I. Examining Common Pollutants in U.S. Industrial Manufacturing

We wanted to know, first, what chemicals are most commonly released due to industrial manufacturing in general, and which industries are the greatest sources of these releases. In doing so, we wanted to provide a baseline against which we could compare pollution from semiconductor manufacturing, while also providing information that may be helpful in identifying other industries or areas that might benefit from our work in the future.

---

A. Finding the Data

---

Various U.S. regulations require that manufacturing facilities that meet certain threshold conditions must submit a Toxics Release Inventory (TRI) report that, among other things, details what toxic chemicals are held in inventory and/or released from their facility (TRI Laws and Regulatory Activities. (n.d.)). Under the 1986 Emergency Planning and Community Right-to-Know Act (EPCR), passed in the aftermath of the 1984 Bhopal Disaster at the Union Carbide India Limited plant in Bhopal, India and a second, albeit less catastrophic, accident at a Union Carbide plant in West Virginia the following year, the U.S. Environmental Protection Agency (EPA) is directed to make these reports available to the public in efforts “to provide communities with information about toxic chemical releases and waste management activities in their communities, and to support informed decision making at all levels by industry, government, non-governmental organizations, and the public” (Toxics Release Inventory (TRI). (n.d.)).

We collected all TRI reports for all TRI chemicals released in all industries for 2022 (the most recent year for which complete information is available). We then identified the 20 most commonly released chemicals (by weight). We then identified the 15 facilities reporting the greatest release for each of these 20 chemicals and retrieved the individual TRI report for each of these facilities’ chemical releases (n=300). We extracted data from these TRI reports to characterize the nature and extent of chemical pollution from industrial manufacturing.

---

B. Results

---

There were 522 chemical types listed in the 300 reports we examined. The 20 chemicals with the largest total releases are listed in Table 1.

One of our most surprising and significant findings concerns how localized TRI chemical releases are in the United States. We learned that most chemical releases were attributable to a relatively small number of facilities overall and that most releases occur within a small subset of industrial subsectors.

Overall, the total amount of toxic chemicals released in the U.S. by facilities covered by TRI reporting requirements was 2.84 billion pounds. While we found that while there was considerable variation in the percentage of release of each of the top 15 facilities for each individual chemical type (some of the facilities released far less than others), the combined releases for the 300 facilities in our sample accounted for 70.6% of the total amount of chemicals in our study that were released nationwide. The range in amounts released among the top 15 facilities in each category, along with the fact that over 2/3rds of all chemical releases in the U.S. were attributable to 249 of the 300 facilities in our sample, indicates that there is a significant minority of U.S. industrial facilities responsible for the vast majority of chemical releases.

---

Table 1. 20 Most Commonly Released Chemicals. Source: EPA Toxic Release Inventory Dataset

---

The TRI data identifies each facility according to the North American Industry Classification System (NAICS) is a six-digit coding system that categorizes economic activity into 20 industry sectors. Each industry sector is divided into further subsectors and industry groups. Using the NAICS codes, we were able to characterize the industry areas that are responsible for the majority of toxic chemical releases in the U.S.

There were a total of 74 unique NAICS codes represented by the 300 facilities in our study. Grouping these according to the NAICS Industry Sector and Subsector designations resulted in 16 sub-sectors represented in the sample (Table 2). For each subsector, we calculated: a. the total number of chemicals released, b. the total number of facilities reporting releases, c. the average release for each facility, and d. the percentage that Industry Subsectors release relative to both the total reported industrial releases in the U.S. and the total reported releases for the 300 facilities in our sample (Table 2).

---

Table 2. Industrial Sub Sector Average Releases. Source: EPA Toxic Release Inventory (TRI) Dataset

---

In addition to the fact that the majority of chemical releases in the U.S. are attributable to a subset of the 300 facilities in our study, we also found that chemical releases tend to be localized to a relatively small subset of industrial sectors. In fact, a considerable majority of the total U.S. industrial-related toxic chemical releases (68.49%) were due to activities in only half of the Industry Subsections we examined. In considering this, we think it is worth emphasizing again that our sample of Industrial Subsections was drawn from focusing on the 20 most commonly released chemicals in the TRI list of 557, and we selected only the top 15 releasing facilities for each of these chemicals. We were surprised to learn that such a large percentage of all industrial pollution could be traced to such a small subset of industries and to a small subset of actors within those industries. It is also worth noting that the localization of chemical releases is due in no small part to the fact that the mining industry alone accounts for nearly half (45.81%) of all chemical releases nationwide.

