Solar

Solar cell manufacturing and robot automation: Right fit, right robot

How do robot automation types and kinematics fit each unique solar cell manufacturing process? Which areas offer the greatest return opportunities? Rush LaSelle from Adept Technology discusses how the solar industry can best maximize factory throughput, drive down costs, and improve efficiencies with robotic automation.

by Rush LaSelle, Adept Technology Inc.

May 5, 2010 – The US has set 2015 as a goal to reach grid parity (i.e., where solar electricity is equal to grid electricity) while other nations predict reaching it as soon as 2010. No matter your thoughts on regulatory involvement, it is clear there will be a resurgence in investment, development, and innovation within the PV manufacturing community throughout the world — and it will largely be driven by technology.

Finding the most effective tools and processes to gain more productivity and decrease costs within a set capital plan is paramount. While the significance of robot automation in the manufacturing of solar cells is obvious, which robot types and kinematics fit each unique process may not. Which solar manufacturing areas offer the greatest return opportunities for robotic automation? Which robot type is best for a given solar application task and how does vision fit it? This article is a primer targeting these issues and discusses how the solar industry can best maximize factory throughput, drive down costs, and improve efficiencies with robotic automation.

Robotic automation’s impact

Robots in the photovoltaic manufacturing process are important due to their ability to significantly reduce costs while continuing to increase their attractiveness compared to manual labor. Richard Swanson, CTO of SunPower Corporation a leading manufacturer of solar technology, framed automation’s impact in a very interesting light by discussing the economies of PV manufacturing in terms of labor. He explained that to produce one gigawatt of solar power it requires 250 to 500 laborers to produce polysilicone, 250 to 500 laborers to process ingots, 3000-6000 people to manufacture the cells, 1500-3000 for the panel lamination and associated applications and 2500-5000 for the solar system integration. In total that’s 8000-16000 laborers required to produce 1GW of photovoltaic capacity. Therefore, to produce 500GWs of solar power per year that equates to roughly 4 million people who could be adding tremendously more value in other capacities. With more automation, inclusive of appropriately applied robotics, the solar industry can cut that labor to 1 million people realizing a 75% savings in direct labor costs alone. Given this magnitude it is critical robots receive ample consideration in line design.

Click to Enlarge
Figure 1. Average robot process relative to labor compensation. SOURCE: R. Swanson/SunPower



Considerations when selecting a kinematic solution

A handful of considerations will provide good direction in selecting the correct robot. First and foremost, what is the payload requirement for the robot? Frequently people only consider the products that are being handled. However it is important to also consider the tooling solution or end of arm tool (EOAT).

Evaluating the motion requirements is also critical. Not only the simple motion of picking and placing, but also what interferences exist between the robot, its linkages, as well as other items that may be in dynamic motion within the cell.

Consideration must also be given to how parts are produced and throughput requirements. Repeatability is also an important factor and it should be understood that robot manufacturers tend to speak in terms of repeatability, while engineers and designers tend to look at it from the standpoint of accuracy. A robot’s repeatability outlines the machine’s ability, once taught, to return to that taught position. Accuracy references the ability to input a given location digitally and have the robot move to that point in space “accurately.” Accuracy, therefore, encompasses offsets and other digitally inputted motion parameters and often varies within a given mechanical unit’s work envelope. Thus, a good understanding of a process’s requirements in combination with the capabilities of a given robotic solution requires careful evaluation.

Other concerns when selecting/evaluating a robot include special environmental requirements, particulate generation that could cause product degradation, and the need for protection from process-specific elements such as slurry ingot processing.

Major robot types

Robot kinematics can be divided into four major categories: Cartesian, SCARA (selective compliance assembly robot arm), articulated and delta/parallel.

Cartesian. The Cartesian kinematic solution is highly configurable as the platform includes everything from a single degree of freedom or unidirectional travel, to numerous axes of motion. Given the simplicity of this kinematic, adjusting strokes or lengths and configuration is relatively easy when compared to this model’s counterparts. Multiple drive trains exist that are optimized to provide high throughput or precise motion as characterized by whether the drive might be a ball screw or a belt driven mechanism. Platforms that accommodate small part assembly to extremely large part transfer such as overhead cranes that might be observed overhead in a manufacturing facility are also available.

