Introduction

Recent research and global decisions have led to the development of advanced techniques and practical solutions to address the global concern of plastic waste and the reuse of these valuable resources. Many companies and organizations have taken action to address sustainability and embrace the circular economy by investing resources to remove plastics from waste streams. To achieve a circular economy in the plastics industry, effective collection of plastic waste [1, 2], improved sorting techniques [3, 4], and sustainable product design are necessary [5]. The main approaches are chemical and mechanical recycling, individually or combined [6]. Chemical recycling involves breaking down used polymers into their building blocks or restoring them to a purified state as new feedstock materials [7]. On the other hand, mechanical recycling is a cost-efficient approach involving incorporating recycled polymers and virgin polymers into new products at different ratios [8]. In various applications, such as film materials, blends of low-density polyethylene (LDPE) and linear low-density polyethylene (LLDPE) are commonly used, which is why the use of recycled materials in such applications is desired.

LDPE and LLDPE resins are commonly combined to improve film processing over different applications [9, 10]. LDPE long-chain branching (LCB) enhances melt strength and offers a high degree of shear-thinning, improving extrudability and flowability [11]. On the other hand, LLDPE is a widely used material in flexible packaging with a density range of 0.915–0.935 g/cm3 [11]. It is produced through catalytic copolymerization of alpha-olefins like 1-butene, 1-hexene, and 1-octene with ethylene [11]. Incorporating LLDPE in such blends reduces density and increases tie chains, forming amorphous rubbery domains [12]. These amorphous domains bridge the adjacent crystalline lamellar regions, enhancing the film's large deformation properties, such as impact, tear, and stretchability [13, 14].

Commercially available post-consumer recycled (PCR) materials may contain LLDPE, LDPE, and polypropylene (PP) resulting from mixed films in the recycling stream [15]. PP might be present in small quantities due to contamination. Identifying and determining the composition of these PCR blends is crucial for targeted applications [12]. The composition of LLDPE/LDPE blends can be determined by analyzing the inherent differences in melting and crystallization characteristics between LDPE and LLDPE during polymer crystallization, either from a molten or solution state [16,17,18,19]. The density of LLDPE and LDPE strongly affects their thermal properties, mainly due to branching content (and chain length) and how chain branches are distributed among polymer chains [12]. Secondarily, the average molecular weight (Mw) and molecular weight distribution (MWD) also play a role. Blends of LDPE and Ziegler–Natta LLDPE (ZN-LLDPE) are phase-segregated in the crystalline state when analyzed using DSC experiments [16]. LLDPE tends to crystallize from its molten state at higher temperatures, and linear crystallizable sequences of similar lengths may be able to co-crystallize in blends with LDPE [20]. To quantify the composition of the LLDPE/LDPE blend, relative peak sizes (peak height or area) can be used to estimate the blend's composition. Prasad [18] proposed a technique that uses DSC heating scans to evaluate the LDPE content in a neat ZN-LLDPE/LDPE blend. Calibration curves were generated using known C4-LLDPE/LDPE, C6-LLDPE/LDPE, and C8-LLDPE/LDPE blends, with the α-olefin type identified using Fourier transform infrared (FTIR) spectroscopy before quantification [18]. However, this method is challenging in estimating LDPE/LLDPE content in post-consumer and post-industrial recycled materials since identifying the comonomer type through FTIR is difficult due to contamination [12]. Furthermore, in PCR materials, the presence of multiple comonomers makes the proposed DSC method impractical for quantifying the PCR blend composition, but it is suitable for comparison purposes.

Other methodologies have been developed to investigate the quantification of LLDPE and LDPE in different blends. Recent research introduced a technique based on nuclear magnetic resonance (NMR) to determine the composition of LLDPE/LDPE blends [21]. This method involves comparing the 13C NMR spectra of the blend with those of neat LDPE, using integrated peak areas derived from specific resonance peaks present in both LDPE and blended samples. By quantifying the total integrated area of specific spectral regions within the LLDPE/LDPE blend and then comparing it to the corresponding LDPE measurements, an estimation of the average mass percentage of LDPE in the blend can be obtained [21]. However, the accuracy of this approach critically relies on having a reference LDPE sample with a closely matched density and processing characteristics, which poses a significant challenge when dealing with mixed post-consumer and post-industrial recycled materials [21]. In addition, implementing this methodology requires excluding specific outlier data points from the analysis to achieve an acceptable level of variance. This data filtering process may impact the accuracy of the estimated LDPE content [21].

