Wafer Manufacturing

As the semiconductor industry moves to processing at lower nanometer ranges, the margin for error gets smaller even as the opportunity for significant error grows. Clearly, these changes need to be accompanied by more stringent calibration and monitoring of the manufacturing process. Innovative new tools are now helping fabs achiever greater manufacturing consistency, reduce bottlenecks, improve productivity, reduce wafer scrap and ultimately maximise their return on investment.

Recognising a number of the inefficiencies inherent in legacy maintenance methods, CyberOptics Semiconductor (CSI) has developed wireless technologies that are helping keep process equipment aligned to tighter specifications, teach robot handoff positions, and measure gaps between pedestals and shower heads in the deposition process. Because these new tools also make system data available in real time, fab personnel can take immediate corrective action to significantly reduce wafer damage. In addition, the tools shorten the time it takes for fab engineers to troubleshoot equipment while significantly reducing the need for human interference and the opportunity for contamination. Their ability to quickly teach accurate robot handoffs ensure the greater manufacturing consistency that today’s processes require.

Wafer Manufacturing Issues and a Wireless Response
Until now, calibration routines to ensure precise equipment levelling, transfer alignment, or gapping distances have been accomplished using bubble levels or wired methods at best, and by estimation and eyeballing at worst. Verification of alignment often requires the technician to install various jigs, read dial indicators or meters, and directly view the interactions among various parts of the automation system as it moves through its sequence. Fixture installation and removal, coupled with technician intrusions into the ultra-clean tool environment, create contamination that compounds downtime and jeopardises yields. In addition, because most current alignment methods do not allow for the collection of data that can be trended and used to control the automation process, this manual calibration never gets easier. It must be repeated periodically during the manufacturing process.

In response, CyberOptics Semiconductor developed a family of innovative wireless measurement devices, called WaferSense, that are designed in a wafer format and can travel through process equipment much like a wafer to measure and monitor critical parameters in 200mm and 300mm semiconductor processes and equipment. Their wafer-like form factor, remote readout and data logging vastly simplifies the job of keeping equipment level, aligned and coplanar. They allow fabs to operate more efficiently, reduce wafer scrap and contamination, and increase the production of good devices.

Control parameters and measure tighter tolerances
One critical factor in today’s wafer manufacturing is maintaining precise gaps (distance) between shower heads, pedestals and heaters used in processes such as thin film deposition, sputtering and etch. During the deposition phase, gas is distributed onto the wafer from holes in the showerhead. Plasma then helps deposit a film onto the wafer that sits on a pedestal or heater. Because the gap between the showerhead and the wafer needs to be precisely controlled for even gas and power distribution, equipment is monitored to ensure optimal alignment following maintenance.

The WaferSense Auto Gapping System (fig. 1) uses three noncontact distance sensors to measure gaps at three points and return live gap measurements. Whether fabs need to set a gap that is perfectly level or slightly tilted, the system’s three sensors each report separate readings in numerical and graphical form to achieve the exact gap needed. Each graphic is colour-coded in the application software to make it easier to see when the gap is above, below or within the user-defined target gap range.

The Auto Gapping System allows fab engineers to set optimum gaps and continually monitor these gaps and compare them to the original settings. Data then is reported to the process engineer for quick remediation of equipment out of gap, thus improving process uniformity and increasing yield.

Automate levelling of semiconductor process equipment
Problems with levelling are another area in the semiconductor process that leads to wafer errors and increased wafer scrap and costs. Many facilities are still using bubble levels or ‘eyeball’ visual inspection to try to verify that their equipment is level, but this method is imprecise and time consuming.

WaferSense Auto Levelling Systems (ALS) (fig. 2) brings equipment-levelling solutions into the modern age. Unlike manual or wired methods of measuring co-planarity, the WaferSense levelling system doesn’t require equipment disassembly or disruptions to vacuum chambers. This can reduce calibration time by 80 percent or more. Since process equipment tends to change shape when evacuated, in-site levelling is much more effective than atmospheric levelling.