Cartesian solutions have numerous applications within the PV industry. They can be applied to both small and large workspaces. Cartesian robots are typically called upon to serve applications where the substrate remains in the same plane. This is to say, that if you were to pick a product off a table or a conveyor, it does not need to be flipped or change its configuration other than a rotation in the same plane as the table or conveyor (X-Y plane). An example of a job using a small Cartesian robot might be dispensing sealing material on the flange of a junction box. The sorting and placement of solar cells in a large rectangular is also an optimal application for a Cartesian solution. Solar cell sorting into multiple stacks in a large work area and processes such as stringing up and lay up within a large cubic area where robots are required to reach with good repeatability are optimum applications for a Cartesian robot.

SCARA robots. The next robot is the selective compliance assembly robot arm (SCARA) robot. It offers a cylindrical work envelope and this category of robot typically provides higher speeds for picking, placing and handling processes when compared to Cartesian and articulated robotic solutions. They also deliver greater repeatability by offering positional capabilities that are superior in many cases than those of articulated arms. This class of robot is usually used for lighter payloads in the sub-10kg category for applications such as assembly, packaging and material handling.

Within solar manufacturing processes, these robots are best suited for high speed and high repeatability handling of cells in smaller workspaces. Where the workspace is constrained sufficiently, the SCARA is an excellent selection. For example, junction box handling and assembly of panels are good applications for this robot group. Stringing is a process that with its increasingly tight tolerances is unmanageable with manual labor. As wafers migrate to thicknesses of 150μm and thinner, the propensity for damage is greatest when labor is applied. As wafer thicknesses decreases over time as forecasted, the thermal expansion of the silicon will also become an issue while soldering. So it’s going to become increasingly important to maintain yields in stringing by controlling and automating the soldering operations even in low cost labor markets through the use of mechanisms such as SCARA robots.

Articulated robots. Articulated robots have a spherical work envelope. These arms offer the greatest level of flexibility due to their articulation and increased numbers of degrees of freedom (DOF). This is the largest segment of robots available on the market and therefore offers a very wide range of solutions from tabletops to very large 1000kg plus solutions. Articulated robots are frequently applied to process intensive applications where they can utilize their full articulation and dexterity for applications such as welding, painting, dispensing, loading, assembly and material handling.

Click to Enlarge
Figure 2. An example of an articulated robot.



Articulated robots are applied to a wide variety of solar applications. Examples include handling heavy silicon ingots which are also in an area where the robots might require industrial protection, and handling wafer cassettes where the orientation of the carrier might differ from pick to place utilizing the full dexterity of the robot. Handling glass, sub assemblies and assemblies where the products are introduced to the cell in a different configuration than they are presented to the system again take advantage of an articulated arm’s flexibility. This is to say that articulated robots permit the optimum introduction of product into a cell which may be in a vertical orientation to maximize floor space while the assembly process is most efficient in a horizontal orientation. Edge trimming and module assembly where tool change and other process considerations dictate the use of articulated arms is yet another use of this class of robots within the PV manufacturing process.

Delta/parallel robots. Parallel robots round out the fourth classification of robots. This kinematic solution provides a cylindrical work envelope and is most frequently applied to applications where the product again remains in the same plane from pick to place. The design utilizes a parallelogram and produces three purely translational degrees of freedom driving the requirement to work within the same plane. Base mounted motors and low mass links allow for exceptionally fast accelerations and therefore greater throughput when compared to their peer groups. The robot is an overhead mounted solution which maximizes its access but also minimizes footprint. These units are designed for high-speed handling of lightweight products and offer lower maintenance due to the elimination of cable harnesses and cyclical loading.

Click to Enlarge
Figure 3. An example of a delta/parallel robot.



Parallel robots are deployed into many solar cell processing steps. Again they offer high-speed transfer of solar cells through manufacturer lines and a multitude of processes. Three examples are diffusion of process equipment, wet benches and PECVD anti-reflective coating machines. In these applications the tables and trays have large placement opportunities which could be equally serviced by a Cartesian however the parallel robot out performs the Cartesian from a throughput standpoint. The Quattro parallel linked product from Adept Technology, Inc recently achieved 300 cycles per minute illustrating the capabilities for this class of machine to handle products at high rates.

