Making sense of plastics recycling

Making sense of plastics recycling

Emma van Bruggen

Student at Delft University, Delft, Netherlands

Rolf Koster

Delft University, Faculty of Industrial Design Engineering, Delft, Netherlands

Kim Ragaert'^, Ludwig Cardon'^, Marcel Moerman^

^Centre for Polymer and Material Technologies CPMT, Associated Faculty of Applied Engineering Sciences, University College Ghent, Ghent, Belgium

^Department of Materials Science & Engineering, Ghent University, Ghent, Belgium

Epco Blessing

Will & Co., Badhoevedorp, Netherlands

ABSTRACT: Major benefits of plastics recycling are reduced depletion of non-renewable resources and re-duction of world-wide waste. Traditional thermo-mechanical recycling causes rere-duction of mechanic£ll proper-ties for most thermoplastics. Down-cycled materials may nevertheless be suited for certain useful applications. Developing such applications may be a step towards more effective future use of resources. Increased applica-tion of recycled thermoplastics is generally preferable over landfills, or waste lying around in some parts of the world. Different recycled high-impact polystyrenes (HIPS) from TVs were mechanically tested, as well as mixtures of them. Mixtures of a low-density polyethylene (LDPE) and a polypropylene (PP) were tested as well, representing a large portion of mixed plastics waste. A contribution was made to a process improvement for creating products from street waste in Kenya. Mixing different recycled HIPS grades gave no significant changes in melt flow index and impact strength. One mixture showed a significant reduction in tensile strength. In PP-LDPE mixtures there was an indication of improved mechanical impact resistance of the PP by addition of LDPE in several percentages. In Kenya a separation process utilizing simple equipment was intro-duced which enabled constant and improved quality of final products made of plastic waste.

1 INTRODUCTION

Thermoplastics can be recycled in many ways, each with different and partly unique opportunities and challenges. There is no simple guideline for estab-lishing which principle of recycling makes most sense in a certain context, or whether it makes sense at all. The work reported here focuses on recycling in situations where plastics waste is abundant but sophisticated material separation facilities are not (yet) available and exact material grades cannot be traced anymore, i.e., representative for a considera-ble part of the world. This activity is a contribution to creating a set of criteria beyond technological considerations only, for supporting decision making related to recycling and re-use as part of over-all product life cycle optimization.

Different recycled high-impact polystyrenes (HIPSs) from television cabinets have been mechan-ically tested, as well as mixtures of these recyclates. Some components of HIPS, ABS and similar materi-als are Icnown to be miscible to a certain extent.

Mixtures of a low-density polyethylene (LDPE) and a polypropylene (PP) were tested as well, repre-senting a large portion of non-miscible mixed plas-tics waste world-wide.

A contribution was made to a separation process for plastics from street waste in Kenya, enabling

bet-ter quality of the mabet-terials for products manufac-tured locally. The design of a toy car was improved as well, as part of another activity not reported here.

2 BACKGROUND

The main benefits of plastics recycling are reduced depletion of non-renewable resources and reduction of world-wide waste. Traditional thermo-mechanical recycling, often termed "mechanical recycling", causes reduction of mechanical properties for most thermoplastics because during product manufactur-ing it involves re-granulation as well as re-meltmanufactur-ing. The latter exposes thermoplastics melts to a high de-gree of shearing deformation and associated flow-induced stresses, particularly during injection mold-ing, causing reduction of average macromolecular masses and broader molecular mass distributions. In addition, some thermoplastics will experience oxida-tion or hydrolysis during melt processing. These phenomena result in reduced mechanical properties, particularly impact strengths, and a worse balance between processability and properties.

Down-cycled materials may nevertheless be suit-ed for certain useful applications. Developing appli-cations for down-cycled plastics may be considered as an intermediate solution for the time being, as

long as more effective use of resources has only been implemented on a small scale world-wide. Increased application of recycled thermoplastics is generally preferable over sending used plastics products to landfills or accumulation of uncollected waste in some parts of the world (Figure 1).

