Optimising the use of machine readable inks for food packaging sorting

Transcription

1 Final report Optimising the use of machine readable inks for food packaging sorting A report on the technical and commercial viability of an identification technique based on fluorescent pigments that can be applied to labels and packaging to enable the automatic separation of a wide range of target materials such as PET, HDPE and PP for the sorting of food contact packaging to enable closed loop recycling. Project code: IMT Date: 19 th September 2014 Research date: 12/02/14 19/09/14

2 WRAP s vision is a world where resources are used sustainably. We work with businesses, individuals and communities to help them reap the benefits of reducing waste, developing sustainable products and using resources in an efficient way. Find out more at Document reference: WRAP Project: IMT Report prepared by: NEXTEK Limited; Edward Kosior, Kelvin Davies, Dr Martin Kay, Dr Jon Mitchell, Dr Rafi Ahmad, Edwin Billiet and Brunel University Prof. Jack Silver Front cover photography: Fluorescent tagged packaging and UV-LED array for sorting trials While we have tried to make sure this report is accurate, we cannot accept responsibility or be held legally responsible for any loss or damage arising out of or in connection with this information being inaccurate, incomplete or misleading. This material is copyrighted. You can copy it free of charge as long as the material is accurate and not used in a misleading context. You must identify the source of the material and acknowledge our copyright. You must not use material to endorse or suggest we have endorsed a commercial product or service. For more details please see our terms and conditions on our website at

3 Executive summary A major challenge in producing recycled plastics from post-consumer recycling feedstock is to sort the material into high purity mono-material streams since small quantities, often as little as 0.1%, of a non-compatible plastic can significantly reduce the quality of the final product. In order to keep costs low while processing high volumes, sorting is typically realised with automatic sorting machines using Near Infrared detection. One of the initial challenges common to bottle-to-bottle food-grade recycling technologies used for PET and HDPE is to ensure that the input material is made from plastics that comply with the EU Directive 10/2011. This requires the material as being approved for food applications and also that it has been used for food applications in its last use to ensure that unexpected contamination is excluded. The development of automatic sorting technology that can identify packaging as prior food use would greatly enhance the purity of the input materials and enable the sorted materials such as PET, HDPE and PP to achieve the high standards required for food contact applications. Polypropylene (PP) from packaging is a significant polymer in the mixed plastics waste stream and recycling it back into packaging would enable considerable carbon savings to be realised and generate significant revenues. Recent estimates suggest that 143,000 tonnes of the total PP packaging 1 is used for food-grade applications in the UK. One of the remaining barriers to closing the loop on the recycling of PP food packaging back to food grade applications is the absence of an automated method for sorting mixed PP packaging waste into a separate stream that consists of at least 99% PP previously used for food applications, from other non-food PP packaging that could be present at levels as high as 50% in the input stream. Machine readable inks (including fluorescent pigments) have shown great potential 2 for the identification and separation of plastic packaging. Unlike existing NIR sorting practices, these technologies are not polymer specific and could be applied to targeted streams like food contact plastic packaging, using commercial labelling and decoration methods and sorted using MRF infrastructure with only minor modifications. This report demonstrates the technical and commercial viability of an identification technique which is based on fluorescent pigments that can be applied to labels and packaging as a machine readable ink (MRI) to enable the automatic separation of a wide range of target materials to allow recycling back into food applications. Different grades of the same polymer type may also be able to be specifically marked to allow further separation based on physical properties adding further value to the recycled plastic. A review of existing intellectual property suggested there is significant activity in the development of detectable compounds such as fluorescent inks and other marker systems, however the focus is mainly on security and anti-counterfeiting applications. Of the relevant patents and previous publications, most notable are those by Dr Rafi Ahmad and Edwin Billiet at Cranfield University, both participants in this study, who initially developed the use of fluorescent compounds for sorting recycled packaging in The search for commercial dyes or pigments with adequate emission strength at the required wavelengths, provided a number of candidates that were evaluated with a laboratory spectrophotometer prior to large scale testing. The selected pigment was dispersed in a solvent lacquer system and a controlled coating of the lacquer was applied to a PP or PET 1 UK market compositional data of polypropylene packaging. WRAP July S Rafi Ahmad, A new technology for automatic identification and sorting of plastics for recycling, Environmental Technology, 25, , Optimising the use of machine readable inks for food packaging sorting 1