In a recent investigation, Hashemnejad [12] proposed a composition analysis of virgin and post-consumer recycled LLDPE and LDPE blends. A crystallization elution fractionation (CEF) technique was proposed for accurate composition analysis. Calibration curves for CEF were fabricated using blended LLDPE and LDPE resins, showing consistent trends. Fourier transform infrared spectroscopy and nuclear magnetic resonance were used to identify comonomer types in LLDPE. In this case, they report that CEF and DSC techniques offer similar LDPE content results for virgin polymer blends, depending on density. However, CEF proves more precise and reliable for estimating LDPE content in PCR LLDPE/LDPE blends, particularly those potentially containing inorganic compounds [12].

In the present investigation, self-nucleation (SN) and self-nucleation and annealing (SSA) protocol employing differential scanning calorimetry (DSC) were used to determine the LDPE/LLDPE composition and quantify PP contamination within commercially post-consumer recyclates. The methodology enables the fractionation of the materials during the thermal treatment, separating such fractions by temperature ranges and then associating the different fractions with the corresponding fractions in neat materials. Calibration curves are constructed using model blends, mimicking the recyclates' composition, and the evaluation of LDPE/LLDPE content is addressed by selecting the appropriate temperature range from the resulting melting trace from the SSA protocol. When quantifying PP content, the self-nucleation protocol is used in Domain III. In this case, unmolten crystals present during the thermal protocol act as nucleating agents to boost PP crystallization and enhance its melting enthalpy. This enables an approximated quantification of the PP content in the post-consumer recycled materials. The results can be compared with the nuclear magnetic resonance (NMR) technique.

Experimental

Different resins of linear low-density polyethylene (LLDPE), low-density polyethylene (LDPE), and polypropylene (PP) were selected for the fabrication of the model blends. Borealis Polyolefine GmbH kindly provided them. The polymer's density and Melt Flow Index (MFI) are listed in Table 1. The selected LLDPE grades covered the density range typically found in recycled PE streams. Melt blending was used to prepare model PP/LLDPE/LDPE and LDPE/LLDPE blends in a microcompounder Xplore MC15HT at 190 °C, 50 rpm, and 2-min residence time, with applied nitrogen to prevent degradation. Based on previous experiments, melt blending conditions were chosen adequately to achieve homogenization while preventing polymer degradation. The model blend formulation is composed of two groups. The first one encompasses the model blends with PP, denoted as MODEL_PP-X, where X stands for the amount of PP incorporated in the blend as 1, 3, 5, and 7%, the remaining composition is equal amounts of LLDPE and LDPE, for example, MODEL_PP-1 is composed of 1% PP, 49.5% LLDPE, and 49.5% LDPE. The second group of model blends is denoted as, for instance, MODEL_LDPE + LLDPE-Y, where Y indicates LLDPE density. For each LLDPE density, three compositions were fabricated (LDPE/LLDPE 30/70, 40/60, and 50/50) for the creation of the calibration curves. LLDPE and LDPE components are listed in Table 1. For a further indication of the blend's name and specific composition, refer to Table S1 in the supplementary information. In addition, six different materials from post-industrial waste (denoted as REC_0 to REC_5) were selected to evaluate the proposed method's effectiveness in estimating PP and LDPE/LLDPE content via self-nucleation and self-nucleation and annealing (SSA) methodology.