By doing this analysis, we better understood what kinds of toxins were released from industrial activities in the U.S., and thus we were better equipped to think about what kinds of problems our work might focus on. Moreover, seeing the concentration of releases within small sets of industries and facilities in those industries strengthened our conviction that there needs to be more sustainable methods developed for industrial manufacturing and dealing with the waste produced by it, as well as our hope that targeted improvements for treating industrial waste could have significant and widespread benefits.

Interestingly, the semiconductor manufacturing subsector was not represented in our sample of the top TRI chemical-releasing industry sectors. We realized that there may be various explanations for the absence of semiconductor manufacturing from our sample. For example, given the fact that the semiconductor manufacturing industry in the U.S. is relatively small at the present time, it is not surprising that it did not show up in our sample, since larger and more established industrial sectors could be expected to release more toxic waste simply due to the scale of the enterprise. Therefore, with our baseline understanding of toxic chemical release in U.S. industry, we turned to examine the emerging sector of semiconductor manufacturing.

II. Analysis of Semiconductor TRI Information

Having learned about chemical releases from industrial manufacturing in the U.S. in general, we wanted to focus on better understanding the current status of semiconductor manufacturing in the U.S. and chemical releases from U.S. semiconductor manufacturing facilities in particular.

In order to examine the Toxic Release Inventory information for U.S. based semiconductor manufacturers, we retrieved the TRI Basic Data Files for the year 2022 from the EPA database (TRI Basic Data Files: Calendar Years 1987-Present. (n.d.)). This file contains the 100 most used data fields from the TRI Reporting in all submitted TRI forms. We then extracted all data associated with facilities reporting themselves with the primary NAICS code associated with semiconductor manufacturing (NAICS code 334413).

---

C. Overview of U.S. Facilities and TRI Chemical Use

---

We found that in 2022, 90 distinct facilities filed TRI reports to the EPA indicating semiconductor manufacturing as their primary NAICS code. These facilities were distributed across 27 different states in the continental United States. As expected, we observed slightly higher concentrations of facilities in the Northeast and Southwest (see Map 1).

---

---

In total, there were 397 distinct reports totaling 3507953.374 pounds of toxic chemical releases across these 90 facilities. In total, there were 38 different TRI chemical categories mentioned in these reports (see Table 3). While industries using chemicals or compounds that are protected by trade secrets or patent laws are not required to publicly disclose the names of those materials, if those materials meet EPA reporting requirements, they must nevertheless report those inventories, using a special masking code that protects the privileged information. We were interested to discover that no facility reported the use of any chemical under the special masking code.

---

D. Health-Related Information Contained in TRI Information

---

Six of the chemicals/compounds in the list [1,3-Propane sultone, Cadmium And Cadmium Compounds, Cobalt compounds, Lead, Nickel compounds, and Tetrachloroethylene] were reported as carcinogens. There were 6 chemical groups [Lead, Lead and lead compounds, Lead compounds, Mercury, Polycyclic aromatic compounds, Tetrabromobisphenol A, and 1,3-Propane sultone] identified as “Persistent Bioaccumulative and Toxic” (PBTs). No reportable PFAS chemicals were included in any of the inventories. The absence of any reported PFAS chemical releases or inventories in the EPA dataset eventually became a serious concern in our Human Practices work, as discussed in our Civic Engagement Project.

---

Table 3: Reported Chemical Inventories at U.S. Semiconductor Manufacturing Facilities by Amount Released

---

E. Wastewater Release

---

We were particularly concerned with identifying the chemicals and compounds released to waterways and wastewater from the semiconductor manufacturing facilities, as this information was important for helping us identify what might be the most important targets for our device.

Facilities covered by TRI reporting are required to report water-related releases in four ways: 1. onsite injections to wells, 2. onsite discharge to surface water, 3. transfers for release to publicly owned treatment works (POTWs), 4. transfers to POTWs for treatment, 5. transfers for wastewater treatment not involving POTWs.

The use of injection wells was almost never reported. Only one facility reported onsite injection well release in 2022, involving 350 pounds of n-methyl-2-pyrrolidone. Onsite release to surface water was only slightly more common. Across all 90 facilities, 1227596.1 pounds of TRI chemicals were released onsite as surface water discharge. Almost all releases to surface water (99.5%) were nitrate compounds. The remaining .05% of surface water discharge of TRI chemicals from semiconductor manufacturing facilities was comprised of ammonia (4294 lbs), copper and copper compounds (316.1 lbs.), ethylene glycol (820 lbs), and n-methyl-2-pyrrolidone (1110 lbs).

