MEMORANDUM

To:            Prof. N. Cheung, Microlab Faculty Director

From:        K. Voros, Microlab Operations Manager

Subject:     6" Upgrade – Final Report

Date:         20 December 2004

Cc:            T. King, C. Spanos,

                W. Flounders, R. Hamilton, S. Parsa, R. Spivey

INTRODUCTION

In February 1995, a group of colleagues from several universities met to discuss in a workshop atmosphere the “Future of Academic Microfabrication Facilities”. As a result, a loose organization, Academic Labnetwork was formed, to provide a discussion forum and common representation for academics and staff involved in supporting university microelectronics facilities. Among the agenda items were two of special concern to us:

►    wafer size and compatibility, including the need and possibility to upgrade to handle 6" wafers, and

►    upgrading laboratory management software.

A year and a half later, in December 1996, at another Labnetwork meeting, UCB and MIT presented upgrade plans for equipment to handle 6" wafers and Stanford presented their plans for software upgrade.

Motivating factors for the 6" upgrade included the need to:

►    promote interaction with industrial researchers affiliated through consortia;

►    promote academic-to-industry technology transfer;

►    validate device research in a more advanced technology environment;

►    maintain process relevance and improve performance with better equipment;

►    to take advantage of the availability of 6" equipment donations;

►    standardize wafer size among university laboratories so that we could leverage our processing capabilities;

►    avoid possible lack of 4" wafers.

PLANNING

Phasing and Scheduling of Upgrade Plans

We developed a three-phase plan based on discussions with UCB Microlab faculty and industrial affiliate members, focusing on CMOS device and MEMS processing capabilities.

Phase I:  limited 6" capability: CMOS process with specific component modules:  litho, LPCVD  nitride, poly-Si, oxide; basic wet and dry etch.

Phase II:  simultaneous 4"/6" capability, MEMS processing.

Phase III: new facility, fully 6" and selected 8" modules.

The three-phase upgrade strategy was designed to minimize impact on labmembers' projects. In our proposal to faculty we made timing conditional on funding; however, we ventured the overly optimistic projection of:

           Phase I:    1997-98,

           Phase II:   by 2000,

           Phase III:  development parallel to Phase II and beyond.

We completed Phase I during 1998-2000 and Phase II by the end of 2004. Phase III will become a reality with the construction of the new engineering building, CITRIS, which will house a new nanofabrication facility. 

Development of Funding

In my 1996 Microlab Year End Report I wrote: "Most university laboratories have upgrade plans but none have funding in place, including Berkeley."  After considering the available options to us – none too promising – we arrived at a realistic plan relying on bootstrapping, equipment donations and procuring targeted tools through faculty grants.

Berkeley Microlab Affiliates (BMLA)

In 1997, with the leadership of Professor Costas Spanos, we established the Berkeley Microlab Affiliates program (BMLA), as a vehicle to give access to our industrial partners to equipment in the Microlab. Companies pay a membership fee, according to the number of employees working in the Microlab. (See details at http://microlab.berkeley.edu/text/bmla.html.)

In addition, BMLA members pay standard lab fees plus a 50% overhead charge on lab fees. While lab fees are part of the operating budget, the overhead and membership fees are retained for laboratory improvement and upgrade. After allowing for the College of Engineering and departmental gift tax withholdings, the collected BMLA fees were saved and fully applied to provide for the 6" upgrade. The total amount collected from 1998 through 2004 and applied to Phases I-II was just over $2M, (cash layout for equipment and facilities upgrade + FTEs).

MEMS Exchange

This was another program to leverage our Microlab facilities investments.

We participated from the start (1998) in the DARPA-funded MEMS Exchange fabrication program administered by the Corporation for National Research Initiative (CNRI) utilizing the resources of several academic and industrial participants (http://www.mems-exchange.org). Since we offered full capability processing modules from the start of the program, CNRI funded 2 FTE's for 5 years. Staff provided the processing service at the approved module charges, which included overhead. While all operating expenses for MEMS Exchange processing were recoverable from the collected module charges, we were also able to apply the overhead portion of the MEMS Exchange income to the 6" upgrade.