4 substrate for testing at both laboratory and commercial sorting trials. Pigment concentrations were adjusted to optimise the system and determine limits of performance of the pigment and instrumentation. An audit of PP packaging waste was conducted to estimate the expected level of sorting performance if existing label formats used machine readable inks. Direct print, in-mould, pressure sensitive and glued labels, can be made from different materials and respond differently through the supply and recycling logistics chain. It was determined that if MRI s were used on existing label formats approximately 55% of the PP packaging would be able to be effectively identified and sorted. The remaining samples were either not labelled or the label had become detached or otherwise destroyed. This is an important consideration for future developments and highlights that alternative or additional labelling or marking may be required to improve yields and recoveries. It is expected that the fluorescent compound can be added to existing lacquer and ink products and applied by conventional means to an existing label or print design, or alternatively as a separate and purpose designed modification for improved sorting recovery. It was determined that for a commercial sorting system, the existing halogen lighting used for optical/nir sorting was not suitable and a UV-LED array was purpose-built for the commercial sorting trial. The UV-LED array increased sensitivity so pigment concentrations down to 3,000ppm could be reliably detected under commercial sorting conditions of 3m/s and 1.9 tonne per hour. MRI labelled lid without UV and with UV exposure. MRI Labelled pots under UV exposure The modification of a one metre wide automated sorting unit in an R&D MRF in Germany enabled a range of packaging samples and label types to be evaluated in a sorting trial under realistic commercial sorting conditions. MRI samples were positively sorted from mixed PP packaging waste with a yield of 74% and a purity of 93% in a single pass. There were some minor handling issues that related to samples rolling on the sorting belt due to lack of flatness and with the size of the labels on larger packs. If these samples are removed from the calculation, it is estimated that yield would increase to 93% and purity increases to 94%. The material cost for the additional labelling was estimated at between to pence per label as a factor of pigment costs, concentrations and label size. For a non-optimised system, the additional pigment costs are estimated at 0.4% to 15% of the 0.75pence price of an applied label. The modification of existing sorting NIR/optical equipment is relatively straight forward and practical. The cost of the modification cannot be accurately estimated at this time as it will vary depending of state of existing equipment, it is anticipated that it will be a relatively small capital cost (of the order of 10 to 20% of the cost of the full unit). In addition, given the lower power usage and long life of LED s, operational costs are also expected to be manageable. Optimising the use of machine readable inks for food packaging sorting 2

5 Based on the audit of current PP waste, results indicated 55% detectability of existing packages. Together with potential yields of 98% from the sorting trials, it can be estimated that of the 143,000 tonnes of PP food packaging in the market each year, theoretically 77,077 tonnes could be recovered and processed back to food applications each year in the UK if all of the PP was captured by the recycling infrastructure and sorted at MRF s. In practice if a recycling rate of 50% was assumed, recovery levels would reach 38,538 tonnes per year. This would increase dramatically if label design and application was modified according to protocols, to better accommodate identification and sorting requirements. Also with higher value applications for the recycled PP materials, collection and recovery rates may also increase. Recommendations: It is recommended that further work be completed to optimise the performance and cost effectiveness of a label-based fluorescent machine readable ink technology. Further development is required to optimise the active fluorescent compounds being used. Ideally lower cost and/or higher performance materials could be identified. Evaluate stability of the compound in the supply chain to ensure that it remains effective but does not otherwise have any negative impact throughout the supply chain, through recovery and final sorting. Assess the ability of the MRI labels and compounds to be effectively removed during the cleaning and decontamination process, particularly for direct print and in-mould labels that tend to remain with the packaging. Conduct a large scale trial with brand owners, converters, retailers, and recyclers to identify any limitations for full commercialisation once preliminaries are optimised. Conduct a full feasibility (business case) study on the costs and benefits for the supply chain based on requirements and outcomes of the large scale trial. Identify and evaluate other target materials that could be better recovered for closed loop recycling by using MRI labels. As an example black plastic packages with a clear MRI label were successfully identified by this technique. Additionally different packages that require positive or negative differentiation could be sorted after a primary polymer sort such as multilayer bottles and trays, milk bottles, bio-plastics or bottles used for household chemicals and garden products that contain harmful contents. Develop a comprehensive standard set of criteria for sorting using MRI technology so that sorting into generic food grade types including PET, HDPE and PP and other target materials can be conducted uniformly across the UK and European packaging and recycling sector. Optimising the use of machine readable inks for food packaging sorting 3

