Automatic method to eliminate first-wafer effect

Details Automatic method to eliminate first-wafer effect Automatic method to eliminate first-wafer effect

1999-06-30

2001-09-18

Bowers, Charles (Department: 2813)

Semiconductor device manufacturing: process

Including control responsive to sensed condition

C029S025010, C414S217000

Reexamination Certificate

active

06291252

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to the manufacture of high performance semiconductor integrated devices. More specifically, this invention relates to the control of tools utilized in the manufacture of high performance semiconductor integrated devices. Even more specifically, this invention relates to the avoidance of the first-wafer effect during the manufacture of high performance semiconductor integrated devices.
2. Discussion of the Related Art
The manufacture of high performance semiconductor integrated devices is done on and in semiconductor wafers that are processed through a multitude of processes. Some of these processes include ion implantation, thermal annealing, the formation of photoresist layers, developing the photoresist layer, etching layers, planarizing the surface of the semi-processed semiconductor wafer and depositing layers of different types of materials. As is known in the semiconductor manufacturing art, the above-delineated processes are just a few of the many processes necessary to completely process a semiconductor wafer.
In some of the above processes wafers are placed in a wafer cassette so that they can be handled and moved expeditiously. Typically, a wafer cassette holds about 25 wafers.
During the manufacturing process of the semiconductor wafers there are occasions when one or more of the tools in the process either breaks down or needs maintenance. This causes the manufacturing processes up to the tool that requires maintenance to be stopped. As can be appreciated, the processes in tools that are still operational are completed and the tools are put on idle. The idle state for the different tools affects the various tools in different ways.
For example, when an etch tool sits idle for an extended period of time, the first few product wafers that are processed through the tool end up with less than optimal processing often necessitating scrapping at least the first product wafer, if not the first few product wafers. These wafers need to be scrapped because they may have the critical dimensions sufficiently outside control limits, bridged features, and/or other yield killing problems. In the semiconductor manufacturing industry, this phenomenon is known as the first-wafer effect.
The first-wafer effect occurs when a chamber in a processing tool is left under idle conditions with varying amounts of materials such as residual gas mixtures and/or polymer forming materials. During the idle state, the chamber remains under vacuum and continuously purges polymer building residual gas mixtures, which were left over from previous wafer processing. This causes the chamber that was left idling to be in a different state than a chamber that has been undergoing continuous processing. The problem with this is that processing parameters are characterized with etch tools that are running in a continuous mode. So at least the first wafer processed in a tool such as a plasma etcher that has been idle will undergo processing that is not tuned for the actual conditions of the etch chamber.
One of the techniques to avoid the first-wafer effect is to manually insert one or more conditioning wafers prior to the product wafers to stimulate and exercise the tool before the product wafers actually get etched. Although this measure is effective in some cases, it is not efficient or reliable to have it done manually. Some of the reasons that it is not effective are:
1. The tool operators often forget to insert the conditioning wafer.
2. Extra manual conditioning wafer insertions in front of the product wafers may cause random killer defects such as scratches and particle contamination to the product wafers and to the tool.
3. The tool may be paused for a long period of time when unattended due to soft error such as temperature and/or pressure stability and will repeat the phenomenon when resumed.
4. The tool operators often forget to reload the proper recipe for the product wafers after the conditioning wafer(s) have been completed.
5. It is difficult for the tool operator to know for sure that a conditioning wafer is needed.
A common and serious problem of the first-wafer effect occurs when the tool is paused after partially processing the lot. It requires the tool operator to make the decision whether or not to repeat the conditioning cycle. To repeat the cycle, the processed wafers must be retrieved manually from the processing chamber and returned to the cassette, mixing them with unprocessed wafers. Since processed wafers should be kept in a constant vacuum, the prolonged exposure to the atmosphere will allow condensation and corrosion to occur that will degrade the reliability of the devices. In addition, there is a high likelihood that misprocessing will occur when the lot resumes processing.
To reduce cost, the conditioning wafers are typically made from previously used, non-patterned wafers, known as dummy wafers. The dummy wafers have a single film layer over the substrate, bare silicon and photoresist coated to simulate as closely as possible the multiple layers of the product wafers.
For product wafers, etch endpoint detectors are assigned to detect when a change of a layer or material occurs before the next etch step is introduced. Because the conditioning wafer has a different film, for example bare silicon rather than an aluminum metal layer, the tool will alarm and the conditioning cycle will pause since the appropriate endpoint signal will not have been detected. Because of the built-ill endpoint alarms, the tool operator has to wait in front of the tool to intervene after each processing step that includes an endpoint. This reduces the throughput and increases the cost of manufacturing. For this reason, it is difficult to process both conditioning wafers and product wafers with a single recipe. To resolve this problem, a modified process recipe without the use of endpoint detectors has been employed to prevent an alarm situation. However, this practice can lead to the problems listed above and result in wafers that have to be scrapped.
Another remedy is to have conditioning wafers run automatically at the beginning of each lot of product wafers. However, this is very time consuming, especially when lots are continuously loaded one after another. This is even more wasteful when each lot contains only a few wafers instead of a complete batch of 25 wafers. As discussed above, a single conditioning wafer and sometimes multiple conditioning wafers are required before each product lot when the tool has been idle. If the tool had been idle for an extended period of time, insufficient conditioning can result in scrapping wafers while over conditioning increases cost and cycle time.
Therefore, what is needed is a method to prevent the first-wafer effect that does not run conditioning wafers unnecessarily and does not over condition the processing tool.
SUMMARY OF THE INVENTION
According to the present invention, the foregoing and other objects and advantages are obtained by a method of manufacturing semiconductor wafers in which product wafers and conditioning wafers are loaded into a processing tool and in which it is determined if the processing tool has been on idle for a predetermined period of time.
In one aspect of the invention, a product wafer is automatically processed if the predetermined period of time has not been exceeded.
In another aspect of the invention, a conditioning wafer is automatically processed if the predetermined period of time has been exceeded.
In another aspect of the invention, product wafers are automatically processed if the predetermined period of time has not been exceeded prior to the next wafer being processed.
In another aspect of the invention, conditioning wafers are automatically processed if the previous conditioning wafer does not fully condition the processing tool.
The described method thus provides a method of manufacturing semiconductor devices automatically without running unnecessary conditioning wafers thus saving processing time and increasing throughput.
Th

Affiliated with

Also associated with

Rating

Automatic method to eliminate first-wafer effect does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Automatic method to eliminate first-wafer effect, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automatic method to eliminate first-wafer effect will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2540029