ll

OPEN ACCESS

Article
Techno-economic, life-cycle, and
socioeconomic impact analysis of enzymatic
recycling of poly(ethylene terephthalate)
Avantika Singh, Nicholas A.
Rorrer, Scott R. Nicholson, ...,
Alberta C. Carpenter, John E.
McGeehan, Gregg T. Beckham

gregg.beckham@nrel.gov

Highlights
Enzymes expand PET-recycling
options beyond chemo-catalytic
and thermal methods

Process analyses reveal key

economic drivers for enzymatic
PET recycling

Enzymatic PET recycling can be

cost-competitive with virgin PET
manufacturing costs

Circularity of the PET economy

offers sustainability and
socioeconomic benefits

This study presents a comprehensive process, economic, environmental, and

socioeconomic analysis of the enzymatic recycling of poly(ethylene terephthalate),
which is the most widely used synthetic polyester. The analyses predict that PET
deconstruction using enzymes can achieve cost parity with terephthalic acid
manufacturing as well as substantial reductions in both supply chain energy use
and greenhouse gas emissions relative to virgin polyester manufacturing. This
study also highlights key research areas for further impactful development of
biocatalysis-enabled plastics recycling.

Singh et al., Joule 5, 2479–2503

September 15, 2021 ª 2021 The Authors.
Published by Elsevier Inc.
https://doi.org/10.1016/j.joule.2021.06.015
 ll
OPEN ACCESS

Article
Techno-economic, life-cycle, and socioeconomic
impact analysis of enzymatic recycling
of poly(ethylene terephthalate)
Avantika Singh,1,2 Nicholas A. Rorrer,1,3 Scott R. Nicholson,1,4 Erika Erickson,1,3 Jason S. DesVeaux,1,2
Andre F.T. Avelino,4 Patrick Lamers,4 Arpit Bhatt,4 Yimin Zhang,4 Greg Avery,4 Ling Tao,2
Andrew R. Pickford,1,5 Alberta C. Carpenter,1,4 John E. McGeehan,1,5 and Gregg T. Beckham1,3,6,*

SUMMARY Context & scale

Esterases have emerged as important biocatalysts for enzyme- Chemical recycling and upcycling
based polyester recycling of poly(ethylene terephthalate) (PET) to of plastics will be critical
terephthalic acid (TPA) and ethylene glycol (EG). Here, we present technologies to address the
process modeling, techno-economic, life-cycle, and socioeconomic plastic pollution challenge. Given
impact analyses for an enzymatic PET depolymerization-based recy- multiple process options for
cling process, which we compare with virgin TPA manufacturing. We recycling plastics, rigorous
predict that enzymatically recycled TPA (rTPA) can be cost-compet- process analysis is necessary to
itive and highlight key areas to achieve this. In addition to favorable identify challenges that must be
long-term socioeconomic benefits, rTPA can reduce total supply overcome for a technology to
chain energy use by 69%–83% and greenhouse gas emissions by reach an industrial scale. For PET
17%–43% per kg of TPA. An economy-wide assessment for the US recycling, several chemical
estimates that the TPA recycling process can reduce environmental recycling strategies have been
impacts by up to 95% while generating up to 45% more socioeco- proposed and, in some cases—
nomic benefits, also relative to virgin TPA production. Sensitivity an- chemo-catalytic and thermal
alyses highlight impactful research opportunities to pursue toward approaches—are being scaled
realizing biological PET recycling and upcycling. up. Given that PET exhibits labile
ester bonds that are also common
in natural biological systems, the
INTRODUCTION
research community is vigorously
Despite widespread application in daily life, there are limited options for the recy- pursuing the engineering of
cling of synthetic plastics, which presents an opportunity for the development of esterase enzymes to
critical new technologies. As plastics accumulate both in landfills and in the natural depolymerize PET. This study
environment, these materials are causing a global pollution crisis due to their recal- applies process analysis to
citrance against abiotic and biological breakdown.1–3 Compounding the issue, the highlight drivers that the
current mechanical recycling industry often produces lower-value products and is community can focus on to
unable to recycle many types of plastics. This confluence of environmental concerns accelerate the development of a
and the limitations of current recycling technology have catalyzed renewed interest biological PET depolymerization
from the global research and industrial communities to pursue recycling strategies process and also provides a basis
that rely on depolymerization of polymers into their constituent monomers or other to compare current and future
processable, non-polymeric intermediates, namely via chemical recycling.4 enzyme-based approaches for
PET-recycling to chemo-catalytic
Poly(ethylene terephthalate) (PET) is the most abundantly produced synthetic poly- and thermal methods.
ester in circulation today,1,5 produced globally at 82 million metric tons (MMT) per
year.6 The majority (54%) of PET is used in the production of textiles and fibers,
with 24 MMT (or 29%) undergoing further polymerization to produce resin for rigid
containers and single-use bottles, whereas the rest is used in films and other appli-
cations.7 Looking at the PET consumed annually in the United States (US),5 which is
approximately 3 MMT, only 18.5% is currently processed by mechanical recyclers.8

Joule 5, 2479–2503, September 15, 2021 ª 2021 The Authors. Published by Elsevier Inc. 2479
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 ll
OPEN ACCESS Article

Therefore, to achieve higher rates of PET recycling and potentially include PET sub-
strates such as textiles that are not currently recycled, multiple chemical recycling
strategies for PET have been proposed, developed, and, in some cases, scaled up
over the last several decades.4,9,10 Methanolysis, hydrolysis, aminolysis, glycolysis,
and thermal depolymerization strategies are among the most well-studied chemical
recycling processes for PET.4,9,10

PET is synthesized mainly via polycondensation of two building blocks: terephthalic

acid (TPA) and ethylene glycol (EG). The inter-monomer chemical linkage in PET is an
ester bond, which is a prevalent linkage in biomolecules. For example, cutin, su-
berin, hemicellulose, and lignin are abundant plant-derived polymers, all of which
exhibit ester bonds.11–13 Given the prevalence of aromatic and aliphatic building
blocks in ester-linked natural polymers, it is perhaps not surprising, in hindsight,
that esterase enzymes can cleave ester bonds in PET and other synthetic polyes-
ters.14–19 Considering the potential for esterases to deconstruct PET under mild con-
ditions, the global research community is vigorously pursuing efforts in prospecting
for new PET-degrading enzymes (PET hydrolases), solving crystal structures thereof,
and engineering and evolving these enzymes for improved PET degradation capac-
ity, with the aim of developing efficient biocatalysts for use as a chemical recycling
approach to address this common polyester.20–36

Given the growth in research related to biocatalyst development for PET depolymeriza-
tion, it is critical to understand the projected economic and sustainability impacts that
such a process could have toward enabling PET circularity. To that end, we present
here a rigorous, comprehensive modeling effort for a conceptual enzymatic process
to depolymerize PET to recycled TPA (rTPA) and EG, including all utilities required for
an integrated process. Techno-economic analysis (TEA) of this process enables predic-
tions of the capital and operating costs to project the minimum selling price (MSP) for
rTPA, including the sale of the two co-products, EG and sodium sulfate (SS). Sensitivity
analyses are employed to highlight the importance of the biocatalytic rate, enzyme
loading, and enzyme cost, along with multiple additional, tunable process variables
that are important for further process improvement and optimization. We employ the
Materials Flow through Industry (MFI) tool37 to estimate the total supply chain energy
and greenhouse gas (GHG) emissions for rTPA production from reclaimed PET via
this enzyme-based PET deconstruction process and compare this with virgin TPA
(vTPA) production. We expand this MFI analysis to include the use of a top-down, envi-
ronmentally extended input-output (EEIO) model38 to evaluate the US economy-wide
impacts of implementing the rTPA process and compare it with the production of
vTPA across multiple environmental and socioeconomic indexes.

Taken together, the results in this study highlight the most crucial process steps to 1BOTTLE Consortium, Golden, CO 80401, USA
improve enzyme-based recycling technologies for PET. In addition to the economic 2Catalytic
Carbon Transformation and Scale-Up
Center, National Renewable Energy Laboratory,
factors, this work identifies important sustainability drivers for realizing environ-
Golden, CO 80401, USA
mental benefits (i.e., reduced resource use and byproduct release) and socioeco- 3Renewable Resources and Enabling Sciences
nomic potential (i.e., added economic value and number of jobs) possible through Center, National Renewable Energy Laboratory,
an enzyme-based PET-recycling strategy. Golden, CO 80401, USA
4Strategic
Energy Analysis Center, National
Renewable Energy Laboratory, Golden, CO
Process and economic model construction 80401, USA
A process model for enzymatic PET depolymerization was developed by Aspen Plus to 5Centrefor Enzyme Innovation, University of
explore how the assumptions and process requirements impact the process economics Portsmouth, Portsmouth PO1 2DY, UK
6Lead
and sustainability. Figure 1 presents a simplified process flow diagram (PFD) for the base contact
case, whereas a detailed PFD indicating all model inputs is provided in Figures S1–S4. *Correspondence: gregg.beckham@nrel.gov
The recycling facility is modeled on the scale of 150 Mt of PET flakes processed per day https://doi.org/10.1016/j.joule.2021.06.015

2480 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

Figure 1. Enzymatic PET-recycling process design and representation of supply chain and economy-wide analysis
(A) Simplified process flow diagram of the PET enzymatic depolymerization process, divided into three sections. In the first section, polyester flakes are
pre-treated (e.g., extrusion, size reduction, etc.) and subsequently enzymatically depolymerized in the second section. Following depolymerization,
rTPA, SS (salt), and EG are recovered via downstream processes in the third section. PET flakes and other raw material inputs are shown in red,
intermediate, recycle, and waste streams are shown in black, and the product and co-product streams are shown in blue. Case studies were performed
to evaluate key variables and process operations in each process section, labeled at the top of the upper panel.
(B) A representation of the supply chain modeled for the bottom-up MFI tool assessment in orange, and the top-down life cycle assessment by the
BEIOM tool in green. BEIOM considers the supply chain for the production of rTPA, as well as production, distribution use, and environmental and
socioeconomic factors. The detailed process flow diagram of the base case design as modeled in Aspen Plus is presented in Figures S1–S4.

(50,000 Mt/year). For reference, the average annual capacity of a US PET production
plant is 88,000 Mt/year.6 The plant size modeled here represents 1.7% of all PET
currently consumed in the US and 7% of the PET volume currently recycled in the US.
Post-consumer PET recycling in the US is primarily conducted for single-use beverage
bottles and, in total, accounts for 0.68 MMT (1,503 million pounds) of PET per year,
which is 29% of all PET bottles consumed annually in the US.39 The modeled facility
is assumed to be in close proximity to another manufacturing plant, such as a PET poly-
merization plant or a petroleum refinery, such that high-pressure steam, cooling water,
and wastewater treatment are assumed to be over-the-fence and priced accordingly as
utility operating expenses. Capital investment to build the auxiliary infrastructure for util-
ities is accounted in the outside battery limit costs.

In the baseline case, we assume that clean PET flakes, consisting of approximately
30% colored flakes, are obtained directly from a recycler, priced at $0.66/kg.40,41
The recyclable fraction (i.e., PET fraction, denoted by F) of the total feed is assumed
to be 0.95, with the remainder being contaminants such as caps, labels, adhesives,

Joule 5, 2479–2503, September 15, 2021 2481

 ll
OPEN ACCESS Article

dirt, etc. Although enzymatic recycling is of potentially greater significance for dirty,
comingled streams, we had access to more reliable data on market prices of clean
PET flakes. Hence, the base case considered clean, colored flakes, which serves as
a conservative projection of the rTPA price. Other scenarios with different PET feed-
stock prices and PET purity are analyzed as sensitivity cases. Prior to
depolymerization, feedstock pretreatment steps designed to improve enzymatic
conversion are modeled in which clean PET flakes are fed to an extruder that heats
the plastic above its melting point (Tm260 C). The molten PET is then quenched to
yield PET with lower crystallinity, followed by size reduction via a microgranulator.42
This series of pretreatment steps results in a PET powder with particle sizes of less
than 1 mm prior to enzymatic depolymerization, in line with recent literature.35,42 Un-
like mechanical recycling, drying is not required in this sequence.

Following mechanical pretreatment, the plastic is fed to a series of stirred-tank reac-

tors for depolymerization to rTPA and EG via enzymatic hydrolysis. Specifically, the
PET stream is conveyed to a series of stirred tanks (950 m3 vessels) with a modeled
solids loading of 15% in the base case.35 The depolymerization process is catalyzed
by a generalized PET hydrolase that is produced offsite and purchased directly for
use in this process. Products of PET hydrolysis are recovered in this process scheme
upon reaching a 90% PET degradation extent.35 Additionally, this model ignores
other monomers that may be present in PET (e.g., isophthalic acid, 1,4-cyclohexane-
diol, diethylene glycol, etc.).

After enzymatic depolymerization, the hydrolysate is passed through a filter press

where the remaining solids are separated from the reaction solution, followed by
ultrafiltration to remove the enzymes. The filtrate is subsequently passed through
an activated carbon column to remove additives such as dyes and pig-
ments.35,43,44 These steps inherently assume that rTPA at pH > 7 remains soluble
in the aqueous phase. The solution is then acidified by the addition of sulfuric
acid to lower the pH to 2.5 and cooled to precipitate rTPA.44,45 The precipitate
is passed through a continuous crystallizer and dried to recover rTPA crystals at
a purity of >98%.44,46 The remaining liquor is neutralized to pH 7 by the addition
of caustic (e.g., NaOH) and passed through a membrane unit with 96.5% rejection
for SS and 72% rejection for EG.47–49 The permeate, which is predominately wa-
ter, is sent to the adjacent wastewater treatment, and the retentate is sent to a
continuous crystallizer for SS recovery. The SS is assumed to be crystallized as
Glauber’s salt (the most stable hydrate form) with 98.5% purity and sold at a mar-
ket price of $0.15/kg50 as a co-product (overall, 76% mass recovery, adjusted for
anhydrous form). The model assumes that any remaining SS salt is lost during the
processing steps, such that the mother liquor from salt crystallization primarily
consists of EG and water. EG is subsequently recovered via a series of distillation
columns to obtain EG at a purity of 99%, enabling it to be sold at market price
($0.96/kg).50 The overall recovery of rTPA and EG in the process is modeled at
90% and 50%, respectively, and these values are varied as parameters in the
sensitivity analysis.

Material and energy balances from the model were used to estimate the required
equipment size and capital investment, whereas information on raw materials, utili-
ties, etc., was used to estimate the variable operating expenses. A discounted cash
flow analysis approach with certain financial parameters (see Table S1) was applied
to project the rTPA MSP produced by this facility, with EG and SS salt sold at their
respective market prices as co-products. A summary of the results from this eco-
nomic analysis can be found in Table S2.

2482 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

RESULTS
Base case scenario
The base case scenario, as outlined earlier, is a 150-metric-tons-per-day (MTPD)-
sized plant that considers depolymerization of clean PET flakes (F = 0.95, cost =
$0.66/kg, 30% color) to yield rTPA as a primary product, while EG (50% recovery,
selling price $0.96/kg) and SS (76% recovery, selling price $0.15/kg) are sold as
co-products. Co-product selling prices are based on a 5-year average of historical
prices in the US. For reference, historic prices for vTPA and the co-products are pro-
vided in Figures 2A and 2B. Including revenue from co-products, rTPA achieves an
MSP of $1.93/kg with a majority of the cost attributed to the feedstock supply, shown
in blue in the simplified cost breakdown chart (Figure 2C). The base case results for
rTPA MSP can be further delineated by the process sections. The detailed MSP cost
breakdown that combines the capital and operating contributions from each pro-
cess step is shown in Figure 2D.

The total capital investment for a plant designed to process 150 MTPD of PET flakes
is projected to be $67M (Figure 2E). The highest contribution (29%) to the total
capital investment for constructing this plant is from product recovery steps, i.e.,
rTPA crystallization, salt recovery, and EG distillation, primarily owing to continuous
crystallizers and distillation columns. Feedstock pretreatment, which consists of
extrusion and cryo-micronization, and the depolymerization section, which consists
of a series of batch hydrolysis reactors, contribute nearly equally to the capital ex-
penses, at 20% each. Additionally, a 25% contribution is assumed from the outside
battery limit (OSBL) investment, which consists of additional capital expenditure,
such as development of piping, instrumentation, etc., required to integrate the
over-the-fence utilities into the plant. Detailed information on the specific unit oper-
ations and the capital requirements for each process section is shown in Table S3.

In addition to the initial capital expenditures, the plant will require $44M per year in
operating costs. As shown in Figure 2F, these operating expenses are dominated by
the feedstock cost, which is priced for clean PET flakes (containing 30% colored
flakes) sourced from bottle recyclers.40,41 Electricity consumption is the other major
driver in the pretreatment section, primarily required for melt extrusion and micro-
grinding. In other process areas, the cost of chemicals such as caustic for pH main-
tenance, particularly in hydrolysis reactors, and sulfuric acid for rTPA crystallization
are primary drivers. The cost contributions for these chemicals are largely offset
by the recovery of SS, and salt recovery is being explored as a variation in the
base case process design (vide infra). Similarly, steam usage in the distillation step
is attributed to the EG recovery, which requires substantial water evaporation. In
addition, maintenance costs for filter, activated carbon bed, and membrane replace-
ment are accounted for in the economics. Further details of the operating costs are
described in Table S4.

Additional analysis, conducted by the MFI tool, estimates that rTPA has a 69% lower
supply chain energy requirement (Figure 2G) and produces 17% less GHG (Fig-
ure 2H) than vTPA production. As shown in Figure 1, we divide the base case into
three sections: (1) feedstock pretreatment, (2) enzymatic PET depolymerization,
and (3) product and co-product recovery (rTPA, EG, and SS).

To further examine the process sections, we considered three primary case studies
focusing on variables within each section. In doing so, this work highlights the poten-
tial for further research and development across multiple disciplines and highlights

Joule 5, 2479–2503, September 15, 2021 2483

 ll
OPEN ACCESS Article

Figure 2. Detailed cost breakdown of rTPA in the base case

(A and B) The US historic price data for (A) vTPA and (B) the co-products (EG and SS).
(C and D) The summary of the base case results for the (C) minimum selling price from the combined process, (D) minimum selling price, both the capital
and operating contributions, shown across each process section.
(E) Total capital investment of $67M in the base case split into different steps of the recycling process, including outside battery limit (OSBL), which is
assumed to be 25% of battery limit investment.
(F) Annual operating expenses for the proposed PET-recycling design, disaggregated by process step (total = $44M/year). Raw materials include PET
feedstock cost as well as other chemicals such as enzymes, caustic, and sulfuric acid, and co-products are EG and SS.
(G and H) The base case (G) supply chain energy, and (H) GHG emissions of rTPA are shown in comparison with vTPA. Numerical data in this figure are
reported in Tables S5–S10.

the most important factors to be overcome to enable the commercial success of this
technology.

Case study I: Feedstock pretreatment

In order to understand the opportunities for process improvement, we examined
various parameters. As noted in the base case, feedstock cost is one of the largest con-
tributors to rTPA MSP. Most feedstock costs are associated with the cleaning of PET
bales and shredding into flakes.41 Due to the variable nature of post-consumer waste
composition, and the lack of uniformity in sorting schemes, the purity grade and relative
contaminant percentages of PET bales vary between material recovery facilities

2484 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

Figure 3. TEA sensitivity results for process parameters associated with the feedstock, pretreatment, and plant size
(A) Historical market prices of PET flakes.
(B) MSP of rTPA as a function of feedstock cost.
(C) Sensitivity analyses demonstrating variations in the pretreatment process section that exhibit the highest impact on rTPA selling price.
Abbreviations and symbols used—F PET : fraction of PET (by weight in the feedstock, base case value = 0.95), MTPD, metric tons per day (base case
value = 150 MTPD); NMP, no mechanical pretreatment. Numerical data in this figure are reported in Tables S11 and S12.