Table 1 Materials used to prepare model blends along with their density and MFR
Full size table

The thermal properties of neat LLDPE and LDPE, along with their model blends, were studied using a TA Instruments DSC (model 2500), calibrated with indium, zinc, and tin under a nitrogen atmosphere. Film sheets were prepared from the samples, and circular sections were loaded into a standard DSC pan. The samples from the first group of model blends (MODEL_PP-X) were heated from room temperature to 200 °C at 10 °C min–1 and kept for 3 min at that temperature to remove the thermal history. Immediately, the blends were cooled to 0 °C and then heated to 200 °C at 10 °C min–1. The estimation of PP, in this case, is based on the obtained endotherms from the second heating step.

As for the second group of model blends (PE model blends), the samples were heated from room temperature to 180 °C at 10 °C min–1 and kept at 180 °C for 3 min to remove any thermal history. Afterward, the materials were cooled at 10 °C min–1 to 0 °C, followed by a second heating step at 10 °C min–1 until 180 °C. Quantitative analysis was based on the observed endotherms during this second heating step.

To estimate the PP content in model and recycled blends, the adopted self-nucleation (SN) protocol is based on the method devised initially by Fillon et al. [22], with subsequent reviews by Michell et al. [23] and Sangroniz et al. [24]. The process consists of six sequential steps conducted at a controlled heating rate of 10 °C min–1. The steps applied are as follows: In Steps 1 and 2, thermal history was removed by heating the material well above its melting temperature (typically 30 °C above the melting peak) and then cooling it down at a fixed rate of 10 °C min–1 to establish a "standard" thermal history. Step 3 consists of heating the sample to a specified temperature, Ts. Step 4 is a 5-min thermal treatment at the chosen Ts. The selection of these Ts might result in different possible outcomes: Domain I when the temperature is high enough for complete melting, Domain II when a lower temperature is chosen, facilitating partial or complete melting, which promotes self-seeding or melt memory, respectively, and finally, Domain III with a temperature low enough that results in the annealing of the unmolten crystals at Ts. In step 5, the sample is cooled from Ts to room temperature, and its crystallization temperature is monitored. Finally, in step 6, a final heating run identifies changes in the material's melting temperature due to the self-nucleation treatment, revealing the actual Domain based on the Ts selection.

This self-nucleation temperature program addressed the different domains in MODEL_PP-X blends. As already mentioned above, in Domain III, the unmolten crystals boost the epitaxial crystallization of the PP phase. Therefore, in the last heating, step 6, it is possible to quantify the enthalpy associated with the self-nucleated PP, which gives a good approximation to the content of PP in the sample. This was used to estimate the PP content of the recycled materials and the model PP materials to compare the accuracy of the prediction. In all experiments, the values of specific parameters in the SN protocols were kept constant: heating and cooling rates at 10 °C min–1, a thermal treatment time at Ts of 5 min, the initial temperature of 200 °C for crystalline-memory erasure, and a final temperature of 20 °C at the end of the cooling step. These specific scan rates and annealing time values were chosen based on previous self-nucleation studies [25,26,27,28]. As part of this investigation, the only variable that changed in each experiment was the value of Ts.

As for the determination of LDPE/LLDPE content in model and recycled materials, the successive self-nucleation and annealing (SSA) protocol was employed. The temperature program utilized for the SSA fractionation was developed by Müller et al. [25, 29, 30]. More details are given in the Supplementary Information (Figures S1 and S2). In brief, this protocol starts with steps resembling the previously mentioned standard SN protocol. In this case, steps 1 to 4 of the SN were repeated, where the first Ts will be the Ts,ideal of the material. Ts,ideal is the temperature leading to the highest nucleation density without annealing of un-molten crystals, i.e., the minimum Ts temperature in Domain II. This temperature corresponds to the lowest temperature within Domain II. Consequently, each material will fractionate from the same initial ideally self-nucleated condition. For comparison purposes, Ts,ideal of recycled materials (i.e., Tsideal = 128 °C) was evaluated in all the samples. It is essential to mention that subjecting the material to 5-min thermal conditioning at the first Ts value does not cause annealing. No thermal fraction is produced since this Ts value lies within Domain II. The subsequent steps are crucial as they determine sample fractionation. After step 4, a cooling ramp is applied to 20 °C, followed by heating to Ts2, which is 5 °C lower than the initial Ts. This process is repeated by decreasing Ts by 5 °C (i.e., 5 °C fractionation window) in each step until 11 fractions are created (12 selected temperatures).

