5‑(Chloromethyl)furfural from Polycotton Waste Textile in a Biphasic System: Different Process Designs

Abstract

Due to underdeveloped waste management technologies for textile waste, 85 wt % ends up in either incineration plants or landfills. With our process, we are able to fully valorize polycotton waste, the largest fraction of textile waste, by separating both components. Cotton is converted into 5-(chloromethyl)­furfural (CMF) through a one-step batch process, using a biphasic system containing superconcentrated hydrochloric acid (HCl) (44 wt %) and chlorobenzene. The reaction temperature, reaction time, stirring speed, effective cotton loading, and volume of the organic layer of the biphasic system were first optimized using a central composite design, rendering 91 °C, 1 h, effective 8.3 mg of cotton/mL HCl, and 4 mL of organic layer/mL HCl to be the (apparent) optimal reaction conditions, leading to a CMF yield of 62.6%, which was in line with the predicted outcome of the model. The effect of adjustments on process design in regard to process structure (one-step vs two-step) and process type (batch vs semicontinuous) was investigated. This resulted in a final CMF yield of 73.3% in the two-step semicontinuous process, a method suitable for large-scale textile valorization.

Introduction

The global textile industry has grown immensely over the past decade, and even greater growth is expected in the coming years. Especially with the rise of fast fashion, and now even ultrafast fashion, the production and consumption rate is sky high. In 2024, the global textile production reached 92 million tons, equivalent to 11.3 kg per capita. Although the amount of textile waste continues to increase, proper waste management still remains underdeveloped and lags behind. In 2024, only 7 wt % of postconsumer textile waste was recycled. Of this fraction, merely 1 wt % is (closed-loop) recycled into new textiles, and the other 99 wt % was downcycled into low-value applications. Twelve wt % of textile waste was reused, meaning that, unfortunately, 81 wt % of the textile waste was incinerated or sent to landfills. The low recycling rate is predominantly caused by the complexity of textile fabrics. The majority of the textiles is a blend of multiple fibers, with polycotton, i.e., textiles made from polyester and cotton, being the most common blend with more than 50 wt % of the global textile waste. The use of blended fibers in textiles makes the complete valorization of both fractions extremely challenging. Additionally, most garments contain zippers, seams, buttons, labels, and prints, which further complicate recycling. − As a result, mechanical recycling is currently the most viable option for blended textiles. However, as mechanical recycling leads to a degradation of the fiber properties, chemical recycling methods have gained interest over the past few years. , Furthermore, chemical recycling of polyesters provides a more circular end-of-life solution, as it retains carbon to a greater extent compared to mechanical recycling.

The Dawn Technology, a promising technology to valorize polycotton waste textiles, is treating the textile with superconcentrated hydrochloric acid (>37 wt % aqueous HCl). − This allows for full separation into (unaffected) polyester and (hydrolyzed) cotton, whereafter both fractions can be repurposed. Polyester can chemically be regenerated into new polyester, which again can be used for, e.g., the production of new fibers. Cotton is hydrolyzed by the use of superconcentrated hydrochloric acid (typically 40–44 wt % HCl), resulting in an acidic glucose stream that can be used (without acid-glucose separation) for the production of platform chemicals such as 5-(chloromethyl)­furfural (CMF) and its derivatives. , These furan derivatives are a biobased alternative for petroleum feedstocks and can be used for the production of monomers, pharmaceutical precursors, biofuels, herbicides, solvents, plasticizers, and other chemicals. − The conversion of cellulose from polycotton waste to CMF is preferably performed in a biphasic system. This HCl induced cascade reaction involves cellulose dissolution and hydrolysis (i.e., cellulolysis), followed by glucose isomerization, dehydration, and chlorination to yield CMF. − The biphasic system consists of an aqueous hydrochloric acid solution and a nonmiscible organic solvent to convert the carbohydrate to CMF at an elevated temperature. The moderately elevated temperature (80–100 °C) is needed for the dehydration of glucose to 5-(hydroxymethyl)­furfural (HMF), as the reaction has a relatively high activation barrier. The HCl is used for (cellulose) dissolution, as hydrolysis and dehydration catalyst, and as a chloride source for the halogenation of HMF to CMF. ,

While the treatment of polycotton waste textiles with 44 wt % HCl and the use of a biphasic system for converting polysaccharides to CMF are both well-established technologies, ,,, their combination (i.e., treating polycotton waste in a biphasic system with 44 wt % HCl) has, to our knowledge, not been previously described in the scientific literature. Several studies have explored the use of other solvent systems such as a methanol and benzoic acid–based deep eutectic solvent system and a 1,8-diazabicyclo(5.4.0)­undec-7-ene, dimethyl sulfoxide, and carbon dioxide switchable solvent system to treat polycotton waste textiles. However, these processes merely focus on the initial separation of cotton and polyester and do not include the subsequent valorization of the cotton component, as attained in a biphasic system.

Therefore, this research investigates the optimal reaction conditions for a one-step batch reaction of polycotton to CMF in a biphasic system by performing a design of experiments (DoE). Thereafter, the effect of the other process designs is investigated, using the optimized reaction conditions from the one-step batch process. Additionally, the quality of the residual polyester is investigated, and recommendations for large-scale deployment are made.

Plausible Reaction Paths of Cellulose to CMF

For the synthesis of CMF from cotton-based cellulose, a cascade of reactions must occur (Scheme ). This cascade sequence starts by dissolving cotton, i.e., cellulose, with the use of superconcentrated aqueous HCl, and depolymerizing this cellulose (via various oligomers) to glucose. Thus, for every n molecules of (anhydro)­glucose, (n-1) water molecules are needed. Next, the general assumption is that glucose is dehydrated into HMF, via either a cyclic or acyclic mechanism, and with or without the isomerization to fructose. − During this dehydration of glucose to HMF, three water molecules are produced. Lastly, HMF is chlorinated to CMF via a nucleophilic substitution reaction, liberating one equivalent of water. Overall, the conversion of cellulose to CMF results in a net production of three molar equivalents of water.