(MRFs),51,52 which, in turn, will affect the quality and price of reclaimed PET. Market re-
ports indicate that the average price of mixed, baled PET in the US varies from $0.14–
0.48/kg, translating to clean PET flakes available at $0.55–0.90/kg (Figure 3A).40,41 Due
to the variability in cost, sensitivities of PET prices ranging from $( 0.11) to $1.10/kg are
modeled (equivalent to 5 cents to 50 cents per pound of PET), where negative feed-
stock costs indicate a scenario where a landfill tipping fee is potentially avoided.40,53,54
As observed, the PET feedstock cost drives the greatest variability in selling prices,
which range from $0.71/kg to $2.63/kg as modeled and exhibit a linear relationship
with MSP (Figure 3B). Additionally, to accommodate the variability in relative PET con-
tent, PET fraction sensitivities of F = 0.99 (e.g., clean mechanically recyclable grade)
and F = 0.91 (e.g., post-consumer curb side pickup grade but enriched in PET) are as-
sessed.40,51,52 The relative PET content has minimal effect on MSP ($0.17/kg).

Additional sensitivity cases explored in this case study include varying the plant size
and modifying the pretreatment configuration (Figure 3C). The latter case considers
an alternate design scenario where no mechanical pretreatment (NMP) is required,
relying on the assumption that enzymes capable of activity on both crystalline and
amorphous domains of PET are available, thereby rendering the pretreatment steps
of extrusion, amorphization, and micronization unnecessary. By avoiding the me-
chanical pretreatment (NMP case) and achieving the same rate of depolymerization,
the rTPA MSP can be lowered to $1.69/kg.

Case study II: Enzymatic PET depolymerization

Next, we considered key variables in the enzymatic depolymerization section. Previ-
ous work from several research groups indicates that cost of enzyme production can
range widely from $25–$110/kg depending on enzyme type, expression strategy, ti-
ters, extraction methods, and production scale.35,55,56 This is higher than the cost of
fungal cellulases estimated in previous studies on biochemical conversion of ligno-
cellulosic biomass, which are $5/kg.57–59 For the base case model, enzyme pur-
chase cost is set as $15/kg and is varied in the sensitivity analysis (Figure 4A).60–63

Aside from enzyme production, additional sensitivity parameters explored include

enzyme loading and bioreactor residence time (t). The base case design considers

Joule 5, 2479–2503, September 15, 2021 2485

 ll
OPEN ACCESS Article

Figure 4. TEA sensitivity results for process parameters associated with the enzymatic depolymerization step
(A) Cost breakdown of the rTPA MSP in the base case process design and as a function of process variables including enzyme loading (mg/g PET
feedstock, base case value = 5), enzyme price ($/kg, base case value = 15) as well as sensitivities related to the reactor configuration, including extent of
conversion (x base case value = 0.9), solids loading, wt % (4 base case value = 0.15), and residence time in hours (t base case value = 96 h).
(B) Multivariate sensitivity analyses demonstrating the combination of enzyme cost and enzyme loading in which a decrease in both parameters exhibits
a non-linear decrease in the selling price of rTPA.
(C) Multivariate sensitivity analyses demonstrating the combination of solids loading and enzyme loading, which affects the selling price of rTPA. Higher
solids loading and lower enzyme loading results in reduction of the rTPA MSP.
(D) Multivariate sensitivity analyses demonstrating the combination of solids loading and extent of conversion in which lowering both parameters
results in a decrease of TPA MSP. The contour lines serve as references to interpolate the MSP of rTPA ($/kg). Numerical data in this figure are reported
in Tables S15–S18.

these variables at an intermediate value of 5 mg enzyme/g of PET input and 96 h,

respectively. Products of PET hydrolysis are recovered in this process scheme upon
reaching a 90% PET degradation extent (the reaction scheme is shown in Figure S1).35
Additionally, the generalized PET hydrolase modeled is assumed to have maximal
degradation activity at 60 C and pH 8. PET hydrolases have been reported to function
over a range of temperatures and pH values,64 as well as a range of enzyme loading
values; therefore, these values were also explored as sensitvity parameters. Tempera-
ture and pH had a negligible impact on MSP in the ranges explored—data provided in
Table S13. The two factors with largest impact on MSP include solids loading (4) for
which a change from 10% to 20% results in a $0.47/kg decrease in MSP between
the two cases, and the extent of conversion (x) for which a change from 0.80 to 0.99
results in a $0.46/kg decrease in MSP between the two limits.

2486 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

In several cases, single-point sensitivity analyses are not sufficient to fully reveal process
modeling insights, namely, when process variables are strongly coupled. Thus, several
single-point sensitivity analyses were complemented with multivariate sensitivity anal-
ysis to understand the interaction of process variables and reveal examples of strongly
non-linear relationships that affect the overall rTPA selling prices. Examples of signifi-
cantly correlated variables studied here include enzyme loading and enzyme cost (Fig-
ure 4B), solids loading and enzyme loading (Figure 4C), as well as solids loading and
extent of depolymerization (Figure 4D). As an illustrative example, solids loading
coupled with the extent of depolymerization can not only drive the MSP of rTPA as
low as $1.60/kg in the range examined at the highest loadings/extents but can also drive
costs to >$2.50/kg at the lowest loadings and extents. An additional multivariate sensi-
tivity exploring enzyme loading and depolymerization is shown in Table S14.

Case study III: Product and co-product recovery

Lastly, we investigated the process variables in the downstream processing section.
Increasing the recovery of EG to 65% or rTPA to 98% will lower the MSP to $1.87/kg
(reduction by $0.05/kg) or $1.83/kg (reduction by $0.10/kg), respectively, whereas
reducing the selling price of EG and/or SS will lead to an increase in the overall
MSP (Figure 5A). Changing the market price of EG by approximately 30% in either
direction of the base case value ($0.96/kg50) changes the selling price of rTPA by
$0.12/kg between the two bounding cases. Since EG recovery leads to a significant
steam usage in distillation, a case where more EG selective membranes49 (99% EG
retention selectivity, while SS selectivity remains the same) are depolyed is modeled,
which could potentially reduce the capital and operating expenses to run the distil-
lation columns. MSP in this case can be lowered to $1.75/kg and would have indus-
try-wide implications in other EG/water recovery scenarios (EG Selective membrane,
Figure 5A).

The base case, reported in the previous case studies, assumes that both EG and SS
are recovered. Therefore, sensitivity cases around downstream process design were
explored where recovery of co-products was changed (Figures 5B and 5C). Two sce-
narios, one in which no EG is recovered (NEG) and one where neither co-product is
recovered (NCP), were considered. In the first scenario (NEG), when EG is not recov-
ered but SS is, the EG-containing solution from the SS crystallizer is sent to waste-
water treatment and the co-product credit associated with EG recovery is lost.
The rTPA MSP in this scenario (NEG) corresponds to $2.08/kg as shown in Figure 5B.
In the second scenario (NCP) where neither EG nor SS is recovered, the mother li-
quor from the TPA crystallizer is sent to wastewater treatment and the revenue op-
portunity from both co-product sales is discounted. As a result, the capital invest-
ment and utilities required for SS and EG recovery are omitted in this scenario. A
similar impact on the rTPA MSP is observed in the scenario exploring no co-product
recovery (NCP), resulting in an increased MSP of $2.10/kg (Figure 5B). Details on the
additional wastewater treatment, co-product credit losses, and the capital expense
tradeoffs can be found in Table S19.

Summary of TEA results

Based on the process variables examined in the three case studies/sections earlier, the
results of those sensitivity analyses from Figures 3, 4, and 5 were summarized to rank
cost drivers on rTPA MSP and are presented in the tornado plot (Figure 6; Table S21).

Within the parameter ranges chosen, feedstock cost is the largest contributor to the
rTPA MSP. The feedstock modeled in this study is clean PET that meets quality stan-
dards for mechanical recycling52 and drove the greatest variability in rTPA MSP as

Joule 5, 2479–2503, September 15, 2021 2487

 ll
OPEN ACCESS Article

Figure 5. TEA sensitivity results for process parameters associated with product and co-product recovery
(A and B) (A) rTPA MSP for the base case (base case values: rTPA recovery = 90%; EG recovery = 50%; EG MP = $0.96/kg; SS MP = $0.15/kg) to indicate
the effect of product and co-product recoveries as well as the co-product market prices (B) MSP as a result of different co-product recovery schemes
without the co-product(s), illustrating that although operational expenses may decrease, MSP will increase because of the lack of a co-product credit.
(C) Process flow diagram of the modified separation processes. Numerical data in this figure are reported in Table S20. Abbreviations used—MP, market
price (for co-products, averaged over 5 years); NEG, no EG is recovered; NCP, no co-product is recovered.

also noted before (Figure 3A). Next, the effect of plant scale emerged important. A
facility size of 50 MTPD would indicate a decentralized approach (with smaller collec-
tion radii) for building such enzymatic recycling facilities for PET, which increases the
MSP from the base case by 26%, whereas 300 MTPD is being investigated to see the
effect of economies of scale, which decreases the MSP by 5%.

Besides exogenous factors, such as feedstock price and plant size, the relatively high
importance of depolymerization-related sensitivities such as solids loading and the
extent of conversion stand out as the dominant factors affecting MSP, followed by

2488 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

Figure 6. Tornado plot summarizing the effect of different process variables investigated on the
rTPA MSP
Univariate analysis summarizing how process variables affect rTPA MSP across the three case
studies of feedstock pretreatment, depolymerization, and product and co-product recovery. The
rTPA MSP in the base case is $1.93/kg. Factors not shown have a minimal effect in a univariate
sensitivity analysis. Numerical data for this figure are reported in Table S21.

overall TPA recovery. Residence time for batch hydrolysis and enzyme loading would
be of particular interest to the research community, and these parameters could
change the MSP by 5%–9% within the ranges examined. Variations in the total cap-
ital expense (+/ $15M relative to the base case fixed capital investment) is also
studied. Finally, the market price of co-products, varying over a 30% range, has a
relatively minor impact on the overall MSP of rTPA.

Note that this tornado plot only includes key process variables in the base design
and does not include the different process design (i.e., NMP, NEG, and NCP), but
those design changes do have a significant impact on MSP, as noted. Additionally,
the impact of variations in process design on the overall supply chain energy and
GHG emissions, as well as broader socioeconomic effects, is discussed in the
following sections.

Bottom-up analysis of supply chain energy and GHG emissions

To quantify the energy and GHG emissions associated with the modeled supply
chain, we employed MFI, a bottom-up, supply chain modeling tool.5,37,65 Specif-
ically, MFI was used to compare the energy requirements and GHG emissions

Joule 5, 2479–2503, September 15, 2021 2489

 ll
OPEN ACCESS Article

associated with rTPA produced using enzymatic depolymerization with those of the
conventional fossil feedstock-derived vTPA supply chain (see Figure S5). The results
of the TEA modeling were integrated into the MFI database by connecting the pro-
cess inputs from the Aspen Plus model to their corresponding representations in the
MFI database of process inputs. The analysis conducted using MFI includes select
sensitivities from the previously mentioned case studies.

When comparing rTPA and vTPA, it is necessary to consider the effect of the first life of
the TPA (i.e., procurement, production, and polymerization into PET). To do so, we
define the functional unit (our basis of comparison) to be the total number of PET ‘‘life-
times’’ that can be obtained from repeatedly utilizing the enzymatic depolymerization
process on the feedstock obtained from an initial quantity of 1 kg of fossil-derived
PET. Given an overall yield from the enzymatic depolymerization process of 0.63 kg
TPA/kg of depolymerized PET (current process design), and assuming that the requisite
amount of rTPA required to manufacture new PET is approximately 0.87 kg TPA/kg PET
(MFI process data), the recycled content of the rTPA-derived PET is approximately 72%.
An infinite summation series can be constructed for multiple loops through the enzy-
matic recycling process, which converges to the ‘‘lifetime’’ amount of rTPA-derived
PET that could be achieved from a given amount of fossil-derived waste PET feedstock.
The total amount of PET is calculated to be 3.6 kg in the enzymatic depolymerization
supply chain (see supplemental experimental procedures, MFI methodology and Fig-
ure S6 for the derivation), of which, 1 kg is fossil-derived and the remaining 2.6 kg is ob-
tained from the enzymatic depolymerization process. The supply chain level impacts on
this supply chain were compared with the analogous supply chain to produce 3.6 kg of
PET from the conventional fossil-derived route. The results were then normalized to 1 kg
of PET for simplicity.

Results from the base case and select MFI analyses are presented in Figure 7 for
rTPA and PET, which indicate that there is a significant estimated energy savings poten-
tial from enzymatic depolymerization-based supply chains. The conventional
manufacturing supply chain for PET bottles requires 127 MJ/kg, (as calculated by
the MFI tool), whereas the enzymatic depolymerization supply chains require 74–
82 MJ/kg, a reduction of 35%–42%. This reduction is primarily due to a sharp reduction
in fossil feedstock energy requirements, from 56 MJ fossil feedstock/kg PET for the
conventional route to 18 MJ fossil feedstock/kg PET for the enzymatic depolymeriza-
tion route. In the latter route, fossil feedstocks arise from the production of other
required supply chain inputs (e.g., sulfuric acid, caustic, etc.). Aside from supply chain
energies, the base case and the NMP case are estimated to emit 8% and 21% lower
GHG emissions per kg of PET, respectively, compared with the fossil-derived PET base-
line, or 4.3 kg carbon dioxide equivalent (CO2e)/kg PET and 3.7 kg CO2e/kg PET,
respectively. The reason for the discrepancy between percent energy reduction and
percent GHG reduction is that fossil feedstock use, by definition, does not entail
GHG emissions because the feedstock is not combusted, but converted into products
(hence the blue bar does not appear in the bottom row of charts in Figure 7).

As also shown in Figure 7A, when the system boundary is narrowed to exclude the
polymerization step, monomer rTPA production from enzymatic depolymerization
is estimated to consume 69%–83% less energy than conventional production of
vTPA, mostly due to differences in feedstock energy requirements. In fact, the enzy-
matic depolymerization monomer production entails a negative feedstock energy
contribution due to the offset from EG and SS co-production. GHG emissions im-
pacts for monomer rTPA production exhibit a similar trend, with 17%–43% lower
GHG emissions for enzymatic depolymerization compared with fossil-derived vTPA.

2490 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

Figure 7. Supply chain energy and GHG emissions analysis

(A–D) Supply chain energy requirements for (A) TPA and (B) PET production. Supply chain GHG emissions for (C) TPA and (D) PET production. Supply
chain energy and GHG emissions are reported for the case of vTPA manufacture, the base case of this work (Base), and the three process design
sensitives of NMP, NEG, and NCP. In all of the cases reported here, the use of enzymatic processes results in a lower supply chain energy requirement
and GHG emissions. The reduction in supply chain energy requirement and GHG emissions is more pronounced for the case of TPA production over PET
production as the PET modeled here uses fossil-derived, polymerization-grade EG. Figure S5 illustrates the system boundary for MFI and numerical
data for this figure is reported in Tables S22 and S23. Abbreviations used—NMP, no mechanical pretreatment; NEG, no EG is recovered; NCP, no co-
product is recovered.

Top-down analysis of economy-wide impacts

The bio-based circular carbon economy EEIO model (BEIOM)66 was used to put the
supply chains into a US economy-wide context.38 BEIOM is a top-down, macro-level
model that assesses the economy-wide impacts of emerging technologies or port-
folios thereof. It uses process-level data to create new sectors within the US

Joule 5, 2479–2503, September 15, 2021 2491

 ll
OPEN ACCESS Article

economic structure reflecting the respective technologies at an industry scale. Orig-

inally, BEIOM was developed to quantify the effects of an expanding US bio-
economy, describing the potential benefits and unintended consequences of new
bio-based technologies across several environmental and socioeconomic impact
categories from a holistic perspective. Here, BEIOM is applied to compare the pro-
duction of conventional vTPA with rTPA produced via enzymatic depolymerization.
The inclusion of the rTPA production processes as a new sector in BEIOM followed
the same procedure as previous work evaluating different bio-based processes.38
Specifics on how each TPA pathway was integrated into BEIOM can be found in sup-
plemental experimental procedures and Figure S7, with detailed results illustrated in
Figures S8–S32.

The impacts of producing 1 kg of vTPA compared with 1 kg of rTPA include direct

impacts (generated by the plant itself and its supply chain) and indirect impacts
generated by related activities across all other sectors of the US economy (Figure 8).
To quantify impacts on a per kg basis, the total impacts (direct and indirect) from
either TPA production were divided by the total TPA output of the respective indus-
try. Additional co-products (EG and SS) are accounted for and their impacts are allo-
cated via substitution (on an economic basis). In the rTPA base case, EG and SS are
both recovered. Two sensitivity cases are also modeled to evaluate scenarios where
no EG is recovered (NEG), and where neither EG nor SS are recovered (NCP). We
also model a scenario that recovers both co-products but excludes mechanical pre-
treatment (NMP).

We estimate that rTPA generates socioeconomic benefits for both added economic
value and jobs, except in the NMP scenario where the exclusion of mechanical pretreat-
ment reduces overall economic effects (Figure 8A). Further, the socioeconomic effects
are more concentrated for rTPA in the conversion plant and its direct supply chain,
whereas they are spread out more broadly across the economy for vTPA. Two thirds
of the jobs from rTPA production are concentrated in the conversion process and the
direct inputs for the process. Production of vTPA sustains more jobs further upstream
in the production chain, where one third of the jobs are in the service sectors.

In the resource use metrics category (Figure 8B), rTPA shows a much higher water with-
drawal than vTPA, except when mechanical pretreatment is excluded (NMP case). This is
exclusively caused by water use accounted for in the US national electricity generation
mix for 2017 and is a consequence of higher electricity consumption modeled to pro-
duce rTPA as compared with vTPA. It suggests that rTPA plants sourcing electricity
with a low water footprint could alleviate this difference. Further, we find that rTPA re-
duces overall land occupation by at least 40% compared with vTPA. In both cases,
the resource use impacts are mainly attributable to economic activities, related to,
but beyond the rTPA plant and its supply chain, i.e., are indirect effects generated by
the system within which the supply chains are operating in. Most land occupation effects
are indirect (i.e., not attributable to the specific conversion facilities) and related to agri-
culture and the oil extraction sectors (land leases). For rTPA, 34% comes from forestry,
27% from grain farming, 12% from oil and gas extraction. For conventional TPA, oil and
gas extraction effects increase to 25%, whereas forestry and grain farming effects are
relatively similar to those from rTPA at 33% and 26%, respectively (Figure S1).

Across all (midpoint) life-cycle impact categories evaluated, rTPA shows an improve-
ment in the domestic footprint by 10%–95% over vTPA (Figure 8C). vTPA impacts are
mainly caused by the TPA production step across all metrics, whereas for rTPA, only
the freshwater ecotoxicity is dominated by the production process. Instead, rTPA

2492 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

Figure 8. Socioeconomic and environmental impacts of producing 1 kg of TPA when accounting for the US economy-wide effects
(A–C) Impact from the production of 1 kg of TPA for each scenario presented for (A) socioeconomic, (B) resource use, and (C) environmental impact
categories. Scenarios include the production of vTPA, rTPA (Base), rTPA with no mechanical pretreatment (NMP), rTPA with no EG recovery (NEG), and
rTPA with no EG or SS recovery (NCP). Numerical data in this figure are reported in Table S24.

impacts are largely caused by upstream processes. Electricity generation also con-
tributes notably to smog formation, acidification, and respiratory effects in the
rTPA cases.

