Investment in Midsize Facilities
The primary investors in midsize materials research facilities are the government agencies (the National Science Foundation [NSF] and the Department of Energy [DOE]) and the universities themselves—the latter through start-up packages for new faculty and provision of matching support when needed. Other investors include the Department of Defense (DOD), some foundations (the Keck Foundation, for example), and state governments, especially in state-affiliated schools. It should be noted that the DOD is a primary supporter of domestic materials research, but the majority of the awards that it makes are to single investigators. The DOD does not, in general, establish or specifically support midsize materials research facilities.1
In this chapter, the committee comments on the programs of support for midsize materials research facilities. Selected specific programs are described in more detail in Appendix F. Investments include the initial capital expense of equipment, the building or modifying of suitable space for installation, and yearly operating expenses (staff, maintenance, consumables, and so on).
GENERAL SCOPE OF SUPPORT
The Division of Materials Research (DMR) at NSF is by far the major investor in the smaller and midsize facilities used in materials research at universities. DMR participates in several programs that provide most of the equipment. Two programs—Major Research Instrumentation (MRI) and Instrumentation for Materials Research (IMR)—only fund equipment purchases and, until recently, typically required a 30 percent match (now reduced to zero) from nonfederal sources. Some NSF/DMR funding of equipment is also available in center programs, such as the Materials Research Science and Engineering Centers (MRSECs). Finally, a small proportion of individual-investigator grant funding is used to purchase less-expensive equipment. Generally, NSF does not provide direct support for yearly operations expenses, expecting that these funds will derive from user fees obtained from other sources, such as individual-investigator grants, center grants, or university subsidies.
The DOE typically purchases major equipment for use at its own facilities, only a few of which are at or associated with a university (the University of California at Berkeley, the University of Illinois, Iowa State University, Kansas State University). There is no general equipment program at the DOE that supports the purchase of equipment for central facilities at non-DOE-affiliated universities. However, the budgets for operations of DOE-funded facilities are included in the yearly DOE budgets. Consequently, user fees are typically very small to nonexistent at DOE facilities.
In distributing its facilities questionnaires, the committee identified more than 270 candidate facilities. (See Appendix C for a list of these facilities.) While many of the facilities fit the definition of midsize materials research facilities, not all did. In the summer of 2003, the committee visited nearly 50 facilities in five regions of the country (these facilities are listed in Appendix D). On the basis of responses to its questionnaires, the site visits, a review of the awards issued by NSF, discussions at the committee’s town hall meetings, and the committee’s own best judgments, it seems likely that there are on the order of 500 facilities nationwide that provide essential instrumentation support for materials research. This estimate is perhaps generous to a certain degree, but it provides a good basis from which to understand the role of midsize facilities in the overall enterprise. If every such facility chose to purchase the next generation of instrumentation, it would well exceed any plausible budget scenario.
Based on the survey responses, the total capital investment (purchase price) in the 56 responding midsize facilities is $529 million (average = $10.2 million, median = $6.8 million). The annual operations budgets average $1.1 million per year, with a median of $0.51 million per year. The current annual investment in
new equipment for these facilities is also difficult to estimate, since the investment by universities, foundations, and states is not compiled in any one source.2 A conservative estimate would put the total capital investment in these midsize facilities at several billion dollars, with operations budgets totaling several hundred million dollars per year.
National Science Foundation
Although the NSF’s funding programs are not a conclusive indicator, an examination of their internal consistency provides some insights about the features of federal support for midsize facilities. DMR had an annual budget near $250 million in 2003.3 This budget supported individual investigators (53 percent), centers (24 percent), large facilities (15 percent),4 and focused research groups (8 percent). Centers include the MRSECs and the NanoScience and Engineering Centers. Since there is no explicit midsize facilities program within DMR, the focus here is on support for instrumentation. In addition to the IMR program, DMR also, more recently, initiated the Instrumentation for Materials Research—Midscale Instrumentation Program (IMR-MIP). Some statistics about the program follow:
In FY 2003, the IMR program supported about $6 million in awards.
The IMR-MIP was expected to award up to a total of $3.5 million across 3 to 4 awards in FY 2004.
In general, about 12 percent of the DMR budget is directed toward capital equipment purchases ($30 million in FY 2003).
Instrumentation for materials research can also be supported at NSF by so-called allied funding. For instance, some equipment that is used for materials research as
well as for other sciences and engineering is purchased through programs in divisions of NSF other than DMR. For example, nuclear magnetic resonance (NMR) instruments are used to characterize many molecular substances as well as polymers. The latter topic is generally considered materials science. Thus, NMR apparatus funded by the Division of Chemistry (CHE) at NSF may also support some materials research activities. Indeed, some equipment funding is done jointly between divisions. CHE also participates in the Major Research Instrumentation program at levels similar to DMR’s participation in the program.5 The committee’s crude estimate is that perhaps as much as 20 percent of the CHE MRI program may be considered as supporting materials research.
