Matt Smith from Olympus America reviews the applications, benefits, and utility of defect review in the photovoltaic manufacturing, from yield analysis to post-electrical testing.
by Matt Smith, Olympus America Inc.
April 26, 2010 – Semiconductor manufacturers worldwide have for decades been using defect review and analysis to help improve yields and understand the effects of various process changes. With efficiency demands on the photovoltaic (PV) industry increasing, it can be effective to apply the benefits of defect review to the manufacture of solar cells as well. Indeed, standards for efficiency in PV manufacturing have been rising and are projected to continue to rise, eventually matching the cast and yield demands of those found in the memory industry. A review of the applications, benefits and utility of defect review in the PV industry’s typically smaller-scale manufacturing facilities reveals that solar cell manufacturers can enjoy the benefits of yield analysis with relatively low capital expenditure. Utility can also be found by using defect review for other applications such as post-electrical testing, where optical inspection can be used to help investigate device failures.
Background on defect review
The need for high throughput and efficiency is a basic requirement in the increasingly competitive PV manufacturing industry. In essence, the defect review function provides a systematic means of retrieving valuable manufacturing information through the examination of defects found by automated instruments or human inspection. Inspection and review is vital for providing feedback at all phases of the manufacturing process, from confirming the use of the correct layer or registration, to discovering the source of process errors so that they can be addressed and eliminated. Defect identification, classification, and elimination are key to building systems that enhance production yields.
In the semiconductor manufacturing field, defect review is routine and is normally performed either online via automated detection machines or offline using specially designed cassette-to-cassette machines. While such tools are powerful, their cost-of-ownership sometimes places them outside of the reach of PV manufacturers.
PV manufacturing requirements
The requirements of PV manufacturers differ in two key ways from current semiconductor manufacturers. First, PV devices typically are much less complex than semiconductor devices, both from a geometric density (x, y planes) point of view, and with regard to z-stack (depth) complexity regarding the number, thickness, and types of layers that go into the devices being built. This means that PV manufacturers, unlike their semiconductor neighbors, can utilize much less expensive defect detection tools — tools that forego detection of sub-micron events. One cost-effective option open to them is using fast and far less expensive detection tools that rely on optical technologies. The inspection and review of PV devices can easily be performed using microscopes that utilize conventional white light illumination. These well-understood optical technologies easily allow for accurate detection of events down to ~0.35μm in size.
Microscope-based systems can go on a tabletop in the manufacturing or quality control (QC) area. These small-footprint systems allow an operator to do defect detection and classification easily, while accommodating file formats used by automated detection instruments or utilizing data generated by the tabletop system itself in a standardized inspection mode. The yield enhancement engineer can look at all events, or choose to filter the data by size or other selected attributes. In many cases, a single set of test areas can be examined to control process quality.
Tabletop implementation also allows for seamless coordination with visual inspections being performed by the operator. Inspection is required for layer verification, lithography performance and post etch evaluation. When combined with an automated defect detection system, any areas of concern seen with the inspection and review system can be entered into the data file with the click of a mouse. When tabletop systems are used for inspection using a user-defined routine, the results can be combined with those of the automated defect detection system. Defects can then be logged and revisited by the inspection and review system as the subsequent layers are built up on the device. As the process continues, this technique allows the yield engineer or an automated software system to determine the number of failed devices prior to electrical testing. This defect review process can be extended to the electrical test area, where failed dies are inspected using the optical microscope. The die location or inter-die failure data can be retrieved from the electrical test system via a data output function, and then transferred automatically to the inspection and review system, where the operator can inspect the failed areas and categorize each failure, make basic dimensional measurements, or grab an image for future reference. This type of systematic review and analysis can be an important part of justifying and implementing process improvements.
Some optical microscope systems have another useful capability for PV manufacturing: many are capable of imaging in the near-infrared (near-IR) range, which enables them to image events and defects beneath the surface of PV devices. The silicon material used in PV devices is highly transparent to near-IR wavelengths (from 800 — 2500nm). The ability to peer beneath visually opaque surfaces, combined with the inspection and review capabilities of the system, allow the yield engineer to better understand such events as sub-surface particulate contamination, layer registration issues and delaminations. Just as it does with post-electrical test inspection, the defect review system can capture images and allow for the entry of data or comments into the system.