Comparing the rTPA base case with the two co-product disposal sensitivity cases
(NEG, NCP) in Figure 8, we find that co-product recovery reduces the environmental
effects across resource use and impact categories per kg of rTPA. Excluding the me-
chanical pretreatment step (NMP) reduces the environmental impacts of the base
case (rTPA) by up to 65%; however, it also lowers the socioeconomic benefits by
10%–20%.

Joule 5, 2479–2503, September 15, 2021 2493

 ll
OPEN ACCESS Article

Capturing economy-wide effects expands the system boundary beyond the supply
chain and generally increases the effects per metric (see Figure 8 ‘‘remaining supply
chain’’ segments for vTPA and rTPA). Although the inclusion of indirect effects does
not favor either one of the two products analyzed, its exclusion would alter results
toward a further improvement of environmental benefits from rTPA while dramati-
cally reducing the socioeconomic benefits attributable to vTPA.

DISCUSSION
In response to the pollution crisis caused by waste plastics, new recycling processes
are needed for which requisite analyses will be critical to understanding the key cost
drivers, supply chain impacts, barriers to scale-up, and market implementation. Here,
we present a comprehensive analysis of enzyme-based PET recycling that includes
rigorous process modeling, economic impacts of key process variables, supply chain
energy, and GHG emissions impacts compared with fossil-based incumbent
manufacturing, and examination of life cycle and socioeconomic impacts. These an-
alyses demonstrate that enzyme-based PET recycling exhibits substantial promise,
which can be improved with further process advancements, as detailed here.

Feedstock considerations
The current, global approach to PET recycling nearly universally employs mechanical
recycling of post-consumer single-use PET bottles, and less than 20% of the eligible
waste stream is processed in the US (less than 7% worldwide67). Furthermore, me-
chanical recycling imposes stringent contamination tolerances,52 necessitating
extensive preprocessing and cleaning steps, significantly impacting reclaimed PET
flake prices.68 Accordingly, plastic recycling technologies that are agnostic to the
feedstock quality and expand the waste streams eligible for recycling will be effec-
tive in changing the recycling landscape.

The enzyme-based chemical recycling process presented here is one such prospec-
tive technology. Enzymatic recycling offers the advantage of high selectivity for the
target PET substrate, thereby reducing the need for stringent contamination toler-
ances, opening up the pool of eligible waste feedstocks, and creating an opportu-
nity for lower feedstock costs. Multiple feedstocks and conversion pathways should
be considered for their economic and environmental impacts, and compared with
mechanical recycling, which is already efficient but limited by the purity of the feed-
stock required. In the end, a combination of different technologies, ranging from
mechanical recycling to more substrate selective chemical recycling, will likely
need to be deployed to enable higher PET-recycling rates.

Size of the PET-recycling facility

Aside from feedstock cost, there are other engineering considerations that are key
cost drivers in realizing an enzymatic recycling process for PET, including the facility
size. The facility size will likely be dictated by local and regional collection rates and
whether any incentives exist that encourage recycling participation (e.g., the US
states with bottle deposit programs demonstrate higher collection volumes).39 In
this study, a plant with the capacity to process 150 MTPD was modeled. However,
the operation of any recycling plant is dependent on securing a steady and suffi-
ciently large feedstock supply. Considering that the largest MRFs in the US can pro-
cess about 1,000 MTPD of post-consumer waste,69 if the incoming waste stream is
about 3.7% PET by weight70 and a recovery of 91% PET can be obtained,71 only
33.7 Mt of PET could be secured from such a facility per day. This estimates that
the largest MRFs in the country are unlikely to obtain more than 50 MTPD of PET
that is suitable for mechanical recycling.

2494 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

Although the advent of new technologies, including the enzymatic process pre-
sented here, may lead to an enhanced recycling capacity, this analysis highlights
that it will still be necessary for any chemical recycling facilities to obtain PET-based
feedstock from multiple MRFs, or to include additional sources of PET previously
excluded from consideration for mechanical recycling, such as textiles, fibers, and
layered packaging materials. It is important to note that the average distance for
waste PET transport significantly varies throughout the US72 and in other parts of
the world;73–77 therefore, additional factors may influence the economics if the plant
size is increased. For example, in some regions, it might be necessary to establish
centralized recycling facilities to increase the collection volumes of post-consumer
PET78 at the expense of greater transportation costs. Securing a consistent feed-
stock supply by analyzing statistical and geospatial distribution of plastic waste
will be critical to de-risk the operation of any recycling facility.

Substrate considerations
Any requirement for size reduction of the feedstock and/or change in PET crystal-
linity will contribute to the economics, energy demands, and GHG emissions of
the overall process. This is exemplified by the NMP case, which lowers rTPA MSP
to $1.70/kg (12% less than the baseline MSP), generates 38% lower GHG emissions,
and requires 45% lower supply chain energies. Pretreatment that is akin to mechan-
ical recycling has been proposed to enable greater enzymatic process efficiencies.
Specifically, the recent study by Tournier et al. reports an engineered PET hydrolase
enzyme capable of up to 85% product yield from post-consumer waste PET within
20 h of treatment,35,79,80 where the PET was pre-treated by extrusion, amorphiza-
tion, and micronization to reduce the substrate crystallinity and size, respectively.
Size reduction has also been proposed to enhance the extent of depolymerization
by increasing the accessible surface area for enzyme action,81 however, systematic
investigation is necessary to understanding the influence of these factors on enzyme
performance. Additionally, the community will need to find an optimum size reduc-
tion that enables a high extent of depolymerization, desirable for increasing
throughput, and beyond which the additional cost of size reduction may not be
offset by incremental product yield.81–85

Biocatalyst performance
Enzyme performance and optimization remains an open area of opportunity with
important implications for the economics of an enzyme-based recycling process.
Here, we highlight that solids loading, enzyme loading, and enzyme cost are key
cost drivers; however, this is predicated on other assumptions in this model, the
key among them being that the enzyme can achieve a high extent of depolymeriza-
tion of the PET feedstock. To achieve a high extent of depolymerization, the global
research community is investigating multiple parameters including the optimization
of enzyme performance, specificity for substrates of different crystallinities, toler-
ance to inhibition by products and contaminants, and improved stability for
enhanced catalytic lifetime.

The best-studied PET hydrolases to date are seemingly most active on the amor-
phous regions of the polymer.34 Continued optimization and engineering of enzyme
performance on PET substrates of different crystallinities will be critical, as the two
largest applications of PET (bottles and fibers) exhibit high (30%–40%) crystallinity.85
By leveraging enzyme design and exploring the use of thermophilic enzymes, im-
provements across multiple factors can be achieved. For instance, enzymes active
at approximately 70 C may exploit the reduction in crystallinity as PET approaches
its glass transition temperature. Enhanced thermostability may be engineered into

Joule 5, 2479–2503, September 15, 2021 2495

 ll
OPEN ACCESS Article

PET-degrading enzymes as shown by the stabilizing effect of glycosylation on leaf-

compost cutinase (LCC),28 or the replacement of a potential calcium-binding site
in LCC with a disulfide bond35; in each case, the thermal stabilization was accompa-
nied by enhanced PET hydrolysis activity at an elevated temperature.

Enzyme design can also enable faster turnover, product selectivity, and inhibition
resistance, all of which are ideal for mixed feedstock streams (e.g., textiles with
dyes). Recently, accumulation of additives has been documented to inhibit cutinase
activity,81 and therefore, methods to overcome inhibition, or remove contami-
nants,86 could improve enzyme performance. In addition, given the inevitable acid-
ification of the solution that occurs upon ester bond hydrolysis, any acid-tolerant
PET-degrading enzymes that exhibit a broad pH optimum range would reduce
the need to continually neutralize the depolymerization reaction, thus saving on
caustic additive. Interestingly, acid-tolerant cutinases have been identified87–89
with an enzyme from the fungus Thielavia terrestris exhibiting esterase activity
over the pH range 4–8 and detectable PET-degrading activity at pH 4.0 and 50 C.87

Combining all of the properties mentioned earlier—thermostability, acid- and prod-

uct-tolerance, and a high turnover of PET of varied crystallinities—represents a major
challenge for future enzyme engineering efforts.

Enzyme production
Beyond enzyme function, the work presented here assumes that the enzyme is pro-
duced offsite, which leads to the MSP exhibiting a linear dependence with enzyme
cost. Factors that drive enzyme production costs include the origin of the enzyme,
the expression host, and the level of purification required for adequate activity.
The price of generic hydrolase used in our base design is modeled after extracellular
secretion from the host organism and concentration of the enzyme in the crude
broth, without any elaborate purification. For intracellular protein production, puri-
fication requirements are key cost drivers.56,59,90–92 However, many industrial en-
zymes are secreted from the host therefore avoiding the processing steps associated
with cell lysis.56,59,90 Stabilizing agents are also frequently used to ensure the
longevity of the enzymes. However, consistent supply and compatibility of any sta-
bilizers with enzyme function are important considerations in the development of an
enzyme product.

The production of cellulase enzymes in filamentous fungi offers an analogous system

for comparison and benchmarking, and this approach was used as the baseline for
enzyme costs in the current work. In the case of cellulases, fungi such as Trichoderma
reesei have been evolved and engineered over decades to secrete native and het-
erologous cellulases to extremely high levels (protein concentrations on the order
of 100 g/L).93,94 Thus, filamentous fungi offer an excellent starting point for future
development of PET-active enzyme production. Alternative eukaryotic expression
hosts may come to the forefront, for example, the methylotrophic yeast Pichia pas-
toris, which has been utilized for the recombinant expression of a variety of esterase
enzymes.28,89,95–100 To date, laboratory-scale fermentations of P. pastoris have pro-
duced recombinant cutinases at yields up to 10.8 g/L.95,96,100

Product and co-product recovery

Although pretreatment and enzyme optimization may influence the kinetics of PET
hydrolysis, efficient product recovery will also have a major impact on the rTPA
MSP, as this determines which products can be sold and at what purity. The process
modeled here assumes that dyes, pigments, and adhesives are removed by the solid

2496 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

filtration and activated carbon column.35,42–44 Spent enzymes are removed by ultra-
filtration in the base case, which may not be a necessary step if there is sufficient
removal by the activated carbon column,101 thereby resulting in a lower MSP.
Although this design is aligned with reported practices, alternate and emergent sep-
aration techniques for product recovery can be potentially explored. For example,
TPA could be recovered by sublimation in a process akin to that developed by East-
man.102 The recovery of EG is contingent on the fact that all the added SS salt is
recovered from the solution prior to sending it to distillation columns, which remains
to be verified. As the recovery of EG is an energy-intensive process,103,104 alternative
separation schemes, (e.g., liquid-liquid extraction, membrane-based separations,
condensed phase chromatographic separations techniques, reactive distillation fol-
lowed by hydrolysis, etc.) could also be explored, which may eventually reduce the
MSP and process energy requirements.44,103,105–113

Sustainability impacts
In terms of the environmental impact analysis of the supply chain modeling, it is clear
that a reduction in feedstock energy (in the form of natural gas and crude oil) is a pri-
mary contributor to the estimated reduction in supply chain energy for the enzymatic
depolymerization relative to fossil-derived routes to PET. However, feedstock en-
ergy alone does not offer an opportunity to reduce GHG emissions (since feedstocks
are, by definition, not combusted). Process-based energy requirements (e.g., from
steam utilization) are shown to be significantly reduced in the enzymatic depolymer-
ization routes at the expense of increased electricity usage. GHG emissions associ-
ated with the supply chain’s electricity requirements are highly dependent on the
assumed grid mix of generation technologies. MFI supply chain models reflect cur-
rent industrial practices within the US industry; for electricity specifically, MFI as-
sumes a national grid mix as of 2016.114 Since the modeled enzymatic depolymeriza-
tion supply chain has been shown to be relatively more electricity-intensive than
conventional PET production, additional GHG emissions reductions could be real-
ized with an underlying electricity grid mix having a higher penetration of lower-
or non-GHG-emitting, renewable (e.g., solar and wind) generation sources.

Furthermore, analysis with BEIOM highlights that, within the rTPA plant design, the crys-
tallization process might warrant a more in-depth assessment to reduce process emis-
sions. In the present design, it shows the highest share of process emissions and result-
ing relative contributions to the impact categories assessed (Figure S30). Sulfuric acid
emissions are driving ecotoxicity and acidification impacts; EG (a regulated hazardous
air pollutant) is driving human toxicity; both EG and sulfuric acid affect smog formation,
along with volatile organic compounds from equipment leaks.

Social and environmental equity

The economy-wide comparison of vTPA and rTPA shows that the environmental and so-
cioeconomic effects of the two processes are distributed unequally across their supply
chain tiers (see also Figure S10). Environmental impacts from vTPA largely result from
the production process itself, whereas its socioeconomic benefits mainly occur across
the broader economy and are not necessarily regional. The distribution of effects
from rTPA are reversed with socioeconomic benefits largely occurring at the facility itself
and its direct suppliers, whereas most environmental impacts result from indirect activ-
ities further away from the rTPA supply chain. To account for and detail these social and
environmental equity considerations, future analysis should further specify effects in
regional contexts and recommend process alterations, if possible, to achieve an appro-
priate balance.

Joule 5, 2479–2503, September 15, 2021 2497

 ll
OPEN ACCESS Article

Beyond conventional substrates for recycling

Given the considerations presented, this work provides targets for process develop-
ment toward realizing a more comprehensive recycling strategy than is available
today. Leveraging the selectivity of biological depolymerization offers the opportu-
nity to process mixed waste plastics and expand the eligible range of feedstocks.
This concept can be extended to an integrated depolymerization refinery where
the action of a cocktail of enzymes could break down the mixed plastic waste stream
in a one-pot or tandem fashion. Changing feedstock, especially to materials that are
not recycled today, would also allow for expanded plant sizes to offer the benefit of
economies of scale for processing varied plastic types at higher volumes, particularly
since PET bottles comprise only 30% of all PET produced globally. Other forms of
PET waste, including discarded fabrics, textiles, and waste carpets (polyester fibers)
comprised 44 MMT (or 54%) of global PET demand in 20196 also present an oppor-
tunity for the application of enzyme-based recycling.

Importantly, the recycling approach described here will likely mirror the enzyme-
based processing of other polymers, especially those that contain linkages that
mirror natural polymers (e.g., nylons, polyurethanes, polycarbonates).115,116
Although each polymer will exhibit its own unique challenges, they may have
similar characteristics that increase their ease of recycling. As an illustrative
example, polyurethanes frequently feature crosslinking, which may limit access
to enzyme activity or slow reaction rates; however, polyurethanes may also
have a lower glass transition temperature and present a higher surface area, fea-
tures that may accelerate reaction kinetics.19 The enzymatic recycling process
model and subsequent analyses described here can therefore provide a founda-
tion for evaluating other new enzymatic technologies, as well as broad guidance
for the developing field.

Conclusions
Here, we describe the process design of an enzyme-catalyzed PET-recycling pro-
cess and discuss the key factors impacting the process economics, supply chain
energy, GHG emissions, socioeconomic impacts, and job creation potential for
the enzyme-based recycling strategy as compared with the fossil-derived produc-
tion currently employed. The TEA results estimate the potential for rTPA produc-
tion at $1.93/kg, and the study outlines certain scenarios to further reduce costs.
In terms of pretreatment, feedstock prices and mechanical preprocessing are the
largest contributing steps to cost and energy requirements. Meanwhile, in the
depolymerization process, the extent of PET breakdown, solids loading, enzyme
price, and enzyme loading are all key cost drivers. Importantly, the yields of
recovered TPA and co-products (i.e., SS and EG) dictate the overall throughput
of the plant and significantly impacts the economics. Supply chain analysis shows
that PET recycling via the enzymatic route has lower supply chain energy require-
ment and lower supply chain GHG emissions compared to vTPA. An economy-
wide assessment shows the potential socioeconomic (and largely regional) bene-
fits of adding a plastics recycling sector and details environmental impacts per
process and supply chain steps to highlight potential improvement areas for
reducing unintended environmental consequences. There are multiple technolo-
gies in development as an alternative to the current mechanical PET-recycling
strategies (e.g., chemo-catalytic, pyrolysis, etc.) of which enzymatic recycling of-
fers the benefit of providing substrate specificity for complex mixed feedstock
streams. Although the framework provided here is focused on PET reclaimed
from an MRF, the insights presented here apply more broadly to polymeric sys-
tems that are susceptible to enzymatic depolymerization.

2498 Joule 5, 2479–2503, September 15, 2021

 ll
Article OPEN ACCESS

EXPERIMENTAL PROCEDURES
Resource availability
Lead contact
Further information and requests for resources and materials should be directed to and
will be fulfilled by the lead contact, Gregg T. Beckham (gregg.beckham@nrel.gov).

Materials availability
This study did not generate new unique materials.

Data and code availability

The datasets in this article are provided in full in the supplemental information. The
Aspen Plus models can be provided upon request and can be replicated by the in-
puts shown in the detailed PFD in the supplemental information.

The MFI tool is freely available to interested users on the NREL website. Accounts
can be created there to use the tool. There are restrictions to the availability of
the MFI code and database due to the proprietary nature of several process inven-
tory datasets used by the tool.

Methods are described throughout the main text and in detail in the supplemental
information where appropriate. In general, Aspen Plus models were constructed
for all process and TEA analysis. MFI and BEIOM were used to quantify various sup-
ply chain and economy-wide effects.

SUPPLEMENTAL INFORMATION
Supplemental information can be found online at https://doi.org/10.1016/j.joule.
2021.06.015.

ACKNOWLEDGMENTS
This work was authored in part by Alliance for Sustainable Energy, LLC, the manager
and operator of the National Renewable Energy Laboratory for the U.S. Department
of Energy (DOE) under contract no. DE-AC36-08GO28308. Funding was provided by
the US Department of Energy, Office of Energy Efficiency and Renewable Energy,
Advanced Manufacturing Office (AMO) and Bioenergy Technologies Office (BETO).
This work was performed as part of the Bio-Optimized Technologies to keep Thermo-
plastics out of Landfills and the Environment (BOTTLE) Consortium and was supported
by AMO and BETO under contract no. DE-AC36-08GO28308 with the National Renew-
able Energy Laboratory (NREL), operated by Alliance for Sustainable Energy, LLC. The
BOTTLE Consortium includes members from the University of Portsmouth, funded un-
der contract no. DE-AC36-08GO28308 with NREL and additionally supported by
Research England (E3 scheme). The views expressed in the article do not necessarily
represent the views of the DOE or the U.S. Government. We thank our colleagues in
the BOTTLE Consortium for helpful discussions and Ryan Davis and Emily Newes for
a critical reading of the paper.

AUTHOR CONTRIBUTIONS
A.S. and J.S.D. developed the process and economic models. N.A.R., A.R.P., J.E.M.,
E.E., and G.T.B. contributed to the process design and analysis. S.R.N. and A.C.C.
conducted the supply chain modeling. A.F.T.A., P.L., Y.Z., A.B., and G.A. conducted
the economy-wide modeling. A.S., N.A.R., S.R.N., E.E., P.L., and G.T.B. wrote the
manuscript with contributions from all authors.