Another example of allied funding is through agency-wide programs, such as the Major Research Instrumentation program. The MRI program assists institutions in the acquisition or development of major research instrumentation that is, in general, too costly for support through other NSF programs. About 10 percent of the annually awarded MRI funds are directed through DMR, about $10 million in FY 2003. The maintenance and technical support associated with these instruments are also supported during the period of the award (usually less than 3 years), although typically more than 90 percent of each award goes directly toward capital costs. Figure 4.1 shows recent trends in support for the equipment portion of the DMR budget and the DMR-captured MRI awards. The average size of an MRI award within DMR in FY 2003 was $235,000 per year, with an average duration of less than 2 years; there are typically only several awards each year made for more than $1 million (including matching funds) through DMR.
In the late 1990s, there was widespread recognition at NSF that investment in major instrumentation was insufficient. Increases in the MRI budget are reflected in the increases in DMR spending. Thus, the NSF annual investment in instrumentation for materials research, including the required matching of 30 percent from nonfederal sources, can be estimated at nearly $40 million per year in 2003—up from about $30 million per year in 1999. As discussed above, the capital equipment investment in the responding 56 midsize facilities is $529 million (purchase price). The committee did not attempt to determine the average age of the equipment in these facilities, but based on the collective experience of the committee at the members’ home institutions and on observations made during the site visits, the average age is over 10 years. Using the average U.S. inflation rates over the past 10 years (about 23 percent, a period of historically low inflation), the committee estimates the replacement cost of that equipment to be about $650 million. It is
not clear if this inflation rate is applicable, since exact models of the older equipment are no longer available. Generally, newer equipment also has capabilities that are enhanced or even entirely new. However, for present purposes, the exact rate is not an essential component of discussing future challenges. If the committee’s estimates of the total cost of all equipment in all 500 or so materials-related facilities are reasonable, the total replacement cost is a factor of 2 to 4 larger (about $1 billion to $2 billion).
Department of Energy
DOE’s materials research programs are managed by the Office of Basic Energy Sciences (BES). BES does not have a specific program for funding instrumentation at universities. Support for acquisition of some instrumentation is provided as part of the normal grant process; in general, it is a small amount, although major equipment can be requested if justified by and required for the proposed research. At the national laboratories, however, the DOE provides some support for instrumentation purchases in conjunction with research program funding.
While the DOE does not have a targeted program to fund significant equipment purchases in non-DOE-affiliated laboratories, it plays an indirect but important role in supporting the operation of many facilities in such venues by providing support for user fees. In 2003, BES provided support for about $175 million worth of single-investigator research at universities.6 This research is largely materials-related, and the “Materials and Supplies” portion of the single-investigator award is not significantly different from those at NSF. These budgets are in part used to pay user fees at midsize facilities. Other agencies that directly support materials research, such as the DOD and the National Aeronautics and Space Administration (NASA), also support facilities through the provision of user fees. Clearly, a relatively modest, consistent investment by DOE in a university-based facility has had significant positive impact, however, such as at the Center for Microanalysis of Materials, Frederick Seitz Materials Research Laboratory, at the University of Illinois at Urbana-Champaign.
DOE’s BES also directly operates and fully supports a suite of smaller and midsize facilities—including four existing electron-beam microcharacterization centers, as well as five nanoscale science research centers (transitioning from construction to operations in FY 2006)—at which there are no user fees for nonproprietary work. The electron-beam characterization centers are well established.
Their status was reviewed in 2000 in a report by the DOE Basic Energy Sciences Advisory Committee.7 In addition, facilities for materials synthesis are being created, such as the Molecular Foundry at the Lawrence Berkeley National Laboratory (LBNL), which should become available in the next year or so.
Construction of the new nanoscience centers is a $350 million DOE initiative to construct and equip five regional facilities to operate in the manner of a large facility. Each center is expected to have an operating budget of $15 million to $20 million per year and is designed to take full advantage of the significant national laboratory infrastructure in which it is embedded. The technical and conventional construction projects are managed much as a large facility would be planned and reviewed. The process involves multiyear planning, formal management plans, four critical design reviews by professional staff, detailed budgets, standardized program management and accounting, and so on.
Future access to these facilities by external users will be based on the scientific merit of user submissions, evaluated during a cyclic, formal proposal process. As in the case of synchrotron users, there will be no charge for access to the facilities and expert staff. Therefore, the long-term user support and maintenance of the equipment will be highly dependent on the constancy of the operations budget from the DOE, which is both owner and operator. The next few years will provide valuable new lessons as these facilities complete their construction and develop a user community.
The national laboratories are generally well stewarded, both in terms of instrumentation and long-term staff, especially when compared with university-based facilities. Some are close to universities and industrial complexes (e.g., LBNL and the Lawrence Livermore National Laboratory in the San Francisco Bay Area, and the Argonne National Laboratory [ANL] in the Chicago area). Others are more remote (e.g., the Oak Ridge National Laboratory [ORNL] in Tennessee and the Ames National Laboratory at Iowa State University). The committee’s observations suggest that these laboratories play an important role in the nation’s research infrastructure but that they cannot, because of their mission and limited number, provide more than a nucleus to address current national needs. Nonetheless, there are major opportunities for significantly enhancing their connectivity to universities and industry.