Results and discussion

Determination of polypropylene content as contamination in recycled blends.

The quantification of polyolefin blend components has been reported extensively using various methodologies [17, 18, 21, 31,32,33,34,35]. One of the main techniques depends on the crystallinity obtained from the melting enthalpy in the standard heating–cooling–heating thermal protocol using DSC equipment. This methodology consists of determining the area under the curve for the specific blended material in the adequate temperature range and comparing it with the neat polymer enthalpy value. In this way, it is possible to estimate the composition of the material [33] roughly. However, the accuracy of using enthalpy to determine polymer crystallinity from DSC measurements has been questioned because they are obtained under non-equilibrium conditions [17, 31]. In addition, the melting transitions are influenced by sample crystallinity, and the thermal history of samples could also impact the analysis, potentially limiting the technique's accuracy for quantitative determinations [36]. Nevertheless, this technique is commonly used for its simplicity and easiness in quality control laboratories as the first approach to qualitative characterization. It is essential to mention that PP has melting temperatures (Tm) that are different from LLDPE and LDPE, resulting in different transition peaks in the DSC scans and allowing their proper identification [35].

A conventional DSC protocol (heating–cooling–heating) was initially employed to determine PP composition in recycled blends. The resulting second heating endotherms from the standard protocol in the recycled materials are shown in Fig. 1. The recycled materials possess three different melting peaks assigned to different components in the blend. Starting from the low temperatures, the first peaks located at around ~ 110 °C are associated mainly with LDPE, while those at higher temperatures, such as around ~ 123 °C, are due to LLDPE. These thermal melting transitions agree with those reported in the literature [37]. However, it is crucial to highlight that, as recognized in the literature, the crystallization behaviors of LLDPE and LDPE frequently involve co-crystallization [38]. These intricate interactions depend on variables such as the specific fabrication method (involving catalysts like Ziegler–Natta types) and the inherent comonomer composition within LLDPE. It is important to note that due to these complex interplays, the resulting crystalline peaks have the potential to overlap during the DSC standard measurements [39]. In addition, at much higher temperatures (~ 160 °C), a small melting transition appears. At this temperature, the peak is associated with the PP component, as the melting temperature coincides with the melting temperature for PP and incidentally is much higher than the equilibrium melting temperature of PE. In this case, the content of PP is low, as indicated by the small peak height and area; therefore, it is associated with contamination in the PE recycled material. This contamination is because the sorting step is not 100% efficient during recycling, allowing slight material contamination [40]. Quantifying the PP content to propose applications to the final recycled material and quality control assurance is crucial.

Fig. 1
figure 1

DSC second heating scans of the post-consumer recycled blends

Full size image

As mentioned before, to address the quantification of crystalline polymer materials with standard DSC, the method is based on the comparison between the enthalpy of melting (ΔHm) from the obtained DSC scans in both virgin and blended material. This method allows the composition calculation, which is then used here to compare further with the SSA and NMR methodologies. Table 2 presents the obtained results from the PP estimation via DSC.

Table 2 Comparison of PP quantification in recycled materials with different techniques
Full size table

In addition, the different recycled materials were studied using the self-nucleation methodology to obtain the different thermal characteristics of the components in the blend and its distinctive domains and quantify the PP content. The different self-nucleation domains for the PP phase within the recycled blends were successfully obtained by applying a range of selected self-nucleation temperatures (Ts). In Domain I, a lack of change in the crystallization temperature of the PP phase indicates the complete melting of crystals, as evident from the red curves in Fig. 2. In addition, in Domain II, a noticeable increase in the PP crystallization temperature (Tc) was observed during cooling scans originating from the selected Ts, as shown in the blue curves in Fig. 2A. This change is due to self-nucleation, which significantly increases the density of nucleation sites. Consequently, an increase in the crystallization temperature is obtained [41]. Finally, an annealing peak appeared in Domain III at temperatures higher than the primary fusion endotherm, highlighted by the green curves in Fig. 2B. The distinctive high-temperature peak that emerged initially at Ts = 165 °C is attributed to the annealing of un-molten crystals.