Equipment Donations

There is no such thing as free equipment. Donations come in all shapes and forms and accepting them is a tricky business. Because we had a definite idea of what we needed for the upgrade we selected only specific tools from the long lists that were offered from several decommissioned plants. In most cases we had to cover crating, shipping, transportation insurance costs, followed by facilities modifications and installation expenses. In some cases purchase of missing parts, (i.e. gas purge panels and vacuum pumps) and software upgrades were also required. Still, if the tools matched our needs we accepted.

 

Three companies provided cash donations to enable use of the donated tools: Intel and Renesas Technology Corp. gave cash to assist with installation and start-up and FANUC with decommissioning and storage. These funds we leveraged with in-house talent for most efficient use and to cover costs of items in Table 4.

Equipment through Research Grants

The most successful equipment acquisitions are research driven. This method requires close coordination between lab management and PIs, which in our case was not a problem – research needs and lab upgrade plans matched.

We received our most expensive 6" tools through grants obtained by the device, process/manufacturing, and BSAC research groups. As early as 1992, when BSAC grants paid for a low-stress nitride LPCVD furnace, we were able to install a 6" compatible furnace bank, with the plan that the rest of the tubes would be financed from other sources. In 1998 the device group paid for the Si-Ge system, in the same bank, and we filled up the rest of the slots during Phase II of the 6" upgrade.

By far the most expensive equipment was the ASM Lithography tool, donation value $2.37M. This was obtained through the UC SMART program, which also provided partial matching research funds for equipment donations. The Small Feature Reproducibility (SFR)grant in which several of our process/ manufacturing PIs participated, with Professor Spanos at the lead, arranged for the donation of the tool from ASML and provided the facilities improvement, installation, and start up from matching funds. The SFR has been supplemented by the Feature Level Compensation and Control (FLCC) effort, led by Professor Neureuther; this partially funded the donation of the Centura etcher and its associated installation and start-up costs. Donations are detailed in Table 1.

Budgeting                                                                                                                                                                                         

In addition to equipment donations and procuring targeted tools through faculty grants, we estimated the following cash layout, to be covered from our own funds, BMLA membership fees, overhead on industrial use:

Phase I:  est. $850K plus 1 additional FTE

litho - 6" wafer stepper, resist track/developer, 2 sinks

          furnaces - 2 LPCVD, 1 atmospheric tube, RTP

          etch - conversion of 2 Lam etchers

          misc. - insp. microscope, dicing saw mod., pallets for cpa

          computers - new LAN terminals

Phase II: est. $ 1,325K plus two additional FTEs (total 2 FTE for equipment upgrade and 1 FTE for software upgrade)

          litho - second 6" wafer stepper, wafer track for thick PR

          furnaces - 4 LPCVD, 4 atmospheric, clean/etch sinks

          etching - 3 Lam etchers converted, Centura added

          misc. - profilometer, handling equipment

          computers - 1 programmer FTE

Phase III: 1 FTE for 3 years

THE UPGRADE PROCESS

The progress of the 6" upgrade project is shown in Table 2. The work spanned six years: 1998-2004. During this time the Microlab operated at full capacity, with the exception of the machines that were being worked on at any one time. Microlab use and equipment hours remained relatively level throughout the 6” upgrade effort. We kept the 4" system up until its replacement was installed and fully characterized. Finally, we added equipment for capacity and for easing process specification demands on some tools. (Ex: asml and svgcoat6 for DUV litho vs. gcaws6 and svgcoat3 for I-line litho.)

Facilities Upgrade

New Construction

Before installation of 6" equipment, extensive renovation of facilities had to take place. We constructed separate rooms on the 1st floor of Cory Hall for the Planarization Lab (190), with the CMP, the cmp wafer cleaner and P5000 TEOS/Ozone PECVD tool. Room 144 was built for the Novellus cluster tool for thin films deposition. The space in each case had to be made available by decanting old equipment, rerouting power and other utilities and constructing walls and mini-cleanroom environments.

Within the Microlab (420 Cory Hall)

There was no room untouched by the upgrade. Starting with decanting GL4 to accept the ASML stepper, for which the laser is located in 432A, (and piped through the wall to GL4!) we had to play musical chairs with equipment. 6" equipment was mostly added, only a few tools were decommissioned; thus, space is utilized to maximum.  For the addition of the Centura cluster tool we had to eliminate walls and combine service chases with the room GL1.