For instance, the committee’s site visit to LBNL’s National Center for Electron Microscopy (NCEM) revealed that it was a well-run and successful facility, serving
Basic Energy Sciences Advisory Committee, Review of the Electron Beam Characterization Centers, Washington, D.C.: Department of Energy, 2000. Available online at http://www.sc.doe.gov/bes/BESAC/e-beamreport.pdf; last accessed June 1, 2005.
a large number of users, the majority of whom came from outside the immediate geographic area. However, NCEM was unable to obtain capital funding for a much-needed “workhorse” instrument but was able to secure support for a “racehorse” instrument. A workhorse instrument at the nearby University of California at Berkeley campus had been installed, but access to the broader community was limited. Clearly, rather straightforward arrangements could be made to adopt a hub-and-spoke model among both of these instruments, thereby serving both constituencies well.
Likewise, the excellent Shared Research Equipment (SHaRE) User Facility and Program at ORNL allows students from various locations in the country to work with experienced researchers on advanced transmission electron microscopes (TEMs) to obtain high-quality advanced data. However, because of the limited time available on the instruments, the overall number of users served this way is very constrained. Budget constraints have also forced the travel support to be eliminated from this program. The parallel electron microscope facilities at ORNL are generally open to multiple users as midsize facilities; they operate either in a mission-oriented fashion or at a higher level of research capability. Accordingly, these excellent facilities can only be made available to a limited number of experienced researchers within the time commitments and accessibility of a national laboratory. Thus, while providing much-needed facilities for excellence in research, it is not possible for them to service the broad materials community, which requires immediate and ongoing characterization capability.
No discussion of the DOE’s involvement in midsize facilities and instrumentation would be complete without reference to the Transmission Electron Aberration-corrected Microscope (TEAM) project, undertaken by five electron microscopy centers with support from BES (see Appendix F for background information). The TEAM project concept has been endorsed by a subcommittee of DOE’s Basic Energy Sciences Advisory Committee. Funding at the level of $25 million has been approved for research, development, and testing, and the construction of an initial instrument to be located at LBNL. Data gathered by the committee suggests that several different U.S. institutions are scheduled to receive Cs-aberration-corrected TEMs and monochromated TEMs in the next year or so.
The National Institutes of Health (NIH) has become an important contributor to midsize facilities in particular and has even begun to support several such facilities in areas related to materials research (see Appendix F for further discussion). About 50 biotechnology research resource centers are supported under the P-41
grants program (grants for Biomedical Informatics/Bioinformatics).8 Each of these centers is required to participate in five different types of activities (listed below). It is standard practice for these facilities to “spin off” the service-work user facility component as a stand-alone entity for accounting reasons and institutional preferences. The types of activities are these:
Technological research and development,
Perhaps surprising to some, NASA has become increasingly involved in materials analysis using sophisticated instrumentation. Its recently established program has arisen out of an emphasis on the microanalysis of extraterrestrial samples returned to Earth: the Sample Return Laboratory Instrument and Data Analysis Program (SRLIDAP) at NASA was specifically formed to foster the creation of midsize facilities dedicated to this type of analysis. Funded SRLIDAP projects include an automated aerogel mining device (for recovery of Stardust particles), a next-generation superscanning transmission electron microscope, extremely low-blank rare-gas mass spectrometers, a resonance ionization mass spectrometer, NanoSIMS (high-spatial-resolution secondary ion mass spectrometers), laser scribing (to cut Genesis wafers) focused-ion beam microsample preparation, and synchrotron x-ray fluorescence microanalysis facilities. Although the entire scale of NASA’s investment in facilities-based materials research is small compared with the more traditional agencies’ investments, the surge of public enthusiasm and interest in extraterrestrial sample analysis will continue to drive NASA’s support forward.
The operations budgets of the 56 respondent midsize facilities totaled $66 million per year. On average, these expenses are covered by user fees (35 percent), direct support from federal programs (e.g., center programs) (35 percent), support from the host institutions (27 percent), and a small contribution from state or other sources (3 percent). There is considerable variation in these proportions among
facilities, with some running almost entirely on user fees and others running almost entirely with some form of university support. The patchwork of support at a given facility depends on many local factors, which are often unique to that institution. Conservatively, the operations cost of all materials-related facilities is several hundred million dollars per year. Given the committee’s findings during its site visits, this expenditure is too low for efficient operation. To place it in perspective, this figure is roughly equivalent to the entire DMR budget ($250 million per year in 2003) of NSF.
Some of the needs of midsize facilities have already been recognized by the National Science Board (NSB) in its report Science and Engineering Infrastructure for the 21st Century. For instance, that report states:
While there are special NSF programs for addressing “small” and “large” infrastructure needs, none exist for infrastructure projects costing between millions and tens of millions of dollars. This report cites numerous examples of unfunded midsize infrastructure needs that have long been identified as high priorities. NSF should increase the level of funding for midsize infrastructure, as well as develop new funding mechanisms, as appropriate, to support midsize projects.9
The NSB report focuses on the challenges associated with “chronic underinvestment,” however, and the present report focuses on the opportunities for optimizing the current investment. That being said, however, the committee recognizes and applauds DMR’s response to the NSB report with the introduction of the IMR MIP (see further discussion in Appendix F).