6 Contents 1.0 Introduction Objectives Patent and Literature Review Relevant Patents and Literature summary Machine readable inks development Sourcing of pigments Bench scale test equipment and preparation of label samples Summary of bench scale testing results CCL Labels Supplier Supplier Detectability of print and labels on recycled packaging Sorting trials with commercial equipment Initial Static Scan Commercial simulation dynamic scan Commercial sorting trial Sample preparation Trial 1: Impact of package type on machine readability Trial 2: Impact of pigment concentration on machine readability Trial 3: Impact of label size on machine readability Trial 4: Impact of pack opacity on machine readability Trial 5: Sorting of samples mixed with PP waste packaging Sorting trial conclusions Evaluation of cost implications Estimation of incremental pigment costs and requirements Sorting equipment modification Sorting Protocols for Fluorescent markers Standardisation protocol for machine readable ink marker technology Conclusions Recommendations for further work Appendix 1 Patent and Literature search Appendix 2 Audit PP articles in the waste stream Appendix 3: Cost evaluation summary Optimising the use of machine readable inks for food packaging sorting 4

7 Figures Figure 1: 70µm PET sheet with wet draw down of lacquer sample together with K1.5 bar.. 10 Figure 2: Fluorescence spectra of different pigments prepared in two lacquer systems Figure 3: Concentration dependence or SR-1 dye Figure 4: Impact of outdoor exposure on emission intensity of selected dyes Figure 5: Intensity of response from Supplier-3 in two lacquer systems Figure 6: Ranking categories whether readable inks can be used in detectable labels Figure 7: Examples of PP packaging found in each of the detection groups Figure 8: Effect of light source distance and ink concentration Figure 9: UV LED array over conveyor (left) and excitation of label samples (right) Figure 10: The experimental setup at the Tomra sorting facility in Germany Figure 11: PP containers used in the sorting trials Figure 12: Unlabelled, 0ppm control and 12,500ppm pots used in Trial Figure 13: Fluorescent inherent in packaging and small label example Figure 14: Different concentrations of MRI on label Figure 15: Dots and stripes labels used to show the effect within the sorting system Figure 16: Thermoformed (left) and Injection (right) pots with MRI labels on their base Figure 17: Samples used in the Trial 5 sorting Figure 18: PP waste and marked containers used in Trial Figure 19: The spectra of two tracers A and B shown individually and in combination Figure 20: How four fluorescent markers could be used to code package classification Tables Table 1: Companies investigated for the provision of pigments Table 2: Fluorescence of SR-1, SL-1 and SC-1 at different concentrations Table 3: Summary of audits carried out at two UK PRFs (% weight of total) Table 4: Classification of PTT samples received and prepared with machine readable labels Table 5: Dimension details of machine readable samples for sorting trials Table 6: Impact of pot or lid type and classification on machine readability Table 7: Sorting efficiency machine readable labels with reduced concentration of pigment Table 8: The effect of label design with florescent stripes and dots Table 9: The effect of container opacity on machine readability Table 10: Composition of mixed material for sorting Table 11: First and Recovery pass sorting efficiency of mixed material Table 12: Overall sort efficiency and yield after addition of recovery pass Optimising the use of machine readable inks for food packaging sorting 5

8 Table 13: Estimation of pigment costs and volumes Glossary MRI LED NIR UV Fluorescent Phosphorescent Food Contact Polymer or Packaging - MRF PRF PTT PE PET PP ppm ms Machine Readable Ink Light Emitting Diode Near Infrared Ultra Violet A substance that can emit longer wavelength radiation(visible light) while being excited by shorter wavelength radiation (non-visible UV) Similar to fluorescent, however the substance keeps emitting the visible light after excitation stops. That which has been used in contact with food or has been tested and approved for use in contact with foods in compliance with the requirements of EU Regulation 10/2011 Material Recycling Facility Plastics Recycling Facility Pots, Trays Tubs Polyethylene Polyethylene Terephthalate Polypropylene Parts per million milliseconds µm micrometre nm nanometre Acknowledgements Professor Jack Silver (Brunel University); Dr Rafi Ahmad and Edwin Billiet (Cranfield University, retired); Paul Gibb and Shaun Nash (CCL Labels); Ralph Uepping and Michael Mayer (Tomra); Terry Widdison and Peter Shanely (RPC). Optimising the use of machine readable inks for food packaging sorting 6

