Research ArticleMATERIALS SCIENCE

# Leveling the cost and carbon footprint of circular polymers that are chemically recycled to monomer

See allHide authors and affiliations

Vol. 7, no. 15, eabf0187

## Abstract

### Primary PDK resin production

Triketone monomer synthesis. The triketone monomer required for PDK resin production is synthesized with dimedone and sebacic acid (13), where the reaction occurs in the presence of DCC as a condensation reagent and DMAP as a catalyst in dichloromethane. The triketone monomer forms along with N,N′-dicyclohexylurea as a by-product. The reaction occurs at room temperature for 4 hours. The crude triketone monomer is recovered by adding 3 weight % (wt %) HCl and subsequently recrystallized using ethyl acetate at 80°C. Here, recrystallization was carried out to estimate analytical purity and may be skipped in manufacturing without affecting final PDK purity. On the basis of the experimental results, we used 90% conversion efficiency for modeling purposes. N,N′-dicyclohexylurea is a known irritant and modeled as a hazardous waste with appropriate cost assigned for disposal. The monomer synthesis stage includes mixing tanks, a continuous blending tank, and a solvent recovery system involving filtration, centrifugation, and distillation for dichloromethane recovery. The schematic of process flow diagram with major process equipment is provided in figs. S4 and S5.

Polymer synthesis. The polymerization of triketone monomer occurs through spontaneous condensation, in this case monomer with TREN to form thermally processable PDK resins. We modeled TREN as an externally sourced chemical. In this case, the synthesized triketone monomer and TREN were added to a ball mill and milled for 45 to 60 min. The resulting powder was subsequently dried to remove water. The uniqueness of the chemistry lies in solvent-free synthesis that occurs at room temperature with no additional heat requirement. The dried powder is pressed at 20,000 psi pressure and 190°C temperature for 60 s to yield transparent PDK material of desired shape. The polymer synthesis stage includes mixing tanks for reactants, a ball mill, a drying unit, and a thermal pressing unit.

### Circular PDK resin production

Depolymerization. Recycling involves three steps: collection, sorting, and chemical/mechanical recycling. We modeled PDK resins as use-phase agnostic, and therefore, we did not make any assumption whether the collection and sorting would be postconsumer or postindustrial. We only consider the third step of chemical recycling and include transport of PDK waste from sorting facility to chemical recycling, and the chemical recycling/depolymerization facility in our system boundary. In the facility, PDK waste is hydrolyzed in 5.0 M H2SO4 to recover triketone and TREN monomers at room temperature for 12 hours. The crude triketone monomer was recovered via filtration and by solubilizing the solid with aqueous K2CO3 and was subsequently precipitated in the presence of aqueous H2SO4. TREN was recovered using a regenerative resin process where it underwent ion exchange in the presence of a strong anionic resin. We used 50 wt % NaOH for column regeneration. Because of an excess of H2SO4 used for TREN recovery and subsequent ion exchange process, our effluent contains ~340 g/liter Na2SO4 and accounts for 28 wt % of the final effluent. Such a high concentration would require either pretreatment (39, 40) or valorization of the salt before the effluent undergoes traditional wastewater treatment. Particularly, treating Na2SO4 as a coproduct rather than waste may generate additional revenue opportunities including purifying and selling as is with the current market value of $100/MT (41), or further treated to potassium sulfate (42) and sold at a higher price of$414/MT as fertilizer (43). In addition, unreacted TREN and monomer are assigned to solid waste category to account for cost and emissions; however, our future work will explore additional ways to recover or minimize the efficiency loss. The depolymerization stage includes a storage tank for acid, a reactor for depolymerization, an ion exchange unit, and a vacuum filtration unit.

### Auxiliary facilities

For each of the plants producing primary, circular, and mixed PDK resins, we assumed that required energy and heat are generated on-site using a natural gas-fired boiler. On-site energy generation unit includes a boiler to generate steam, which is sent to an extraction turbine coupled with an electric generator. If additional electricity is generated, it is exported and sold to the grid. In addition, we modeled a separate utilities section that cycles cooling water with additional makeup cooling water and process water sourced from outside with an assigned cost.

### Mixed PDK resin production

To understand the impact of current recycling infrastructure on the quantity of PDK recovered, we modeled a combined plant that produced both circular and makeup primary PDK resin materials, which we termed as “mixed PDK resins.” We modeled the recycling of PDK waste each successive year and calculated the amount of makeup raw material required to produce 20,000 MT of resin consistently each year. To this end, we modeled several scenarios with each incorporating a combination of different recycling rate and product life span. We conducted this exercise to measure number of years and the level of recycling required to recover sufficient PDK waste for the plant so it primarily operates the depolymerization unit using PDK waste as the main input.