1. Plausible Reaction Pathways for the Cascade Reaction of Cellulose to 5-(Chloromethyl)­furfural (CMF).

Despite being a ketose, it is mentioned in the literature that the dehydration of fructose to HMF is much faster compared to the dehydration of glucose. , However, Brønsted acids, which are used in many cases, are known to facilitate dehydration reactions over isomerization, which is the predominant role of a Lewis acid. − Therefore, the exact mechanism of hexose conversion to HMF is still not clear. Several chemical pathways have been proposed in the literature, and they may be correct and occurring simultaneously.

The hydrolysis of cellulose, the isomerization of glucose to fructose, and the dehydration of glucose and/or fructose to HMF all take place in the aqueous phase. As the aqueous layer consists of HCl, it is most likely that the substitution of the hydroxyl group of HMF with chloride also occurs in the aqueous layer. However, there is still the possibility that, instead of occurring in the aqueous layer, HMF is extracted into the HCl-saturated organic phase, where chlorination occurs. It is also suggested that all of the above-mentioned reactions, although not necessarily to the same degree, occur simultaneously in a biphasic system and that the characteristics of this biphasic system and the applied physical properties dictate the preferred mechanism of CMF formation.

Unfortunately, the acid-catalyzed conversion of cellulose is accompanied by the formation of side products, such as levulinic acid (LA) and formic acid (FA). LA and FA are formed in equal amounts via rehydration of HMF. ,,,− Parallel to these reactions, humins can be formed via various reactions such as rehydration of HMF, aldol condensation, and oligo/polymerization reactions. ,, Humins are dark-colored polymeric solids, and their formation is random and uncontrollable from compounds such as glucose and HMF, which are converted into complex macromolecular structures. , The humin concentration increases with an increase in monosaccharide and furan derivative concentrations as well as with higher reaction temperatures. ,,

In addition, especially at a high monosaccharide concentration, reoligomerization can also lead to the formation of disaccharides. This can be advantageous as this can protect the sugars from degradation to byproducts. However, it is proposed that the glucose disaccharides can in themselves also engage in the humin formation process. All these degradation paths lead to a low overall selectivity for CMF formation. Fortunately, the use of a biphasic system reduces these side reactions as the furan derivatives are constantly extracted into the organic layer, away from the acid, preventing the standard decomposition routes of CMF and its conversion to HMF and subsequently to LA, FA, and humins due to the limited presence of water. ,

Materials and Methods

Materials

Blue postconsumer polycotton (cotton/polyester 41/59) waste textiles were supplied by the Dutch workwear company Groenendijk Bedrijfskleding, now Lyreco. Hydrochloric acid in water (37 wt %) and chlorobenzene (PhCl) (99.8%) were purchased from Fisher Scientific Solvents. 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) was purchased from Biosolve. Superconcentrated HCl in water (44 wt %) was produced by absorbing 100% HCl gas into a 35–36 wt % HCl solution. The HCl concentration was determined using a 716 DMS Titrando unit (Metrohm AG) using 10 times diluted HCl with 0.1 M sodium hydroxide at 25 °C. We want to emphasize the difference in concentration between the terms “concentrated HCl” (37 wt %) and “superconcentrated HCl” (44 wt %) used in this article to prevent confusion and misinterpretation of the results.

Cotton Removal at Different HCl Concentrations

To see the effect of different HCl concentrations on the separation of cotton and polyester from polycotton textile, 20.2 (azeotropic), 37, and 44 wt % aqueous HCl solutions in the presence of chlorobenzene as a biphasic system were tested. The polycotton textile was treated at a solid loading of 50 mg textile/mL HCl, 2 mL PhCl/mL HCl at 90 °C and 600 rpm for 2 h. Thermogravimetric analysis (TGA) was used to conclude whether the residual textile contained any cotton as published previously in Leenders et al. All reaction conditions were tested in triplicate.

Process Designs

The cascade of reactions for the formation of CMF can be done in several process designs, where a distinction is made in process structure (one-step or two-step) and process type (batch and semicontinuous (vide infra)) (Figure ). A schematic representation of the process designs can be found in Figure S1. In a one-step process, the cellulolysis, dehydration, and chlorination are performed simultaneously in a biphasic system at an elevated temperature. In a two-step process, the first step involves cellulolysis in a superconcentrated (40–44 wt %) aqueous hydrochloric acid solution at room temperature (the so-called Bergius process), whereafter the nonmiscible organic solvent is added, and the hydrolysis, dehydration, and chlorination are performed at elevated temperatures. For both designs, these processes can be performed in a batch or semicontinuous design as shown before. In a semicontinuous design, the nonmiscible organic solvent is frequently refreshed.

1.

Process designs for the conversion of cellulose from polycotton waste to CMF in a biphasic system.

Synthesis of CMF from Polycotton Textile in a Biphasic System

A standard one-step batch design consisted of a 9 mL Ace glass pressure tube, filled with 50 mg of polycotton textile, 3 mL chlorobenzene, and 1 mL of an aqueous 44 wt % HCl solution, and a rare earth core stirring egg. The reactor was heated to 90 °C for 2 h under magnetic stirring at 600 rpm. For the one-step semicontinuous design, a similar procedure was followed, however, the organic layer was replaced half way through the reaction. For the two-step designs, a 9 mL Ace glass pressure tube was filled with textile and 1 mL of an aqueous 44 wt % HCl solution, and a rare earth core stirring egg. The cellulolysis was performed at room temperature for either 1.5 or 24 h, whereafter the residual polyester was removed and 3 mL chlorobenzene was added. Then, the reactor was heated to 90 °C for 2 h with magnetic stirring at 600 rpm. After completion, the reactor was cooled, the organic layer was removed with a syringe, and the aqueous layer was washed 3 times with 2 mL chlorobenzene to maximize CMF collection. The organic layer and the organic washing layer were analyzed with gas chromatography (GC) to quantify the CMF concentration, and the cumulative values were used to calculate the overall CMF yield. The aqueous layer was diluted and neutralized first with 5 mL of 1.6 M sodium hydroxide, then with 2 mL of 1 M sodium bicarbonate and sequentially analyzed with high-performance liquid chromatography (HPLC). All reaction conditions were tested in triplicate.