An additional emerging application of optical defect inspection and review has been found through the use of confocal imaging for device inspection or analysis. Confocal microscopes such as the Olympus LEXT OLS4000 system allow for traceable x-, y- and z-axis measuring on the surface of the device using a visible wavelength (405nm) of light, or under-the-surface imaging using an IR or near-IR wavelength (1310nm). Confocal technology also handles surface roughness measurements, a key attribute in understanding the overall potential efficiency of devices. In addition, the ability to measure thin films without cross sectioning is a powerful tool when manufacturing thin-film solar cells.
Typical system components
A typical tabletop defect review system for PV applications consists of three major sub-assemblies. The optical microscope is first, since it is the main instrument required to perform the inspection. It is important that the focus system and optical configuration of the microscope be automated in order to remove inconsistencies in illumination and field of view found with most manual instruments. Automation vastly enhances the repeatability of the inspection and defect review process.
The second key component is an automated stage system. The automated stage allows for highly repeatable positioning of the device in order to provide fast, reproducible defect location. Among the many types of precision stages, linear encoded stages offer the highest accuracy and most repeatable results. Linear encoded stages not only help increase throughput; they offer the capability to make measurements that are not limited to pixel-counting techniques.
Both the microscope and the stage are controlled and coordinated with software, which comprises the third key sub-assembly. The software application should be designed to coordinate not only the microscope and stage, but also any automated defect detection tools or import data coming from other inspection and review stations. For fast and efficient operation, the system should include control of the optical attributes and stage location, and handle retrieval and update of applicable data fields.
Some of today’s software applications allow basic inspection to be handled through user-created inspection routines that allow a human operator to view the device onscreen via the graphical user interface (GUI). These applications may also display the site or location under inspection, defect categorization if available, the sizes of any defects found and any comments entered by previous inspectors, all right on the display. Where an automated system is not used, good inspection applications will allow for some basic measurements to be taken; this data can then automatically or manually be entered into the data file.
A further time-saving use of inspection and review software in PV manufacturing is a semi-automatic function where the system moves to predetermined (pre-taught) areas of interest, capturing images that are stored for later examination on the system or elsewhere. This implementation is useful for R&D activities where in-depth analysis is required. Stored images can be examined on any networked or web-enabled computer and the data file can be updated with comments, defect code data or other useful input.
Some PV manufacturers are finding automated wafer handling systems useful as part of the process. These subassemblies communicate with application software to allow sampling and retrieval of specific wafers along with complete cassette-to-cassette handling, eliminating potential product damage that can sometimes occur with manual wafer handling.
Implementation and training
The implementation of a defect review or inspection system is straightforward. Ideally, a single source provider will allow users to implement a turnkey solution. By using a single source provider, manufacturers also can be sure that their hardware and software components are optimally matched.
Training is an important aspect of getting the most out of a defect review system, but thankfully, learning to use these systems is rarely time consuming. Typically, training time required for a comprehensive inspection and review system can take anywhere from a few minutes for simple defect review functions, up to an hour to learn to create inspection routines and set up defect code definitions. A key point when implementing these systems is to take advantage of the flexibility available with the systems to handle both inspection and yield management functions.
Adapting semiconductor manufacturing and QC techniques is a logical and straightforward process for today’s PV manufacturers. With careful choices, efficient inspection and yield improvement plans can be adapted for reduced cost and consistent repeatability. Proven optical techniques provide a cost-efficient means to inspect devices on both the surface and below, delivering traceable data for improved yields.
Matt Smith’s undergraduate studies were in imaging science and instrumentation at the Rochester Institute of Technology; he holds a master of science degree from the U. of Maryland in management and is director of sales and marketing, industrial microscopes, at Olympus America Inc., Scientific Equipment Group, 3500 Corporate Parkway, Center Valley, PA, USA; ph.: 484-896-5300; e-mail firstname.lastname@example.org.