Joule 5, 2479–2503, September 15, 2021 2499

 ll
OPEN ACCESS
DECLARATION OF INTERESTS
G.T.B., N.A.R., E.E., A.R.P., J.E.M., and G.T.B. have submitted patent applications
on enzymes for PET recycling. N.A.R. and G.T.B. have submitted patent applications
on upcycled materials from PET.
INCLUSION AND DIVERSITY
One or more of the authors of this paper self-identifies as an underrepresented ethnic
minority in science. One or more of the authors of this paper self-identifies as a member
of the LGBTQ+ community. While citing references scientifically relevant for this work,
we also actively worked to promote gender balance in our reference list.
Received: March 16, 2021
Revised: May 16, 2021
Accepted: June 16, 2021
Published: July 15, 2021
REFERENCES
1. Geyer, R., Jambeck, J.R., and Law, K.L. (2017).
Production, use, and fate of all plastics ever
made. Sci. Adv. 3, e1700782.
2. Jambeck, J.R., Geyer, R., Wilcox, C., Siegler,
T.R., Perryman, M., Andrady, A., Narayan, R.,
and Law, K.L. (2015). Marine pollution. Plastic
waste inputs from land into the ocean.
Science 347, 768–771.
3. Law, K.L., Starr, N., Siegler, T.R., Jambeck,
J.R., Mallos, N.J., and Leonard, G.H. (2020).
The United States’ contribution of plastic
waste to land and ocean. Sci. Adv. 6,
eabd0288.
4. Rahimi, A., and Garcı́a, J.M. (2017). Chemical
recycling of waste plastics for new materials
production. Nat. Rev. Chem. 1, 0046.
5. Nicholson, S.R., Rorrer, N.A., Carpenter, A.C.,
and Beckham, G.T. (2021). Manufacturing
energy and greenhouse gas emissions
associated with plastics consumption. Joule
5, 673–686.
6. Bescond, A.-S., and Pujari, A. (2020). PET
polymer. In Chemical Economics Handbook
(IHS Markit), p. 32. https://ihsmarkit.com/
products/chemical-economics-handbooks.
html.
7. Nayak, A., Pujari, A., and Sharma, S. (2020).
Polyethylene terephthalate (PET) solid-state
resins. In Chemical Economics Handbook
(IHS Markit)), p. 10. https://ihsmarkit.com/
products/polyethylene-terephthalate-resins-
chemical-economics-handbook.html.
8. Ormonde, E., Loechner, U., Yoneyama, M.,
and Zhu, X. (2019). Plastics recycling.
Chemical Economics Handbook (IHS Markit),
p. 56. https://ihsmarkit.com/products/
plastics-recycling-chemical-economics-
handbook.html.
9. Sinha, V., Patel, M.R., and Patel, J.V. (2010).
PET waste management by chemical
recycling: a review. J. Polym. Environ. 18,
8–25.
10. Paszun, D., and Spychaj, T. (1997). Chemical
recycling of poly (ethylene terephthalate). Ind.
Eng. Chem. Res. 36, 1373–1383.
2500 Joule 5, 2479–2503, September 15, 2021
11. Pollard, M., Beisson, F., Li, Y., and Ohlrogge,
J.B. (2008). Building lipid barriers:
biosynthesis of cutin and suberin. Trends
Plant Sci. 13, 236–246.
12. Ralph, J. (2010). Hydroxycinnamates in
lignification. Phytochem. Rev. 9, 65–83.
13. Scheller, H.V., and Ulvskov, P. (2010).
Hemicelluloses. Annu. Rev. Plant Biol. 61,
263–289.
14. Ronkvist, Å.M., Xie, W., Lu, W., and Gross,
R.A. (2009). Cutinase-catalyzed hydrolysis of
poly (ethylene terephthalate).
Macromolecules 42, 5128–5138.
15. Loredo-Treviño, A., Gutiérrez-Sánchez, G.,
Rodrı́guez-Herrera, R., and Aguilar, C.N.
(2012). Microbial enzymes involved in
polyurethane biodegradation: a review.
J. Polym. Environ. 20, 258–265.
16. Sulaiman, S., Yamato, S., Kanaya, E., Kim, J.J.,
Koga, Y., Takano, K., and Kanaya, S. (2012).
Isolation of a novel cutinase homolog with
polyethylene terephthalate-degrading
activity from leaf-branch compost by using a
metagenomic approach. Appl. Environ.
Microbiol. 78, 1556–1562.
17. Roth, C., Wei, R., Oeser, T., Then, J., Föllner,
C., Zimmermann, W., and Sträter, N. (2014).
Structural and functional studies on a
thermostable polyethylene terephthalate
degrading hydrolase from Thermobifida
fusca. Appl. Microbiol. Biotechnol. 98, 7815–
7823.
18. Yoshida, S., Hiraga, K., Takehana, T.,
Taniguchi, I., Yamaji, H., Maeda, Y., Toyohara,
K., Miyamoto, K., Kimura, Y., and Oda, K.
(2016). A bacterium that degrades and
assimilates poly (ethylene terephthalate).
Science 351, 1196–1199.
19. Magnin, A., Pollet, E., Phalip, V., and Avérous,
L. (2020). Evaluation of biological degradation
of polyurethanes. Biotechnol. Adv. 39,
107457.
20. Han, X., Liu, W., Huang, J.W., Ma, J., Zheng,
Y., Ko, T.P., Xu, L., Cheng, Y.S., Chen, C.C.,
and Guo, R.T. (2017). Structural insight into
Article
catalytic mechanism of PET hydrolase. Nat.
Commun. 8, 2106.
21. Wei, R., and Zimmermann, W. (2017).
Biocatalysis as a green route for recycling the
recalcitrant plastic polyethylene
terephthalate. Microb. Biotechnol. 10, 1302–
1307.
22. Wei, R., and Zimmermann, W. (2017).
Microbial enzymes for the recycling of
recalcitrant petroleum-based plastics: how far
are we? Microb. Biotechnol. 10, 1308–1322.
23. Austin, H.P., Allen, M.D., Donohoe, B.S.,
Rorrer, N.A., Kearns, F.L., Silveira, R.L.,
Pollard, B.C., Dominick, G., Duman, R., El
Omari, K., et al. (2018). Characterization and
engineering of a plastic-degrading aromatic
polyesterase. Proc. Natl. Acad. Sci. USA 115,
E4350–E4357.
24. Danso, D., Schmeisser, C., Chow, J.,
Zimmermann, W., Wei, R., Leggewie, C., Li, X.,
Hazen, T., and Streit, W.R. (2018). New
insights into the function and global
distribution of polyethylene terephthalate
(PET)-degrading bacteria and enzymes in
marine and terrestrial metagenomes. Appl.
Environ. Microbiol. 84, 02773–02717.
25. Fecker, T., Galaz-Davison, P., Engelberger, F.,
Narui, Y., Sotomayor, M., Parra, L.P., and
Ramı́rez-Sarmiento, C.A. (2018). Active site
flexibility as a hallmark for efficient PET
degradation by I. sakaiensis PETase. Biophys.
J. 114, 1302–1312.
26. Joo, S., Cho, I.J., Seo, H., Son, H.F., Sagong,
H.Y., Shin, T.J., Choi, S.Y., Lee, S.Y., and Kim,
K.J. (2018). Structural insight into molecular
mechanism of poly (ethylene terephthalate)
degradation. Nat. Commun. 9, 382.
27. Nikolaivits, E., Kanelli, M., Dimarogona, M.,
and Topakas, E. (2018). A middle-aged
enzyme still in its prime: recent advances in
the field of cutinases. Catalysts 8, 612.
28. Shirke, A.N., White, C., Englaender, J.A.,
Zwarycz, A., Butterfoss, G.L., Linhardt, R.J.,
and Gross, R.A. (2018). Stabilizing leaf and
branch compost cutinase (LCC) with

Article
glycosylation: mechanism and effect on PET
hydrolysis. Biochemistry 57, 1190–1200.
29. Su, A., Shirke, A., Baik, J., Zou, Y., and Gross,
R. (2018). Immobilized cutinases: preparation,
solvent tolerance and thermal stability.
Enzyme Microb. Technol. 116, 33–40.
30. Danso, D., Chow, J., and Streit, W.R. (2019).
Plastics: environmental and biotechnological
perspectives on microbial degradation. Appl.
Environ. Microbiol. 85, 01019–010195.
31. Furukawa, M., Kawakami, N., Tomizawa, A.,
and Miyamoto, K. (2019). Efficient
degradation of poly (ethylene terephthalate)
with Thermobifida fusca cutinase exhibiting
improved catalytic activity generated using
mutagenesis and additive-based
approaches. Sci. Rep. 9, 16038.
32. Palm, G.J., Reisky, L., Böttcher, D., Müller, H.,
Michels, E.A.P., Walczak, M.C., Berndt, L.,
Weiss, M.S., Bornscheuer, U.T., and Weber,
G. (2019). Structure of the plastic-degrading
Ideonella sakaiensis MHETase bound to a
substrate. Nat. Commun. 10, 1717.
33. Son, H.F., Cho, I.J., Joo, S., Seo, H., Sagong,
H.-Y., Choi, S.Y., Lee, S.Y., and Kim, K.-J.
(2019). Rational protein engineering of
thermo-stable PETase from Ideonella
sakaiensis for highly efficient PET
degradation. ACS Catal. 9, 3519–3526.
34. Taniguchi, I., Yoshida, S., Hiraga, K.,
Miyamoto, K., Kimura, Y., and Oda, K. (2019).
Biodegradation of PET: current status and
application aspects. ACS Catal. 9, 4089–4105.
35. Tournier, V., Topham, C.M., Gilles, A., David,
B., Folgoas, C., Moya-Leclair, E., Kamionka,
E., Desrousseaux, M.L., Texier, H., Gavalda,
S., et al. (2020). An engineered PET
depolymerase to break down and recycle
plastic bottles. Nature 580, 216–219.
36. Knott, B.C., Erickson, E., Allen, M.D., Gado,
J.E., Graham, R., Kearns, F.L., Pardo, I.,
Topuzlu, E., Anderson, J.J., Austin, H.P., et al.
(2020). Characterization and engineering of a
two-enzyme system for plastics
depolymerization. Proc. Natl. Acad. Sci. USA
117, 25476–25485.
37. Hanes, R.J., and Carpenter, A. (2017).
Evaluating opportunities to improve material
and energy impacts in commodity supply
chains. Environ. Syst. Decis. 37, 6–12.
38. Lamers, P., Avelino, A.F.T., Zhang, Y., Tan,
E.C.D., Young, B., Vendries, J., and Chum, H.
(2021). Potential socioeconomic and
environmental effects of an expanding U.S.
bioeconomy: an assessment of near-
commercial cellulosic biofuel pathways.
Environ. Sci. Technol. 55, 5496–5505.
39. NAPCOR (2018). 2017 PET recycling report
(National Association for PET Container
Resources (NAPCOR)). https://napcor.com/
reports-resources/.
40. Recycling Markets (2021). Secondary
materials pricing. https://www.
recyclingmarkets.net/.
41. Closed Loop Partners (2017). Cleaning the
rPET Stream: How We Scale Post-Consumer
Recycled PET in the US (Closed Loop
Partners). https://www.mitchellwilliamslaw.
com/webfiles/2%20Cleaning%20your%
20rPET%20Stream%20.pdf.
42. Chateau, M. (2020). Method for producing
terephthalic acid on an industrial scale (World
Patent No. WO 2020/094661 A1) (World
Intellectual Property Organization (WIPO)).
43. Karp, E.M., Cywar, R.M., Manker, L.P., Saboe,
P.O., Nimlos, C.T., Salvachúa, D., Wang, X.,
Black, B.A., Reed, M.L., Michener, W.E., et al.
(2018). Post-fermentation recovery of
biobased carboxylic acids. ACS Sustainable
Chem. Eng. 6, 15273–15283.
44. Brown, G.E., Jr., and O’Brien, R.C. (1976).
Method for recovering terephthalic acid and
ethylene glycol from polyester materials (U.S.
Patent No. US 3952053) (U.S. Patent and
Trademark Office).
45. Fard, M.H., and Vakili, M.H. (2010). Chemical
recycling of poly ethylene terephthalate
wastes. World Applied Sciences Journal 8,
839–846.
46. Lamparter, R.A., Barna, B.A., and Johnsrud,
D.R. (1985). Process for recovering
terephthalic acid from waste polyethylene
terephthalate (U.S. Patent No. US 4542239)
(U.S. Patent and Trademark Office).
47. Jehle, W., Staneff, T., Wagner, B., and
Steinwandel, J. (1995). Separation of glycol
and water from coolant liquids by
evaporation, reverse osmosis and
pervaporation. J. Membr. Sci. 102, 9–19.
48. Li, W., Van der Bruggen, B., and Luis, P. (2014).
Integration of reverse osmosis and
membrane crystallization for sodium sulphate
recovery. Chem. Eng. Technol. 85, 57–68.
49. Biancari, A., Palma, L.D., Ferrantelli, P., and
Merli, C. (2003). Ethylene glycol recovery from
dilute aqueous solution. Environ. Eng. Sci. 20,
103–110.
50. Chang, R.J., and Lacson, J. (2020). Process
Economics Tool - PEP Yearbook (IHS Markit).
51. Eriksen, M.K., Damgaard, A., Boldrin, A., and
Astrup, T.F. (2019). Quality assessment and
circularity potential of recovery systems for
household plastic waste. J. Ind. Ecol. 23,
156–168.
52. The Association of Plastics Recyclers – APR.
(2021). PET Resin specifications. https://
plasticsrecycling.org/markets/model-bale-
specs/pet-resin-specifications.
53. Moore, P., and Staub, C. (2020). Slipping
through the Cracks. Plastics Recycling
Update. https://resource-recycling.com/
plastics/2020/05/19/slipping-through-the-
cracks/.
54. Enviros consulting. (2009). MRF quality
assessment study. Project Code: MRF011.
https://docplayer.net/20953314-Mrf-quality-
assessment-study.html.
55. Ferreira, R.D.G., Azzoni, A.R., and Freitas, S.
(2018). Techno-economic analysis of the
industrial production of a low-cost enzyme
using E. coli: the case of recombinant beta-
glucosidase. Biotechnol. Biofuels 11, 81.
56. Mcvey, T.F. (2002). Industrial enzymes process
economics program report - 139A (IHS Markit,
SRI Consulting). https://ihsmarkit.com/
ll
OPEN ACCESS
products/chemical-technology-pep-
industrial-enzymes-2002.html.
57. Davis, R.E., Grundl, N.J., Tao, L., Biddy, M.J.,
Tan, E.C., Beckham, G.T., Humbird, D.,
Thompson, D.N., and Roni, M.S. (2018).
Process design and economics for the
conversion of lignocellulosic biomass to
hydrocarbon fuels and coproducts: 2018
biochemical design case update; biochemical
deconstruction and conversion of biomass to
fuels and products via integrated biorefinery
pathways (National Renewable Energy
Laboratory (NREL)), NREL/TP-5100-71949.
https://www.nrel.gov/docs/fy19osti/71949.
pdf.
58. Humbird, D., Davis, R., Tao, L., Kinchin, C.,
Hsu, D., Aden, A., Schoen, P., Lukas, J.,
Olthof, B., and Worley, M. (2011). Process
design and economics for biochemical
conversion of lignocellulosic biomass to
ethanol: dilute-acid pretreatment and
enzymatic hydrolysis of corn stover (National
Renewable Energy Laboratory(NREL)),
Technical Report NREL/TP-5100-47764.
https://www.nrel.gov/docs/fy11osti/47764.
pdf.
59. Klein-Marcuschamer, D., Oleskowicz-Popiel,
P., Simmons, B.A., and Blanch, H.W. (2012).
The challenge of enzyme cost in the
production of lignocellulosic biofuels.
Biotechnol. Bioeng. 109, 1083–1087.
60. Biddy, M.J., Davis, R., Humbird, D., Tao, L.,
Dowe, N., Guarnieri, M.T., Linger, J.G., Karp,
E.M., Salvachúa, D., Vardon, D.R., and
Beckham, G.T. (2016). The techno-economic
basis for coproduct manufacturing to enable
hydrocarbon fuel production from
lignocellulosic biomass. ACS Sustainable
Chem. Eng. 4, 3196–3211.
61. Gregg, D.J., Boussaid, A., and Saddler, J.N.
(1998). Techno-economic evaluations of a
generic wood-to-ethanol process: effect of
increased cellulose yields and enzyme
recycle. Bioresour. Technol. 63, 7–12.
62. Sassner, P., and Zacchi, G. (2008). Integration
options for high energy efficiency and
improved economics in a wood-to-ethanol
process. Biotechnol. Biofuels 1, 4.
63. Wingren, A., Galbe, M., and Zacchi, G. (2003).
Techno-economic evaluation of producing
ethanol from softwood: comparison of SSF
and SHF and identification of bottlenecks.
Biotechnol. Prog. 19, 1109–1117.
64. Kawai, F., Kawabata, T., and Oda, M. (2020).
Current state and perspectives related to the
polyethylene terephthalate hydrolases
available for biorecycling. ACS Sustainable
Chem. Eng. 8, 8894–8908.
65. Rorrer, N.A., Nicholson, S., Carpenter, A.,
Biddy, M.J., Grundl, N.J., and Beckham, G.T.
(2019). Combining reclaimed PET with bio-
based monomers enables plastics upcycling.
Joule 3, 1006–1027.
66. Lamers, P. Bio-based circular carbon
economy environmentally-extended input-
output Model (BEIOM). https://
bioenergymodels.nrel.gov/models/42/.
67. World Economic Forum, Ellen MacArthur
Foundation, and McKinsey & Company
(2015). Project MainStream – a global
Joule 5, 2479–2503, September 15, 2021 2501
 ll
OPEN ACCESS
collaboration to accelerate the transition
towards the circular economy: status update.
http://www3.weforum.org/docs/
WEF_Project_Mainstream_Status_2015.pdf.
68. Awaja, F., and Pavel, D. (2005). Recycling of
PET. Eur. Polym. J. 41, 1453–1477.
69. Powell, J. (2018). Sortation by the numbers.
https://resource-recycling.com/recycling/
2018/10/01/sortation-by-the-numbers/.
70. Combs, A.R. (2011). Life cycle analysis of
recycling facilities in a carbon constrained
world. https://repository.lib.ncsu.edu/
handle/1840.16/7808.
71. Raymond, A.G. (2017). Modeling of material
recovery facility performance with
applications for life cycle assessment, PhD
thesis (Massachusetts Institute of
Technology).
72. Kuczenski, B., Geyer, R., and Bren, D. (2011).
Life cycle assessment of polyethylene
terephthalate (PET) beverage bottles
consumed in the state of California
(Department of Resources Recycling and
Recovery). https://www2.calrecycle.ca.gov/
Publications/Details/1487.
73. Merrild, H., Larsen, A.W., and Christensen,
T.H. (2012). Assessing recycling versus
incineration of key materials in municipal
waste: the importance of efficient energy
recovery and transport distances. Waste
Manag 32, 1009–1018.
74. Chen, X., Xi, F., Geng, Y., and Fujita, T. (2011).
The potential environmental gains from
recycling waste plastics: simulation of
transferring recycling and recovery
technologies to Shenyang, China. Waste
Manag. 31, 168–179.
75. Nakatani, J., Fujii, M., Moriguchi, Y., and
Hirao, M. (2010). Life-cycle assessment of
domestic and transboundary recycling of
post-consumer PET bottles. Int. J. Life Cycle
Assess. 15, 590–597.
76. Shen, L., Worrell, E., and Patel, M.K. (2010).
Open-loop recycling: a LCA case study of PET
bottle-to-fibre recycling. Resour. Conserv.
Recycl. 55, 34–52.
77. Martin, E.J.P., Oliveira, D.S.B.L., Oliveira,
L.S.B.L., and Bezerra, B.S. (2020). Dataset for
life cycle assessment of pet bottle waste
management options in Bauru, Brazil. Data
Brief 33, 106355.
78. Lonca, G., Lesage, P., Majeau-Bettez, G.,
Bernard, S., and Margni, M. (2020). Assessing
scaling effects of circular economy strategies:
A case study on plastic bottle closed-loop
recycling in the USA PET market. Resour.
Conserv. Recycl. 162, 105013.
79. Boisart, C., and Maille, E. (2014). Method for
recycling plastic products (World Patent No.
WO 2014/079844 A1) (World Intellectual
Property Organization (WIPO)).
80. Maille, E. (2017). Process of recycling mixed
PET plastic articles (U.S. Patent Application
Pub. No. US 2017/0114205 A1) (U.S. Patent
and Trademark Office).
81. Gamerith, C., Zartl, B., Pellis, A., Guillamot, F.,
Marty, A., Acero, E.H., and Guebitz, G.M.
(2017). Enzymatic recovery of polyester
2502 Joule 5, 2479–2503, September 15, 2021
building blocks from polymer blends. Process
Biochem 59, 58–64.
82. Kadhum, H.J., Rajendran, K., and Murthy, G.S.
(2017). Effect of solids loading on ethanol
production: experimental, economic and
environmental analysis. Bioresour. Technol.
244, 108–116.
83. Humbird, D., Mohagheghi, A., Dowe, N., and
Schell, D.J. (2010). Economic impact of total
solids loading on enzymatic hydrolysis of
dilute acid pretreated corn stover.
Biotechnol. Prog. 26, 1245–1251.
84. Modenbach, A.A., and Nokes, S.E. (2013).
Enzymatic hydrolysis of biomass at high-solids
loadings – a review. Biomass Bioenergy 56,
526–544.
85. Kawai, F., Kawabata, T., and Oda, M.
(2019). Current knowledge on enzymatic
PET degradation and its possible
application to waste stream management
and other fields. Appl. Microbiol.
Biotechnol. 103, 4253–4268.
86. Ügdüler, S., Van Geem, K.M., Roosen, M.,
Delbeke, E.I.P., and De Meester, S. (2020).
Challenges and opportunities of solvent-
based additive extraction methods for plastic
recycling. Waste Manag 104, 148–182.
87. Yang, S., Xu, H., Yan, Q., Liu, Y., Zhou, P., and
Jiang, Z. (2013). A low molecular mass
cutinase of Thielavia terrestris efficiently
hydrolyzes poly(esters). J. Ind. Microbiol.
Biotechnol. 40, 217–226.
88. Nyyssölä, A., Pihlajaniemi, V., Järvinen, R.,
Mikander, S., Kontkanen, H., Kruus, K., Kallio,
H., and Buchert, J. (2013). Screening of
microbes for novel acidic cutinases and
cloning and expression of an acidic cutinase
from Aspergillus niger CBS 513.88. Enzyme
Microb. Technol. 52, 272–278.
89. Nyyssölä, A., Pihlajaniemi, V., Häkkinen, M.,
Kontkanen, H., Saloheimo, M., and Nakari-
Setälä, T. (2014). Cloning and characterization
of a novel acidic cutinase from Sirococcus
conigenus. Appl. Microbiol. Biotechnol. 98,
3639–3650.
90. Johnson, E. (2016). Integrated enzyme
production lowers the cost of cellulosic
ethanol. Biofuels Bioprod. Bioref. 10,
164–174.
91. Klein-Marcuschamer, D., Oleskowicz-Popiel,
P., Simmons, B.A., and Blanch, H.W. (2010).
Technoeconomic analysis of biofuels: a wiki-
based platform for lignocellulosic
biorefineries. Biomass Bioenergy 34, 1914–
1921.
92. Kazi, F.K., Fortman, J.A., Anex, R.P., Hsu, D.D.,
Aden, A., Dutta, A., and Kothandaraman, G.
(2010). Techno-economic comparison of
process technologies for biochemical ethanol
production from corn stover. Fuel 89,
S20–S28.
93. Le Crom, S., Schackwitz, W., Pennacchio, L.,
Magnuson, J.K., Culley, D.E., Collett, J.R.,
Martin, J., Druzhinina, I.S., Mathis, H., Monot,
F., et al. (2009). Tracking the roots of cellulase
hyperproduction by the fungus Trichoderma
reesei using massively parallel DNA
sequencing. Proc. Natl. Acad. Sci. USA 106,
16151–16156.
Article
94. Cherry, J.R., and Fidantsef, A.L. (2003).
Directed evolution of industrial enzymes: an
update. Curr. Opin. Biotechnol. 14, 438–443.
95. Duan, X., Liu, Y., You, X., Jiang, Z., Yang, S.,
and Yang, S. (2017). High-level expression and
characterization of a novel cutinase from
malbranchea cinnamomea suitable for butyl
butyrate production. Biotechnol. Biofuels 10,
223.
96. Duan, X., Jiang, Z., Liu, Y., Yan, Q., Xiang, M.,
and Yang, S. (2019). High-level expression of
codon-optimized Thielavia terrestris cutinase
suitable for ester biosynthesis and
biodegradation. Int. J. Biol. Macromol. 135,
768–775.
97. Hong, R., Sun, Y., Su, L., Gu, L., Wang, F., and
Wu, J. (2019). High-level expression of
Humicola insolens cutinase in pichia pastoris
without carbon starvation and its use in cotton
fabric bioscouring. J. Biotechnol. 304, 10–15.
98. Gamerith, C., Vastano, M., Ghorbanpour,
S.M., Zitzenbacher, S., Ribitsch, D., Zumstein,
M.T., Sander, M., Herrero Acero, E., Pellis, A.,
and Guebitz, G.M. (2017). Enzymatic
degradation of aromatic and aliphatic
polyesters by P. pastoris expressed cutinase 1
from Thermobifida cellulosilytica. Front.
Microbiol. 8, 938.
99. Kwon, M.A., Kim, H.S., Yang, T.H., Song, B.K.,
and Song, J.K. (2009). High-level expression
and characterization of Fusarium solani
cutinase in pichia pastoris. Protein Expr. Purif.
68, 104–109.
100. Seman, W.M., Bakar, S.A., Bukhari, N.A.,
Gaspar, S.M., Othman, R., Nathan, S., Mahadi,
N.M., Jahim, J., Murad, A.M., and Bakar, F.D.
(2014). High level expression of Glomerella
cingulata cutinase in dense cultures of pichia
pastoris grown under fed-batch conditions.
J. Biotechnol. 184, 219–228.
101. Daoud, F.B., Kaddour, S., and Sadoun, T.
(2010). Adsorption of cellulase Aspergillus
niger on a commercial activated carbon:
kinetics and equilibrium studies. Colloids Surf.
B Biointerfaces 75, 93–99.
102. Tustin, G.C., Pell, T.M., Jr., Jenkins, D.A., and
Jernigan, M.T. (1995). Process for the recovery
of terephthalic acid and ethylene glycol from
poly(ethylene terephthalate) (US Patent No.
5413681) (U.S. Patent and Trademark Office).
103. Spychaj, T. (2005). Chemical recycling of PET:
methods and products. In Handbook of
Thermoplastic Polyesters (John Wiley & Sons),
pp. 1252–1290.
104. Dye, R.F. (2001). Ethylene glycols technology.
Korean J. Chem. Eng. 18, 571–579.
105. Shahverdi, M., Mohammadi, T., and Pak, A.
(2011). Separation of ethylene glycol-water
mixtures with composite poly(vinyl alcohol)-
polypropylene membranes. J. Appl. Polym.
Sci. 119, 1704–1710.
106. Nik, O.G., Moheb, A., and Mohammadi, T.
(2006). Separation of ethylene glycol/water
mixtures using NaA zeolite membranes.
Chem. Eng. Technol. 29, 1340–1346.
107. Huang, R.Y.M., Shao, P., Feng, X., and
Anderson, W.A. (2002). Separation of ethylene
glycol water mixtures using sulfonated
poly(ether ether ketone) pervaporation