Before a product is available for end-of-life management, it may go through various stages of use, repair, refurbishment, and exchanging hands. The accumulation of material in the use phase is represented by product life span and can vary based on type of product, region of use, and demographic practices. Following Geyer et al. (8), we modeled product life spans as ranges represented by log-normal distribution. We explored two cases: one where PDK resin is used as packaging with a shorter average life span of half a year, and second, where PDK resin is used as a generic consumer and institutional product with an average life span of 3 years. Recycling rates represent the efficiencies in collection and sorting of waste and vary with products. For example, lead acid batteries have a recycling rate of 99%, and PET and HDPE plastics have a recycling rate of ~30% in the United States (5). For this scenario, we selected recycling rates of 100 and 44%. Our selection of 100% represents a theoretical maximum, and 44% represents a baseline case from 2050 projections for plastic recycling provided by Geyer et al. (8).

### Techno-economic analysis

The techno-economic model for a PDK resin production facility was developed using process simulation software SuperPro Designer (44). The facility operates 24 hours/day and 330 days/year. The plant produces 20,000 MT of PDK resin (primary, circular, or mixed depending on the specific scenario) annually. The plant life is assumed to be 30 years. The assumptions for calculating the MSP are based on techno-economic reports on similar precommercial processes for bioenergy conducted by National Renewable Energy Laboratory and provided in table S10 (38). The costs for production were estimated by including capital and operating costs. For capital cost contribution, the total cost of equipment was determined based on equipment purchase price, required equipment size, and number of equipment from the process simulation. We used an average multiplier of 1.7 with purchased equipment costs to determine the installed equipment cost. The capital cost estimates also include additional costs such as cost of setting up warehouse, site development, additional piping, project land contingencies, and permits. Each cost is approximated as a percentage of the installed and total direct costs, and the percentages are provided in the table S10. The annual operating costs include cost of material, maintenance, labor, transportation, and waste management and are obtained from SuperPro Designer. The prices associated with each raw material purchase and electricity were obtained from the literature (38, 45) and e-commerce websites using request for bulk quotes (29, 46). The cost of labor was estimated from the U.S. Bureau of Labor Statistics and the literature (47, 48). The MSP was calculated by conducting a discounted cash flow analysis using a 10% internal rate of return. The assumptions regarding tax rate, internal rate of return, financing, and depreciation are further detailed in fig. S6.

### Life cycle assessment

The scope of the LCA is cradle–to–facility gate with a functional unit of 1 kg of resin produced, which is compared against 1 kg of a range of commodity and specialty polymers in use today. We also included the recycling process as it is an integral part of sourcing the raw material for circular PDK resins. The life cycle inventory data for input materials and commodity polymers are obtained from peer-reviewed literature (49, 50) and LCA databases including ecoinvent (31), U.S. Life Cycle Inventory (USLCI) (51), The Greenhouse gases, Regulated Emissions, and Energy use in Technologies Model (GREET) (52), and Waste Reduction Model (WARM) (53). The life cycle GHG footprints of TREN and dimedone were estimated using neural network models that identified compounds with similar molecular characteristics and existing production data because no data specific to the production of these compounds were available (5456). When these models could not identify any similar compounds, as was the case for DCC, we obtained their industrial synthesis route through available patents and created a separate process simulation to estimate material and energy use, which was then used to calculate life cycle GHG emissions (5759). A detailed description of the modeling process for GHG emissions along with data sources is provided in the section S3. Next, we combined the data gathered in the life cycle inventory with the Intergovernmental Panel for Climate Change 100-year global warming potential characterization factors to arrive at final life cycle GHG emissions. We used a hybrid LCA approach that combined process-based model with IO matrix using physical units and created a process matrix to compute total requirements for production (60, 61). The hybrid approach overcomes the disadvantage of the economic sector aggregation issues of IO modeling while still retaining the benefit of a larger system boundary and avoiding cutoffs in system boundary compared with traditional process-based LCA Because of the closed-loop nature of the process, we do not assign environmental burdens to PDK waste as recommended by ISO 14044 (62).

### Sensitivity analysis

To understand the variation in our results, we explored two cases: a pessimistic case and an optimistic case. The optimistic case represents a combination of lower prices, lower GHG emissions, and higher reaction yields. Similarly, the pessimistic case represents a combination of higher prices, higher GHG emissions, and lower yields compared with the baseline case. For the sensitivity analysis, we selected the dominant contributing factors for GHG emissions and MSP and varied the parameters based on available range of data or by varying the data by ±20%. The data are provided in table S4.