Modeling of Optimal Conditions

To obtain the optimal reaction conditions for the direct CMF production from cotton (from polycotton waste) textile in a biphasic system, a DoE was carried out. A central composite design (CCD) was set up within a response surface methodology (RSM) to model and optimize the CMF yield as a function of temperature, reaction time, stirring speed, effective cotton loading, and volume of the organic layer. The CCD uses factorial, axial, and center points to estimate the linear and quadratic effects as well as the interactions of the five factors of interest. In Table , the boundaries of the tested conditions are shown, which were obtained after preliminary screening. The experimental data were fitted to a second-order polynomial using the standard least-squares method. All of the variables were mean-centered to improve numerical stability and reduce multicollinearity. To model the optimal conditions for the highest CMF yield, the software JMP, version 17 developed by SAS Institute Inc., was used for statistical analysis.

1. Boundaries of Tested Reaction Conditions to Determine the Response Surface of the CMF Yield from Polycotton Textile.

	Temperature (°C)	Reaction time (h)	Stirring speed (rpm)	Effective cotton loading (mg/mL)	Volume organic layer (mL)
Lower boundary	60	1	300	8	2
Upper boundary	100	3	900	30	4

Analytical Methods

Thermogravimetric Analysis

To determine the composition of polycotton samples and to establish the composition of residual textile, the samples were subjected to thermogravimetric analysis. The analysis was performed utilizing a Mettler Toledo TGA/DSC 3+ heating from 25 to 600 °C with a heating rate of 2.5 °C/min under a nitrogen flow of 80 mL/min. Approximately 10 mg of samples was used for the measurement.

Textile Composition Analysis

The postconsumer waste textile used in this study was characterized for its cotton/polyester composition using thermogravimetric analysis. In previous work, we showed that using TGA is a reliable method for analyzing polycotton textiles. For the composition analysis, approximately 10 mg of textile was placed in an aluminum crucible and heated from 25 to 600 °C at a rate of 2.5 °C min^–1 under a nitrogen flow of 80 mL min^–1; the resulting TGA and derivative thermogravimetric (DTG) curves were used to determine the moisture, cotton, and polyester content. TGA analysis indicated a composition of 40% cotton, 57% polyester, and 3% moisture.

Gas Chromatography

CMF produced during the cascade reaction was quantified by gas chromatography with a Thermo Scientific Trace 1310 GC with flame ionization detection. An Agilent J&W DB-624 UI (30 m × 0.25 mm × 1.4 μm) column was used with helium as the carrier gas. The CMF yield was calculated based on the initial quantity of anhydroglucose moles present in the cellulose fraction of the polycotton textile (eq ).

$$\mathrm{CMF\; yield}(\%)=\left(\frac{\mathrm{moles\; of\; CMF\; after\; reaction}}{\mathrm{moles\; of\; anhydroglucose\; present\; in\; the\; starting\; material}}\right)\times 100\%$$ 1

High-Performance Liquid Chromatography

Residual glucose, HMF, LA, and FA were quantified using an Agilent 1260 Infinity HPLC system with diode-array detection and refractive index detection. An Aminex HPX-87H (300 × 7.8 mm; dp 9 μm) separation column was used, operating at 50 °C using 0.6 mL/min of 5 mM H₂SO₄ as eluent. The glucose, HMF, LA, and FA yields were calculated based on the initial quantity of anhydroglucose moles present in the cellulose fraction of the polycotton textile (eq , eq , eq and eq , respectively).

$$\mathrm{Glucose\; yield}(\%)=\left(\frac{\mathrm{moles\; of\; glucose\; after\; reaction}}{\mathrm{moles\; of\; anhydroglucose\; present\; in\; the\; starting\; material}}\right)\times 100\%$$ 2

$$\mathrm{HMF\; yield}(\%)=\left(\frac{\mathrm{moles\; of\; HMF\; after\; reaction}}{\mathrm{moles\; of\; anhydroglucose\; present\; in\; the\; starting\; material}}\right)\times 100\%$$ 3

$$\mathrm{LA\; yield}(\%)=\left(\frac{\mathrm{moles\; of\; LA\; after\; reaction}}{\mathrm{moles\; of\; anhydroglucose\; present\; in\; the\; starting\; material}}\right)\times 100\%$$ 4

$$\mathrm{FA\; yield}(\%)=\left(\frac{\mathrm{moles\; of\; FA\; after\; reaction}}{\mathrm{moles\; of\; anhydroglucose\; present\; in\; the\; starting\; material}}\right)\times 100\%$$ 5

Gel Permeation Chromatography

Gel permeation chromatography (GPC) was performed to analyze the impact of the reaction conditions on the molecular weight of the polyester. The analysis was performed with an Agilent 1260 Infinity II GPC/SEC System with a PSS PFG guard column (50 × 8 mm, dp 7 μm), PSS PFG analytical column (300 × 8 mm, dp 7 μm), and a VWR Hitachi RI detector 5450. The samples were dissolved in HFIP, and the system was calibrated with poly­(methyl methacrylate).