Article
membranes: membrane relaxation and
separation performance analysis. Ind. Eng.
Chem. Res. 41, 2957–2965.
108. Chen, Fengrong, and Chen, Hongfang (1996).
Pervaporation separation of ethylene glycol/
water mixtures using crosslinked PVA/PES
composite membranes. Part II. The swelling
equilibrium model of the dense active layer in
ethylene glycol/water mixtures. J. Membr. Sci.
118, 169–176.
109. Feng, X. (1996). Pervaporation with chitosan
membranes. I. Separation of water from
ethylene glycol by a chitosan/polysulfone
composite membrane. J. Membr. Sci. 116,
67–76.
110. Chen, F.R., and Chen, H.F. (1996).
Pervaporation separation of ethylene glycol-
water mixtures using crosslinked PVA-PES
composite membranes. Part I. Effects of
membrane preparation conditions on
pervaporation performances. J. Membr. Sci.
109, 247–256.
111. Mohammadi, T., and Akbarabadi, M. (2005).
Separation of ethylene glycol solution by
vacuum membrane distillation (VMD).
Desalination 181, 35–41.
112. Garcia-Chavez, L.Y., Schuur, B., and de Haan,
A.B. (2012). Liquid–liquid equilibrium data
for mono ethylene glycol extraction
from water with the new ionic liquid
tetraoctyl ammonium 2-methyl-1-
naphtoate as solvent. J. Chem. Thermodyn.
51, 165–171.
113. Garcia-Chavez, L.Y., Hermans, A.J., Schuur,
B., and de Haan, A.B. (2012). COSMO-RS
assisted solvent screening for liquid–liquid
ll
OPEN ACCESS
extraction of mono ethylene glycol from
aqueous streams. Sep. Purif. Technol. 97,
2–10.
114. U.S. Energy Information Administration (EIA)
(2018). Annual Energy Outlook 2018, Table 8,
Table 16. https://www.eia.gov/outlooks/
archive/aeo18/.
115. Brueckner, T., Eberl, A., Heumann, S., Rabe,
M., and Guebitz, G.M. (2008). Enzymatic and
chemical hydrolysis of poly(ethylene
terephthalate) fabrics. J. Polym. Sci. A Polym.
Chem. 46, 6435–6443.
116. Wei, R., Tiso, T., Bertling, J., O’Connor,
K., Blank, L.M., and Bornscheuer, U.T.
(2020). Possibilities and limitations
of biotechnological plastic degradation
and recycling. Nat. Catal. 3, 867–
871.
Joule 5, 2479–2503, September 15, 2021 2503
Joule, Volume 5

Supplemental information

Techno-economic, life-cycle, and socioeconomic

impact analysis of enzymatic recycling
of poly(ethylene terephthalate)
Avantika Singh, Nicholas A. Rorrer, Scott R. Nicholson, Erika Erickson, Jason S.
DesVeaux, Andre F.T. Avelino, Patrick Lamers, Arpit Bhatt, Yimin Zhang, Greg
Avery, Ling Tao, Andrew R. Pickford, Alberta C. Carpenter, John E.
McGeehan, and Gregg T. Beckham
Supplemental Items

Part I – Base Case Model Construction

• Figure S1: Detailed PFD of feedstock pretreatment and depolymerization process sections (base
case).
• Figure S2: Detailed PFD of clarification and product crystallization process sections (base case).
• Figure S3: Detailed PFD of co-product crystallization process sections (base case).
• Figure S4: Detailed PFD of ethylene glycol recovery process section (base case).
• Table S1. Financial parameters used in discounted cash flow (DCF) analysis.
• Table S2. Economic summary of recycled terephthalic acid (rTPA) production.
• Table S3. Capital investment breakdown by process section (base case).
• Table S4: Yearly operating cost breakdown (base case).
• Table S5: Historical prices of chemicals (related to Figures 2A, 2B) and common utilities.
• Table S6: Simplified breakdown of the minimum selling price of rTPA in the base case, related to
Figure 2C.
• Table S7: MFI results on supply chain energy and GHG emissions for base case, related to Figures
2D, 2E.
• Table S8: CAPEX Breakdown for the base case, related to Figure 2F.
• Table S9: Annual operating cost by process section for the base case, related to Figure 2G
• Table S10: Minimum selling price of rTPA by process section for the base case, related to Figure
2H.
• Table S11: PET Bale and flake conversion data, related to Figure 3A.
Part II – Technoeconomic Case Studies Supporting Data
• Table S12: Plant, feedstock, and pretreatment sensitivity results, related to Figure 3B and 3C.
• Table S13: Results of analysis on reactor temperature and pH.
• Table S14: Multivariate analysis of MSP ($/kg rTPA) as a function of depolymerization extent and
enzyme loading (mg/g PET) multivariate.
• Table S15: Depolymerization sensitivity results, related to Figure 4A.
• Table S16: Multivariate analysis of MSP ($/kgrTPA) as a function of enzyme loading (mg/g PET) and
enzyme cost ($/kg), related to Figure 4B.
• Table S17: Multivariate analysis of MSP ($/kgrTPA) as a function of enzyme loading (mg/g PET) and
solids loading (wt. %), related to Figure 4C.
• Table S18: Multivariate analysis of MSP ($/kg rTPA) as a function of depolymerization extent and
solids loading (wt. %), related to Figure 4D.
• Table S19: Summary of capital expense, waste treatment cost, and co-product recovery credits for
the no mechanical pretreatment (NMP), no ethylene glycol recovery (NEG), and no co-product
recovery (NCP) process design variations.
• Table S20: Product and co-product recovery sensitivity results, related to Figure 5A and 5B.
• Table S21: Univariate sensitivity summary, related to Figure 6.
• Table S22: MFI supply chain energy requirements for both a monomer and polymer basis, related
to Figures 7A and 7B.
• Table S23: MFI supply chain GHG emission results for both a monomer and polymer basis, related
to Figures 7C and 7D.
• Table S24: Normalized BEIOM Data, related to Figure 8.
Part III – MFI Methods and Supplemental Data
• Figure S5. Materials Flows through Industry Tool System Boundary for conventional (fossil-derived)
PET and rPET produced via enzymatic depolymerization.
• Figure S6. Cumulative lifetime quantity of PET (in kg) obtained from processing an initial 1kg
quantity of fossil-derived PET through repeated cycles through the enzymatic depolymerization and
subsequent repolymerization process.
Part IV – BEIOM Methods and Supplemental Data
• Figure S7. Inputs and outputs of conventional and virgin TPA pathways.
• Figure S8. Supply chain tiers: Value added.
• Figure S9. Value added contribution by sector for each fuel pathway.
 • Figure S10. Supply chain tiers: Employment.
• Figure S11. Employment contribution by sector for each fuel pathway.
• Figure S12. Supply chain tiers: Water withdrawals.
• Figure S13. Water withdrawals contribution by sector for each fuel pathway.
• Figure S14. Supply chain tiers: Land occupation
• Figure S15. Land occupation contribution by sector for each fuel pathway.
• Figure S16. Supply chain tiers: Smog formation potential.
• Figure S17. Smog formation potential contribution by sector for each fuel pathway.
• Figure S18. Supply chain tiers: Eutrophication potential.
• Figure S19. Eutrophication potential contribution by sector for each fuel pathway.
• Figure S20. Supply chain tiers: Acidification potential.
• Figure S21. Acidification potential contribution by sector for each fuel pathway.
• Figure S22. Supply chain tiers: Freshwater ecotoxicity potential.
• Figure S23. Freshwater ecotoxicity potential contribution by sector for each fuel pathway.
• Figure S24. Supply chain tiers: Human toxicity potential.
• Figure S25. Human toxicity potential contribution by sector for each fuel pathway.
• Figure S26. Supply chain tiers: Respiratory effects potential.
• Figure S27. Respiratory effects potential contribution by sector for each fuel pathway.
• Figure S28. Supply chain tiers: Ozone depletion potential.
• Figure S29. Ozone depletion potential contribution by sector for each fuel pathway.
• Figure S30. Contribution of rTPA processes to the total impacts per plant.
• Figure S31. Influence of commodity inputs to the impacts of the rTPA process
• Figure S32. Economy-wide impacts from the production of 1 kg of TPA per pathway from an
industry perspective (i.e., no economic separation/allocation of impacts across products, Eq. 8).
Virgin TPA: vTPA, Recycled TPA (base): rTPA, Recycled TPA (no EG recovery): rTPA_S1,
Recycled TPA (no EG/salt recovery): rTPA_S2, Recycled TPA (no mechanical pre-treatment):
rTPA_S3.

Supplemental Experimental Procedures

MFI Methodology: Derivation of Functional Unit Used in MFI Calculations
Economy-wide analysis (BEIOM): Data and Methods

Supplemental References
Part I – Base Case Model Construction

Figure S1: Detailed PFD of feedstock pretreatment and depolymerization process sections (base case). PET flakes and other raw material inputs
are shown in red; intermediate, recycle, and waste streams are shown in black; and product and co-product streams are shown in blue.
Figure S2: Detailed PFD of clarification and product crystallization process sections (base case). PET flakes and other raw material inputs are
shown in red; intermediate, recycle, and waste streams are shown in black; and product and co-product streams are shown in blue.
Figure S3: Detailed PFD of co-product crystallization process section (base case). PET flakes and other raw material inputs are shown in red;
intermediate, recycle, and waste streams are shown in black; and product and co-product streams are shown in blue.
Figure S4: Detailed PFD of ethylene glycol recovery process section (base case). PET flakes and other raw material inputs are shown in red;
intermediate, recycle, and waste streams are shown in black; and product and co-product streams are shown in blue.
Supplemental Tables

Table S1. Financial parameters used in discounted cash flow (DCF) analysis.