Fig. 2
figure 2

DSC scans for the recycled material REC_0: (a) cooling and (b) subsequent heating after treatment at Ts. Domains are color-coded: red for Domain I, blue for Domain II, and green for Domain III

Full size image

When the Ts value reaches 160 ºC, annealing affects a large population of the existing PP crystals, producing a sharp melting endotherm well separated from the PE melting endotherms. The obtained endotherm at Ts = 160 °C is then used to estimate the PP composition more accurately than the untreated DSC standard melting scan. Furthermore, the observation of these self-nucleation Domains is extended to other recycled materials, confirming their consistent occurrence and enabling the methodology to estimate the PP content and compare it with the DSC standard method and with the self-nucleation in Domain II.

An interesting additional observation emerged when examining the influence of self-nucleated PP on the PE phase. As the crystallization temperature increased in the PP phase, a corresponding increase occurred in the PE phase crystallization temperature. This trend aligns with our earlier studies in different model systems [27, 28, 42,43,44,45], and it is extended here to real recycled materials, as illustrated in Figures S3 and S4. Since the selected Ts value is high enough to melt the PE phase completely, the PP phase acts as a nucleating agent for the LLDPE, increasing its crystallization temperature.

Table 2 presents the polypropylene content prediction in post-consumer recycled materials using the standard and the self-nucleation methods. Additionally, the quantification results from NMR are provided for comparative analysis. Notably, the conventional DSC protocol underestimates the polypropylene composition compared to the NMR measurements. In addition, the self-nucleation (SN) protocol demonstrates closer alignment with the calculated NMR values. Consequently, the self-nucleation method emerges as a resourceful and time-effective alternative to the conventional standard protocol, thus showcasing its potential for application in environments where time is of the essence, e.g., quality control procedures as an initial step for quantification.

Quantification of LDPE/LLDPE in recycled blends

The SSA thermal fractionation protocol was applied according to the recommendations by Muller et al. [26, 29, 30] to analyze the behavior of each material individually. Before that, the selection of the Ts ideal was performed. Using the SN protocol specifically for the PE component of model and recycled materials, the Ts, ideal values were obtained and are given in Table 3.

Table 3 Obtained values of Ts, ideal from the application of the self-nucleation protocol (SN)
Full size table

The correct application of the SSA protocol demands the use of the Ts, ideal, as a starting point because this is the temperature that causes maximum self-nucleation (maximum increase in Tc) without producing annealing, as the sample is ideally self-nucleated; it contains the maximum possible nucleation density [26]. This is necessary when the results of the SSA method will be used to quantitatively determine short-chain branching distribution data through suitable calibration curves obtained by FTIR and NMR. Nevertheless, in this case, the comparison at different Ts,ideal temperatures would be difficult since the melting peak distribution and the areas under the different melting peaks will be different.

As suggested in the literature [26, 27], in recycled materials, the selection of a common highest Ts, ideal for all materials under study, is a convenient way to compare a series of samples using the same thermal fractionation procedure. Together with a constant fractionation window (5 °C) and the same thermal history, the tree parameters make the SSA results fully comparable. For this reason, a Ts = 128 °C was selected as the highest value; therefore, all the SSA procedures were carried out at this Ts. When selecting a constant Ts for performing SSA in different materials, the resulting melting curve will match all valley positions, thus making the comparison easier. To fractionate the individual components, the standard SSA protocol was used with the selected constant Ts and incorporating 12 steps (128, 123, 118, 113, 108 °C; 103 °C; 98 °C; 93 °C; 88 °C; 83 °C; 78 °C; and 73 °C).