Results and Discussion

Parameter Screening: Cotton Removal with Different HCl Concentrations

To maximize polycotton valorization, it is essential to fully separate both components. This is particularly important for polyester, as any residual cotton will pollute the chemical recycling process of PET. Thus, to valorize both fractions, the polyester residue must be cotton free. The concentration of HCl in solution has a considerable influence on the rate of dissolution, and thus separation, of cotton from polycotton waste textile. In previous work, we reported that 37 wt % HCl for 24 h at ambient temperature was insufficient to consistently separate cotton from the polyester residue under ambient conditions, due to the recalcitrant crystalline structure of cellulose I, the only polymorph present in cotton. As the one-step batch reaction of cellulose to CMF is performed at a higher temperature, the effect of different HCl concentrations at an elevated temperature on the removal of cotton was reinvestigated. The polycotton textile was treated with 20.2, 37, and 44 wt % HCl solutions with a solid loading of 50 mg textile/mL HCl and 2 mL PhCl/mL HCl at 90 °C and 600 rpm for 2 h. The residual solid textiles were analyzed with TGA to determine the cotton content after the reaction. In Figure .a, the starting material is compared with the samples treated with the different HCl concentrations. Here, it can be seen that only 44 wt % HCl fully removes cotton, as only one major decrease in mass is observed, corresponding to the decomposition of polyester alone. This can also be concluded when considering the DTG curves (Figure .b), as only one peak was present between 200 and 500 °C in the DTG curve of the sample treated with 44 wt % HCl. The samples treated with 20.2 and 37 wt % HCl both display two peaks between 200 and 500 °C, indicating the presence of both polyester and residual cotton. These residual materials are unfit for chemical recycling of polyester, as the cotton residue will degrade and contaminate the PET recycling stream. Therefore, 44 wt % HCl was chosen as the aqueous acidic medium for the rest of the experiments to ensure full separation of cotton and polyester.

2.

a Thermogravimetric analysis results of the starting material and the treated samples after cascade reaction under biphasic conditions with different hydrochloric acid concentrations (HCl) (20.2, 37, and 44 wt %) (heating rate: 2.5 °C/min under N₂ flow). b Derivative thermogravimetric (DTG) analysis of the starting material and the treated samples after cascade reaction in a biphasic system with different HCl concentrations (20.2, 37, and 44 wt %) (heating rate: 2.5 °C/min under N₂ flow). Reaction conditions: 1 mL HCl, 2 mL chlorobenzene (PhCl), 90 °C, 2 h, 600 rpm, and an effective cotton loading of 20 mg/mL.

Optimalization of CMF Yield in One-Step Batch Process

A preliminary screening was performed to determine the range of reaction parameters needed as an input for the DoE to model CMF production in the cascade reaction within a one-step batch design. The goal of the model is to maximize the CMF yield by finding the optimal reaction conditions, varying the temperature, reaction time, stirring speed, effective cotton loading, and volume of the organic layer. This study is the first to report, and optimize, the one-step batch conversion of polycotton waste to CMF in a 44 wt % HCl biphasic system focused on high CMF yield and cotton-free polyester fit for industrial recycling. The boundaries of the reaction conditions are listed in Table . The DoE modeled the response surface of the CMF yield using 58 different combinations of reactor conditions, either performed as a single experiment or in duplicate. In the design, factorial, center, and axial points were investigated to determine main and quadratic effects as well as interactions between the factors.

For this study, all effects (main, interactions, and quadratic) were tested on significance and effects with a p-value >0.01 were considered statistically not significant and were removed. An exception to this rule was made for individual parameters with a p-value >0.01 being used in an interaction effect with a p-value <0.01. The effect summary of the reaction parameters on the CMF yield (Table ) showed that the quadratic temperature effect has the biggest impact on the CMF yield, as it has the highest logWorth (i.e., −log10­(p-value)). It also suggests that temperature itself, in combination with the reaction time, as well as with the volume of the organic layer has a significant effect on the CMF yield. Furthermore, it indicates that the effective cotton loading has a main (first-order) effect on the CMF yield, and that the volume of the organic layer has a quadratic (second-order) effect on the CMF yield. All the effects stated in the effect summary are used in the model (eq ) to predict the CMF yield from the one-step batch reaction of cotton from polycotton waste textile in a biphasic system with 44 wt % HCl. An overview of the HMF, FA, LA, and glucose yields during CMF production with the one-step batch reaction can be found in Figure S2.

$$\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{C}\mathrm{M}\mathrm{F}\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}(\%)=0.09996+0.004202T+0.003194t-0.002114C+0.036008V-0.000608{(T‐81.03)}^2-0.004605(T‐81.03)(t-2.093)+0.002652(T-81.03)(V-3.065)+0.02641{(V‐3.065)}^2$$ 6

where T is the temperature (°C), t is the reaction time (h), C is the effective cotton loading (mg/mL), and V is the volume of the organic layer (mL).

2. Effect Summary of All the Statistical Significant Factors (Main, Interactions, and Quadratic) in the Conversion of Cotton from Polycotton Materials to CMF in a One-Step Batch Process .

Effects	logWorth	P-value
Temperature (°C) × Temperature (°C)	40.856	0.00000
Temperature (°C)	22.240	0.00000
Temperature (°C) × Reaction time (h)	20.353	0.00000
Temperature (°C) × Volume org. layer (mL)	8.366	0.00000
Volume org. layer (mL)	8.073	0.00000
Effective cotton loading (mg/mL)	3.266	0.00054
Volume org. layer (mL) × Volume org. layer (mL)	2.023	0.00949
Reaction time (h)	0.247	0.56668

^aP-values <0.01 are statistically significant.

Interestingly, the stirring rate appears to be a nonsignificant effect in this model and is therefore not incorporated into the model (Table and eq ) even though increasing the stirring rate is associated with an increased interfacial area between two phases, promoting mass transfer. Lane et al. also investigated the relationship between the stirring rate and CMF yield in a biphasic system with 20 wt % HCl. They found that above 840 rpm significantly more CMF (29% yield) was produced compared to lower stirring rates (19% yield at 180 rpm). It is also possible that the tested range of stirring rates was not broad enough to significantly affect the CMF yield.

The accuracy of the model for the conversion of cotton to CMF is assessed by evaluating the actual versus predicted plot. The actual versus predicted plot indicates the accuracy by comparing the model’s predictions with the obtained experimental data. The red line represents perfect prediction where the predicted data is equal to the actual data. The blue line represents the regression null hypothesis, which states that none of the variables have a significant effect. The model is not significant when the blue line is contained by the red 95% confidence region and vice versa. In the actual versus prediction plot (Figure ), based on the used model effects (Table ), it can be seen that the blue line is not contained within the red 95% confidence region, indicating the model is significant. This is also confirmed by the low p-value of <0.0001. Additionally, the low root mean square error (RMSE) indicates a high predictive accuracy. Furthermore, the high coefficient of determination (R² = 0.93) shows that the variance in the CMF yield is explained by the model’s variables, thus the reaction temperature, the reaction time, the volume of the organic layer, and the effective cotton loading.