Facilities’ maintenance of their sophisticated instruments includes the cost of replacement parts and maintenance contracts with the manufacturer. Because of tight budgets, many facilities do not have replacement-parts reserves. Often they seek contributions from every conceivable source, but if the contributions are not forthcoming, the instruments sit idle. Grant programs are not a viable recourse for replacement parts, because the application and awards process is much too slow. Maintenance contracts have become very expensive, often exceeding a yearly cost of about 5 to 10 percent of the initial capital cost. In some sense, budgeting for the maintenance costs has been unrealistically low for many years, because 5 to 10 percent is a typical figure for maintenance of capital items in industry and at universities in general. In addition, for many manufacturers the profit margin on equipment sales is low, while on maintenance contracts it is high. This tendency puts considerable pressure on university and government facilities. A large fraction
TABLE 4.1 Key Instruments in Materials Research and the Initial Capital Investment Necessary to Acquire Them
Today’s Purchase Price ($ million)
2.5−5.0 (50 kV vs. 100 kV)
Dual-beam focused-ion beam
Field-emission gun transmission electron microscope (TEM) (fully equipped)
Nanoscale secondary ion mass spectrometer
of facilities take the risk of operating without full or even any maintenance contracts, since they cannot afford them.
These analyses underscore the scope of the budget problem for midsize facilities. Replacement costs for current aging equipment in the 500 or so midsize facilities are estimated to be $1 billion to $2 billion. The management of this enterprise is larger than any one agency’s program in materials research. For instance, NSF support (including matching funds) is a little more than $30 million per year for such equipment. Assuming a standard 10 year depreciation timescale and a goal of a nominal 10 year replacement of the inventory, NSF can only service up to 30 percent of the perceived needs of the existing midsize facilities. The investment needed to replace current equipment is far too great to be realistic—leading to the conclusion that sharing facilities and resources is increasingly necessary for the future.
Table 4.1 gives illustrative costs of some of the impressive new instrument capabilities. Comparing these figures with the average NSF-MRI award in DMR ($0.24 million in 2003),10 it can be appreciated that there is no federal governmental mechanism through which research institutions can obtain these state-of-the-art machines. Only very few would be able to acquire them, with huge demand for their capabilities. Once again, this problem can be addressed to a large extent by the establishment of facility centers open to a community of qualified users, both internal and external. Not every institution can have all or even some of these instruments. However, there is a general belief that they should all expect to have reasonable access to a shared facility for the most groundbreaking research.
The diversity of funding sources for midsize facilities is both a boon and a bane. It allows facilities to develop some autonomy by not being embedded within a single agency, but it also allows them to fall between the cracks of agency
programs. For instance, when a midsize facility (or a significant instrument acquisition) is proposed, no one agency program manager (or even peer review panel) has sufficient expertise or resources to evaluate the proposal against other activities in the same region. It is often up to the proposal author to consider these types of issues. Likewise, because midsize facilities have no clear program steward, they often do not formally close down or release resources when the facilities are no longer effective. (The committee did encounter several facilities which, owing to a lack of operating funds that year, were unable to operate or share equipment with users.) In an era of constrained resources, focusing operations on the most effective facilities is a serious consideration.
In summary, the committee estimates that the midsize materials research facilities enterprise in the United States numbers more than 500 separate facilities and represents a domestic capital investment in excess of $1 billion, with an annual operating cost of more than $100 million. Not only does the scope of this enterprise represent a significant national investment, its management is something that exceeds the capability and resources of any one federal funding agency.
The United States is unique in many respects, including in its scientific enterprise. Compared with its major technological competitors (Japan, Germany, the United Kingdom, France11), it is significantly larger geographically, has a larger population and economy, and is more ethnically diverse. Moreover, no single, overriding paradigm dictates how research is conducted: there are a variety of funding agencies, the possibility of local (state) support, a number of charitable foundations, and an approximately equal proportion of private and state research universities. The problems that the nation faces and needs to address are therefore largely relevant only to the U.S. environment. However, quoting from the National Academies report Experiments in International Benchmarking of US Research Fields: “There continues to be concern among top university researchers that facilities and equipment for materials research in several foreign universities now outclass those at most universities in the United States.”12
Some important features are revealed by considering how these same issues are approached in other countries. Japan has the extremely impressive National Institute for Materials Science in Tsukuba, with about a thousand researchers and a remarkable array of equipment (e.g., over 35 advanced TEMs, including two high-voltage, high-resolution microscopes that no longer exist in the United States after the decommissioning of the NCEM microscope in 2004).13 The Japanese facilities, like those in the major national universities, reside largely in the groups of individual investigators rather than being multiuser operations.
The Max Planck and Fraunhofer Institutes form the backbone of the German research system. Again, each is better equipped than the U.S. research universities but is comparable to the U.S. national laboratories. The committee’s experience here is that the permanent technical support staff is a key component of these facilities, an aspect that is strongly supported by the German education system.
In the United Kingdom, excellent facilities are found at the elite institutions (e.g., Oxford and Cambridge Universities). These are continually upgraded (e.g., Oxford already has an aberration-corrected TEM) and are well supported with technical and scientific staff. However, there tend to be fewer users from outside those institutions.
