Desorption by Ion-Assisted Desorption
One possible mechanism for pillar formation during GaAs etching is through the creation of sub-chlorides (GaCl2, GaCl, AsCl2 or AsCl). Lower volatility sub-chlorides (i.e., GaCl2-b.p. 535° C.), once formed, may act as an etch mask initiating pillar formation. Sub-chloride formation is favored in a chlorine deficient process regime. Thus sub-chloride formation, and hence pillar formation, will be reduced in a Cl-excess environment. In order to supply an excess of reactive Cl at the wafer surface, a variety of process changes may be implemented as discussed in more detail below.
For a given reactor volume and process pressure, increasing the total process gas flow decreases the gas residence time. Decreasing the residence time reduces the efficient use of chlorine and results in an excess of reactive chlorine at the wafer surface.
Introducing higher process gas flows into the reactor decreases residence times, but may not result in the higher concentrations of reactive Cl species required for minimizing pillar formation during GaAs etching in all situations. Therefore, in conjunction with higher Cl2 flows 28, relatively higher ICP powers 26 are preferably utilized to effectively dissociate the relatively higher amounts of Cl2 into reactive Cl species. This is confirmed in experiments that show that even at higher Cl2 flows 28 (low residence times) pillar formation 24 can still occur without increased ICP power 26. For example, FIGS. 3(a) and 3(b) show that pillar 24 density increases at lower ICP powers 26, even at flows 28 in excess of 500 seem. Therefore, increased ICP power 26 is utilized in conjunction with the preferred embodiments of the present invention to fully dissociate the etch gas and prevent the formation of pillars.
Placing an aperture between the ICP and the wafer surface to force the reactive species generated in the ICP over the surface of the wafer also decreases pillar densities. This is shown in
In addition to the use of higher process gas flows and higher ICP powers as discussed above, the present inventors discovered that using a low RF bias voltage during etch initiation also reduces pillar densities during GaAs via etching. Prior to entering the GaAs via etch process, the substrates are mounted to a rigid carrier (typically sapphire) and thinned to ˜100 microns. During the course of the thinning operation, the surface of the GaAs is degraded (e.g., embedded abrasive materials, damage in the crystal structure, etc.). One approach to overcoming contamination on etched surfaces is to initiate the etch using a physical etch that will be able to remove both volatile and non volatile etch products. Nam et. al. recommend an Ar sputter etch for this purpose.
Through designed experiments, we have discovered that a more chemically driven etch initiation (high Cl2 flows, high ICP powers, low RF bias) results in lower pillar densities. This is shown in
A typical GaAs via etch process consists of multiple steps (i.e., Etch initiation, Main Etch, Over etch, etc.). These different steps often require different RF bias powers. During the transition between two process steps with different RF bias powers, there is often a momentary decrease in RF bias power, due to the time needed for the matching network to respond. This RF bias decrease is accompanied by a corresponding “spike” in RF bias reflected power. Experiments have shown that pillar densities can increase due to these RF bias decreases. RF bias changes between process steps are usually step functions (digital changes). Thus, these RF bias decreases (and the associated reflected power spikes) can be eliminated by ramping the RF bias between steps, allowing time for the matching network to follow the change. Alternatively the response time of the RF impedance matching can be improved through the use of a variable RF frequency (i.e. frequency matching) instead of a mechanical RF impedance matching network.
The above discussed reflected power spikes are illustrated in
The step change in the setpoint illustrated in
Using the above preferred method, a 150 mm GaAs wafer, patterned with an array of 40 micron diameter vias as part of a pattern comprising 15% of the wafer surface, was etched using the process conditions detailed below.
The wafer etched at an etch rate of 9 micron/minute with a selectivity ratio to photo-resist of >13:1 and had substantially no pillars over the entire wafer surface.
It will be understood that the specific embodiments of the invention shown and described herein are exemplary only. Numerous variations, changes, substitutions and equivalents will now occur to those skilled in the art without departing from the spirit and scope of the present invention. Accordingly, it is intended that all subject matter described herein and shown in the accompanying drawings be regarded as illustrative only and not in a limiting sense and that the scope of the invention be solely determined by the appended claims.
This application claims priority from and is related to commonly owned U.S. Provisional Patent Application Ser. No. 60/371,056, filed Apr. 9, 2002, entitled: Improved Method for Etching Vias, this Provisional Patent Application incorporated by reference herein.
| Number | Name | Date | Kind |
|---|---|---|---|
| 5169487 | Langley et al. | Dec 1992 | A |
| 5441596 | Nulty | Aug 1995 | A |
| 5456796 | Gupta et al. | Oct 1995 | A |
| 5624529 | Shul et al. | Apr 1997 | A |
| 5637189 | Peeters et al. | Jun 1997 | A |
| 5785877 | Sato et al. | Jul 1998 | A |
| 5849639 | Molloy et al. | Dec 1998 | A |
| 5902494 | Gupta et al. | May 1999 | A |
| 5948703 | Shen et al. | Sep 1999 | A |
| 5968845 | Chino et al. | Oct 1999 | A |
| 6008130 | Henderson et al. | Dec 1999 | A |
| 6057244 | Hausmann et al. | May 2000 | A |
| 6139679 | Satitpunwaycha | Oct 2000 | A |
| 6178919 | Li et al. | Jan 2001 | B1 |
| 6191021 | Fuller et al. | Feb 2001 | B1 |
| 6214720 | Sill et al. | Apr 2001 | B1 |
| 6254746 | Subramani et al. | Jul 2001 | B1 |
| 6255221 | Hudson et al. | Jul 2001 | B1 |
| 6258723 | Takeichi | Jul 2001 | B1 |
| 6287974 | Miller | Sep 2001 | B1 |
| 6303512 | Laermer et al. | Oct 2001 | B1 |
| 6326307 | Lindley et al. | Dec 2001 | B1 |
| 6391787 | Dhindsa | May 2002 | B1 |
| 6399515 | Tao et al. | Jun 2002 | B1 |
| 6415198 | Nallan et al. | Jul 2002 | B1 |
| 6433484 | Hao et al. | Aug 2002 | B1 |
| 6451157 | Hubacek | Sep 2002 | B1 |
| 6462419 | Asai | Oct 2002 | B1 |
| 6475889 | Ring | Nov 2002 | B1 |
| 6492277 | Ono et al. | Dec 2002 | B1 |
| 20010036744 | Taravade et al. | Nov 2001 | A1 |
| 20010044212 | Nguyen et al. | Nov 2001 | A1 |
| 20020055265 | Ring | May 2002 | A1 |
| 20020088542 | Nishikawa et al. | Jul 2002 | A1 |
| 20020098708 | Kumihashi et al. | Jul 2002 | A1 |
| 20020123229 | Ono et al. | Sep 2002 | A1 |
| 20020137351 | Chiou et al. | Sep 2002 | A1 |
| 20020179913 | Shirakawa | Dec 2002 | A1 |
| 20020180055 | Takahashi et al. | Dec 2002 | A1 |
| 20030003724 | Uchiyama et al. | Jan 2003 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20030216034 A1 | Nov 2003 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60371056 | Apr 2002 | US |