3.

Actual versus predicted CMF yield values based on the response surface model by JMP software.

The prediction profiler (Figure ) shows the significant effects of the model and their relation to the CMF yield, together with the optimized reaction conditions for maximal CMF production. Here, the quadratic effect of the temperature and the volume of the organic layer, and the main effect of the reaction time and cotton loading can be seen. As the stirring speed did not show any significance to the model, it is not included in the prediction profiler. The prediction profiler also includes the desirability of the model. The desirability was maximized to obtain the highest CMF yield possible for the current system. The optimal reaction conditions were found to be a reaction temperature of 91 °C, a reaction time of 1 h, an effective cotton loading of 8.3 mg/mL HCl, and 4 mL of organic layer/mL HCl.

4.

Prediction profiler for the maximum CMF yield based on the response surface model in JMP software. The highest molar CMF yield predicted is 65.0% at 91 °C, 1 h, an effective cotton loading of 8.3 mg/mL and an organic volume of 4 mL.

From the prediction profiler, it can be seen that the optimal values of reaction time, effective cotton loading, and volume of the organic layer are at the boundaries of the system. This could indicate that the optimum found is an apparent optimum and that the actual optimum can be outside the measured conditions. For effective cotton loading and the volume of the organic layer, it is not surprising that the optimum was found at the boundaries. For the effective cotton loading, or more generally, the sample loading, it is generally accepted that a lower concentration leads to a higher yield. Similar results were found by Mascal et al., Shi et al., and Howard et al. However, low sample loading is unfavorable when considering commercializing such a process, as it will negatively affect the economics. The prediction profiler does, however, indicate that increasing the loading will only lead to a slight decrease in CMF yield, which would be beneficial for an eventual business case of large-scale production.

Concerning the volume of the organic layer, higher yields were achieved by increasing the amount of organic solvent as this improves the extraction and, therefore, protects the CMF from degradation in the aqueous phase. As the optimal value was at the boundary of the design, extending the boundary of this factor would help to increase the CMF yield, though in theory, this should be infinitely high to maximize mass transfer. A (semi-) continuous process could facilitate this need, with a constant renewal of the organic layer. A similar observation was done by Shi et al., who found that a high organic-to-aqueous phase ratio led to a higher HMF yield compared to a low organic-phase/aqueous-phase ratio. Furthermore, a reaction time of 1 h seems to be too long at the optimal reaction temperature as the optimal reaction time is at the boundary of the system and for further optimalization shorter reaction times are suggested. Lastly, a reaction temperature above 91 °C leads to a decrease in CMF yield, possibly by decomposition of CMF.

The recommended reaction condition was tested (in quintuplicate), and the reaction was performed at 91 °C for 1 h with an effective cotton loading of 8.3 mg (20 mg polycotton textile), 1 mL 44 wt % HCl, 4 mL of organic layer/mL of HCl at a stirring speed of 900 rpm. The reaction was stopped after 1 h by removing the organic layer and washing the system thrice with 2 mL of chlorobenzene at RT. All the organic layers were collected and analyzed for CMF. The analysis of the reaction and washing layers showed a total CMF yield of 62.6% with a standard deviation of 2.3%, which is in line with the predicted CMF yield. Additionally, traces of glucose, HMF, LA, and FA were detected in the aqueous layer after synthesis, however, at the optimal reaction conditions, all these concentrations were below the limit of quantification. The chromatograms of the aqueous and organic layers are shown in Figure S3 and Figure S4, respectively. Besides glucose, HMF, LA, and FA, humins were visually present in the reactor after synthesis. Bueno Moron et al. investigated the one-step batch design of cotton to CMF at 80 °C with a solid loading of 15 mg in 1 mL of HCl (37 wt %) with 3 mL PhCl, which after 1 h yielded 42.4% CMF. The difference in yield can be related to the lower reaction temperature (91 vs 80 °C) and the lower acid concentration (44 vs 37 wt %). To verify the cause of the lower CMF yield, the optimized one-step batch reaction was performed with 37 wt % HCl, which led to a CMF yield of 43.9% (SD 2.4%), confirming the diminished effect of the lower acid concentration on the CMF yield.

The Effect of Different Process Designs

One-Step Batch versus One-Step Semicontinuous Design

Since the response surface model showed that a higher ratio of organic phase positively influences the CMF yield, the process type was changed from batch to semicontinuous to investigate the effect on the CMF yield. This design was also performed by Bueno Moron et al., which showed an 18.6% increase in overall CMF yield when switching from a one-step batch to one-step semicontinuous design. The optimal reaction time of the one-step batch design (1 h) was used for the total reaction time in the one-step semicontinuous process, by dividing the reaction into two blocks of 30 min with a solvent removal and replenishment between the two time blocks. As expected, the overall CMF yield of the semicontinuous process increased and resulted in a total CMF yield of 74.5% (Figure ). It is interesting to see that after the first 30 min in the semicontinuous design, which up until this point has been running as a batch system, there is no significant difference between the CMF yield of the organic layer after 30 min and 1 h. This confirms that the reaction time of 1 h at 91 °C is an apparent optimum, as similar results are obtained at shorter reaction times. The stagnation of the CMF production could indicate reaching equilibrium, as removing the organic layer with the formed CMF and adding fresh chlorobenzene further drives the CMF production. It can also be possible that the CMF formation in the second half hour is equal to the CMF decomposition. Traces of glucose, LA, and FA were detected in the aqueous layer, however, below the limit of quantification, making quantification inaccurate. HMF was not detected in the aqueous layer, likely due to the refreshment of the organic layer, which drives the conversion of HMF to CMF and/or avoids the hydroxylation of CMF to HMF.

