Feature: Novel Deposition System Designs For Thin Film Materials Research
Next generation vacuum deposition systems must evolve in order to keep pace with the ongoing evolution of thin film materials and devices. Researchers seeking to pursue new areas, such as biomedical devices, 2D materials, specialized magnetics and oxide-based films need new tools to support their work. The frontiers of materials science, particularly at the intersection of biology and thin film deposition, have brought new materials into the vacuum space that were never intended to be there (Ref. 1).
Novel new system designs must also be based on thoughtful understanding of operation and maintenance functions as well as integration issues that affect system access in clean room and chase environments. Many existing NanoFabs are already filled to capacity so that an additional tool must also have the abilities of a contortionist.
With a fixed physical footprint, enhancements to vacuum performance, such as reduced pump-down times and improved ultimate base pressures, may be the only available route to increased capacity in today's modern NanoFab. Improved vacuum system performance is also key to achieving the ever more precise thin film line widths, nanometer-scale thicknesses, and a minimization of defects required for most next generation devices.
Re-thinking vacuum system design for flexibility and performance
For several years engineers at the Kurt J. Lesker Company; have had the opportunity to interact with a global network of expert multi-user facilities managers and principal investigators. This long term dialog has provided the inputs for a novel new system platform that meets many of the great challenges of NanoFab facility managers and their clients.
Figure 1. Technicians at Penn State University's Nanofab Facility
The premium cost of clean room space
The cost of building new clean room space has increased rapidly over the last couple of decades. The new build cost for class 100 clean room space was estimated at $894 "“ 1,100 per square foot back in 2013. By the end of 2016 those costs are expected to increase another 10 percent. In its "2015 lab construction outlook,"Laboratory Design predicted that R&D build costs have increased 3 percent - 4 percent per year since 2013 and expect that trend to continue through 2016 (Ref. 2). With the cost and limited availability of clean room space combined with continuous pressure to increase capabilities and tool density, managers and facilities designers have to be mindful of the impact on NanoFab infrastructure from new tools.
Maximum user interaction in tool and process design
While expectations of users in thin film materials and device development tends toward complete customization for each new tool, this often leads to considerable investments in design and poor understanding of the functionality (including ergonomics), of the system which may result in unintended process outcomes. Transition to a more modular design enables more direct interaction of end users in the process of system specification and configuration. By utilizing a common platform, the wealth of experience developed through feedback from a large community of thin film researchers is readily incorporated into each new system (Figs. 2,3)
Figure 2. Two configurations for clean room-integrated PVD systems: system (a) includes clean room accessible user interface and load lock, system (b) also includes clean room access to deposition chamber. Access to the balance of the system is through the chase.
Figure 3. Alternative system configurations which optimize space utilization in the chase of a clean room where L = Load Lock, C = Vacuum Chamber and Ele = Electrical panel.
Optimized pump-down performance achieved through improved conductance
Lessons learned about the interplay between the passive and active components of a vacuum system has led to the optimization of factors like conductance, which has dramatically improved system performance factors such as shorter times-to-pressure and also lower attainable base pressures.
Figure 4. Vacuum system performance can be substantially improved by eliminating un-necessary piping and increasing conductance using large diameter interfaces between active and passive components
As an example of the performance improvement achievable with compact system design principles, consider a simple system where a 790 litre per second turbo pump is connected to a vacuum chamber with a gate valve and a pumping port. If the gate valve and port have internal diameters of 8" and lengths of 4" and 6.5" respectively the effective pumping speed of the system is about 650 liters per second, enabling a base pressure on the order of 2.24 x 10-7 Torr. If the pumping port is eliminated, the effective pumping speed increases 11 percent to about 730 liters per second and the base pressure improves to 2 x 10-7 Torr (Fig. 4; Ref. 3 "“ Dushman and Lafferty, 1962).
For a system equipped with a 1,500 liter per second cryopump the effective pumping speed increases 22 percent from 1,065 liters per second to 1,300 and the base pressure of a properly conditioned system improves from 6.5 x 10-8 to 4 x 10-8.
Pump-down curves comparing the performance of conventional system designs with the new modular platform suggest that the optimization of conductance between the chamber and the high vacuum pumps (cryopump, in this case) has improved both the time-to-pressure and ultimate pressure of the new system (Fig. 5).