Shaped like a wafer, the tool can be placed in cassettes, front opening unified pods (FOUPs), on end effectors, aligners, in load locks, transfer pins or process chamber pedestals. It offers real-time feedback of equipment measurements, via sensors in the hardware and Bluetooth connectivity, to quickly ensure that all stations are level and coplanar. The graphical user interface software displays an intuitive bubble level, and users can refine Go/No Go regions to ensure compliance with the acceptable range The device has an inclination measurement range of ±14 degrees from horizontal, and provides precise pitch and roll measurements, accurate to ±0.03 degrees, that can be logged to relate co-planarity with yield and determine the ideal tool set-up for optimal yields.

The WaferSense Levelling System also comes in a format ideal for use in reticle handling equipment such as reticle stockers that, until now, were challenging to measure accurately because of space constraints.

Wireless levelling devices reduce setup time and preventative maintenance downtime by promptly delivering reliable, real-time measurements. As a result, alignment problems can be diagnosed and remedied quickly, equipment is returned to productive use faster, and fabs realise better tool-to-tool process uniformity.

Wirelessly teach wafer transfer positions
Yet another problem in wafer production occurs as the result of misaligned wafer transfer. During wafer transfers, the cooperating supports must be as parallel to one another as possible. If they are not parallel, the wafer may slide as the weight of the wafer shifts from the first to the second support. When wafers are transferred slightly off centre, non-uniform processing may occur. The wafer edge may rub on the shoulders of one support or the other, dislodging particles or breaking the wafer. Although most of these may be completed without system-stopping faults, misalignment may lead to intermittent and difficult-to-diagnose defects.

Teaching transfer locations via a wireless device offers another opportunity to reduce downtime and raise equipment availability and overall performance. The traditional robot ‘teaching’ process (determining wafer transfer coordinates and loading them into the robot’s controller software) requires technicians to manually direct the robot’s motion and estimate wafer transfer positions. This process is not only labour intensive, but inaccurate because coordinates for most teach positions are ‘eyeballed’ and subjective. The current procedures may also require that equipment be taken off the production line and partially dismantled during maintenance.

The WaferSense Auto Teaching System (fig. 3) provides accurate wafer transfer coordinates for robotic handling. As it moves inside semiconductor equipment, the teaching system uses machine vision technology to ‘see’ targets that mark wafer transfer locations, such as load locks and process chambers, and transmits their digital coordinates to the user. Because there is no need to dismantle the equipment or cool it below maintenance temperatures, fabs can save qualification time and consumables. The teaching system relays live video and real-time coordinate data via a graphical user interface making equipment checks and adjustments fast and easy. The live video allows engineers to look for lost or broken wafers without opening the equipment. The system is accurate to +/-0.1 mm (x and y positions) and +/-0.5 mm (z position).

Easy-to-use software logs data for later comparison and analysis, and helps the user ‘train’ the teaching system to recognise new types of targets. Hard data and repeatable coordinates reduce technician-to-technician variations in calibration technique to eliminate wafer scrap due to mishandling.

Wirelessly view vibrations in wafer transport and handling
Another yield limiter in the fab is particle defects caused by rough handling of wafers, with the vibrations causing particle generation and defects.

Fingerprinting locations and handling of process and material handling equipment via a wireless device offers another opportunity to reduce particulates and raise equipment availability and overall product yields. The traditional method for troubleshooting particulate or robot vibration problems required technicians to manually direct the robot’s motion and expend numerous test wafers to determine particle sources. This process is not only labour intensive, and may also require that equipment be taken off the production line and partially dismantled during the troubleshooting.

The WaferSense Auto Vibration System provides vibration levels inside process equipment and material handling systems. As it moves inside semiconductor equipment, the vibration system transmits the data to a PC that corresponds to the wafer locations, such as load locks and process chambers, and allows the user to view this real-time. Because there is no need to dismantle the equipment or cool it below maintenance temperatures, fabs can save qualification time and consumables. The vibration system relays live data in x, y and z dimensions via a graphical user interface making equipment base lining and troubleshooting faster and easy.

Extending semiconductor wafer manufacturing
The cost of wafer scrap and equipment downtime, coupled with the greater sensitivity of today’s semiconductor geometries, makes it critical for fabs to find new maintenance tools equal to the task. Problems that were insignificant at 100nm can have a serious impact on yield at lower nanometer levels. New technologies, such as wireless precision metrology tools from CyberOptics Semiconductor, will be critical if the semiconductor industry is to reach their ambitious wafer yield and throughput goals.