The final heating curves after the SSA protocol for neat, model, and recycled materials at a specific density are shown in Fig. 3. The sum of the neat DSC traces at different compositions producing theoretically predicted curves is also represented. We will denote these traces as "unmixed blends," as they are a simple weighted superposition of the neat polymers’ experimental DSC plots [25]. The SSA curve obtained for neat LDPE shows only eight melting peaks because no annealing was produced at the highest Ts temperature since Ts = 128 °C is within Domain II, where no annealing occurs. Regarding the LLDPE, the characteristic melting peaks of the produced fractions were obtained. For each LLDPE density, the corresponding SSA is presented in Figures S5–S7 in the supplementary information. The unmixed blends, in this case, were generated by adding each neat material's corresponding DSC trace normalized by the selected composition; for instance, when generating the unmixed LDPE_LLDPE_931 (70/30) curve, the neat LDPE curve was multiplied by its corresponding fraction (0.7). Then, it was summed with the corresponding fraction of the LLDPE curve (0.3). In this way, the specific contributions of different compositions were obtained. This is the theoretical curve one would obtain if there were no interactions between the components, i.e., the unmixed blend.

Fig. 3
figure 3

Final DSC heating scan after applying the SSA thermal protocol to: (a) neat LDPE and LLDPE, (b) the sum of neat curves at different compositions to create the model predicted, (c) the experimental model blends at different compositions, and d) the resulting DSC scan from recycled material number 1

Full size image

Nonetheless, as it is generally known [46], the different types of polyethylenes can have different interactions, such as co-crystallization or dilution of the higher-temperature crystallizable phase. Therefore, experimentally, the obtained fractionated SSA curves will differ from the unmixed blends when such interactions arise. The unmixed blend curves in this context are meaningful since they provide insights into which theoretical thermal fractions differ from the experimental curve.

Three different unmixed blends were constructed per each selected LLDPE. Figure 3 also presents the model experimental blends (these are blends made with virgin materials, not recycled ones), and in this case, the density of 931 kg m–3 is shown as an example. The first observation that can be made from the experimental curves of the model blends is that they differ from the unmixed blends. This means the peaks do not follow the same pattern as in the unmixed blends but instead have a characteristic behavior. When the first three high-temperature peaks from unmixed and model blends curves are compared, it is observed that the melting peaks corresponding to the model blends fractions are narrower and that the melting temperature varies among the peaks. As previously noted, this distinct behavior is attributed to the co-crystallization effect of the different polyethylene chains in the model blends.

Typically, polyethylene blends, which have a similar melting range and are miscible in the melt, often undergo co-crystallization between specific chain populations where the methylene sequence length (MSL) is of comparable length [25]. In fact, this co-crystallization can be observed during the second heating scan in a standard heating–cooling–heating thermal protocol in DSC, as presented in Fig. 4. In this case, it comes together using the already mentioned SN and SSA protocol to avoid most of the co-crystallization effects obtained in standard DSC for the LLDPE composition estimation.

Fig. 4
figure 4

DSC second heating scans of neat, model predicted, model experimental, and recycled material number 1, showing the effects of co-crystallization

Full size image

We now turn to the design of calibration curves obtained from the model blend that can correlate areas under specific thermal fractions and the quantity of LLDPE in the blends. The evaluation was conducted using the TRIOS software in the following manner: Initially, the total area was derived from the SSA melting curves for each material. The temperature limits for integration were set and consistently applied across all model and recycled blends to guarantee precise comparisons. Specifically, the selected temperature range covered from 25 to 140 °C. Next, the area under each thermal fraction within the final SSA heating run was determined. It is crucial to note that the integration limits chosen for each fraction remained consistent across all materials.

Given that the fractionation window and the selected seeding temperature (Ts) were identical, the valleys in each fraction appeared at precisely the same values, corresponding to the applied Ts temperatures. Details regarding the integration limits for every analyzed peak can be found in Table S2 in the supplementary information. For a more accurate representation, the area of the first two and three peaks (counting from high to low temperatures; the first peak always corresponds to the melting of the highest temperature thermal fraction) was normalized against the total melting area. This step of normalization was undertaken to remove any potential sources of variation. Two sets of calibration curves were subsequently established. The first set centered on the combined normalized area of the initial two high-temperature peaks, spanning a temperature range from 140 to 119.7 °C. Four distinct calibrations were derived, each corresponding to the different LLDPE densities used (see Fig. 5A). For the second set of calibration curves, the temperature range was extended from 140 to 114.9 °C to include the first three high-temperature fractions (see Fig. 5B).