Many of the midsize facilities in France are supported by the Centre National de la Récherche Scientifique (CNRS). Thus, many of the scientists are permanent government employees themselves (see Box 4.1, “Materials Research Facilities in France,” for more information).
The smaller European countries with on the order of 10 million population each (e.g., Belgium, the Netherlands, Sweden) tend to have a few highly funded, well-supported centers that are national resources and are extensively used by many colleagues on a national and international level. Examples include the high-resolution electron microscope (HREM) laboratory at the Middelheim Campus, University of Antwerp, Belgium; the Dutch National Center for HREM in Delft, Netherlands; the Swedish National Center for HREM at the Lund Institute of Technology; Interuniversity MicroElectronics Center in Leuven, Belgium; and Materials Analysis at Chalmers University of Technology, Goteborg, Sweden. The model of these smaller countries is one to note especially. These are stable, well-funded facilities that serve a large number of users. They are successful because of a combination of recognized need, enthusiastic collaboration, and continued oversight from the government and scientific community. It is also undoubtedly advantageous that these countries are geographically small, so that national facilities are never more than a few hours’ drive away.
The materials research system in France offers an interesting counterpoint to that in the United States. Outside of the corporate laboratories, research in France is performed at universities or national laboratories, which are directed at a specific technology, such as atomic energy (Commissariat à l’Energie Atomique [CEA]), or in a series of laboratories working on a broad base of topics called Centre National de la Récherche Scientifique (CNRS). CNRS is a large, 25,000-employee operation with many sites around the country. The main characteristic of CNRS is that the scientists are government employees who benefit from lifetime employment. There is a management body in charge of the evaluation of research teams, but the management also includes union heads whose main focus is on stability of employment.
In recent years, in an effort to decentralize the entire government, the French government has created 22 regions and has passed on some of the funds and administrative responsibility to these regions. Thus, today the CEA is still centrally administered, but the CNRS laboratories have more and more of their activities locally operated. The university faculties are paid by the Ministry of Education, and the universities are run by a central government body. Occasionally, regional governments fund large equipment investments to give the region a good reputation and to attract executives and high-tech companies. Except for synchrotron research, the committee was not able to identify central facilities for the support of materials research funded by the French national government. However, there are a few regionally funded materials-related facilities established to support universities and industry in that region. There are a microscopy facility in the Grenoble area that supports university and industrial research in that region and the well-known laboratory in Toulouse. Another facility, to be located in an industrial park near Rousset in Provence, is in the planning stages. These are targeted directly at the aerospace and semiconductor industries in the region.
GENERAL COMMENTS ON FEDERAL AGENCY POLICIES
Midsize facilities for materials research do not have a programmatic or thematic home across the agencies or in any particular agency. The lack of a specific program of support is only part of the problem, however. When the lack of a specific program of support is combined with a general lack of communication and coordination among facilities, users, and agencies, issues of a more serious nature can arise.
As described in this chapter and Appendix F, midsize facilities have many sources of partial programmatic support across the federal research agencies. However, as pointed out earlier, the important challenges that midsize facilities face are not well met by any specific program. In particular, there is no strong support for the long-term infrastructure that a midsize facility requires in order to be successful continually. The closest model may indeed be the P-41 centers of NIH or the SHARE Facility component of the NSF MRSEC. For the P-41 centers, the five
explicit components of technological research and development, collaborative research, service, training, and dissemination combine to make a facility that serves the need of a diverse user base and addresses a broad set of research avenues.
During the committee’s site visits, several facility directors characterized the NSF model of support for instrumentation as “fund and forget.” That is, awards for important instrumentation often provided sufficient sums of money to acquire the instrumentation, but resources were typically insufficient to sustain operations and maintenance of the instrument. And, there was typically no follow-up after an award was received, to certify that the instrument was meeting the needs of the researchers and being managed effectively. As many members of the community were concerned by the lack of follow-up as by the insufficient resources.
This lack of stability and predictability of support for research facilities and instrumentation has recently been recognized and discussed by the National Science and Technology Council’s (NSTC’s) Research Business Models (RBM) Subcommittee. This subcommittee has held a series of four workshops in order to identify issues through community input. One of the subcommittee’s early observations about facilitating multidisciplinary research concerned the need for “stability and predictability of support for research facilities and instrumentation independent of individual projects.”14 The RBM Subcommittee will continue to report to the NSTC and to recommend improvements to the federal research process.
Another agency-specific lesson became apparent to the committee: DOE’s BES is a strong supporter of materials research, most notably at the national laboratories. Despite DOE’s significant commitment and investment in research infrastructure and workforce development, BES supports only one materials research facility in the nation that is not housed in a national laboratory.15 The most significant example of a DOE-supported materials research facility based at a university (not affiliated with a national laboratory) is the Center for Microanalysis of Materials (CMM) at the Frederick Seitz Materials Research Laboratory of the University of Illinois at Urbana-Champaign. It is a striking example (see Box 2.1 in Chapter 2). The CMM is nationally famous for its world-class instrumentation, first-rate support staff, and well-maintained facility. Some (non-Illinois) facility directors referred to CMM as “paradise” and a “utopia” for materials microanalysis. For an agency’s single investment in a midsize facility at a university to result in
Research Business Models Subcommittee, “Progress and Next Steps,” Washington, D.C.: National Science and Technology Council, March 16, 2004. Also available online at http://rbm.nih.gov/20040316_RBM_status_report.ppt; last accessed June 1, 2005.