More common than injection wells and surface water release was released to publicly owned treatment works (POWTs) –i.e., municipal wastewater treatment plants. We identified a total of 5859945.62 pounds of chemicals and compounds transferred to POWTs. Transfers to POTWs are recorded as either (a) transfers for release or (b) transfers for treatment.

Slightly over one million pounds of TRI chemicals were sent directly to publicly owned treatment works (POWTs) for release. Again, nitrate compounds (46.91%) were most commonly released this way, followed by ammonia (43.23%). The remaining 9.8% of the amount released contained 17 different chemicals or compounds. (See Table 4)

---

Table 4. Transfers to POTWs for Release

---

Facilities sent an additional 4830074.66 pounds of 7 different chemicals and chemical compounds to POTWs for treatment. Again, nitrate compounds accounted for the majority of transfers for treatment (67.79% or 3274260.42 lbs.), followed by ammonia (13.93%), ethylene glycol (9.53%), n-methyl-2-pyrrolidone (6.42%), certain glycol ethers (2.33%), nitric acid (.01%) and less than one pound of hydrogen fluoride.

III. A Fuller Picture – Analysis of Information Regarding Pre-Release Wastewater

While the EPA’s TRI information provided us with a valuable understanding of both the current state of toxic chemical use and release in U.S. industry generally and in semiconductor manufacturing particularly, there are a number of limitations to the analysis we were able to do using only this information. There are two major limitations worth mentioning.

First, facilities are only required to report specific chemicals that appear on the TRI reportable chemical list. While this list is extensive and constantly reviewed and updated, it is possible that there may be contaminants or pollutants in semiconductor wastewater that are not included in these reports. Secondly, since most of the information in TRI reports concerns the release of chemicals, the TRI system provides an important but only partial picture of the manufacturing process as it relates to our project. Since semiconductor facilities must also treat their wastewater onsite before it is released to groundwater or to POTWs, we could not be sure that, looking only at TRI release information, we would be able to fully characterize chemicals present in the full cycle of wastewater usage. Because finding more efficient, sustainable, and potentially less expensive methods of wastewater treatment than those presently available would be an important way of ensuring that our project was beneficial and good for the world, we wanted to know more about the waste streams within the manufacturing facilities.

As noted, there are widespread concerns about the lack of available information about what chemicals are used in the manufacturing of semiconductors and about the reluctance of the industry to disclose this information. However, representatives we met with from one local fab were very forthcoming and helpful in describing their operations and waste streams. In that meeting, and in a meeting with another stakeholder representing the NY Chapter of the environmental advocacy group The Sierra Club, we learned that we might be able to find out more about the chemicals used and released by examining the wastewater permits issued by local municipalities to the facilities. (see our Stakeholder Interviews)

In investigating this option, we learned that while municipal wastewater permits, like the TRI reports, are principally concerned with what is in waste that is released from the manufacturing process, there are important differences that make these documents potentially valuable sources of information for anyone interested in learning about what is happening within the manufacturing process.

First, these permits describe broader treatment and monitoring requirements than the reporting required for TRI compliance. We were interested, for example, with the pH levels required for wastewater transfer, and such information is not contained in TRI reporting.

Secondly, unlike TRI requirements, the specific requirements set out in municipal wastewater permits for monitoring and releasing wastewater are tailored to the individual facility applying for it and the processes and materials they employ. Thus, by examining the specifics of these permits, it is possible to form a fuller picture of what the wastewater prior to release to POTWs might contain.

While there is no publicly accessible database containing all industrial wastewater permits, we did learn in interviews with both industry representatives at NY Creates and Dr. Hughes of the Sierra Club that these permits are issued by public offices in New York State thus can be requested by the public under New York’s Freedom of Information Law (FOIL), a law which gives the public the right to access records held by state government agencies.

We filed a FOIL (Freedom of Information Law) request with five New York State counties requesting wastewater permits for existing New York State semiconductor manufacturing facilities. Surprisingly, only one of these counties met the legal obligation to provide the requested documents. This problem with the existing regulatory system further contributed to our eventual shift to a Civic Science approach to our project, as discussed in our Civic Engagement section.

In reviewing the available wastewater permit, we were interested in answering two important main questions:

Question #1: What is the pH range permitted for the discharge of wastewater to the local POTW?