Fig. 5
figure 5

Calibration curves obtained from the model DSC experimental curves considering (A) the first two peaks and (B) the first three peaks from the SSA fractionation melting curve

Full size image

A linear trend is observed in the calibration curves. Specifically, higher-density materials display a higher normalized value from the sum of the initial peaks, while in the lower-density LLDPEs, a lower value is observed. Given that LLDPE at lower densities typically has a higher branching content, as a result of the SSA protocol, fractionation at lower temperatures is greater because of the high branching content, implying that the corresponding peaks at these temperatures will possess a larger area. Since the calibration curve evaluation involves normalizing all peaks, an increase in the total area relative to the first two or three peaks results in a reduced normalized value. In contrast, for higher-density LLDPE, which has a lower short-chain branch content and exhibits, on average, longer linear chain sequences, the analysis outcome with the normalized first two or three peaks will be a higher value in the calibration curves. Table 4 presents the fitting parameters derived from the calibration curves. We have chosen to compare the results obtained with both calibration curves (two and three peaks) to differentiate the effect of the co-crystallization over the estimation of LLDPE. As predicted, the calibration curves for the two peaks are less affected by co-crystallization with LDPE, thus representing the LLDPE content more closely.

Table 4 Numerical results and correlation coefficient of the linear calibration curves based on SSA peak area determination
Full size table

Estimation of the composition of recycled blends

For the analysis of post-consumer polymer feedstocks, recognized for their heterogeneous composition and contamination [47, 48], the SSA methodology mentioned above was employed to estimate the LLDPE content. We investigated several post-consumer recyclate resins commercially marketed as blends of LLDPE and LDPE. Initially, the LDPE content was evaluated using a methodology previously developed by our group that utilizes a temperature-modulated DSC (TMDSC) [16]. For comparative purposes, the values obtained through this method were contrasted with those acquired using NMR. These values are presented in Table S3 in the supplementary information. A notable discrepancy was observed between the estimation from TMDSC and NMR in some materials. This can be attributed to the fact that recycled materials have a diverse composition, and the TMDSC method was initially designed for virgin materials. This may cause TMDSC to provide altered results when applied to recycled materials. For this reason, SSA is introduced as an alternative methodology. It aims to deliver precise results and offers a practical, cost-effective approach suitable for quality control assurance.

Within the context of this research, the LLDPE content in recycled materials was estimated using the previously established calibration curves (Fig. 5) obtained through the SSA methodology, considering the specific temperature range covering the first two and three peaks, which result from the last heating step in the SSA thermal protocol. For each recyclate material, the normalized sum of these peaks was determined. This value was then correlated with the respective calibration curve. This procedure was applied in each calibration curve, covering all density ranges. Refer to the supplementary information (Figure S8) for a more detailed calculation example in a specific recycled material.

Figure 6 compares the LLDPE composition determined from the calibration curves by the SSA methodology with the estimations derived from the NMR methodology. This comparison aims to evaluate the SSA efficacy against the results from the NMR, a known method for its robustness and precise prediction capabilities in polymer and recycled materials. The alignment of results from SSA with those of NMR highlights the potential of the former as a viable predictor of LLDPE composition. As observed in Fig. 6, the values obtained from the calibration curves agree with the NMR outcomes. Specifically, materials REC1 to REC5 display a good estimation, ideally approaching the diagonal dotted line, particularly in the 918–923 kg m–3 density range. However, when considering the estimations with the higher densities (Fig. 6C and D), one can notice that only the REC0 material from the SSA analysis is adjusted correctly. In addition, Figure S9 presents the results considering the calibration curves calculated with the first three peaks.