Utilities

Major activity was invested in utility upgrades. Six inch capable equipment and the addition of equipment without equivalent decommission means more of all utilities are required:

Power: new transformer, additional feed and break out panels

           Cooling water: a separate new cooling loop with its own chiller

           DI water: increased number of RO membranes

           Drains: major re-plumbing of drain lines; elimination of untenable vacuum drain system

           Acid waste neutralization: new system installed (outside Cory Hall)

           Compressed air: building compressor replaced by Campus Services

Exhaust: duct lines redesigned, separated for each bank of furnaces, rebuilt in non-    corrosive material

Nitrogen gas: increased capacity vessel (9000 gal) installed; new 1" nitrogen gas line plumbed to Microlab

Specialty gases: storage space rebuilt outside Cory Hall meeting tighter regulatory requirements

           HAZCOM: second toxic gas monitoring system installed, display moved outside of Microlab; new HAZCOM (blue) alarm system for local evacuation added

Equipment Upgrade

No new tools were obtained for the 6" upgrade. All equipment new to the Microlab came in as upgrade of previously owned Microlab equipment or refurbished used tools, including all tools donated directly by companies. Some tools were purchased on the used market then sent to local refurbishing vendors who modified them to Microlab staff defined specifications.

We have installed, modified, or improved for the 6" upgrade 16 furnace tubes and 44 other tools, or about half of the total equipment in the Microlab. As a result we have full 6” CMOS and MEMS processing capability. Equipment installation time line is shown in Table 3.

The list of equipment and size compatibility is available on-line on the Microlab's web site, http://microlab.berkeley.edu/labmanual/chap1/1.13.html. Of the 116 operating systems in the Microlab, 80 are both 4” and 6” capable, 28 handle only 4” Si wafers and eight can handle only 6” Si wafers.

Role of the Machine Shop

I cannot overstate the importance for the 6" upgrade project the availability of a machine shop. It simply could not have been done or only at an exorbitant expense of time and funds. The ERL Machine Shop became an extension of the Microlab, especially at times of high activity such as installation of furnace banks and various complex equipment. Many of our modifications relied on the fine design and machining capabilities of the staff of the Machine Shop. The availability of additional personnel, tools and materiel was indispensable to the flow of upgrade work and operation of the Microlab at the same time.

One of the lessons of the 6” upgrade is clear: successful completion of Phase III of ongoing Microlab evolution – fitup of the new laboratory, and ongoing flexible operation of the new facility – mandates maintaining the ERL Machine Shop. The Machine Shop is key to enabling the Microlab to respond rapidly to the ever changing needs of faculty and researchers.

Process Characterization and Upgrade

Process Modules

Throughout the 6” upgrade project our approach was that equipment installation and/or upgrade was followed immediately by process characterization on the upgraded system. A standard process was established and operating manuals were updated before equipment was released to members.

►    The first module completed was lithography based on the ASML DUV stepper. The module also included a new 6” resist dispense track and developer, new photoresist, CD-SEM, DUV microscope and upgrading of mask making equipment. This first major effort resulted in a very impressive 0.35 ΅m lithography capability.

►    Substantial effort was expended on characterizing the furnaces. Temperature profiling, boat design and load distribution, for both 4” and 6” processes, required many iterations and careful experimental design. Not only did we have to develop the 6” process in all the atmospheric (7) and LPCVD (9) furnaces, but 4” processes also had to be re-characterized in the larger diameter tubes. At the end, we have all our furnace process modules reestablished, both in the CMOS and MEMS tubes.

►    Upgrading the Lam auto-ethers turned out to be the least disruptive. A clever redesign of the handling mechanism by our technical staff and the Machine Shop, resulted in 4”/6” dual capability with the flip of a switch. On the Rainbows the 4” wafer transport module was replaced with a 6” module, purchased from Lam. All etch processes, however, had to be re-characterized for both 4” and 6” wafers. A major process performance improvement resulted from the installation of the Centura etcher (6” only).

►    Cleaning and wet etch processes were developed in the new sinks, with larger tubs for 6” wafers. Dual size wafer handling equipment went through several modifications before arriving at a user friendly solution, with dedicated MEMS and MOS sinks in the cleaning area.

CMOS Baseline

The Microlab maintains a CMOS baseline, which specifies standard process modules across the operation. During the 6” upgrade the baseline actually pushed the agenda by requiring the completion of upgrades for the next step. The first 6” run (CMOS 150) played an important role in gauging the success of the upgrade.