5.

Effect of process type on CMF production. For the semicontinuous process, 4 mL PhCl was used per 30 min. Reaction conditions: 1 mL HCl (44 wt %), 4 mL PhCl, 91 °C, 1 h, 900 rpm, and an effective cotton loading of 8 mg/mL HCl.

Lastly, humins were also visually present after the reaction. Comparable results were obtained by Mascal and Nikitin, who produced CMF from cotton in a biphasic system with a true (one-step) continuous extraction design. Cotton was treated in a design containing a 5% lithium chloride in concentrated (37 wt %) HCl solution and 1,2-dichloroethane at 65 °C for 30 h, which resulted in a molar CMF yield of 67%. Although these results seem promising, the use of LiCl, long reaction times, and the excessive use of an organic solvent make the process economically unattractive. A higher CMF molar yield (83.5%) was obtained in a subsequent Mascal study on the conversion of microcrystalline cellulose in a biphasic one-step semicontinuous design containing 37% HCl and 1,2-dichloroethane at 100 °C for 3 h. However, it is generally known that microcrystalline cellulose is less recalcitrant than cellulose from native cotton. This difference in recalcitrance may account for the variation in yield. The research of Bhat et al. verifies this postulate, which showed that microcellulose crystalline resulted in a CMF molar yield of 72% whereas cotton led to a 64% CMF yield only under the same conditions.

Although the total CMF yield increased from 62.9% to 74.5% when switching from a one-step batch design to a one-step semicontinuous design, the yield is almost 20% lower than previously published by Bueno Moron et al., where a CMF yield around 95% was achieved. However, their study used a 3 wt % glucose solution, whereas the concentration in this work is almost three times higher. As mentioned above, it is generally known that lower substrate concentrations often lead to higher yields.

M_n Reduction Correlated to Combined Severity Factor

By treating the polycotton textile in the one-step design, the PET is (in contrast to the two-step process) also exposed to the more severe conditions needed for CMF synthesis. These conditions include temperatures above the T _g of PET (70–80 °C). Therefore, the quality of the polyester, in terms of number-average molecular weight (M _n), was investigated after the reaction. The polycotton material was exposed to different reaction conditions ranging from 80 to 100 °C, with reaction times ranging between 1 and 3 h, in a 44 wt % HCl solution. GPC was performed on the starting material and the solid residue after the reaction. To examine the effect of different reaction conditions, the combined severity factor (CSF) was calculated (eq ). The CSF helps to compare different reaction conditions by combining the reaction time, the reaction temperature, and the acid concentration into a singular factor. As the CMF production is dependent on more than the reaction time, reaction temperature, and acid concentration, the CSF cannot be used to investigate the relationship between the CSF and the CMF production.

$$\mathrm{C}\mathrm{S}\mathrm{F}=\log (R_0)-\mathrm{p}\mathrm{H}$$ 7

$${\mathrm{R}}_0=\mathrm{t}·{\mathrm{e}}^{{(\mathrm{T}}_{\mathrm{r}}{-\mathrm{T}}_{\mathrm{b}})/14.75}$$

where t = reaction time (min), T _r = reaction temperature (°C), T _b = base temperature (100 °C), and the constant 14.75 is the activation energy for a first-order system. ,

In Table , the calculated CSF and the measured average M _n of the starting material and the polyester residues are shown. From this table, it is clear that the polyester residue deconstructs.

3. Reaction Conditions of the Biphasic System with Their Corresponding Combined Severity Factor (CSF) and the Molecular Weight M _n of the Polyester Residue after Synthesis.

Reaction conditions	CSF	M n (kDa)
Starting material	-	27.7
90 °C, 1 h	2.65	16.5
80 °C, 2 h	2.66	13.5
90 °C, 2 h	2.95	9.1
90 °C, 3 h	3.13	7.2
100 °C, 2 h	3.25	5.8

Under more severe conditions, as the average M _n decreased from 27.7 to 16.5 and even to 5.8 kDa. The decrease in molecular weight can be directly linked to the increase in CSF, indicating that harsher reaction conditions lead to more decomposition. The decomposition products of PET could contaminate the aqueous and organic layers, which might require additional downstream processing of the aqueous layer to remove the impurities when the hydrochloric acid is reused on a commercial scale. Additionally, these impurities could affect the conversion from glucose to CMF, as both can react with starting material, intermediates, and products in the formation of humins. A significant lowering of the PET M _n can also lead to a change in the material properties of the polyester. However, during PET mechanical recycling, the M _n can be increased in a solid-state polymerization step, which is often included in PET mechanical recycling.

To avoid possible contamination of the reaction mixture and to limit a PET molecular weight drop, a two-step design can be used. With the two-step process, the separation of cotton and polyester with HCl is performed at room temperature, whereafter the polyester is removed from the reactor before adding the chlorobenzene. Previous work showed that acid hydrolysis at room temperature with superconcentrated HCl for 24 h led to only minor degradation of the polyester residue.

One-Step Versus Two-Step Batch Synthesis

To avoid contamination of the reaction mixture with PET decomposition products and to preserve the polyester, the CMF synthesis was performed in a two-step batch design using the optimal conditions found for the one-step batch process. The two-step synthesis started with the cellulolysis, thus dissolution, and hydrolysis of cotton with HCl at room temperature. The polyester residue was removed, chlorobenzene was added, and the reactor was heated to start the CMF production. Cellulolysis of cotton to glucose for 24 h at room temperature leads to the full separation of cotton and polyester and a high concentration of monomeric glucose and a low concentration of oligomers. However, it is not known whether a high concentration of glucose is pivotal to a high CMF yield. In contrast, a low concentration of glucose and a high concentration of oligomeric saccharides could be beneficial, as the oligomers might protect the glucose from degradation. Therefore, research was done on the minimal hydrolysis time required for complete cotton and polyester separation, leading to a high oligomeric saccharide concentration, and the effect of the monomeric/oligomeric saccharide ratio on the CMF yield in a two-step design.