Fig. 6
figure 6

Comparison of the estimated LLDPE composition by NMR and the calculated LLDPE composition by SSA of 6 different recycled materials, considering the first 2 areas of the fractionated melting SSA result at different LLDPE densities as (A) 918 kg m-3; (B) 923 kg m-3; (C) 931 kg m-3; and (D) 935 kg m-3

Full size image

Considering the nature of the different recycled materials under investigation, which exhibit a complex molecular structure as presented in Fig. 7, they can also display different characteristics ranging from a high degree of branching distribution to very linear chain sequences. This diversity in molecular architecture makes the estimation process challenging, as it is impossible to establish a single calibration curve that accurately determines the specific composition of LDPE (low-density polyethylene)/LLDPE (linear low-density polyethylene). It is possible to note in Fig. 7 that the REC0 material presents an increased first peak with respect to the other fractions and also with respect to the other materials. This means that this material possesses more linear chains that are able to crystallize at higher temperatures instead of the broader distribution found in other recycled materials. In addition, to support this statement, CEF measurements of REC0 and REC3 showed a higher elution temperature for REC0 compared to REC3, confirming the structural differences in the chain distribution obtained in the SSA. Refer to the supplementary information in Figure S10.

Fig. 7
figure 7

Final DSC heating scan after applying the SSA protocol to the recycled materials using fractionation windows of 5 °C; the dashed vertical lines correspond to the employed values of Ts starting from Ts = 128 °C

Full size image

We have formulated a qualitative parameter to establish the most appropriate calibration curve for an unknown recycled material. This is based on the understanding that recycled materials that contain longer uninterrupted chain fractions (i.e., higher densities or higher amounts of higher densities LLDPEs) will exhibit a more pronounced initial peak during the final heating phase as part of the successive self-nucleation and annealing (SSA) protocol. Given that these materials are more accurately quantified using calibration curves corresponding to higher densities, we propose the following criterion: If the area of the first peak is greater than 1.5 times that of the second peak, the material's composition is best determined using calibration curves designed for higher densities (931 and 935 kg m–3).

In addition, if this condition is not met, calibration curves associated with lower densities (918 and 923 kg m–3) should provide a more accurate estimation of the linear low-density polyethylene (LLDPE) content. This approach allows for a more tailored analysis of the polyethylene content, optimizing the accuracy of the compositional assessment.

where \({A}_{1}\) is the area under the first peak and \({A}_{2}\) is the area under the second peak. Using this methodology, it is possible to establish the composition of LLDPE in recycled post-consumer materials when considering the mentioned calibration curves. In addition, it is possible to integrate the protocol for estimating PP as contamination in the PCR. In this context, it is possible to estimate the approximate content of PP contamination, as presented in Table 2, and determine the content of LLDPE via one methodology.

Conclusions

This study presents a methodology based on the successive self-nucleation and annealing (SSA) technique for the qualitative analysis of post-consumer polymer feedstocks in the estimation of linear low-density polyethylene (LLDPE) and low-density polyethylene (LDPE) content within heterogeneous commercial recycled PE materials. This methodology offers an advantage over other methods that require more time and resources to be completed or provide an inadequate quantification. We have demonstrated that this approach quantifies LLDPE/LDPE ratio by consistently fractionating the constituent polymers in recycled materials.

Our findings indicate that traditional methods, such as temperature-modulated differential scanning calorimetry (TMDSC), may not provide accurate estimations when applied to recycled materials due to their varied composition compared to virgin blends. The proposed SSA methodology overcomes these limitations. Comparative analysis showed that the SSA protocol yields results that align more closely with those derived from nuclear magnetic resonance (NMR) measurements. This suggests that the SSA protocol could offer a more precise and practical alternative for the quality control assurance of recycled polymers.

Future work should focus on refining the SSA methodology for different types of polymer blends and exploring its applicability on an industrial scale. The ultimate goal is to enhance the sustainability of plastic recycling processes by providing reliable, cost-effective methods for material characterization.