CMOS 150 reproduced our 1 ΅m process, (well established on 4” wafers) on 6” wafers (150 mm). This lot was completed and report published in 2002. (Available at http://microlab.berkeley.edu/baseline/index.html) A great advantage provided by the 6” upgrade was that the performance of the new equipment is several generations better than that of the 1980 vintage tools. Specifically, because the ASML deep UV lithography tool is capable of 0.35 ΅m technology, we were able to skip a few shrinks and design a 0.35 ΅m process. The Centura etcher provided the other key process module, by being able to etch SiO2 uniformly over a 6” wafer, with good selectivity.

The first 0.35 ΅m cmos run (CMOS 161), which included a complete process redesign, produced on 6" wafers was completed in December 2004. Results were excellent and the wafers tested have a high yield. This we attribute to the improved tool set provided by the 6" upgrade.

The data for the run CMOS 161 is posted on-line at: http://microlab.berkeley.edu/baseline/index.html

Our baseline engineering group, Attila Horvath, Sia Parsa and graduate student Hiu-Yung Wong, are developing data for a formal report, including process and device parameters.

Computers

CAPE – Local Area Network

Equipment upgrade included a computer systems upgrade. In 2001 we had completed the migration to all graphic (flat panel) terminals inside the lab. These are running on a Windows server dedicated to the LAN for the Microlab. The Common And Personal Environment (CAPE) application was developed in-house by Tim Duncan. Lab members can now work in a common PC environment, with one window for the original lab control function (ascii).

WIS – Equipment Control

The obsolete Taurus equipment control system was upgraded to an industrial PC server (with Linux OS) and new control boxes at each equipment. The interlocks are turned on and off through the WIS server and activity information is stored in a single database on the central host. The new WIS system is extremely reliable and is protected from compromises with a high level of security measures.

Gasinven – Microlab Gas Management System

With the 6” upgrade and process development enabled by the new tools, specialty gas cylinder inventory increased considerably. This necessitated the in-house development of a gas cylinder management system. Gasinven utilizes MS Access on an SQL server. 56 different specialty gases are managed in this system, accessible to staff on a password protected web site.

RUMS – Resource Utilization Management System

A major step forward in upgrading the Microlab’s computer control system was the release of RUMS in June 2003. The old facilities monitoring system was completely replaced with new hardware and in-house developed software. Thirty-two sensors are monitoring various utilities, such as N2 pressure, air pressure, temperature, etc., which have to be in spec for the lab to operate properly. Collected data with graphs are available not only by directly connecting to the RUMS server but also on the Microlab’s web site. The efforts of our software and hardware engineers Duncan, Chen, Pestal and Merport resulted in a great system, which is described in detail in Memorandum No. UCB/ERL 03/43.

Mercury – New Laboratory Management System

Microlab operations are controlled by BCIMS, Berkeley Computer Integrated Manufacturing System, a set of in-house developed software. The system, which was installed 20 years ago as a research project, consists of components for equipment and facilities control, accounting, purchasing and inventory, reservations and maintenance. Because we kept up with developments in computer technology, BCIMS served us extremely well during the past 20 years; however, by now it needs a major upgrade. In 1998 we embarked on a collaborative effort with Stanford and MIT to develop a new laboratory management and information system, but our diverging needs for local-specific software tools lead us to the conclusion that the Microlab will be best served by developing our own.

In 2002 we started design and development of Mercury, a system using industry standard technologies, with platform and database vendor independence. All functionalities of the well-tested Wand system were retained in Mercury; the accounting module was redesigned based on the double-entry paradigm. We are currently testing the new system; roll-out target is the start of Fiscal Year 2005/2006.

 

Documentation

As part of the 6” upgrade we expended considerable effort to keep documentation up to date. Reason: we expect lab members to use the on-line manual (and hard copy) as the first step in finding operational and process information.

The task of writing new manuals, appending, updating existing ones with the latest information, is shared by staff and expert users of the tools, with process staff carrying most of the burden. Madeleine Leullier, Computer Resource Specialist, has the assignment of document control, i.e. editing and installation of both on-line and hard-copy manuals.