Discounted Cash Flow Financial Parameters

Equity (% of FCI) 40%
Loan Interest 8.0%
Loan Term, years 10
Working Capital (% of FCI) 5.00%
Discount Rate (IRR) 10%
Income Tax Rate 21%
Plant Depreciation Period (Years) 7
Plant Life (Years) 30
Construction Period (Years) 3
% Spent in Year -2 8%
% Spent in Year -1 60%
% Spent in Year 0 32%
Start-up Time (Years) 0.50
rTPA production year 1 (% of Normal Capacity) 50%
Variable Costs (% of Normal) 75%
Fixed Cost (% of Normal) 100%
Land Requirement Acres 10
Land Cost $/acre $14,000
Table S2. Economic summary of recycled terephthalic acid (rTPA) production

Production of recycled Terephthalic Acid (rTPA) from the Enzymatic Depolymerization of PET
PET Pretreatment, Enzymatic Depolymerization, Crystallization of rTPA and SS, and Distillation of
All Values in 2016$
rTPA Minimum Selling Price (MSP) $1.93 /kg
Contributions:
Feedstock $1.26 /kg
rTPA Product and Co-product Conversion $0.97 /kg
EG Co-product Credit -$0.20 /kg
SS Co-product Credit -$0.10 /kg

rTPA Production 31.1 Mkg per year

rTPA Yield 0.66 tonnes rTPA / tonne PET feed
EG Co-product Production 6.5 Mkg per year
EG Co-product Yield 0.14 tonnes / tonne PET feed
SS Co-product Production 21.8 Mkg per year
SS Recovery 0.47 tonnes /tonne PET feed

Feedstock + Handling Cost $600 /dry U.S. ton feedstock

Internal Rate of Return (After-Tax) 10%
Equity Percent of Total Investment 40%

Capital Costs Manufacturing Costs (cents/kgrTPA)

Feedstock Pretreatment $13,974,736 Feedstock + Handling 104.8
Depolymerization $13,487,655 Sulfuric Acid 6.6
Clarification $6,502,702 Caustic 9.1
Crystallization $17,655,708 OSBL Utilities 7.5
EG Recovery $1,893,211 Electricity (import) 8.5
OSBL (25% of ISBL) $13,378,503 Other Raw Materials 1.4
Total Installed Equipment Cost $66,892,515 Enzymes 10.7
Additional Direct Costs (17.5% of ISBL) $6,919,373 Waste Disposal 9.7
Total Direct Costs (TDC) $73,811,889 EG Co-product -19.9
Indirect Costs (60% of TDC) $44,287,133 SS Co-product -10.3
Land $140,000 Fixed Costs 14.2
Working Capital (5.0% of FCI) $5,904,951 Capital Depreciation 12.5
Total Capital Investment (TCI) $124,100,00 Average Income Tax 3.8
Average Return on 34.6
Installed Equipment Cost/Annual kg $2.15 Total 193.2
Total Capital Investment/Annual kg $3.99
Manufacturing Costs ($/yr)
Operating Hours Per Year (On-Stream 7884 (90%) Feedstock + Handling $32,600,00
Loan Rate 8.0% Sulfuric Acid $2,000,000
Term (years) 10 Caustic $2,800,000
OSBL Utilities $2,300,000
Electricity (import) $2,600,000
Carbon Retention Efficiencies: Other Raw Materials $400,000
rTPA Carbon Retention 61.5% Enzymes $3,300,000
% (C in rTPA / C in PET) Waste Disposal $3,000,000
EG Carbon Retention 8.5% EG Co-product -
% (C in rTPA / C in PET) SS Co-product -
Fixed Costs $4,400,000
Product Recoveries (% of Maximum Theoretical Yield): Capital Depreciation $3,900,000
Overall rTPA 90.0% Average Income Tax $1,200,000
Overall EG Recovery 50.0% Average Return on $10,800,00
Overall SS Recovery 76.1% Total $59,900,00
Table S3. Capital investment breakdown by process section (base case).
Equipment Type Installation Capital Cost (2016$)
Factor
Equipment Installed
Cost Cost
Feedstock Pretreatment
Conveyor transfer system 1.70 3,933,582 6,687,089
Truck dumper package 1.70 224,887 382,307
Dust collection system 1.70 56,872 96,683
Feedstock storage dome 1.70 711,158 1,208,969
Shredder 1.70 210,000 357,000
Extruder 1.70 2,409,639 4,096,387
Microgranulator 1.70 674,295 1,146,301
Feedstock pretreatment total ($) 8,220,433 13,974,736
PET Depolymerization
Small particle conveyor 1.00 820,543 820,543
Enzyme storage and transfer 1.81 294,648 534,177
Water recirculation pump 2.30 6,417 14,759
Depolymerization reactors (7) 1.98 5,837,187 11,560,606
Storage tanks 1.93 102,746 198,535
Reactor effluent cooler 3.90 21,905 85,429
Caustic storage and transfer 1.97 138,836 273,606
PET depolymerization total ($) 7,222,282 13,487,655
Clarification
Solids filter package 1.74 2,250,175 3,905,450
Ultra-filtration membrane 2.39 619,877 1,482,089
Activated carbon columns 2.36 472,334 1,115,164
Clarification total ($) 3,342,387 6,502,702
Crystallization
rTPA crystallizer 2.50 2,868,818 7,172,045
rTPA centrifuge 2.30 639,362 1,470,532
rTPA crystals dryer 1.50 923,378 1,383,342
Membrane unit 1.75 173,182 303,068
SS crystallizer 2.50 1,831,833 4,579,582
SS centrifuge 2.30 411,867 947,293
SS dryer 1.58 1,002,284 1,588,497
Caustic storage and transfer 1.56 126,783 197,732
Sulfuric acid storage and transfer 7,970 13,616
1.71

Crystallization total ($) 7,985,475 17,655,708

EG Recovery
Crossflow heat exchanger 5.86 6,095 35,731
Distillation column 1 2.00 379,712 760,362
Distillation column 2 1.84 431,587 794,811
EG storage 2.00 151,153 302,306
EG Recovery total ($) 968,547 1,893,211
OSBL
Outside battery limits capital (25% of 6,933,939 $13,378,503
ISBL)
OSBL total ($) 6,933,939 $13,378,503

TOTAL ($) 34,669,694 66,892,515

Table S4: Yearly operating cost breakdown (base case).

Operating Parameters

Metric/Parameter Value Units

Annual operating hours 7,884 hrs/yr
On stream factor 0.9 %
Feedstock contaminants 5.0% wt %
Feedstock PET 46,811 Tonnes/yr
Total rTPA rate 31,112 Tonnes/yr
rTPA yield 0.66 Tonnes rTPA / tonne PET feed
PET flake feed mass flow 6,250 kg/h
TPA production rate 3,946 kg/h

Variable Operating Costs

Process Hierarchy/Section Raw Material/Utility Mass Flow, kg/hr $M/yr (2016)

Raw Materials
Feedstock Pretreatment PET flake feedstock 6,250 32.59
Caustic (as pure) 2,133 2.78
PET Depolymerization Water Makeup 22,683 0.03
Enzyme Consumption 28 3.34
Ultrafiltration unit replacement - 0.14
Clarification
Activated carbon replacement - 0.25
Sulfuric Acid, 93 wt. % 2,702 2.05
Crystallization Caustic (as pure), reneutralize 51 0.07
Membrane replacement - 0.01
MP Steam, 175°C/114 psig 10,381 1.19
HP Steam, 250°C/560 psig 5,513 0.63
OSBL Utilities Cooling water 1,162,662 0.36
Chiller water, 4°C 2 0.17
Grid electricity 6,929 2.64
Subtotal 46.24
Waste Streams
Clarification Solids disposal 4,649 1.26
Crystallization Brine disposal, permeate 17,330 1.45
Crystallization Brine disposal, SS wastewater 3,707 0.31
Subtotal 3.02
Co-Products and Credits
Crystallization SS co-product 2,769 3.21
EG Recovery EG co-product 8199 6.20
Subtotal 9.41
Total Variable Operating Costs 39.85
Fixed Operating Costs
 Position $M/yr (2016)
Labor & Supervision
Total Salaries (Managers, Supervisors, Engineers, Technicians, Administrative Staff) 1.27
Labor Burden (90% of Total Salaries) 1.14
Other Overhead
Maintenance 3.0% of ISBL 1.19
Property Insurance & Tax 0.7% of FCI 0.83
Total Fixed Operating Costs 4.42
$M/yr (2016)
Total Operating Costs 44.27

Table S5: Historical prices of chemicals (related to Figures 2A, 2B) and common utilities.
Price Terephthalic Ethylene Sodium Sulfuric Sodium Process Cooling High
Year Acida Glycola Sulfateb Acida Hydroxidec Watera Watera Pressure
(vTPA) (EG) (SS) (H2SO4) (NaOH) Steama
$/tonne $/tonne $/tonne $/tonne $/tonne cents/m3 cents/m3 $/tonne
2010 996.50 892.90 148 42.20 494 32.86 3.53 19.84
2011 1478.90 1283.80 154 78.50 632 34.18 3.91 18.30
2012 1381.20 1078.70 154 100.10 672 34.53 4.14 15.65
2013 1405.20 1191.80 154 98.50 659 41.74 4.20 16.76
2014 1075 1000.20 154 94.60 604 36.19 3.91 16.53
2015 928.60 762 171 93.30 567 32.76 3.49 10.80
2016 927 769.90 111 94.60 537 33.55 3.65 12.79
2017 989.20 969.80 106 93.90 699 26.15 3.62 12.35
2018 1215.60 921.50 100 71.90 845 28 3.88 14.99
2019 1021.20 612.20 100 44.10 761 28 3.54 10.14
2020 980.40 587.10 100 45.90 - 28 3.43 8.60
2012-
2016 1143.40 960.54 148.80 96.00 607.80 35.75 3.88 14.51
Averaged
a. IHS Markit PEP Yearbook
b. IHS Markit Sodium Sulfate CEH
c. IHS Markit Chlorine/Sodium Hydroxide (Chlor-Alkali) CEH
d. The five-year price average for 2012-2016; used in the DCF economic analysis to determine rTPA
MSP.
Table S6: Simplified breakdown of the minimum selling price of rTPA in the base case, related to Figure
2C.

Cost Category Cost Contribution ($/kgrTPA)

Feedstock 1.05

Enzyme 0.11
Electricity 0.08
Caustic 0.09
Steam 0.06
Capital Charge 0.51
Operational Cost 0.34
Co-product Credit -0.30
MSP 1.93
Table S7: Minimum selling price of rTPA by process section for the base case, related to Figure 2D.

Cost Category ($M/year)

Process Area
Capital Operational Co-
Feedstock Enzyme Electricity Caustic Steam Total
Charge Costs product
Feedstock
1.05 0.00 0.08 0.00 0.00 0.11 0.03 0.00 1.26
Pretreatment
PET
0.00 0.11 0.00 0.09 0.00 0.10 0.03 0.00 0.33
Depolymerization
Clarification 0.00 0.00 0.00 0.00 0.00 0.05 0.08 0.00 0.13
Crystallization 0.00 0.00 0.01 0.00 0.00 0.13 0.15 -0.10 0.19
EG Recovery 0.00 0.00 0.00 0.00 0.06 0.01 0.04 -0.20 -0.09
OSBL 0.00 0.00 0.00 0.00 0.00 0.10 0.00 0.00 0.10
Total 1.05 0.11 0.08 0.09 0.06 0.51 0.34 -0.30 1.93

Table S8: CAPEX Breakdown for the base case, related to Figure 2E.

Process Section $M % TIC

Feedstock
14.0 20.9%
Pretreatment
PET
13.5 20.2%
Depolymerization
Clarification 6.5 9.7%
Crystallization 17.7 26.4%
EG Recovery 1.9 2.8%
OSBL 13.4 20.0%
Total Installed Cost 66.9 100.0%
Table S9: Annual operating cost by process section for the base case, related to Figure 2F.

Cost Category ($M/year)

Process
Section Raw Waste Co- Fixed Sub-
Electricity Steam Enzyme
Materials Treatment product Cost Total
Feedstock
32.64 2.41 0.00 0.00 0.00 0.00 0.00 35.05
Pretreatment
PET
2.93 0.02 0.00 3.34 0.00 0.00 0.00 6.29
Depolymerization
Clarification 0.39 0.04 0.00 0.00 1.26 0.00 0.00 1.69
Crystallization 2.25 0.17 0.02 0.00 1.76 -3.21 0.00 0.98
EG Recovery 0.24 0.00 1.80 0.00 0.00 -6.20 0.00 -4.16
Fixed Cost 0.00 0.00 0.00 0.00 0.00 0.00 4.42 4.42
Total 38.45 2.64 1.82 3.34 3.02 -9.41 4.42 44.27

Table S10: MFI results on supply chain energy and GHG emissions for base case, related to Figures 2G,
2H.
Supply Chain Energy Requirements (MJ/kg PET)
Fossil
Process Fuel for Renewable Fuel for Total
Case Chemical
Fuel Electricity Electricity Transportation Energy
Feedstocks
Fossil-derived
38.68 4.690 0.340 3.050 50.670 97.430
(vTPA)
Base Case
16.00 18.100 1.300 1.300 -6.100 30.600
(rTPA)
Supply Chain GHG Emissions (kg CO2e/kgPET)
Total
Electricity
Case Process Fuel Transportation GHG
Generation
Emission
Fossil-derived
2.382 0.341 0.262 2.985
(vTPA)
Base Case
1.049 1.321 0.111 2.480
(rTPA)
Table S11: PET Bale and flake conversion data, related to Figure 3A.
Date National Average Bale Pricea Conversion Cost Estimated Average PET
($/kg) to Flakesb ($/kg) Flake Price ($/kg)
Low High Avg
5/23/2014 0.46 0.52 0.48 0.42 0.90
4/30/2015 0.28 0.34 0.30 0.42 0.72
8/7/2015 0.29 0.36 0.31 0.42 0.73
2/11/2016 0.17 0.22 0.19 0.42 0.61
7/30/2016 0.23 0.27 0.24 0.42 0.66
11/30/2016 0.20 0.26 0.22 0.42 0.64
8/11/2017 0.33 0.37 0.35 0.42 0.77
1/3/2018 0.28 0.33 0.30 0.42 0.72
7/31/2018 0.35 0.40 0.37 0.42 0.79
1/18/2019 0.31 0.35 0.33 0.42 0.75
6/7/2019 0.32 0.37 0.34 0.42 0.76
1/17/2020 0.18 0.23 0.21 0.42 0.62
4/10/2020 0.18 0.22 0.20 0.42 0.62
9/30/2020 0.12 0.15 0.14 0.42 0.55
a. Recycling Markets PET bale price (picked-up), includes up to 30% recycled material.1
b. Closed Loop Partners estimated cost of PET bale to flake conversion (converted from 19 cents/lb
to 0.42 $/kg).2
Part II – Technoeconomic Case Studies Supporting Data
Table S12: Plant, feedstock, and pretreatment sensitivity results, related to Figure 3B and 3C.
Case Description Abbre- Cost Category Contribution (cents/kgrTPA) MSP
viation (cents/
Feedstock Enzyme Electricity Steam Caustic Operational Capital Co-Product kgrTPA)
Cost Charge Credits
Base Case
Base Case Base 105 11 8 6 9 34 51 -30 193
Plant Sensitivities
Plant size of 50
105 11 9 11 9 55 75 -30 243
tonnes per day 50 MTPD
Plant size of 300
105 11 8 6 9 29 47 -30 184
tonnes per day 300 MTPD
Capital expenses
105 11 8 6 9 33 39 -30 181
(CAPEX) of $50M
Capital expenses
105 11 8 6 9 34 60 -30 202
(CAPEX) of $80M
Feedstock and Pretreatment Sensitivities
No mechanical pre-
105 11 1 6 9 32 37 -30 169
treatment NMP
Feedstock cost of
-17 11 8 6 9 34 51 -30 71
-0.11 $/kg
Feedstock cost of
0 11 8 6 9 34 51 -30 88
0.00 $/kg
Feedstock cost of
35 11 8 6 9 34 51 -30 123
0.22 $/kg
Feedstock cost of
70 11 8 6 9 34 51 -30 158
0.44 $/kg
Feedstock cost of
140 11 8 6 9 34 51 -30 228
0.88 $/kg
Feedstock cost of
175 11 8 6 9 34 51 -30 263
1.10 $/kg
PET weight fraction
109 11 9 6 9 35 53 -30 202
of 0.91 of feedstock ΦPET = 0.91
PET weight fraction
101 11 8 6 9 32 49 -30 185
of 0.99 of feedstock ΦPET = 0.99
Table S13: Results of analysis on reactor temperature and pH.

Sensitivity Parameter Range Bound MSP ($/kgrTPA)

Sensitivity
Lower Base Higher Lower MSP Higher MSP
Units
Case Case Case Value Value

Reactor temperature °C 40 60 80 1.93 1.93

Reactor pH – 6.0 8.0 10.0 1.93 1.93

Table S14: Multivariate analysis of MSP ($/kgrTPA) as a function of depolymerization extent and enzyme loading (mg/g PET) multivariate.
Enzyme Loading (mg
Depolymerization enzyme/g PET feedstock)
Extent
5 10 15
0.80 2.10 2.20 2.32
0.85 1.97 2.06 2.17
0.90 1.85 1.93 2.04
0.95 1.74 1.82 1.92
0.99 1.66 1.74 1.84
Table S15: Depolymerization sensitivity results, related to Figure 4A.
Case Description Abbrevi- Cost Category Contribution (cents/kgrTPA) MSP
ation (cents/
Feedstock Enzyme Electricity Steam Caustic Operational Capital Co-Product kgrTPA)
Cost Charge Credits
Base Case
Base case Base 105 11 8 6 9 34 51 -30 193
Depolymerization Sensitivities
Enzyme price of 5
5 $/kg 105 4 8 6 9 34 51 -30 186
$/kg
Enzyme price of
25 $/kg 105 18 8 6 9 34 51 -30 200
25 $/kg
Enzyme loading of
1 mg enzyme/g 1 mg/g 105 2 8 6 9 34 51 -30 184
PET feedstock
Enzyme loading of
10 mg enzyme/g 10 mg/g 105 21 8 6 9 34 51 -30 204
PET feedstock
Extent of PET
depolymerization ξ = 0.80 118 12 10 7 9 37 57 -30 220
of 0.80
Extent of PET
depolymerization ξ = 0.99 95 10 8 5 9 31 47 -31 174
of 0.99
Reactor residence
τ = 10h 105 11 8 6 9 32 41 -30 182
time of 10 hours
Reactor residence
τ = 240h 105 11 8 6 9 36 65 -30 210
time of 240 hours
Solids loading of
ϕ = 10% 105 11 9 11 9 41 67 -28 225
10%
Solids loading of
ϕ = 20% 105 11 8 3 9 30 43 -31 178
20%
Table S16: Multivariate analysis of MSP ($/kgrTPA) as a function of enzyme loading (mg/g PET) and
enzyme cost ($/kg), related to Figure 4B. The contour lines are interpolation by a cubic equation to project
minimum rTPA selling price.
Enzyme Loading Enzyme Cost ($/kg)
(mg enzyme/g
PET feedstock) 5 10 15 20 25
1 1.83 1.84 1.85 1.85 1.86
3 1.85 1.87 1.89 1.91 1.93
5 1.86 1.90 1.93 1.97 2.01
7 1.87 1.93 1.98 2.03 2.08
10 1.90 1.97 2.04 2.11 2.19

Table S17: Multivariate analysis of MSP ($/kgrTPA) as a function of enzyme loading (mg/g PET) and solids
loading (wt. %), related to Figure 4C. The contour lines are interpolation by a cubic equation to project
minimum rTPA selling price.
Enzyme Loading Solids loading (wt. %)
(mg enzyme/g
PET feedstock) 10 12 15 17 20
1 2.16 1.99 1.85 1.77 1.69
2 2.18 2.01 1.87 1.79 1.71
3 2.20 2.03 1.89 1.81 1.73
5 2.24 2.07 1.93 1.86 1.78
6 2.27 2.10 1.95 1.88 1.80
7 2.29 2.12 1.98 1.90 1.82
8 2.31 2.14 2.00 1.92 1.84
9 2.33 2.16 2.02 1.94 1.86
10 2.35 2.18 2.04 1.97 1.89

Table S18: Multivariate analysis of MSP ($/kgrTPA) as a function of depolymerization extent and solids
loading (wt. %), related to Figure 4D. The contour lines are interpolation by a cubic equation to project
minimum rTPA selling price.

Depolymerization Solids loading (wt. %)

Extent 10 12 15 17 20
0.80 2.55 2.36 2.20 2.12 2.03
0.83 2.45 2.27 2.11 2.03 1.94
0.86 2.36 2.18 2.03 1.95 1.87
0.89 2.27 2.10 1.96 1.88 1.80
0.92 2.19 2.03 1.89 1.81 1.73
0.95 2.12 1.95 1.82 1.75 1.67
0.99 2.02 1.87 1.74 1.67 1.60
Table S19: Summary of capital expense, waste treatment cost, and co-product recovery credits for the no mechanical pretreatment (NMP), no
ethylene glycol recovery (NEG), and no co-product recovery (NCP) process design variations, other data remains identical to base case. Refer to
Table S3 and Table S9 for complete base case capital and operating expenses, respectively.