DOE’s BES supports two other midsize facilities based at universities—the Notre Dame Radiation Laboratory at the University of Notre Dame and the James R. McDonald Laboratory for Atomic, Molecular, and Optical Physics at Kansas State University—but they are not focused on materials research.
such high praise is surely remarkable and thought-provoking. Embedded within the Ames National Laboratory at Iowa State University, DOE’s Materials Preparation Center also draws impressive praise.
The committee draws a careful distinction between gaps in federal agency programs and the predisposition of peer review panels—composed primarily of members of the materials research community. The committee notes the important role that perceptions of proposal reviewers and funding agencies play when assessing funding requests related to midsize facilities. Applications for workhorse machines for routine characterization are often not as well received as those proposing racehorse types, yet both are important for the national infrastructure. At NCEM, workhorse machines (basic microscopes) have been almost impossible to obtain because of the facility’s focus on advanced microscopy. NCEM has been able to acquire such instruments only through special arrangements with manufacturers during the purchase of racehorse machines. It was noted that two $0.5 million machines for standard characterization are a lot less exciting to a funding agency and its peer review community than is a single $1 million cuttingedge machine. The Center for High Resolution Electron Microscopy at Arizona State University has experienced similar constraints: a basic microscope was purchased using funds awarded by the federal Department of Education.
This observation bears repeating: A large component of the racehorse-versus-workhorse issue is related to the perceptions of value and impact within the materials research community. For example, many researchers’ perception of impact is strongly tilted toward sophisticated rather than basic instruments. Consequently, the former are valued more highly than the latter by the community. This is reflected in the way that the peer community evaluates research proposals. If the materials research community is to seriously exploit the full potential of midsize facilities, it will need to do its part to change the internal culture.
In a similar vein, it was noted that instrument development is likely to be less well received than is instrument purchase in the peer review proposal system, although NSF explicitly supports some instrumentation development under the MRI program (these projects are in the minority in the list of approved awards). Instrument development at a facility can enhance the working environment greatly, as well as providing remarkable innovations. A balanced portfolio combining advanced with basic equipment, and off-the-shelf instruments with instrument development, is important when judging competing facility requests.
It is clear that the survival of many facilities hinges on the recovery of expenses through user fees. However, whenever a grant is awarded, the negotiation of the award between the agency and the principal investigator involves a degree of cost trimming. Since student and postdoctoral salaries are invariable, the category that often is easiest to prune is support for services (i.e., facility costs). As a conse-
quence, the amount of support for facility usage throughout the duration of the grant is reduced to a level that is not in accord with the research needs. The importance of providing adequate funds to cover facility costs, perhaps as a guideline percentage line item, should not be underestimated.
The committee emphasizes that it is not advocating for or against user fees: this cost-recovery mechanism is simply a business tool that many midsize facilities use to generate operating revenue in order to meet the demands of a realistic budget. Some facilities, such as those embedded within the DOE national laboratories, are able to secure this support directly from the agency and therefore do not employ user fees. Of course, by supplying individual investigators with research awards that include funds for the payment of user fees, federal agencies are inevitably underwriting this part of facility operating budgets.
There are some consequences to these different approaches. In a free market, one would expect users to flock to facilities that do not charge user fees because of the perceived bargain. (For instance, as reported at the CMM, eliminating user fees more than tripled usage.) A significant backpressure is supplied, however, through the problems of oversubscribed resources: time-until-access can exceed several months, or on-site training can be compromised. That is, many users prefer to “pay to do it now rather than wait in line to do it for free.” In an environment in which users are not well informed about their consumer choices, however, these trade-offs are not always realized.
Midsize facilities provide access to and support for relatively expensive equipment. It is common practice among U.S. institutions for investigators to negotiate with vendors as individuals for each single instrument. Different facilities will negotiate for different instruments because of different resources and needs, but when these needs overlap, there is potential for both redundancy and inefficiency. One could imagine that cost savings might accrue from a consolidated equipment purchase, spread among several competing institutions over a period of time. In this scenario, one might imagine identifying a national infrastructure need for 10 focused-ion beam machines over the next 5 years; the net cost to the federal program could be much less if the acquisition was proposed as a consolidated purchase at the national level rather than as a series of one-time negotiations. The machines would not be identically configured, of course. The concept of a national strategy would result in a much stronger (and more highly leveraged) infrastructure. The DOE TEAM project echoes some of this type of thinking, as do the DOE Nanoscale Science Research Centers.
However, the committee’s enthusiasm for this possibility is restrained by the question of the feasibility of such intimate coordination and agreement and the risks involved in being overly prescriptive. For instance, the one-time negotiations that secure instruments for midsize facilities are often dominated by very special
and very local circumstances that allow the investigator unique bargaining power. Generalizing equipment purchases to provide for a national strategy would not, in general, take advantage of these possibilities.
