Steve Bishop

ATMI’s Ion Implant Process Efficiency Research Laboratory (IIPERL)

We describe an ion source test stand recently installed by ATMI, and show data illustrative of the research being carried out at the new facility.

The Development of In‐Situ Ion Implant Cleaning Processes

Considerable gains in implanter utilization efficiency can be attained with in‐situ cleaning of deposited material, particularly in and around the ion source. Different methods of in‐situ cleaning are described, and we discuss the relative merits of several chemical reagents. We introduce XeF2, a new and promising reagent for in‐situ cleaning and present some preliminary experiments showing its ability to etch dopant materials. We also show that in some cases etching by XeF2 can be selective with respect to ion source construction materials such as tungsten.

Implanter Source Life and Stability Improvement Using In-Situ Chemical Cleaning

Current 300 mm fabs rely heavily on automation to provide manufacturing efficiency. While implant processes and equipment follow this trend, tool availability and maintenance cycles are often driven by the ion source and adjacent areas which suffer from premature failures due to unwanted material deposits. While working in a high volume production environment, side by side comparative data has been collected on two tools running similar processes, with one tool having integrated in‐situ cleaning cycles and the other with no in‐situ cleaning. This paper will discuss significant improvements achieved in beam stability, glitch rate, ion source lifetime and maintenance cycles which were achieved on the tool with integrated in‐situ cleaning. A program was established at Texas Instruments’ DMOS6 wafer fab in early 2007 to explore and document process and equipment performance. Other critical areas, such as particle and metals contamination will be discussed with inferences as to potential yield improvements.

Investigation into Methods to Improve Ion Source Life for Germanium Implantation

Germanium tetrafluoride has long been the standard dopant gas of choice for germanium implantation processes. While this material maintains several positive attributes (e.g., it is a nonflammable gas that is easily delivered to an ion source), its use can result in extremely short ion source lifetimes. This is especially the case for the situation when an ion implanter runs solely or predominantly GeF4. Presented here is an examination of various potential solutions to the short source life problem, some of which enable significant improvement.

Decreasing Beam Auto Tuning Interruption Events with In‐Situ Chemical Cleaning on Axcelis GSD

Ion beam auto tuning time and success rate are often major factors in the utilization and productivity of ion implanters. Tuning software frequently fails to meet specified setup times or recipe parameters, causing production stoppages and requiring manual intervention. Build‐up of conductive deposits in the arc chamber and extraction gap can be one of the main causes of auto tuning problems. The deposits cause glitching and ion beam instabilities, which lead to errors in the software optimization routines. Infineon Regensburg has been testing use of XeF2, an in‐situ chemical cleaning reagent, with positive results in reducing auto tuning interruption events.

Source and beam performance improvement for carbon implantation with carbon monoxide (CO) gas

Carbon co-implantation has been widely adopted for better Ultra-Shallow Junction formation in the fabrication of advanced semiconductor devices. Currently, carbon dioxide (CO2) is the primary feed gas for carbon implantation. The primary disadvantage of CO2 is the high oxygen content, which causes significant oxidation of the implant source components resulting in rapid degradation of the source performance. Usually the C+ beam starts to degrade quickly while running continuously with carbon dioxide. Also, some other species’ implants, especially boron, are significantly affected after a short duration of CO2 usage. Carbon monoxide (CO) is described here as an alternative carbon source material to CO2. In our testing, CO has demonstrated significant improvements compared to CO2 for both source life and beam performance. Additionally, this paper describes a subatmospheric delivery option for CO. The cylinder package with reliability information is provided.Carbon co-implantation has been widely adopted for better Ultra-Shallow Junction formation in the fabrication of advanced semiconductor devices. Currently, carbon dioxide (CO2) is the primary feed gas for carbon implantation. The primary disadvantage of CO2 is the high oxygen content, which causes significant oxidation of the implant source components resulting in rapid degradation of the source performance. Usually the C+ beam starts to degrade quickly while running continuously with carbon dioxide. Also, some other species’ implants, especially boron, are significantly affected after a short duration of CO2 usage. Carbon monoxide (CO) is described here as an alternative carbon source material to CO2. In our testing, CO has demonstrated significant improvements compared to CO2 for both source life and beam performance. Additionally, this paper describes a subatmospheric delivery option for CO. The cylinder package with reliability information is provided.