Figure 1. Street waste in Kenya.

In an ideal situation all resources would be fully and perpetually cycled in matching time scales such that no shortage of any resource would be created. In such a scenario discarded products and waste mate-rials will always contribute to new resources. New materials from virgin resources would then possibly be needed only to match any increases in world pop-ulation and in affluence. Plastics and other materials from renewable resources will be helpful in the f u -ture for safeguarding new materials, once associated infrastructures will have reached sufficient scale. The need to re-use or recycle renewably-sourced ma-terials from discarded products, however, is basically the same as for non-renewable materials.

The cun-ent situation is obviously much different from an ideal scenario. Synthetic polymeric materi-als are apparently most challenging to recycle, be-cause of the macromolecular structures as described above.

3 METHOD

3.1 High-impact polystyrenes

A l l the TV cabinets used in this study were made of high-impact Polystyrene (HIPS). HIPS is a polysty-rene with a rubbeiy additive of rubber, butadiene, to increase the toughness and impact strength. Two ma-terial mixtures have been made: (1) random TVs produced by JVC (AV-28R475K, unloiown year), Ferguson (FTV28FW1, 2004), Dual (DRF2810, un-known year), Toshiba(32zl3B, 1999) and Mitsubishi (CT-2110B, unknown year), and (2) random TVs

produced by Toshiba produced between 1994 and 1999. To verify whether the TVs had been made from HIPS, all were checked by Fourier-Transform Infra-Red (FTIR) spectroscopy. The spectra were an-alyzed and automatically compared with polymers from a database.

The TVs have been granulated (ZERMA) and ex-truded (PRISM TSE 16TC) into small particles. The two mixtures were made of 100 grams from each TV from both groups and blended homogeneously dur-ing extrusion and injection molddur-ing of the test piec-es. A small part of the recyclate was reserved for the melt flow index testing. The rest was injection molded (Boy 50 T) into five tensile test and five im-pact test specimens. Five specimens of each TV and five specimens of the blends have been tested be-cause of the limited amount of plastic from one TV. The tests were done at Axion Polymers in Man-chester (UK). The Melt Flow Index (MFI) was tested according to ISO 1133. Tests according ISO 527 were conducted to determine tensile strength and strain at break. Notched IZOD sideways impact strength tests were conducted based on ISO 180, with the notch facing the impact hammer.

Due to the small sample size and the explorative character of this study, only descriptive statistics were applied by way of confidence intervals.

3.2 Polyethylene-polypropylene mixtures

Mixtures of an LDPE, Purell grade 2410T, and a PP, DUCOR (Domolen) grade 2370P, were mechanical-ly combined into ten 500 gram batches consisting of mixtures containing 5, 10, 15, 50% by weight of LDPE in PP.

Flexural impact test specimens in accordance with ISO 180 were injection molded (Boy 225) and notched (Notchvis) based on ISO 180, with the notch facing the impact hammer. The non-perfect mixing method was responsible for a clearly course hetero-geneous material structure, partly because each ma-terial has its own ideal molding settings that could not be met by a single setting for the mixture.

A minimum of seven specimens were tested in sideways impact with a Zwick 1 harmnered impact tester, using a hammer suitable for 2.75 J impact en-ergy maximum. During impact testing it was found that most specimens did not break with the standard 0.25 mm notch depth. From the results of some tri-als, reducing the number of specimens available, it was found that 0.75 mm notch depth was more ap-propriate and the remaining specimens have been given such a notch for testing. A l l results reported here are from the specimens with 0.75 mm notch depths.

Many results series had broad distributions. This was most likely caused by the course material struc-ture, with interfacial regions between the two mate-rials acting as weaker spots. For additional

infor-mation a Grubbs' Test was nm to create an addition-al data set without outliers. This can be useful, for example, to separate one effect when more than one dominant effect occurs with different combinations of effects for different specimens [Harris 2010]. Two Grubbs' Test runs were needed to eliminate all outli-ers.