Additionally, the individual and specific needs of different facilities would be challenging to simultaneously identify and coordinate in order to build a competitive national bid. Clearly, though, the nation has identified a need for intense synchrotron radiation sources, and the national light sources are therefore coordinated and connected. Regional facilities (or facilities working together within a region), therefore, could benefit by coordinating their needs for joint acquisition strategies. One example might be the joint purchase of similar instruments from the same vendor by two facilities within a region. Similarly, one facility might work with a larger and more-experienced facility to develop a stronger negotiation with a common vendor. Another possibility might be that of bundled service and maintenance contracts within a region. In general, to the extent that strategic partnerships can be formed with vendors to reduce the costs of acquiring and maintaining equipment, they should be pursued.
THE CASE FOR REGIONAL AND REGIONALLY NETWORKED FACILITIES
The National Nanotechnology Infrastructure Network (NNIN) is evidence that a regionally diverse network of midsize user facilities can be organized, funded, and managed in a way that will make advanced fabrication tools and staff available to hundreds of research programs each year (see Appendix F for details). Furthermore, it testifies to the recognized need for this type of infrastructure. However, the NNIN grew from the smaller National Nanotechnology Users Network, which in turn grew from a single national facility. This evolution spanned more than two decades. With the exciting high-growth path currently occurring in nanofabrication and nanocharacterization, it remains a significant challenge to extract the necessary efficiencies far more rapidly and to organize the materials research infrastructure in a way to meet the demand.
All facilities serve a region of primary users; some facilities serve a national region and are thus national facilities, and some serve an individual investigator’s research group. A “regional facility,” however, is commonly understood to be a facility that serves a relatively large geographic region, perhaps larger than an individual city but smaller than the largest states (somewhat like the smaller nations of Europe).
The primary drivers for regional facilities are improved access to and improved utilization of the resources (instrumentation and personnel). Midsize facilities in materials research are natural candidates for regionalization according to the following analysis:
Appropriate level of investment. The capital and recurring investment costs for midsize facilities are of a scale that can be afforded regionally—that is, dozens of such facilities could be afforded nationally. As a counterexample, synchrotron light sources are simply too expensive to provide each of the 50 states with its own such world-class laboratory.
Sufficient users. On the basis of its site visits, town hall meetings, and questionnaires, the committee believes that a relatively broad and uniform community of users for midsize materials research facilities has been identified.
Untapped potential. As mentioned above, a clear message from users was a lack of knowledge about existing midsize materials research facilities. The committee found that many facilities managers (about 40 percent) indicated that they had resources to serve additional users. High utilization is desirable for maximum scientific output. It helps to generate needed funds for instrument maintenance costs and to justify tool upgrades and replacement. The committee warns against “oversubscribing,” however, since this practice can create long queues and bottlenecks that delay research and is generally unable to absorb or accommodate outages. The committee observed many effective management styles for controlling this issue, including staff ownership of scheduling for certain tools, Web-based reservation systems, e-mail distributions, Web posting of tool status, and so on. The networking model (see below) describes a method of provisioning these available resources to best use.
Networking. The committee identified several examples of facilities that serve relatively large regions, but most impressive were facilities within the same region or in neighboring regions that work together. Such networking between facilities (e.g., between Harvard University and the Massachusetts Institute of Technology, or between the Northwestern University Atomic and Nanoscale Characterization Experimental Center and the Electron Microscopy Center at ANL) greatly benefited the research and the users. As observed above, hub-and-spoke arrangements appear to be very effective. This form of networking allows a central “hub” resource to specialize and capitalize on cutting-edge research, while so-called spokes focus on educating, training, preparation work, and even workhorse services. Each element retains some autonomy and self-direction, but is able to draw on the expert resources of its neighbors.
The committee notes in passing that NSF supported a Regional Instrumentation Facility program in the 1970s and 1980s. These regional facilities were phased out because they failed to develop the important networking relationships with
one another and with their users. That is, the facilities failed to effectively serve their regions because many became entirely focused on the on-site investigators’ goals.
Regional facilities are, therefore, a natural option when considering strategies for optimizing the nation’s facility investments for materials research. By combining resources on a local scale, they retain the flexibility and responsiveness necessary to serve their users’ diverse needs, but they are large enough to significantly leverage individual resources. In fact, midsize facilities are already regional facilities in cases where the region perhaps consists of several campuses (on average).
The committee notes, however, that even the largest midsize facilities are still dominated by local users. That is, the size of facilities does not correspond directly to the scale of the geographic regions that they serve. The committee emphasizes this observation. The scale of a facility is not a linear function of its size, investment, and user base. The largest “regional” facilities are indeed national facilities, and the smallest “regional” facilities simply serve the needs of one principal investigator’s laboratory. At face value, the optimal region to be served by a facility is determined by the distance that users can travel by car in part of a day (echoing the users’ survey responses). However, it is clear that many existing midsize facilities serve regions that are considerably larger and smaller than this scale: the variations depend on the type of instrumentation available and the type of research to be performed.
In order to properly exploit the opportunities of networking midsize facilities at the next level of aggregation, a different type of “regional facility” must be envisioned. The committee believes that a successful regional network requires a minimum level of interconnecting architecture. Through the use of advanced computing and networking infrastructure, professional societies, user groups, individual outreach activities, mutual use agreements, and other cooperative instruments, midsize facilities can better communicate and coordinate their activities—with each other, with the research community, and with the federal research agencies.