Optimization of Xenon Difluoride Vapor Delivery

Xenon difluoride (XeF2) has been shown to provide many process benefits when used as a daily maintenance recipe for ion implant. Regularly flowing XeF2 into the ion source cleans the deposits generated by ion source operation. As a result, significant increases in productivity have been demonstrated. However, XeF2 is a toxic oxidizer that must be handled appropriately. Furthermore, it is a low vapor pressure solid under standard conditions (∼4.5 torr at 25 °C). These aspects present unique challenges for designing a package for delivering the chemistry to an ion implanter. To address these challenges, ATMI designed a high‐performance, re‐usable cylinder for dispensing XeF2 in an efficient and reliable manner. Data are presented showing specific attributes of the cylinder, such as the importance of internal heat transfer media and the cylinder valve size. The impact of mass flow controller (MFC) selection and ion source tube design on the flow rate of XeF2 are also discussed. Finally, cylinder release rate...

Development of “Static” In‐Situ Implanter Chamber Cleaning

Since the introduction of XeF2 in‐situ cleaning, its use in production implanters has been mainly focused on cleaning ion sources by flowing the cleaning vapor through the source arc chamber. This has been called “Dynamic” in‐situ cleaning. “Static” in‐situ cleaning is a different method under development at ATMI which allows an entire vacuum chamber and its contents to be cleaned. The chamber is filled to a pressure of 1–3 Torr of XeF2 vapor, which reacts with deposited material on all internal surfaces, and the reaction by‐products are then pumped away. When applied to the source vacuum chamber, the Static cleaning method allows cleaning vapor to contact components, such as the HV bushing and the manipulator assembly, which may not be adequately cleaned with the Dynamic method. Recently, ATMI has installed a prototype Static in‐situ cleaning system on an in‐house Ion Source Test Stand in Danbury, CT. This paper will describe the prototype cleaning system and process and its applicability to production i...

EnrichedPLUS 72Germanium Tetrafluoride (EnPLUS 72GeF4) and Hydrogen (H2) Mixture for Implanter Performance Improvement

In the manufacturing process of advanced node semiconductor devices, germanium tetrafluoride (GeF4) gas is used for pre-amorphization implantation (PAI) of the silicon crystal structure. This germanium implant is required for controlling channeling and reducing Transient Enhanced Diffusion (TED) for better ultra-shallow junction formation. Fluoride gases, including GeF4, have a long history of poor ion source life performance due to tungsten transport from halogen cycling inside the arc chamber of the implant tool. Also, running long periods of GeF4 negatively impacts the arc chamber, which can affect subsequent dopant processes. This paper will demonstrate the advantages of mixing hydrogen (H2) with EnPLUS 72GeF4 gas to increase implanter source life, extend the length of the beam campaign, improve beam performance, and eliminate post beam purge time.

Carbon Implantation Performance Improvement by Mixing Carbon Monoxide (CO) with Carbonyl Fluoride (COF2) and Carbon Dioxide (CO2)

Co-implantation of impurities such as carbon (C) has been proven to effectively reduce Transient Enhanced Diffusion (TED) of boron during annealing, enabling the formation of high-quality ultra-shallow junctions - a requirement for advanced-node semiconductor devices. Carbon dioxide (CO2) is traditionally used as the feed gas in implant tools for carbon implantation. Recently, carbon monoxide (CO) has been widely adopted as a replacement to CO2 due to its lower oxygen content. To further improve the carbon implant performance, a mixture of CO, carbonyl fluoride (COF2), and CO2 is studied and reported in this paper. The effects of the additional fluorine and extra oxygen content from COF2 or CO2 are investigated. Our experiments show that with the right balance and mixture level of COF2 and CO2, the carbon mixture can achieve improvements in key implant tool performance parameters, such as beam current and source life. Additionally, the CO/COF2/CO2 mixture stability is studied, and a safe delivery package is described.