Figure 5. Effect of optimizing conductance on pump down times and ultimate base pressure on a system using a cryopump for high vacuum
Similar performance improvements are seen when comparing systems utilizing turbomolecular pumps to achieve high vacuum. While the performance improvement is not quite as dramatic, the intrinsic benefits of system optimization are obvious (Fig. 6).
Figure 6. Effect of optimizing conductance on pump down times and ultimate base pressure on a system using a turbomolecular pump for high vacuum
Maximum user-enabled tool customization with minimum down time
Faced with continual pressure to expand the number of deposition materials supported by a tool set, equipment managers must be on the look-out for deposition systems that can easily be re-configured by end users on-site. Different materials require different deposition techniques, ranging from thermal evaporation, e-beam evaporation, and sputtering by DC and RF. In this highly modular system design, five out of the six sides of the box chamber are user configurable for ease of customization enabling the addition of alternative deposition processes, as required (Fig. 7). Because of this highly adaptive approach, a chamber can be easily re-purposed in the field to add or remove tools and capabilities as required. For example, a thermal evaporation system can be transformed into an e-beam or sputter system, or sputter capability could be added to an existing thin film deposition tool dedicated to evaporation. New sides can be engineered and manufactured off-site and then installed in a system in-situ, eliminating the need to remove an entire vacuum system and transport it to a fabricator so modifications can be made.
Figure 7. (a) Exploded view of modular chamber design detailing individual sidewalls, (b) assembled chamber
While the least expensive and least complicated deposition technique for metals and some oxide films is thermal evaporation, many materials, such as Ta, W, and Pt, require much more energetic techniques (Fig. 8). E-beam evaporation is great for most elemental metals and some simple metal oxides, but it is not good for alloys. Techniques such as sputtering are much more accommodating for metal alloys such as NiCr and mixed metal oxides such as LiCoO2, YBa2Cu3O7-x, and ITO.
Figure 8. A system platform configured for four thermal evaporation sources. The modular base plate enables better segregation of individual evaporation sources to minimize cross contamination
Sputtering has proven to be a very flexible deposition technique which enables the fabrication of electrically conductive films using a DC power supply, and insulators, using RF. Alternative microstructures can also be obtained using pulsed DC and HiPIMs power supplies. In addition to the broad spectrum of materials that can be sputtered, advanced materials discovery techniques, like combinatorial synthesis utilizing systems designed for Combinatorial Magnetron Sputtering, have been employed to deposit arrays of several thousand distinct stoichiometries on one substrate during a single deposition. In recent work by researchers at Cal Tech, a six cathode sputter system was demonstrated which produced arrays of 5,000 distinct compounds through the optimization of the cathode-to-substrate angle of incidence and throw distance (Ref. 4).
Figure 9. (a) Interchangeable deposition platform outfitted with six sputter cathodes and close-mount domed shutters; (b) side view of platform with 3" Mag-Keeper cathodes
Conceivably, a multi-user system, which has to support a wide range of deposition materials, could have several deposition platforms (base plates) for thermal evaporation, e-beam evaporation, and sputtering by DC or RF, that could be retrofitted to a common system in the course of a few hours (Figs. 9,10).
Figure 10. Example of a quick-change sputter cathode system that enables tool change from inside the chamber with (a) a cathode installed in base plate; (b) detail of holding screws; and (c) cathode disconnected
Impact of stray deposition material on chamber wall and deposition sources
Cross contamination is always an issue in multi-user facilities. In order to support a reasonable pallet of materials NanoFab operators must be able to clean their chambers and deposition sources during change-overs and maintain reasonable pump-down performance in order to insure the quality of films produced. Traditionally, rigid metal shielding has been employed to protect chamber walls and vacuum grade aluminum foil is available to wrap unused deposition sources and chamber components that are likely to be contaminated when not in use. This approach is not particularly efficient with respect to the multi-stage process that is required to utilize reusable shielding.
An unclean chamber, or one that is not vacuum ready, will exhibit long pump down times and poor ultimate base pressure. Materials deposited on the chamber walls and internal system components will likely be coated with materials that may be gettering water, such as titanium metal and some oxides. Also, the surface area of the system will be substantially higher than on a new, or cleaned system, creating many more resident sites for adsorbed water and other gasses which may result in an unacceptable gas load. Add to that a few stray fingerprints and the performance of the system is substantially compromised (Fig. 11).