The right column headers (Feedstock Pretreatment, Enzymatic PET Depolymerization, etc,) correspond to the process sections (Figure
1). Note that Crystallization section is associated with SS co-product recovery and EG Recovery relates to EG recovery alone. Waste
produced in clarification section is the solids stream (unreacted PET, contaminants, etc.) and in crystallization is the permeate (water,
some EG) from the membrane units.

Design Feedstock Enzymatic PET EG

Expense Cost Category Clarification Crystallization
Case Pretreatment Depolymerization Recovery
CAPEX ($M) Capital Expense 13,975,000 1,348,800 6,503,000 17,656,000 1,893,000
Base Waste Treatment 0 0 1.26 1.76 0
OPEX ($M/yr)
Co-product Recovery 0 0 0 -3.21 -6.20
CAPEX ($M) Capital Expense 8,711,000 1,348,800 6,503,000 17,656,000 1,893,000
NMP Waste Treatment 0 0 1.26 1.76 0
OPEX ($M/yr)
Co-product Recovery 0 0 0 -3.21 -6.20
CAPEX ($M) Capital Expense 13,975,000 1,348,800 6,503,000 17,656,000 0
NEG Waste Treatment 0 0 1.26 1.76 0
OPEX ($M/yr)
Co-product Recovery 0 0 0 -3.21 0
CAPEX ($M) Capital Expense 13,975,000 1,348,800 6,503,000 10,224,000 0
NCP Waste Treatment 0 0 1.26 2.97 0
OPEX ($M/yr)
Co-product Recovery 0 0 0 0 0
Table S20: Product and co-product recovery sensitivity results, related to Figure 5A and 5B.
Case Description Abbrevi- Cost Category Contribution (cents/kgrTPA) MSP
ation (cents/
kgrTPA)
Feedstock Enzyme Electricity Steam Caustic Operational Capital Co-
Cost Charge Product
Credits
Base Case
Base Case Base 105 11 8 6 9 34 51 -30 193
Product and Co-product Recovery Sensitivities
No ethylene glycol and
NCP 105 11 8 0 9 35 42 0 210
no salt recovery
No ethylene glycol
NEG 105 11 8 0 9 36 49 -10 208
recovery
Ethylene glycol co-
EG MP
product selling price of 105 11 8 6 9 34 51 -24 199
0.66 $/kg
0.66 $USD/kg
Ethylene glycol co-
EG MP
product selling price of 105 11 8 6 9 34 51 -36 187
1.26 $/kg
1.26 $USD/kg
Overall ethylene glycol 40% EG
105 11 8 6 9 34 51 -26 197
recovery of 40% Recovery
Overall ethylene glycol 65% EG
105 11 8 6 9 34 51 -36 187
recovery of 65% Recovery
Membrane unit with EG-
ethylene glycol Selective 105 11 9 5 9 33 50 -46 175
selectivity of 0.99 Memb.
Sodium sulfate co-
SS MP
product selling price of 105 11 8 6 9 34 51 -27 196
0.10 $/kg
0.10 $USD/kg
Sodium sulfate co-
SS MP
product selling price of 105 11 8 6 9 34 51 -32 192
0.17 $/kg
0.17 $USD/kg
Overall terephthalic 80% rTPA
118 12 10 7 10 38 57 -34 218
acid recovery of 80% recovery
Overall terephthalic 98% rTPA
99 10 8 6 9 32 48 -29 183
acid recovery of 98% recovery
Table S21: Univariate sensitivity summary, related to Figure 6.
% MSP
Bound MSP Difference
Sensitivity Parameter Range
($/kgrTPA) (from base
Sensitivity case) Justification for Parameter Range
Lower Higher
Base Lower Higher Lower Upper
Units MSP MSP
Case MSP MSP % %
Case Case
Varied +/- 20 cents per pound from
Feedstock Cost $/kg 0.22 0.66 1.10 1.23 2.63 -36.2 36.1 the base case. Studied in more detail
in Figure 3.
Shows 2X and 0.33X of the base
Plant Size MTPD 300 150 50 1.84 2.43 -4.7 25.8
case plant size
Solid Loading % 20 15 10 1.78 2.25 -8.1 16.2 Changed +/- 5% in either direction
Depolymerization Extent - 0.99 0.90 0.80 1.76 2.20 -9.1 13.7 Changed +/- 10% in either direction
Effect of high (98%) and low (80%)
rTPA Recovery % 98 90 80 1.83 2.18 -5.4 12.7
recovery – middle baseline selected
Varied to cover 10 h and 10 days of
Residence Time hrs 10 96 240 1.82 2.10 -5.8 8.6
residence time
Reflect the advances made in
Enzyme Loading mg/g PET 1 5 10 1.84 2.04 -4.6 5.6 polymer degradation and biomass
deconstruction works.
Varied to see the effect of CAPEX in
CAPEX $M 50 66 80 1.81 2.02 -6.3 4.8
the +/- $15M range
Selected to represent the different
PET Feedstock Fraction - 0.99 0.95 0.91 1.85 2.02 -4.4 4.6 grades of PET in the market (+/- 4%
from baseline)
Discussed in the main text, Case
Enzyme Cost $/kg 5 15 25 1.86 2.00 -3.8 3.7
Study II
Ranges from +/- 30 cents/kg of the
EG Price $/kg 1.26 0.96 0.66 1.87 1.99 -3.3 3.2
baseline price
High is +15% and low is -10% from
EG Recovery % 65 50 40 1.87 1.97 -3.1 2.0
the baseline EG recovery
Market price of SS has varied from 10
SS Price $/kg 0.17 0.15 0.10 1.92 1.96 -0.9 1.7 cents/kg to 17 cents/kg in the last 5
years
Table S22: MFI supply chain energy requirements for both a monomer and polymer basis, related to
Figures 7A and 7B.

Monomer Basis (TPA)

Case Supply Chain Energy Requirements (MJ/kgPET)
Process Fuel for Renewable Fuel for Fossil Total
Fuel Electricity Electricity Transportation Chemical Energy
Feedstocks
Fossil-derived
(vTPA) 38.7 4.7 0.3 3.1 50.7 97.4
Base Case 16.0 18.1 1.3 1.3 -6.1 30.6
No Mechanical
Pretreatment
(NMP) 15.7 5.9 0.4 1.0 -6.3 16.8
No Ethylene
Glycol Recovery
(NEG) 3.8 18.6 1.4 1.7 -0.8 24.7
No Ethylene
Glycol/Salt
Recovery (NCP) 4.3 18.9 1.4 2.0 -0.5 26.0
EG-Selective
Memb. 12.1 16.9 1.2 0.7 -10.6 20.4
Polymer Basis (rPET)
Case Supply Chain Energy Requirements (MJ/kgPET)
Process Fuel for Renewable Fuel for Fossil Total
Fuel Electricity Electricity Transportation Chemical Energy
Feedstocks
Fossil-derived
(vPET) 41.2 21.6 1.6 6.5 56.1 127.0
Base Case 26.8 30.1 2.2 4.9 18.4 82.4
No Mechanical
Pretreatment
(NMP) 26.7 22.3 1.7 4.8 18.3 73.7
No Ethylene
Glycol Recovery
(NEG) 19.2 30.4 2.2 5.2 21.7 78.7
No Ethylene
Glycol/Salt
Recovery (NCP) 19.4 30.6 2.3 5.4 21.9 79.5
EG-Selective
Memb. 24.4 29.3 2.2 5.0 15.5 76.4
Table S23: MFI supply chain GHG emission results for both a monomer and polymer basis, related to
Figures 7C and 7D.

Monomer Basis (TPA)

Case Supply Chain GHG Emissions (kg CO2e/kgPET)
Process Electricity Transportation Total
Fuel Generation GHG
Emission
Fossil-derived
(vTPA) 2.4 0.3 0.3 3.0
Base Case 1.0 1.3 0.1 2.5
No Mechanical
Pretreatment
(NMP) 1.0 0.4 0.1 1.5
No Ethylene
Glycol Recovery
(NEG) 0.2 1.4 0.1 1.7
No Ethylene
Glycol/Salt
Recovery (NCP) 0.3 1.4 0.2 1.8
EG-Selective
Memb. 0.8 1.2 0.6 2.6
Polymer Basis (rPET)
Case Supply Chain GHG Emissions (kg CO2e/kgPET)
Process Electricity Transportation Total
Fuel Generation GHG
Emission
Fossil-derived
(vPET) 2.5 1.6 0.5 4.6
Base Case 1.7 2.2 0.4 4.3
No Mechanical
Pretreatment
(NMP) 1.6 1.6 0.4 3.7
No Ethylene
Glycol Recovery
(NEG) 1.1 2.2 0.4 3.8
No Ethylene
Glycol/Salt
Recovery (NCP) 1.2 2.2 0.4 3.8
EG-Selective
Memb. 1.5 2.1 0.4 4.1
Table S24: Normalized BEIOM Data, related to Figure 8.
Value-Added
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 2E-02 0.2 0.2 0.5 1.0
Base 0.5 0.2 0.3 0.1 0.0 0.0 0.0 1.1
NMP 0.4 0.2 0.3 1E-02 0.0 0.0 0.0 0.8
NEG 0.6 0.2 0.4 0.1 0.0 0.0 0.0 1.3
NCP 0.5 0.2 0.4 0.1 0.0 0.0 0.0 1.3
Jobs
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 2E-02 0.1 0.1 0.9 1.0
Base 0.2 0.4 0.8 0.1 0.0 0.0 0.0 1.5
NMP 0.2 0.4 0.6 1E-02 0.0 0.0 0.0 1.2
NEG 0.2 0.5 0.9 0.1 0.0 0.0 0.0 1.7
NCP 0.3 0.5 1.0 0.1 0.0 0.0 0.0 1.9
Water Consumption
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 0.5 0.0 5E-02 0.4 1.0
Base 0.0 0.0 0.3 2.2 0.0 0.0 0.0 2.5
NMP 0.0 0.0 0.3 0.3 0.0 0.0 0.0 0.6
NEG 0.0 0.0 0.3 2.5 0.0 0.0 0.0 2.8
NCP 0.0 0.0 0.3 2.6 0.0 0.0 0.0 2.9
Land-Use
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 2E-03 0.0 3E-04 1.0 1.0
Base 3E-02 6E-03 0.5 8E-03 0.0 0.0 0.0 0.5
NMP 3E-02 5E-03 0.4 1E-03 0.0 0.0 0.0 0.4
NEG 3E-02 7E-03 0.5 9E-03 0.0 0.0 0.0 0.6
NCP 4E-02 7E-03 0.6 9E-03 0.0 0.0 0.0 0.6
Smog-Formation
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 2E-02 0.7 0.1 0.2 1.0
Base 2E-02 2E-02 0.1 0.1 0.0 0.0 0.0 0.3
NMP 1E-02 2E-02 0.1 1E-02 0.0 0.0 0.0 0.2
NEG 3E-02 2E-02 0.1 0.1 0.0 0.0 0.0 0.3
NCP 3E-02 2E-02 0.2 0.1 0.0 0.0 0.0 0.3
Eutrophication
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 9E-03 0.7 3E-02 0.2 1.0
Base 0.0 3E-03 0.3 4E-02 0.0 0.0 0.0 0.4
NMP 0.0 3E-03 0.2 5E-03 0.0 0.0 0.0 0.2
NEG 0.0 3E-03 0.3 4E-02 0.0 0.0 0.0 0.3
NCP 0.0 3E-03 0.3 4E-02 0.0 0.0 0.0 0.3
Acidification
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 3E-02 0.8 4E-02 0.1 1.0
Base 0.1 8E-03 0.1 0.1 0.0 0.0 0.0 0.3
NMP 5E-02 8E-03 0.1 2E-02 0.0 0.0 0.0 0.2
NEG 0.1 9E-03 0.1 0.1 0.0 0.0 0.0 0.3
NCP 0.1 1E-02 0.1 0.1 0.0 0.0 0.0 0.3
Ecotoxicity
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 2E-03 0.2 1E-02 0.8 1.0
Base 0.3 5E-06 0.4 6E-03 0.0 0.0 0.0 0.7
NMP 0.3 5E-06 0.4 9E-04 0.0 0.0 0.0 0.7
NEG 0.4 6E-06 0.5 7E-03 0.0 0.0 0.0 0.8
NCP 0.4 6E-06 0.5 8E-03 0.0 0.0 0.0 0.9
Human Toxicity (Total)
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 3E-03 0.8 2E-02 0.2 1.0
Base 4E-04 2E-04 0.1 1E-02 0.0 0.0 0.0 0.1
NMP 4E-04 2E-04 0.1 2E-03 0.0 0.0 0.0 0.1
NEG 1E-03 2E-04 0.1 2E-02 0.0 0.0 0.0 0.1
NCP 1E-03 2E-04 0.1 2E-02 0.0 0.0 0.0 0.1
Respiratory Effects
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 2E-02 0.8 0.1 0.1 1.0
Base 2E-02 8E-03 0.3 0.1 0.0 0.0 0.0 0.4
NMP 7E-03 8E-03 0.2 1E-02 0.0 0.0 0.0 0.3
NEG 3E-02 1E-02 0.4 0.1 0.0 0.0 0.0 0.5
NCP 3E-02 1E-02 0.4 0.1 0.0 0.0 0.0 0.6
Ozone Depletion
Case rTPA Materials rTPA Electricity TPA Petro- TPA Sum
Production Recycle Remaining Generation Production chemical Remaining
Facility Supply Supply
Chain Chain
vTPA 0.0 0.0 0.0 4E-05 1.0 5E-03 1E-02 1.0
Base 0.0 0.0 9E-03 2E-04 0.0 0.0 0.0 1E-
02
NMP 0.0 0.0 9E-03 2E-05 0.0 0.0 0.0 9E-
03
NEG 0.0 0.0 1E-02 2E-04 0.0 0.0 0.0 1E-
02
NCP 0.0 0.0 1E-02 2E-04 0.0 0.0 0.0 1E-
02
Part III – MFI Methods and Supplemental Data

Figure S5. Materials Flows through Industry Tool System Boundary for conventional (fossil-derived) PET
and rPET produced via enzymatic depolymerization.

3.5

2.5
PET (kg)

Cumulative PET
2 "Lifetimes"
Remaining Fossil-
Derived PET
1.5

0.5

0
0 5 10 15 20
PETase Enzymatic Depolymerization "Cycles"

Figure S6. Cumulative lifetime quantity of PET (in kg, blue) obtained from processing an initial 1kg
quantity of fossil-derived PET through repeated cycles through the enzymatic depolymerization and
subsequent repolymerization process. Remaining fossil-derived PET (in kg, orange) due to process
losses at each cycle.
Part IV – BEIOM Methods and Supplemental Data

Figure S7. Inputs and outputs of conventional and virgin TPA pathways.
Socio-economic metrics

Figure S8. Supply chain tiers: Value added.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S9. Value added contribution by sector for each fuel pathway.
 Figure S10. Supply chain tiers: Employment.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S11. Employment contribution by sector for each fuel pathway.

Resource use metrics

Figure S12. Supply chain tiers: Water withdrawals.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S13. Water withdrawals contribution by sector for each fuel pathway.
 Figure S14. Supply chain tiers: Land occupation

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S15. Land occupation contribution by sector for each fuel pathway.
Impact potentials (midpoint) metrics

Figure S16. Supply chain tiers: Smog formation potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S17. Smog formation potential contribution by sector for each fuel pathway.
 Figure S18. Supply chain tiers: Eutrophication potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S19. Eutrophication potential contribution by sector for each fuel pathway.
 Figure S20. Supply chain tiers: Acidification potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S21. Acidification potential contribution by sector for each fuel pathway.
 Figure S22. Supply chain tiers: Freshwater ecotoxicity potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S23. Freshwater ecotoxicity potential contribution by sector for each fuel pathway.
 Figure S24. Supply chain tiers: Human toxicity potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S25. Human toxicity potential contribution by sector for each fuel pathway.
 Figure S26. Supply chain tiers: Respiratory effects potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S27. Respiratory effects potential contribution by sector for each fuel pathway.
 Figure S28. Supply chain tiers: Ozone depletion potential.

Note: “Rest of the economy” encompasses all other sectors whose contribution is less than 5%.

Figure S29. Ozone depletion potential contribution by sector for each fuel pathway.
Process-level improvement potentials

Figure S30. Contribution of rTPA processes to the total impacts per plant. Within the rTPA plant design,
the crystallization process has the highest share of process emissions and resulting relative contributions
to the environmental impact potentials assessed. Sulfuric acid emissions are driving FEP and ACP;
ethylene glycol (a regulated HAP) is driving HTX; both ethylene glycol and sulfuric acid affect SFP, along
with VOC from equipment leaks.

Note: HAP: hazardous air pollutant, VAD: value added, JOB: jobs, H2O: water withdrawals, LOC: land
occupation, SFP: smog formation potential, ETP: eutrophication potential, ACP: acidification potential, FEP:
freshwater ecotoxicity potential, HTX: human toxicity, REP: respiratory effects potential, ODP: ozone
depletion potential, VOC: volatile organic compound.
Field of Influence Analysis

Figure S31. Influence of commodity inputs to the impacts of the rTPA process. A Field of Influence Analysis
(FOIA) shows which commodity inputs (into the rTPA process) drive the impacts of the rTPA process. We
find that:
• Most of the impacts (ACP, HTP, SFP, H2O) are sensitive to electricity consumption (NAICS code:
221100);
• EUP is primarily influenced by waste streams from the plant (221300) and electricity (221100);
• FEP and ODP are sensitive to chemicals/enzymes consumption (325180, 32519B);
• And economic impacts (JOB, VAD) mainly sensitive to feedstock consumption (MRF set-ups).