Question #2: What specific chemicals are Global Foundries responsible for monitoring and controlling in the wastewater they release to the POTW?

According to the permit, the maximum daily pH effluent limitations are between 5.0 and 10.5.

In addition to requiring that Global Foundries meets waste stream monitoring requirements as determined by pre-determined federal and county regulations, the permit sets out a list of total waste stream monitoring requirements for the facility. This list is more extensive than the combined pre-established regulatory requirements and, thus, is likely the best guide to the facilities' wastewater streams.

The effluent limitations and monitoring requirements are based on a monthly average flow limitation of 6.6 million gallons per day. The requirements identify 12 chemicals with specific monthly average limitations and 15 chemicals that require monitoring only.

---

Table 5. Global Foundries Permit List – Total Waste Stream Monitoring

---

IV. Lessons Learned and Steps Taken

In attempting to obtain information about chemical use and release in U.S. industry and U.S. semiconductor manufacturing, we gained a better understanding of the nature of industrial waste production and disposal in the United States that was important to our thinking about how to design our project in a responsible and effective manner. For example, our early idea was that copper might be a suitable target for a project aimed at improving semiconductor wastewater treatment processes, since copper is (a) commonly known to be used in semiconductor manufacturing, (b) a substance with negative health effects, and (c) might be among the contents of these facilities’ waste streams with sufficient value to incentivize new collection strategies to replace disposal strategies. We learned from this project, however, that the amount of copper released from semiconductor manufacturing is already relatively small, and that there is a range of toxic chemicals released from semiconductor manufacturing facilities that may represent more serious environmental and health concerns. By characterizing toxic release inventories, our Human Practices work helps our team, as well as future researchers, identify the most urgent and most effective research agendas for industrial wastewater treatment initiatives.

In exploring these alternative legal tools for obtaining important information that could help guide our thinking about how to develop products that are responsible and truly good for the world, we also gained an appreciation for how bringing a diverse set of tools and methodologies into our project benefits the process of scientific research and technological development. We learned about how environmental regulations and freedom of information requests can serve to fill information gaps that, given the nature of the semiconductor manufacturing context, we could not fully do through standard tools such as literature reviews, lab analysis, or expert interviews with industry stakeholders.

Finally, our Human Practices work in analyzing and characterizing industrial waste with the use of these tools provided us with a better understanding of the policy and legal contexts of our issue. As a result, we began to better appreciation, first, that our initial approach to thinking about the problems of industrial wastewater through the aperture of improving the efficiency of industrial practices and operations had prevented us from considering the broader ways in which semiconductor manufacturing affects the environment, the public health, and the communities most directly affected by the development of these facilities and the ways in which synthetic biology might be applied to address these concerns. Secondly, reflecting on this aspect of our Human Practices work increased our awareness of the ways in which the regulatory and policy environments contribute to or frustrate effective research and development.

In addition to helping us think better about how to shape the direction of the work in our labs, these important realizations about the relationships between science and society also led us to reflect on our Human Practice work itself, and whether it was enough for our project to focus on our dual goals of designing a product that would be good for the world and communicating our scientific work to the public. Perhaps the most surprising result of our Human Practices work was the way in which we changed the way we thought about what Human Practices work can and should be doing and the responsibilities that scientists have to engage in Human Practices.

The ways in which our thinking about the relationship between science and society changed and about the role of Human Practices in our project changed as a result of the reflections on and discussions about our work are described in our section on Civic Engagement.

References

Kim, S., Yoon, C., Ham, S., Park, J., Kwon, O., Park, D., … Kim, W. (2018). Chemical use in the semiconductor manufacturing industry. International Journal of Occupational and Environmental Health, 24(3–4), 109–118. doi:10.1080/10773525.2018.1519957

Toxics Release Inventory (TRI). (n.d.). Retrieved from https://health.gov/healthypeople/objectives-and-data/data-sources-and-methods/data-sources/toxics-release-inventory-tri#:~:text=The%20Toxics%20Release%20Inventory%20

TRI Basic Data Files: Calendar Years 1987-Present. (n.d.). Retrieved from https://www.epa.gov/toxics-release-inventory-tri-program/tri-basic-data-files-calendar-years-1987-present

TRI Laws and Regulatory Activities. (n.d.). Retrieved from https://www.epa.gov/toxics-release-inventory-tri-program/tri-laws-and-regulatory-activities