Figure 11. Examples of pump down curves for unconditioned and vacuum ready (conditioned) systems.
In order to facilitate better over-all chamber cleanliness a new, disposable, shielding system has been developed for the modular system. This approach, using wire frames covered in user-configurable vacuum grade aluminum foil enables quick change-over by eliminating the need to ablate, wash, rinse and heat treat chamber walls and internal components (Fig. 12).
Figure 12. Example of a wire frame, covered in vacuum grade aluminum foil, designed to protect the pumping port side of the modular chamber.
These undesirable deposits are also on every feed through, weld, optical port, substrate holder, and other chamber features, making the cleaning process very intensive. So for users trying to reach the extremes of high vacuum, cleaning may entail taking every component out of the system for a thorough ablating of all surfaces exposed to vacuum.
In general, to achieve optimal performance, the internal surfaces of a vacuum chamber need to be sufficiently clean so that the gas load to the high vacuum pump is similar to the established outgassing rates of the materials of its construction. For example, unbaked 304L stainless steel can have an outgassing rate on the order of 2 x 10-8 to 5 x 10-12 Torr.liters/cm2.sec. For 6061 aluminum, an acceptable outgassing rate is on the order of 5 x 10-11 Torr.liters/cm2.sec. The disposable shielding approach enables a higher percentage of the system's internal features to be protected against stray deposition materials and the use of vacuum grade aluminum foils insures that much of the system can be kept in pristine, vacuum-ready, condition.
By intentionally accumulating stray deposition material onto the shields it is also possible to recover that otherwise waste material for reuse or recycling. In the case of precious metals such as gold, platinum, ruthenium, and others, it is possible to reclaim those metals and minimize the materials budget for the tool.
The recovery of stray precious metals can yield the facility a significant return on their investment in precious metals. With gold currently at $43.00 per gram, and a shield set (of six) that is 50 x 50 cm on a side, a one micron thick coating on the shields will weigh 28 grams and be worth more than $850 to the tool owner after the reclamation costs of the metal are deducted (Ref. 5).
Need for intuitive user interfaces for safety, process control, troubleshooting, and accountability
Multi-user facilities have the problem of high turnover among users (they graduate!) and limited resources to completely train new users. Information is often passed along in a tribal tradition where the older students are obliged to teach their craft to the youngsters. This potential dilution of process information introduces a lot of risk into a NanoFab and may substantially slow the pace of research. Complex deposition tools can be easily damaged, processes can fail to yield positive results, and people can get hurt. Some NanoFabs may also choose to maintain their own tools without the benefit of expensive service contracts, which greatly increases the importance of high quality phone and internet-based support from equipment manufacturers. All of this begs for a shift in how system software is designed and how the system communicates remotely. Future integrated system interfaces need to robust, intuitive, and address the many complicated parameters that need control in order to produce quality vacuum and deposition results. To address the needs of the multi-user facility, an enhanced user interface should enable less skilled tool users to be able to run the equipment with significantly less training. Users, through a log-in function, can be given access to the tools commensurate with their level of training. This insures that process recipes cannot be unreasonably altered, that new process recipes are built by fully trained users, and that all safety protocols are honored. The system should also provide automatic data logging for all steps in a deposition recipe.
The goal of this enhanced design approach is to provide an intuitive experience for the less demanding user. Designers of vending machines have been able to deliver a highly intuitive user interface such that anyone can walk up to a new CokeÂ® soft drink machine, select their beverage, flavor it with a selection of options, and dispense the beverage of their choice without needing to be trained on the machine. When a novice user can do the same thing with a thin film deposition tool, the goal of a completely intuitive user interface has been achieved.
Vacuum deposition systems for multi-user facilities have to be easily adaptable for new materials and processes while being intuitive to use and practical to maintain. Modular platform designs which encourage and enable maximum user interaction in the design and specification while minimizing the impact of system integration within high cost clean space will help make future tool additions to nanofabrication facilities easier to justify and accomplish. Forward looking system design approaches which anticipate future (unknown) adaptations will extend the useful lifetime of new deposition tools and better support the future of thin film research.