Extent of Documentation

The Microlab’s operating manual consists of 1,653 pages, compiled in 156 chapters, available on-line from the web portal of the Microlab, http://microlab.berkeley.edu/. The listing of

6” Equipment Capability is shown in the Equipment panel.

Reformatting and Standardizing

We started compiling our on-line lab manual in 1982, when the present Microlab was built and computer control was introduced. This first manual was in ascii format, without specific structure. When we started to move information to the WWW the manuals were gradually reformatted. At the same time, Sia Parsa introduced a rigorous standard format for all chapters. As part of the 6” upgrade we established the goal of completion of the transfer of all manual chapters into the new form, with updated information. This in itself was a monumental task; however, it is now complete and we receive comments, requests for permission to copy, from all over the world on a regular basis.

 

New Chapters

With the addition of 6” equipment 30 new chapters were added. Twelve of these were the result of separating manuals for previously grouped equipment, such as LPCVD furnaces, atmospheric furnaces and sinks. Now all 20 furnace systems and 10 sinks have separate chapters. This was necessary because of the different operating procedures, restrictions and policies for each of the tools.

 

Microlab Web Portal–http://microlab.berkeley.edu

As part of the 6” upgrade we also upgraded our web portal in 2003. The most important goal of this change was to streamline the portal to make information easily available. A search function was also added. The end result is a very plain-looking, nine-panel front page, without pictures, pop-ups, or other flourishes.

FINANCIALS

Table 4 shows the expenditures from Microlab funds. We also needed to increase the number of staff for the duration of the 6” upgrade project. Salaries and benefits for the additional FTEs were covered from different sources, shown in Table 5 below.

Fiscal

Year

Operation’s Closing Balance $

Microlab Operations

(FTE)

BMLA Overhead (FTE)

CNRI

(FTE)

SMART

(FTE)

BSAC

(FTE)

Sub

Total

Other

Total

(FTE)

1998/1999

400.00

25.0

–

1.8

–

0.5

2.3

27.3

1999/2000

-19,000

25.0

–

2.0

1.0

1.0

4.0

29.0

2000/2001

-16,000

27.6

–

2.0

0.3

0.5

2.8

30.4

2001/2002

-161,000

29.6

–

2.0

–

0.5

2.5

32.1

2002/2003

-168,000

29.6

2.7

2.0

–

0.5

5.2

34.8

2003/2004

-201,000

25.8

2.2

1.6

0.3

1.3

5.1

30.9

*2004/2005

-120,000

22.9

2.5

1.0

0.8

–

4.3

27.2

*FY 04/05 estimated

Table 5 - Staffing During the 6” Upgrade

Figure 1 shows various indicators during the years of the 6" upgrade.

     a) Balance at fiscal closing each year

     b) Machine Shop expenses (these were included in Table 4 in each item when applicable.)

     c) Staffing changes during the upgrade

Cost of the 6” Upgrade

►    Cash paid out of Microlab funds (BMLA, Overhead)       $1,716,907.00

Summarized in Table 4, including Machine Shop charges                                                          

►    Additional staff Salaries and Benefits

Microlab operations, 2000-2003                                   $   701,547.00

Microlab funds (BMLA, Overhead), 2002-2004              $   162,698.00

Funded by research groups (CNRI, SMART, BSAC)      $1,017,161.00

►    Cash provided by research grants                                 $1,588,974.00

For equipment shown in Table 1a)                                                                                                     

►    Recorded value of donated equipment                            $7,053,499.00

Summarized in Table 1b)                                                                                  

The numbers above represent an added value of $12.25M to the Microlab. Microlab PI’s who were especially interested in and supportive of the 6” upgrade participated by brunting the burden and accepting higher equipment charges on the 6” tools. Increased staff costs showed up as debt at the end of the fiscal year, during the past five years, accumulating to $200K at the end of 2004. This debt, ~2% of the total value of the upgrade will be recovered from increased use during the next five years. Recharge rates during the past 10 years increased only by the rate of inflation, issued by the U.S. Department of Labor, Bureau of Labor Statistics. We intend to adhere to this guideline in the coming years.

SUMMARY

After more than 6 years of continuous work we have completed the 6" upgrade, resulting in full 6” capabilities for both CMOS and MEMS processing. This report was compiled to document the activities, timeline, and costs of this long term effort. The Microlab 6” upgrade demonstrates the extensive finances and the meticulous planning required to accomplish such complex undertaking without disrupting research services.