3.3 Kenya project

A project was done with Muima Toy Industries in Kahawa, Kenya , focusing on separation of different plastics. Muima Toy Industries is a plastics repro-cessing company, an initiative of Wilson Mzungu. Manufacturing plastic toy cars from plastic waste started in 2003. Nine employees in the factory pro-duce 200 toy cars a day. These cars are being sold on Kenyan markets for an average price of € 0.75 each. Wilson buys an average of 100 kg of plastic waste daily from collectors who collect large amounts of plastic from different people, some of whom are homeless.

Current sorting consists of removing PET bottles, which cannot be processed with the equipment, black-colored plastics, which are of lower quality as experience has learned, and non-plastics. The next step is crushing the remaining mixture with a shred-der. This shredder is hand-made and consists of rela-tively soft metal: the knives have to be sharpened every day. The holes in the sieve of the shredder have 16 mm diameter, just small enough to work properly with the available hand-made extruders. The next step is hand-washing the plastics in a tank filled with water and soap.

With a small plastic sieve the particles are re-moved from the water. The employees already noted that some plastic pieces sink in the water and they throw them away, because they do not work with the machines. The washed plastics particles are then sun-dried. The dried plastic is collected in a tank from which it is fed into two sequentially operating extruders.

The extruders rotate very fast and most of the plastic softens by friction in the first extruder. The partly softened m i x l ^ e is then fed into the second extruder. A closed mold is mounted at the outlet opening of the second extruder. During clamping the mold, the rotation wheel is disconnected and no plastic will flow through the opening. The wheel is then reconnected and the mold will be filled with plastic. The molds consist of a large number of com-ponents. The walls of the molds are fixed by hinges, making it possible to easily remove the product. Be-cause of the many hinges and non-perfect fitting of the mold components there is a lot of overflow on the product. The overflow is removed manually, us-ing knives. The final step is assemblus-ing the toy car. Bicycle spokes are used to fix the wheels on the car.

A low-tech separation process was designed as schematically shown in Figure 2. Polyesters (PET), nylons, PVC and non-plastics will be separated first. The products in which these materials occur can eas-ily be recognized by their appearance. The next steps are based on density differences. The fluids which are used are (1) water and (2) oil and watery sah so-lutions.

All kind of plastics

J 3 -Seperation by eye PET Nylon LOPE PVC No plastic

PE, PP, PS, ABS, other

Water

Floa

t

Sin

k

HDPE, PP PS, ABS, other

- a Oil Floa t Sin k

Water with dissolved salt

PP HDPE PS ABS, other

Figure 2. Design of low-tech plastics separation process for sti"eet waste.

PP and PE are the only two common plastics tliat float on water. This floating fraction will go to the oil bath for fiulher separation of PPs and PEs. Sun-flower oil is a low-cost fluid with density 910 kg/m , i.e., between PPs and PEs. There is only an overlap-ping ai*ea of density of PP and LDPE, which was al-ready separated from the waste stream and does not cause difficulties. The sinldng fraction will go to a sah solution. The density of PS is slightly less than the one of ABS, just enough to achieve separation when the fluid's density is 1060 kg/m^ by adding the appropriate amount of salt.

The separation principle was tested in advance with some mixed plastics collected from waste bins. By checking with FTIR spectroscopy it was found that the separation processes by water and oil con-formed for at least 80% to the expectations. The sep-aration with salt solutions did not work properly, probably because the composition was not correct. It was decided to test this again in Kenya, using plas-tics plates with known densities for calibration. The same calibration was to be used for separation with all other fluids as well.