Robot deployment in solar PV process flow

The diagram below shows typical PV process steps. The steps are broken into four basic groups where high concentrations of robots are deployed. The ingot-processing step predominantly uses Cartesian gantries and large articulated arms due to the requirement for heavier payloads and large workspace optimization. Wafer manufacturing uses a variety of arm types depending on volume and process requirements. Cell processing tends to use gantries, SCARAs and parallel linked robots and the decision usually lies with the reach and repeatability considerations. Module build uses a variety of arms with a high concentration of articulated and Cartesian arms for reach and flexibility, but some specific tasks utilize the services of SCARAs and parallel robots.

Click to Enlarge
Figure 4. System build process flow.



Robot comparison

The following sections provide a comparison of four robot categories that can be used in an anti-reflective coating load/unload process. A Cartesian robot it is optimized from a reach standpoint, however, the majority of solutions here would prove too slow and would require in excess of a single head EOAT. This complication would drive the need for pre-alignment and further complications in pre-conditioning the product, and therefore, a Cartesian solution would be considered less flexible.

Cartesian robots. These robots are too slow for loading/unloading using a single-head EOAT. However, because multi-head EOAT is often used, cells require pre-alignment. These robots are also less flexible when reconfiguring for different size wafers is required.

SCARA robots. These robots are faster and more flexible than Cartesian robots when used with vision guidance. Table-mounted versions, however, could limit work space and multiple robots may required to cover pallet/matrix. Articulated robots would be pedestal-mounted and may prove too slow in increasing complexity of the installation.

Articulated robots. These robots are too slow for loading/unloading with single-head EOAT and the spherical work envelope isn’t ideal for covering pallet/matrix. Therefore, a delta or parallel style robot might be optimal for a number of reasons. First, the overhead mount is ideal in reducing the footprint of the automation cell — all places on the PECVD pallets can be reached. And when we combine the benefits of the delta with vision, it provides an exceedingly flexible solution that will meet the throughput requirements. As noted below, vision is an enabler not only for parallel linked robots, but also provides the same benefits to all categories of robots.

Delta/parallel robots. The overhead mount design is ideal for loading/unloading equipment because larger delta robots can cover the width of most PECVD pallets. When used with vision guidance, larger delta robots enable extremely good positioning, excellent flexibility, and are quickly reconfigurable. Additionally, because of their light weight, these robot designs are optimal for handling cells at high speeds.

Flexibility with vision

Vision has become a highly adopted tool to improve the productivity of robot automation in all industries and all facets of placement. Vision systems offer tremendous flexibility for applications that don’t require fixtures or trays for part location. Vision-guidance is a feature that allows the vision system to take a picture and compute a part’s location and orientation and guide the robot to the part using a computed robot-to-camera transformation obtained through an automated calibration process.

Additionally, vision systems allow tremendous flexibility and cost-savings because parts don’t have to be fixtured. Parts can be randomly presented to the robot without pre-orientation or alignment or put into a tray which also reduces cost. These systems frequently incorporate line tracking, which enables the robot to pick these parts from a moving belt that further optimizes the production process. Robot-integrated vision allows inspection to be incorporated into the handling process and this puts the inspection or quality control process in parallel with handling, further reducing the overall cycle time and increasing throughput.

Different part geometries only require vision re-training or the selection of a recipe instead of manual changes in fixtures and tooling, which increases the overall lifetime profit of the equipment by virtue of its optimization and improved throughput. Most robot manufacturers offer packages with multiple cameras and tracking solutions for integration into a single cell, which offers tremendous power and flexibility for solar manufacturing.

Conclusion

The common goal for solar manufacturers is to drive down the cost per watt. As the solar industry strives to achieve grid parity, manufacturers need to be knowledgeable about modern robotics and automation technologies, as well as how they contribute to reducing the cost of solar cells.

History has shown that automation has played a significant role in reducing manufacturing costs in many industries, and when the costs associated with higher quality and yields are considered, the benefits of automation offer an even more appealing value proposition. While robotics and automation may be viewed by some industries as mature technologies, industry leaders are continuing to develop innovative products and new technologies that are ideal for solar manufacturing processes. It would be prudent for solar manufacturers to look outside of their industry for the best practices in high-volume manufacturing with automation and robotics to achieve their cost reduction initiatives.

Biography

Rush LaSelle received his BS in mechanical engineering from Santa Clara U. and his MBA from UC Berkeley’s Haas School of Business. He is director of worldwide sales and marketing at Adept Technology Inc., 5960 Inglewood Dr., Pleasanton, CA 94588 USA; ph.: 925-245-3400; e-mail [email protected]; www.adept.com.