The success of the 6” upgrade was the result of building on our strengths, the talents of our capable staff and the support of our member PIs. Faculty who requested the upgrade participated by investing time and effort in obtaining equipment donations and providing cash from their grants.

The 6” upgrade also serves as a benchmark for the on-going Microlab evolution; it gives a measured taste of the financial, personnel and infrastructure resources needed for start-up of the new laboratory in the CITRIS building, equipment migration from Cory Hall to CITRIS, and decommission of the existing Microlab.

ACKNOWLEDGEMENTS

This major project could not have been accomplished without the creativity and hard work of the staff of the Microlab, the Cory Hall Machine Shop, the support of Microlab PIs and the faculty leadership we are privileged to enjoy.

Jim Bustillo, Technology Manager of the Microlab and BSAC from 1991-2000, was instrumental in developing the upgrade plans, selecting equipment from Intel donation lists and seeing the upgrade through execution, until 2000.

Bill Flounders followed up since mid-2001; he arranged for the FANUC donation, which enabled upgrade of many general-use lab capabilities, in addition to the specialized tools of the Device Group and BSAC. Bill also oversees Phase III of the on-going Microlab evolution.

Bob Hamilton successfully solicited equipment donations from industry in a total value of nearly $1M. He also introduced advantageous parts procurement through the on-line auction site, E-Bay. Bob also provided enthusiastic leadership during the 6" upgrade and with his Equipment Engineering staff carried the brunt of the work – in addition to keeping daily operations going.

The Machine Shop, and Microlab equipment engineer David Lo, designed and built custom wafer handling tools for several of the Lam etchers. The Berkeley Microlab is the only facility in the world with Lam Autoetchers capable of both 4" and 6" wafer handling.

Joe Donnelly managed the installation and start-up the Novellus and P5000 PECVD tools, overcoming various difficulties and the problem of missing and/or wrong parts.

Phill Guillory, with his staff, Bob Connolly, Danny Pestal, Mario Lizardo and student assistants was the driving force behind facilities upgrades, constructions of new space and upgrading utilities, including the HAZCOM alarm system, and RUMS.

Mike Linan, with Brian McNeil, provided gas line upgrades and installation of the additional HAZCOM system. Management of the greatly increased number of vacuum pumping systems is the responsibility of Mike's group. The pump inventory contains 145 units.

Evan Stateler’s outstanding troubleshooting skills were called upon throughout the upgrade projects and continue to be in demand in the expanded operation. In addition, Evan managed to site equipment in the Microlab, which exceeded the dimensions of Cory Hall elevators and hallways.

Patrick Wehrly provided meticulous care during the furnace upgrades.

Sia Parsa and his Process Engineering staff, Kim Chan, Jimmy Chang, Marilyn Kushner, were instrumental in bringing up processes in the newly upgraded systems. They also carried the burden of updating manuals. Baseline Engineer Attila Horvath was instrumental in 0.35 ΅m baseline process development.

Rosemary Spivey provided ongoing financial accounting of all aspects of the 6” upgrade from 1998 to present.

Susan Kellogg-Smith purchased equipment related to 6” wafer processing as well as parts required for equipment installation and facilities upgrade and modification in support of the project. Additionally, inventory was updated to include spare parts for 6” wafer processing equipment and materials required by lab members to process 6” wafers.

Scott McNally, Director of Space, Planning and Facilities in EECS since 2001, provided great support during facilities upgrades which involved the building (Cory Hall); specifically, he guided (through Campus Project Management) the installation of the acid waste neutralization system.

Faculty actively involved in soliciting donations and providing grant funds for installations are included in Table 1. They were also instrumental in keeping us going by the unrelenting demands of their graduate students' projects.

Prof. Spanos provided initiative and leadership during the planning phase, which was continued by Prof. King when she took over as the Faculty Director of the Microlab in 2000. She kept her eyes vigilantly on our upgrade activities and focused us on the path of integrity for device processing.

Finally, I am satisfied and proud of the fact that we were able to complete the tall order of the 6” upgrade, Phase II of the Microlab evolution, without special fundraising. Our next goal is completion of Phase III, building and fit-up of the new nanolab in the CITRIS building. For Phase III we will embark on development of dedicated funds.