4 RESULTS AND DISCUSSION 4.1 High-impact polystyrenes

The FTIR spectra of each T V were compared to the spectra of HIPS. For each material the confidence of agreement was 90%. The tensile strengths, strains at break and notched IZOD impact strengths are shown below with their 95% confidence intervals

For mixture 1 and its components the melt flow indices are shown in Figure 3 and the mechanical properties in Figure 4. 20 15 g/lOmin 10 5 0 .of- .c^ .c?' .c? ^ J> ^&

O = melt flow index

2 0 15

g/lOmin 10 4 -5 0

O = melt flow index

Figure 3. Melt flow indices of mixture 1 ("blend") and its com-ponents. 80 -% 60 40 M p a 20 k J / m 2 0

1

4-

i

• = Strain at break, 0 = tensile strength at break,

^ = impact strength

Figure 4. Mechanical properties of mixture 1 ("blend") and its components.

For mixture 2 and its components the melt flow indices are shown in Figure 5 and the mechanical properties in Figure 6. Based on these results the on-ly significant effect with 95% confidence is a nega-tive effect on the tensile stress at break for mixture 2. The comparisons are based on the average of five times five samples for the individual TVs compared to a single set of five samples for the mixtures. The values for the mixtures, therefore, are less reliable because of the limited sample sizes, causing larger confidence intervals. Nevertheless, the tensile stress at break of mixture 2 showed a significant reduction when compared to the individual samples.

Figure 5. Melt flow indices of mixture 2 ("blend") and its com-ponents. 80 60 % 40 M P a 20 0

i

• = strain at break, 0 = tensile strength at break,

A = impact strength

Figure 6. Mechanical properties of mixture 2 ("blend") and its components.

4.2 Polyethylene-polypropylene mixtures

Figure 7 shows the average impact strengths of the ten mixtures of LDPE in PP with their 95% confi-dence intervals. Figure 8 shows the same results af-ter removal of the outliers by the two Grubbs' test rims.

Considering all impact test results (Figure 7), the only differences found with 95% confidence are (1) between the mixtures containing 5% and 15% LDPE, and (2) between the mixtures containing 5% and 35% LDPE. Considering the impact test results after the removal of outliers (Figure 8), the following mixtures can be considered different from the 5% LDPE mixture but not from each other with 95% confidence: (1) 15%, (2) 30%, (3) 35%, and (4) 50% LDPE. From Figures 7 and 8 it can be concluded that removing the outliers yields lower average im-pact strengths for the mixtures with 10%, 30%, 35% and 40% LDPE, whereas for the mixture with 50% LDPE the value increases.

20 I " la i-15 +¬ 14 i-12 -1-impact strength 10 30 AO % PE in PP

Figure 7. Notched impact strengths as a function of the per-centage of LDPE in PP.

12

impact strength 10

Figure 8. Notched impact strengths as a function of the per-centage of LDPE in PP after removal of outliers.

4.3 Kenya project

Tlie separation was optimized after some trials. After visual sepai'ation the first step was separating based on density by water. This step could be combined with the washing step and could be optimized. Be-fore, tanlcs with a small surface were used. By in-creasing the surface for the new water box, there was more space for the particles to sink or float without sticking together. The new water box is also more user-friendly. The water can be removed easily by opening a tap. Further optimization was achieved by decreasing the particle sizes. Some particles that should sink will float when the surface tension is too high. By decreasing the particle sizes, the surface tension will decrease and the effectiveness of the separation process will increase. This was achieved by decreasing the holes' diameters from 16 to 6 mm in the new sieve, before tlie sluedder. The surface tension was reduced further by the soap used during the washing step.

The next step was the separation by oil. The i n -tended high capacity of separating, however, re-quired a large amount of oil. It took also a long time to reach equilibrium and the oil became polluted rap-idly. The oil was difficult to remove from the plastic.

After some time, however, it was found that there are very few different types of plastics. A reason for this is that most people in Kenya buy their food on the market without packagmg. Only products like shampoo, body lotion and yoghurt are packed. The most common products were tested by the designed separation process. Separation was done just once for each collection of products. Except PET bottles most plastics products are PP and PE. It was decided not to use the small fraction of other plastics.