Another consideration in regional networking is differences in access time and usage costs. As described earlier, some midsize facilities employ user fees to recover expenses, while others do not. How can such obvious differences in upfront cost be addressed in a regional network in which users are better informed about their range of options? Organizing regional networks around differences in scientific capability is one possibility. But the committee is quick to offer a cautionary note: regional networking or consolidation, if inappropriately implemented, can lead to elitism and an increasing gap between facilities. The essential ingredient is a scientific and research-based partnership among the participants in terms of capability and functionality so that the whole remains greater than any one part. Ultimately, of course, regional networking makes the whole greater than the sum of all its parts.
A leading example of regional networking is the hub-and-spoke model. Rather than concentrating resources at one point, a set of midsize facilities can coordinate to form a hierarchical network. The hub facility might focus on the most advanced opportunities (i.e., racehorse equipment and professional staff training), while the spokes might specialize in instrumentation and services that are more conventional and more widely used (i.e., workhorse equipment and training). Alternatively, a spoke facility might specialize in a specific set of techniques or capabilities, whereas the hub might retain a broader suite of capabilities. An advantage of the hub-and-spoke model is that the spoke facilities drive the active and two-way relationship between the hub facility and the broader community. An additional advantage of this style of networked facilities is that institutions with different levels of resources can participate. As evidenced by the success of both the CMM at the Frederick Seitz Materials Research Laboratory at the University of Illinois at Urbana-Champaign on a local level and the NCEM at LBNL on a national level, the DOE national laboratory infrastructure can provide an excellent home for such racehorse and hub facilities. However, the extent of this success can be limited by laboratory site access issues, available instrument time, or just general location.
The hub-and-spoke model offers certain economies. A spoke facility with older or less-sophisticated equipment could handle much of the work and serve as a filter for research that makes use of the most sophisticated equipment at the hub facility. Appropriate use of this model might also result in some cost savings, since not every facility would need to have state-of-the-art equipment. A systematic way of gaining access to such high-end capabilities could lessen the drive for every facility to upgrade.
For instance, a hub facility might incur longer lead times for access because of a higher subscription rate. A spoke facility might offer easier and faster access. Similarly, federal agencies invested in a regional network might offer individual investigators research awards with appropriate incentives; for instance, support for user fees at networked facilities might be included as a separate line item.
A key challenge of regional networking is that of identifying the stewards for managing the network and identifying the incentives for encouraging them to do so. Mission agencies currently do not plan to form any new facilities outside of their current geographic infrastructure, for instance, and yet a relatively modest investment by DOE at the University of Illinois facility has had huge impact. Any new such regional facilities may need to be set up with key participation of the universities. (For instance, the State of Ohio helped create and fund a microstructural characterization facility at Case Western Reserve University, a private institution, to ensure that industry in Ohio had access to state-of-the-art instrumentation.) The difficulty is that the purchase of advanced instrumentation is often associated with individual personalities and institutional pride. It is not clear
how many deans and research provosts will be willing to help support—in cash or in long-term staff—a facility at another site or campus. A federal research agency could assume stewardship through an incentivized role: institutions that collaborate in the siting and long-term staffing, as well as set up a long-term management contract for the running of the facility, could be given priority for support of midsize facilities.
A final important—and rather general—consideration is the role that midsize facilities have in providing research and training opportunities to minorities and traditionally underserved populations—especially those at smaller schools. Because of the large infrastructure burden placed on institutional hosts of midsize facilities, success stories tend to be well correlated with available (but creatively secured) resources. Smaller schools, and especially those without strong research traditions, are increasingly able to obtain sophisticated instrumentation, but because of the combination of the lack of experience, training, and on-site resources, they are often unable to provide the necessary infrastructure to support a successful midsize facility. That is, the infrastructure burden for today’s expensive research tools can discriminate among institutions on the basis of their resources.
A clear advantage of a regional facility network, especially the hub-and-spoke model, is that smaller schools could opt to participate in a larger network, when they choose to create a midsize facility, to fully utilize a sophisticated instrument—thereby leveraging their contribution off the pooled resources and experience of the other participating facilities, especially the hub. By taking on a well-defined role in the region with a specific research and training responsibility and a connection to the hub, a smaller school’s midsize facility initiative could be substantially enhanced—and have broader impact. A well-planned hub-and-spoke model for regional teaming would provide a safety net for smaller schools and empower their participation in the larger research and training enterprise. That is, no school should be too small for a midsize facility if it is well matched to a region’s needs and opportunities.
Implementing a system of regional facilities is challenging in a budget-conscious environment. However, not only is the outlook for federal budgets quite constrained, but the committee was also tasked to consider revenue-neutral solutions. Given these constraints, the committee proposes that facilities participating in the regional network should be chartered with maintaining their capabilities at the state of the art as a whole, and that increased complementarity should be developed at the participating institutions. That is, given the requirement to remain revenue-neutral, the committee identifies formation of and stewardship of a network of regional facilities as a higher priority than that of expanding other single, atomistic facilities.