The cellulolysis for the first step was performed with different textile concentrations (25, 50, and 100 mg/mL HCl) at different reaction time lengths (0.5–3 h in 0.5 h increments), and the residual textile materials were measured with TGA. Information on the separation of cotton and polyester and the glucose yield can be found in Figure and Figures S5–S7. It shows that at a concentration of 25 and 50 mg textile/mL HCl, the cotton and polyester are fully separated after 1.5 h. At a concentration of 100 mg textile/mL HCl, cotton is still present after 3 h; however, this was most likely caused by the incomplete submersion of the textile in HCl due to the large volume uptake from the polyester fraction (Figure S8).

6.

Effect of textile loading (25, 50, and 100 mg/mL) and reaction time on the separation of cotton and polyester and the molar glucose yield. Red indicates incomplete separation of polyester and cotton, while green indicates full separation of polyester and cotton. Reaction conditions: 1 mL HCl (44 wt %), no stirring. Textile ratio: 41% cotton/59% polyester.

Moreover, data showed that with a lower textile, and thus cotton concentration, the glucose yield is higher as relatively more acid is available per mole of cellulose, increasing the reaction rate. This effect seems to be more profound at longer reaction times. Additionally, since the textile residue is free from cotton after 1.5 h for a textile loading of 25 and 50 mg/mL, and the glucose yield is only 10% and 6%, respectively, it can be concluded that a large amount of oligomers is present in the solution. Thus, 1.5 h is the minimal time needed to fully separate cotton and polyester at concentrations of 25 and 50 mg/mL. This cellulolysis time leads to a solution containing a high concentration of oligomers and a relatively low amount of monomeric glucose.

With this information, the two-step batch reaction was performed with a cellulolysis time of 1.5 and 24 h to investigate the effect of a one-versus-two-step process and the effect of a low glucose and high oligomer concentration, and vice versa, simultaneously. The results of the one-versus-two-step CMF synthesis can be found in Figure . The CMF yield improves when performing the reaction in two steps, increasing from 62.6% to 69.8% and 71.8%, respectively, for the two-step process with 1.5 and 24 h of cellulolysis time. Glucose, HMF, LA, and FA were found in the aqueous layer; however, in this case, their concentrations were below the limit of quantification in all the considered process design options. Humins were also visually present in all designs. By intermediate removal of the polyester from the reaction mixture in the two-step synthesis, the reaction mixture is less contaminated, which leads to a higher CMF yield. A similar conclusion was drawn by Subramani et al., who found that increasing the purity of their carbohydrate-rich solution increased the CMF yield.

7.

Effect of the process structure on the molar CMF yield. For the two-step reaction, the hydrolysis step is either performed for 1.5 or 24 h. Reaction conditions: 1 mL HCl (44 wt %), 4 mL PhCl, 91 °C, 1 h, 900 rpm, and an effective cotton loading of 8 mg/mL HCl.

No significant difference was observed between the two-step reaction with either 1.5 or 24 h cellulolysis time, i.e., a low monomer/high oligosaccharide or high monomer/low oligosaccharide ratio, respectively. This indicates that the rate-limiting step of the overall CMF synthesis is not the hydrolysis of cellulose and glucose oligomers to glucose but rather the dehydration of glucose to HMF/CMF. Similar results were found by Mascal and Nikitin, who found nearly identical results for the conversion of glucose as well as cellulose to CMF in a biphasic system using an HCl/lithium chloride solution with 1,2-dichloroethane.

One-Step Batch Versus Two-Step Semicontinuous Process

By going from a batch to a semicontinuous process and going from a one- to a two-step synthesis, both led to an increase in CMF yield. The combination of these two alterations was also investigated to see the effect on the CMF yield. As the cellulolysis time did not have a significant effect on the CMF yield, the first step of the two-step semicontinuous process was performed for 1.5 h. The polyester residue was removed, and PhCl was added to start the second step, which involved the semicontinuous production of CMF. The effect of the change in the process structure and mode on the CMF yield is given in Figure . Here, it can be seen that changing from a one-step batch process to a two-step semicontinuous process led to an increase in CMF yield from 62.6% to 73.3%. While changing only the process type resulted in an 11.9% increase in yield (from 62.6% to 74.5%) and modifying the process structure led to a 7.2% increase (from 62.6% to 69.8%), implementing both changes simultaneously produced a 10.7% increase, suggesting that the effects are not additive. Additionally, glucose, LA, and FA were present in the aqueous layer at the end of the reaction in low quantities, below the limit of quantification. HMF was not detected in the two-step semicontinuous process, which was consistent with the results from the one-step semicontinuous process. This observation could imply that replacing the organic layer drives the conversion of HMF to CMF and/or prevents the hydroxylation of CMF to HMF due to a shift in equilibrium.

8.

Effect of the process structure and mode on the molar CMF yield. For the two-step semicontinuous reaction, the hydrolysis step is performed for 1.5 h, the polyester is removed, and thereafter 4 mL PhCl is used per 30 min during the second step. Reaction conditions: 1 mL HCl (44 wt %), 4 mL PhCl, 91 °C, 1 h, 900 rpm, and an effective cotton loading of 8 mg/mL HCl.

An overview of the performance of the different process designs is provided in Figure . Overall, the semicontinuous designs showed a higher CMF yield compared to their batch counterparts. The refreshment of PhCl drives the reaction toward CMF formation, as demonstrated by the absence of HMF in the aqueous layers of the semicontinuous processes, indicating complete conversion to CMF. Additionally, refreshing the organic phase protected the formed CMF from being degraded.

9.

Overview of the characteristics and performances of the different process designs used for the conversion of cellulose from polycotton waste to CMF in a biphasic system.

When the process structure was changed from a one-step to a two-step configuration, contamination by PET decomposition products was avoided, as the PET was not exposed to higher temperatures, which helps preserve its quality. Therefore, combining a semicontinuous operation with a two-step process structure appears most advantageous, as it results in high CMF yield, complete conversion of HMF to CMF, and prevents contamination of the system by PET decomposition products while preserving the PET quality.