A product catalog was made including a descrip-tion of floating and sinking behavior in a single doc-ument. Employees can now separate plastics by compai'ing the products with the catalogs. This pro-cess is very simple and effective, and easy to learn. The separation by type of plastic can be combined with the separation by color, which was: already common practice, because both are done by hand.

5 CONCLUDE^JG REMARICS

When designing a confirmatory study, such as with the HIPS mixtures, it is recommended to use the same sample size for the mixtures and for the com-bination of individual components, to minimize dif-ferences in confidence of results between series. It is also recommended to fully flush extruders and injec-tion molding equipment, which was not possible in these experiments due to limited amounts available of each material. Having a traceable and known composition of the individual types of HIPS would be helpful too.

Results of an earlier study show the effects of blending HIPS with ABS [Arnold et al. 2010]. The mechanical properties of the blend were significantly worse than those of the individual polymers. This was a blend of two different classes of plastics. The effect on the properties of mixing more than two dif-ferent types of polymers of the same class (HIPS) is not described. A priori, much smaller differences be-tween the properties of various HIPS plastics are ex-pected when compared to different classes, because the fractions of the immiscible components are very small, compared with the fraction of HIPS. In anoth-er study (Osswald & Dattnanoth-er 2005), using mixed plastics from used electronic housings containing at least 70% ABS, the mechanical behavior of repro-cessed material was foimd to be fairly repeatable, except for impact properties, within each group con-taining 70%, 80% end 90% ABS, respectively. In this study an improved compounding process was developed, giving more consistent and better proper-ties of recyclates. The prediction was that com-pounding rubber modifiers with the regranulate should enhance impact strengths.

Addition of an LDPE to a PP, such as may be found in plastics packaging waste, does not neces-sarily decrease all mechanical properties, as was

found in this study for notched impact strengths, one of the more critically sensitive properties with re-peated melt reprocessing. Of course the application may need to be different from the original applica-tions of the virgin materials. It is unlikely, for exam-ple, that an LDPE mixed with PP, such as in the mixtures studied, can be applied again for plastic bags because of reduced melt elasticity.

To conclude, the tests reported here were explora-tory and intentionally low-tech, with the purpose of exploring useful application of mixed plastics waste at locations cun-ently lacking sophisticated infra-structures for collecting discarded plastics products. Additional experiments with larger batch sizes and more specimens in the future should give more pre-cise infoimation on evolution of properties in mate-rials containing mixed plastics obtained by different sorting and reprocessing techniques, as well as more understanding by allowing studies of material struc-tures at visible scale and at micrometer scale

In the Kenya project it was shown that the largest portion of current street waste, consisting of poly ole-fins, can be easily and reproducibly separated mto comparatively pure and clean materials suitable for manufacturing toys. Plastic separation and recycling of the largest portion of street waste is already possi-ble with veiy simple methods and equipment.

6 REFERENCES

Arnold, J.C., Watson, T., Alston, S., Carnie, M., & Glover, C. 2010. The use of FTIR mapping to assess phase distribution in mixed and recycled WEE plastics. Polymer Testing, Vol-ume 29, Issue: 4. Elsevier, 459-470.

Harris, D.C. 2010. Quantitative Chemical Analysis. Freeman, (New York). ISBN-10 1-4292-1815-0.

Osswald, T.A., & Dattner, M.W. 2005. Processing Solutions and Market Applications for Mixed ABS. ANTEC 2005 -Proceedings of the 63"^ Annual Technical Conference & Exhibition, Boston, M A (USA), May 1-5. 3216-3220.

7 ACKNOWLEDGEMENTS

The following persons cooperated and accommodat-ed the first author for preparing the Kenya project and investigating the HIPS recyclates from TV cabi-nets, respectively: Mr. Rien Driessen, director of Hermion BV, Waalwijk (NL), and Mr. Keith Free-gard at Axion Polymers, Manchester (UK).