Legend: Darker colors indicate higher influence

221100 - Electric power generation, transmission, and distribution
221300 - Water, sewage and other systems
325180 - Other basic inorganic chemical manufacturing
32519B - Other basic organic chemical manufacturing
484000 - Truck transportation
562000 - Waste management and remediation services
562920 - Materials recovery facilities

Note: VAD: value added, JOB: jobs, H2O: water withdrawals, LOC: land occupation, SFP: smog formation
potential, ETP: eutrophication potential, ACP: acidification potential, FEP: freshwater ecotoxicity potential,
HTP: human toxicity potential, REP: respiratory effects potential, ODP: ozone depletion potential
Industry-level impacts

Note: VAD: value added, JOB: jobs, H2O: water withdrawals, LOC: land occupation, SFP: smog formation potential, ETP: eutrophication potential,
ACP: acidification potential, FEP: freshwater ecotoxicity potential, HTP: human toxicity potential, REP: respiratory effects potential, ODP: ozone
depletion potential

Figure S32. Economy-wide impacts from the production of 1 kg of TPA per pathway from an industry perspective (i.e., no economic
separation/allocation of impacts across products, Eq. 8). vTPA: virgin TPA; rTPA: recycled TPA; Base: base case design; NEG: no EG is
recovered; NCP: no co-product is recovered; NMP: no mechanical pretreatment.
Supplemental Experimental Procedures

MFI Methodology: Derivation of Functional Unit Used in MFI Calculations

MFI was used to compare the energy requirements and GHG emissions associated with an enzymatic
depolymerization-based PET supply chain to the conventional fossil feedstock-derived TPA supply chain.
In doing so, we define the functional unit (our basis of comparison) to be the total number of PET
“lifetimes” that can be obtained from repeatedly utilizing the enzymatic depolymerization process of 1 kg
of fossil-derived PET. Given the assumed yield of the enzymatic depolymerization process of
approximately 0.63 kg TPA per kg of depolymerized PET and the requisite amount of TPA required to
manufacture new PET, which is approximately 0.87 kg TPA per kg PET, an infinite summation series can
be constructed to determine the “lifetime” amount of PET that could be achieved. This amount is
illustrated in Equations below:

𝐿𝑖𝑓𝑒𝑡𝑖𝑚𝑒 𝑃𝐸𝑇
(𝐸𝑛𝑧𝑦𝑚𝑎𝑡𝑖𝑐 𝑅𝑜𝑢𝑡𝑒)
0.63 𝑘𝑔 𝑟𝑇𝑃𝐴 1 𝑘𝑔 𝑟𝑃𝐸𝑇 0.63 𝑘𝑔 𝑟 2 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 2 𝑃𝐸𝑇
= 1 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 + ( ∗ )+( ∗ )
𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟𝑇𝑃𝐴 𝑘𝑔 𝑟𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟 2 𝑇𝑃𝐴
0.63 𝑘𝑔 𝑟 3 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 3 𝑃𝐸𝑇 0.63 𝑘𝑔 𝑟 𝑛 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 𝑛 𝑃𝐸𝑇
+( ∗ ) + ⋯ ( ∗ )
𝑘𝑔 𝑟 2 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟 3 𝑇𝑃𝐴 𝑘𝑔 𝑟 𝑛−1 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟 𝑛 𝑇𝑃𝐴
2 2
0.63 𝑘𝑔 𝑟𝑇𝑃𝐴 1 𝑘𝑔 𝑟𝑃𝐸𝑇 0.63 𝑘𝑔 𝑟 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 𝑃𝐸𝑇
= 1 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 + ( ∗ )+( ∗ )
𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟𝑇𝑃𝐴 𝑘𝑔 𝑟𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟 2 𝑇𝑃𝐴
0.63 𝑘𝑔 𝑟𝑇𝑃𝐴 1 𝑘𝑔 𝑟𝑃𝐸𝑇 0.63 𝑘𝑔 𝑟 3 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 3 𝑃𝐸𝑇
∗( ∗ )+( ∗ )
𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟𝑇𝑃𝐴 𝑘𝑔 𝑟 2 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟 3 𝑇𝑃𝐴
0.63 𝑘𝑔 𝑟 2 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 2 𝑃𝐸𝑇 0.63 𝑘𝑔 𝑟𝑇𝑃𝐴 1 𝑘𝑔 𝑟𝑃𝐸𝑇
∗( ∗ 2
)∗( ∗ )
𝑘𝑔 𝑟𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟 𝑇𝑃𝐴 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟𝑇𝑃𝐴
0.63𝑛 𝑘𝑔 𝑟 𝑛 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 𝑛 𝑃𝐸𝑇 0.63𝑛 𝑘𝑔 𝑟 𝑛−1 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 𝑛−1 𝑃𝐸𝑇
+ ⋯ [( ∗ ) ∗ ( ∗ )
𝑘𝑔 𝑟 𝑛−1 𝑃𝐸𝑇 0.87𝑛 𝑘𝑔 𝑟 𝑛 𝑇𝑃𝐴 𝑘𝑔 𝑟 𝑛−2 𝑃𝐸𝑇 0.87𝑛 𝑘𝑔 𝑟 𝑛−1 𝑇𝑃𝐴
𝑛
0.63 𝑘𝑔 𝑟𝑇𝑃𝐴 1 𝑘𝑔 𝑟𝑃𝐸𝑇
∗…( ∗ )]
𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87𝑛 𝑘𝑔 𝑟𝑇𝑃𝐴
0.63 𝑘𝑔 𝑟𝑇𝑃𝐴 1 𝑘𝑔 𝑟𝑃𝐸𝑇 0.632 𝑘𝑔 𝑟 2 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 2 𝑃𝐸𝑇
= 1 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 + ( ∗ )+( ∗ )
𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87 𝑘𝑔 𝑟𝑇𝑃𝐴 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.872 𝑘𝑔 𝑟 2 𝑇𝑃𝐴
0.633 𝑘𝑔 𝑟 3 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 3 𝑃𝐸𝑇 0.63𝑛 𝑘𝑔 𝑟 𝑛 𝑇𝑃𝐴 1 𝑘𝑔 𝑟 𝑛 𝑃𝐸𝑇
+( ∗ ) + ⋯ ( ∗ )
𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.873 𝑘𝑔 𝑟 3 𝑇𝑃𝐴 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 0.87𝑛 𝑘𝑔 𝑟 𝑛 𝑇𝑃𝐴
∞
0.63 𝑛
= 1 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 + ∑ ( ) 𝑘𝑔 𝑟1…𝑛 𝑃𝐸𝑇
𝑛=1 0.87
1…𝑛
≈ 1 𝑘𝑔 𝑓𝑜𝑠𝑠𝑖𝑙 𝑃𝐸𝑇 + 2.625 kg 𝑟 PET ≈ 𝟑. 𝟔𝟐𝟓 𝒌𝒈 𝑷𝑬𝑻

Therefore, we make a comparison between 1) a supply chain in which 1 kg of PET is produced from
conventional sources and then 2.625 kg is produced from the enzymatic depolymerization process and 2)
a supply chain in which all 3.625 kg of PET is produced from conventional sources. These supply chains
are then normalized to per 1 kg of PET for simplicity.
Economy-wide analysis (BEIOM): Data and Methods
Data
BEIOM v2.0 was used as the base for this analysis. This model represents the 2017 US economy assuming
the economic structure of 2012 (using the latest available benchmark input-output table from the US Bureau
of Economic Analysis3) with total production and environmental accounts for 2017. The BEIOM model
accounts for conventional TPA production in the more general Other basic organic chemical manufacturing
sector (NAICS 325190). Hence, the conventional TPA sector was disaggregated from the latter sector to
have a specific representation of its production technology in the model (Figure S7). The recycled TPA
industry alongside material recycling facilities (MRF) are included in the model as new sectors representing
the pathway shown in Figure S7 (bottom).i
Operation and maintenance (O&M) costs for conventional TPA were based on Nexant 4 for a typical 350,000
metric tons/yr facility using the Amoco process with para-xylene as feedstock. Crude TPA is produced via
oxidation of para-xylene with air and then purified via hydrogenation. The main inputs for the process are
para-xylene, acetic acid and hydrogen. In 2017, US production of terephthalic acid amounted to 2.3 million
metric tons, mostly consumed by PET manufacturing (97%). 5 TPA sales were allocated exclusively to the
Plastic Resins Manufacturing industry (TPA is the sole output of this industry). The environmental profile
for conventional TPA production was estimated using 2017 point source data from the National Emissions
Inventory, Toxics Release Inventory, Discharge Monitoring Report, and Greenhouse Gas Reporting
Program (Subsection C) for the four major TPA plants in the US that represent 89% of all TPA production
in 2018.ii
O&M costs for recycled TPA were based on NREL’s design case for a 31,100 metric tons/yr facility using
enzymatic depolymerization of PET flakes. This feedstock is supplied by MRF plants. A single recycled
TPA plant is assumed in the analysis, producing TPA, ethylene glycol and sodium sulfate. Similar to the
conventional TPA commodity, recycled TPA is allocated exclusively to the Plastic Resins Manufacturing
industry. Ethylene glycol and sodium sulfate sales were distributed to different sectors according to their
respective US consumption shares based on IHS Markit data. 6,7 For the recycle TPA process, we rely on
the ASPEN model to identify the type of pollutants from unit operations or equipment, and then use EPA’s
AP-42 emission factors or equations, EPA’s TANKS model and mass balance approach to estimate
emissions of regulated pollutants (e.g., criteria air pollutants, hazardous air pollutant). To err on the
conservative side, we assume that the facility producing recycled TPA is operate at the design capacity
(which is a required criterion in estimating potential emissions for air permits).
Material recycling facilities (MRF) were modeled as a single-stream 43,730 metric tons per year plants
following the specifications from GBB.8 We considered a waste stream containing 3.3% of PET and a 98%
recovery rate (capital intensive assumption). Energy consumption (electricity and diesel) and employment
per plant were obtained from Pressley et al.,9 and land occupation from Pyle et al..10 The recycled TPA
industry uses PET recovered from the MRF as feedstock, assuming transportation costs of $8/ton. 11
Similarly, we use AP-42 emission factors to estimate the magnitude of emissions associated with unit
operations for the MRF.
Methodology
To determine the economic and environmental impacts of producing 1 kg of TPA from conventional and
recycle feedstocks we apply a hypothetical extraction method for each pathway. Assume a commodity-by-
D
commodity IO database with 𝑚 commodities, in which 𝑎𝑖𝑗 represents the amount of (domestic iii) commodity
𝑖 required to produce $1 dollar of commodity 𝑗, and 𝑦𝑖 is the total final demand for commodity 𝑖 in 2017.
The following matrices are defined:

i
The final model contains 425 commodities and 415 industries.
ii
BP - Cooper River, SC; DAK Americas - Columbia, SC; Eastman - Kingsport, TN; and PCL - Decatur, AL.
iii
This is a domestic model, i.e., we only account for locally produced inputs (modeling only domestic sectoral
linkages) disregarding economic leakages to foreign countries via imports. Hence, results portray economic impacts
and emissions in the US only.
 D D
𝑎11 … 𝑎𝑚1
D
𝐀 =[ ⋮ ⋱ ⋮ ] (1)
D D
𝑎𝑛1 ⋯ 𝑎𝑚𝑚

𝑦1
𝐲=[ ⋮ ] (2)
𝑦𝑚

Hence, the total output (𝐱) of the economy in 2017 is easily derived by:

𝐱 = (𝐈 − 𝐀D )−1 𝐲 (3)

BEIOM also contains the physical coefficient matrix (𝐏) which shows the amount of emissions per
dollar of production for a given commodity, as well as the LCIA characterization factors matrix (𝐅) derived
from TRACI that convert the physical metrics from 𝐏 into impact potentials (𝐞). Then, the total impact
potentials in the US economy in 2017 are calculated as:

𝐞 = 𝐅 ′ 𝚼(𝐈 − 𝐀D )−1 𝐲 (4)

A comprehensive explanation of these tables and steps is available in Avelino et al. Methodological
Appendix.12,13 Denote the total impact potentials from the US economy in 2017 as 𝐞𝟐𝟎𝟏𝟕 . In the hypothetical
extraction method, the economy-wide contribution of a commodity 𝑘 is calculated by “removing” the
commodity from the local economy and re-estimating the total production and impact potentials for the
D D
same year. In practice, the row and column of commodity 𝑘 in 𝐀D are zeroed (𝑎𝑘𝑗 = 0, ∀𝑗 and 𝑎𝑖𝑘 = 0, ∀𝑘)
generating a new input requirements table (𝐀D∗ )iv. The new impact potentials are then calculated based on
Eq. 4:

𝐞𝑘 = 𝐅 ′ 𝚼(𝐈 − 𝐀D∗ )−1 𝐲 (5)

The contribution of the commodity 𝑘 in the US economy in 2017 for each impact metric (𝐡𝑘 ) is then
derived by:

𝐡𝑘 = 𝐞 − 𝐞𝑘 (6)

Because sectors can produce more than one commodity, e.g. recycled TPA biorefineries produce
ethylene glycol and sodium sulfate besides TPA, two different measures of the impacts of producing 1 kg
of TPA can be derived depending on the perspective chosen: commodity or industry. From an industry
perspective, all impacts attributed to the sector are allocated to TPA, while from a commodity perspective,
those impacts are distributed to the different commodities produced by the sector according to the total
annual value of each (economic allocation). As an illustration, denote by 𝑘, 𝑙, 𝑞 the three commodities
produced by the biorefinery: TPA, ethylene glycol and sodium sulfate, respectively.
The impact potentials of producing 1 kg of TPA from a commodity perspective (𝐦com ) is then
calculated by dividing the total contribution 𝐡𝑘 by the total volume of TPA produced by the industry in 2017
(𝑣):

𝐦com = 𝐡𝑘 (𝑣)−𝟏 (7)

iv
There is an implicit assumption that any industry that requires that locally produced commodity 𝑘 as an input, now
sources it from abroad, therefore not constraining its production.
 The impact potentials of producing 1 kg of TPA from an industry perspective (𝐦ind ) is then calculated by dividing the sum of the total
contributions 𝐡𝑘 , 𝐡𝑙 , 𝐡𝑞 by the total volume of TPA produced by the industry in 2017 (𝑣):

𝐦ind = (𝐡𝑘 + 𝐡𝑙 + 𝐡𝑞 )(𝑣)−𝟏 (8)

Note that if an industry only produces one commodity, then 𝐦com = 𝐦ind , which is the case for conventional TPA. Conversely, for recycled
TPA, 𝐦com < 𝐦ind .
Supplemental References
Useful mechanical recycling facility (MRF) and PET resources. 9,14-35
1. Recycling Markets, Secondary Materials Pricing, https://www.recyclingmarkets.net/, (accessed
2/20/2021).
2. Closed Loop Partners and RRS, (2017). Cleaning the rPET Stream: How we scale post-consumer
recycled PET in the US.
3. Bureau of Economic Analysis, (2018). 2012 Benchmark Input-Output Tables.
4. Nexant ChemSystems (2005). Squeezing Profitability from the PTA-PET Value Chain: Impact of
the Latest Technology Prospectus, p. 23.
5. A. Nayak and A. Pujari, (2019). Dimethyl Terephthalate (DMT) and Terephthalic Acid (TPA) - CEH,
IHS Markit, p. 91.
6. S. S. L. Beraud, A. Gao and C. Funada, (2020). Sodium Sulfate - CEH, IHS Markit, p. 63.
7. M. Monconduit and T. Keel, (2020). Ethylene Glycol - CEH, IHS Markit, p. 145.
8. Gershman, Brickner & Bratton and Inc., (2008). Materials Recovery Facility Feasibility Report—
City of Tucson, AZ p. 54.
9. P. N. Pressley, J. W. Levis, A. Damgaard, M. A. Barlaz and J. F. DeCarolis (2015). Analysis of
material recovery facilities for use in life-cycle assessment. Waste Management 35, 307-317.
10. T. Pyle, C. Bell and B. Wallace, (2018). City of Moscow Local Material Recovery Facility
Assessment, Great West Engineering, p. 33.
11. CB&I Environmental & Infrastructure (2016). Material Recycling Facility Contract Development
Phase I Report—City of Ann Arbor.
12. A. Avelino, P. Lamers, Y. Zhang and H. Chum (submitted). Using a harmonized time series of
environmentally-extended input-output tables to evaluate the effects of an expanding US
bioeconomy. J. Clean. Prod.
13. P. Lamers, T. A. AF, Y. Zhang, D. T. EC, B. Young, J. Vendries and H. Chum (2021). Potential
Socioeconomic and Environmental Effects of an Expanding U.S. Bioeconomy: An Assessment of
Near-Commercial Cellulosic Biofuel Pathways. Environ. Sci. Technol. 55, 5496-5505.
14. G. Giannotta, R. Po, N. Cardi, E. Tampellini, E. Occhiello, F. Garbassi and L. Nicolais (1994).
Processing effects on poly(ethylene terephthalate) from bottle scraps. Polym. Eng. Sci. 34, 1219-
1223.
15. D. Toto (2019). The largest 75 MRFs in North America. Waste Today
16. J. N. Hahladakis and E. Iacovidou (2019). An overview of the challenges and trade-offs in closing
the loop of post-consumer plastic waste (PCPW): Focus on recycling. J. Hazard. Mater. 380,
120887.
17. T. Chilton, S. Burnley and S. Nesaratnam (2010). A life cycle assessment of the closed-loop
recycling and thermal recovery of post-consumer PET. Resour. Conserv. Recycl. 54, 1241-1249.
18. Recyclass, Design for Recycling Guidelines, https://recyclass.eu/recyclass/design-for-recycling-
guidelines/).
19. Plastics Recycling, PET Buyers and Sellers, https://plasticsrecycling.org/markets/buyers-and-
sellers/pet-market-information).
20. S. M. Al-Salem, P. Lettieri and J. Baeyens (2009). Recycling and recovery routes of plastic solid
waste (PSW): A review. Waste Management 29, 2625-2643.
21. The Association of Plastics Recyclers - APR, PET Resin Specifications,
https://plasticsrecycling.org/markets/model-bale-specs/pet-resin-specifications).
22. F. L. Bayer (2002). Polyethylene terephthalate recycling for food-contact applications: testing,
safety and technologies: a global perspective. Food Addit. Contam. 19, 111-134.
23. C. A. Bernardo, C. L. Simões and L. M. C. Pinto (2016). Environmental and economic life cycle
analysis of plastic waste management options. A review. Proceedings of the Regional Conference
GRAZ 2015 – Polymer Processing Society PPS: Conference Papers, DOI:
10.1063/1.4965581140001.
24. A. Chruszcz and S. Reeve, (2018). Composition of plastic waste travelling through the English and
Welsh sorting infrastructure, WRAP, Prepared by Resource Futures, p. 36.
25. A. J. Dubanowitz, A. Prof, N. J. Themelis and A. J. Dubanowitz (2000). Design of a Materials
Recovery Facility (MRF) For Processing the Recyclable Materials of New York City’s Municipal
Solid Waste
26. M. K. Eriksen, A. Damgaard, A. Boldrin and T. F. Astrup (2019). Quality Assessment and Circularity
Potential of Recovery Systems for Household Plastic Waste. J. Ind. Ecol. 23, 156-168.
27. J. N. Hahladakis, P. Purnell, E. Iacovidou, C. A. Velis and M. Atseyinku (2018). Post-consumer
plastic packaging waste in England: Assessing the yield of multiple collection-recycling schemes.
Waste Management 75, 149-159.
28. K. Ip, M. Testa, A. Raymond, S. C. Graves and T. Gutowski (2018). Performance evaluation of
material separation in a material recovery facility using a network flow model. Resour. Conserv.
Recycl. 131, 192-205.
29. M. L. Mastellone, R. Cremiato, L. Zaccariello and R. Lotito (2017). Evaluation of performance
indicators applied to a material recovery facility fed by mixed packaging waste. Waste Management
64, 3-11.
30. P. Moore and C. Staub, Slipping through the cracks, https://resource-
recycling.com/plastics/2020/05/19/slipping-through-the-cracks/).
31. A. Pawlak, J. Morawiec, F. Pazzagli, M. Pracella and A. Galeski (2002). Recycling of postconsumer
poly(ethylene terephthalate) and high-density polyethylene by compatibilized blending. J. Appl.
Polym. Sci. 86, 1473-1485.
32. A. B. Raheem, Z. Z. Noor, A. Hassan, M. K. Abd Hamid, S. A. Samsudin and A. H. Sabeen (2019).
Current developments in chemical recycling of post-consumer polyethylene terephthalate wastes
for new materials production: A review. J. Cleaner Prod. 225, 1052-1064.
33. A. G. Raymond (2017). Modeling of material recovery facility performance with applications for life
cycle assessment (Massachusetts Institute of Technology).
34. Recycling Markets, Recycling Markets - Your Online Recycling Database,
https://www.recyclingmarkets.net/secondarymaterials/index.html).
35. D. P. Shonfield (2008). LCA of Management Options for Mixed Waste Plastics - Project Code:
MDP017 (WRAP).