Pilot Plant Process Implementations

The laboratory results have shown the potential to perform this reaction at the pilot scale and to valorize large quantities of polycotton waste, and decrease its burden on the environment. From all the possible reaction designs (Figure ), a two-step semicontinuous process with a minimal cellulolysis time is most suitable to valorize polycotton waste on a larger scale. A two-step design preserves the polyester residue better compared to a one-step design. In the semicontinuous design, the organic layer is continuously refreshed, which enhances CMF yield compared to the batch process. This was also concluded from the DoE, as the CMF yield increased with increasing organic volume. Additionally, the use of a semicontinuous design fully drives the conversion of HMF to CMF, as no HMF was found at the end of either of these semicontinuous designs. The reaction time of the first step, the cellulolysis of cotton, should be optimized to ensure only the full separation of cotton and polyester. Results showed that a high glucose concentration (i.e., long cellulolysis time) was not necessary to obtain a higher CMF yield and that a high concentration of oligomers (i.e., short cellulolysis time) leads to similar yields. On larger scales, the minimized reaction times will improve process economics. Additionally, it is recommended to perform the cellulolysis in a simulated moving bed design to increase the saccharide concentration in the acidic aqueous solution. The DoE showed that increasing the cotton loading only resulted in a slight decrease in CMF yield, however, from an economical perspective, increasing the textile loading will reduce the recycling and solvent evaporation volumes and make the process more economically viable.

After dissolving cotton in the first step, the acidic sugar solution can be removed from the first section and pumped into a different reactor vessel for further hydrolysis, dehydration, and chlorination of the sugar to CMF. This way, the valorization of both polycotton textile components can be maximized. A process flow diagram (PFD) of the two-step semicontinuous process for the production of CMF from polycotton waste textiles can be found in Figure S9 and Table S1. The PFD consists of a hydrolysis, a CMF production, a flash and extractive distillation, and an HCl absorber section. Key elements of the design include the flash and extractive distillation sections, which enable HCl recycling, and the solvent-recovery unit for PhCl, both of which are essential for enhancing the sustainability of the process. Although chlorinated solvents are often associated with negative environmental impacts, literature has shown that halogenated solvents in general are superior to nonchlorinated solvents, as the yield is significantly lower when using nonchlorinated solvents. ,, To diminish the environmental impact, solvent recovery and recycling are essential and an important part of the process design. The conversion of sugars to CMF can build upon Avantium’s current Dawn Technology pilot plant, where poly­(vinyl chloride) (PVC) lined reactors are used to hydrolyze the cotton from polycotton waste textiles. The PVC-lined reactors are one of the common solutions for the use of concentrated HCl, something that has been long established in the chemical industry.

Conclusions

The complexity of polycotton waste textile, the biggest fraction of textile waste, makes it extremely difficult to completely valorize it into new valuable chemicals. This study is the first to investigate the synthesis of CMF from polycotton waste in a biphasic system with 44 wt % HCl and optimized CMF yield while preserving the polyester. The organic layer of the biphasic system protects the CMF from degradation due to the absence of water. Using a DoE, the (apparent) optimal reaction conditions (91 °C, 1 h, effective 8.3 mg of cotton/mL HCl, and 4 mL of organic layer/mL HCl) were determined for the one-step batch design, which resulted in a CMF yield of 62.6%. The stirring speed did not show any significance to the model in the tested region. Variations to the current design (one-step batch) were tested, and improvements in CMF yield were observed when performing the reaction in a different process design. Switching to a semicontinuous design positively affected the CMF yield, as the yield increased from 62.6% to 74.5% by driving the reactions toward CMF formation and protecting the CMF from degradation. The difference between a one-step and a two-step batch process was also investigated, as this would prevent the deconstruction of polyester residue in the aqueous acidic medium. The cellulolysis time was optimized to ensure solely full separation of cotton and polyester, as results showed that there was no difference between the CMF yield of a process containing a high quantity of monomeric glucose or a process containing a high quantity of oligomeric glucose. By performing the reaction in a two-step manner, the yield improved from 62.6% to 69.8% and 71.8%. Lastly, the two variations were combined, and the synthesis of CMF was performed in a two-step semicontinuous process with a 1.5 h cellulolysis time, which led to a CMF yield of 73.3%. This design has the most potential for large-scale CMF production, the PET is preserved and fit for further mechanical and chemical recycling, while maximizing the CMF yield.

Glossary

Abbreviations

CCD

central composite design

CMF

5-(chloromethyl)­furfural

CSF

combined severity factor

DoE

Design of Experiments

DTG

derivative thermogravimetry

FA

formic acid

GC

gas chromatography

GPC

gel permeation chromatography

HCl

hydrochloric acid

HFIP

1,1,1,3,3,3-hexafluoro-2-propanol

HMF

5-(hydroxymethyl)­furfural

HPLC

high-performance liquid chromatography

LA

levulinic acid

M _n

number-average molecular weight

PEF

poly­(ethylene 2,5-furandicarboxylate)

PET

poly­(ethylene terephthalate)

PFD

process flow diagram

PhCl

chlorobenzene

PVC

poly­(vinyl chloride)

RMSE

root-mean-square error

RSM

response surface methodology

TGA

thermogravimetric analysis

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c02010.

• 1.Additional results of the hydrolyzed textile residues for 25, 50, and 100 mg polycotton/mL HCl solution. Process flow diagram of the one-step semicontinuous design for CMF production from polycotton waste (DOCX)

N.L. devised the experimental program and wrote the manuscript. N.L. and T.G. performed the experimental work. G.v.K., and G.-J.G. conceived the concept. All authors have edited the manuscript and have given approval to the final version of the manuscript.

MIWATEX project, part of research program Circularity (KIC) (project number KICH1.ED01.20.006) which is (partly) financed by the Dutch Research Council (NWO).

The authors declare no competing financial interest.