9 1.0 Introduction A major challenge in producing recycled plastics from post consumer recycling feedstock is to sort the material into high purity mono-material streams since small quantities, often as little as 0.1%, of a non-compatible plastic can significantly reduce the quality of the final product. In order to keep costs low while processing high volumes, sorting is typically realised with automatic sorting machines using Near Infrared (NIR) detection. Efficient collection and sorting enables high quality recycling. By increasing the value of waste, the recycling sector in turn makes collection schemes more economically feasible. One way to enhance the value of recycled plastic is via technologies that turn recovered plastic into food-grade polymer by removing contamination. This technology has been proven for clear polyethylene terephthalate (PET) bottles and high density polyethylene (HDPE) milk bottles and recently polypropylene (PP). One of the initial challenges common to all of these material recycling technologies is to ensure that the input material is made from plastics that comply with the EU Directive 10/2011 as being approved for food contact and also that the material has been used for food applications in its last use to ensure that unexpected contamination is excluded. The development of automatic sorting technology that can identify packaging as prior food applications would greatly enhance the purity of the input materials and enable the sorted materials such as PET, HDPE and PP to achieve the high standards required for food applications. Previous studies have shown that when mixed post consumer plastics are recycled, as much as 60% of the remaining plastic is PP after the PET and HDPE was removed. Significant environmental and economic benefits could be produced if a similar food grade application that is currently available to PET and HDPE were also possible for PP. A decontamination process for PP was developed in an earlier WRAP projects 34, however this required separation of PP packaging that was used for food applications from PP packaging that was used for household chemicals, personal care etc. Grades of virgin PP used for these non-food applications are not certified for food applications, and when recycled, cannot be used for food applications. It is necessary then to further sort food use and non-food use PP packaging to close the loop on the food grade PP packaging fraction. Standard NIR and other commercial sorting equipment cannot distinguish between food and non-food PP materials so a new sorting technology is required. A previous WRAP 5 project evaluated techniques such as polymer additives, surface markings such as diffraction gratings and fluorescent markers. This project further investigated the use of machine readable ink technologies such as fluorescent markers that can be applied to packaging or packaging labels so that existing NIR-optical sorting equipment can be utilised with minimal modification via the addition of specific lighting to make the markers visible to the detectors. The machine readable inks should be invisible to the naked eye under normal lighting so not to affect branding and aesthetics, but be detectable with existing MRF/PRF infrastructure at commercial sorting speeds and sorted at high efficiencies. The positive results show that the use of a suitable dye or pigment applied as a surface varnish on an existing PP package or label used for a food application may provide an adequate means of identification and separation. 3 MDP-039 Development of a food-grade recycling process for post-consumer polypropylene. 4 IMT Food grade decontamination trials of household PP waste. 5 IMT Sorting plastics for food use Optimising the use of machine readable inks for food packaging sorting 7

10 2.0 Objectives The objectives of this project were to investigate and develop the use of machine readable inks that could be applied to the broadest range of materials and packaging types to assist identification of different types of polymers during sorting and recovery for recycling. A specific application example is to be able to mark food grade approved plastic packaging articles so that they are able to be individually identified and separated into a different stream. Existing automated NIR sorting systems can identify many commodity polymers, however some polymer variants like food grade and non-food grade PP or HDPE, as well as bottle grade and thermoforming grades of PET are not able to be automatically separated by existing processes. By applying selected markers such as a fluorescent inks or coatings to the packaging during its manufacture, specific grades and applications of polymer types can be tagged ready for identification and separation by automated sorting units. These spectral markers are not visible to the naked eye with normal lighting and so do not detract from the packages aesthetics, but are excited under specific light sources to emit light at wavelengths that automated sorting units can identify on individual packages and sort them into separate recycling streams. An objective of the project was to utilise inks and dyes that are available commercially and apply them to the packages using standard printing and decoration processes. By selecting appropriate compounds, the need to modify and adjust existing sorting equipment may also be minimised. Successful commercialisation of the technology will depend on minimising disruption and the cost of use to the brand owners, packaging converters and recyclers. 3.0 Patent and Literature Review A patent and literature search was undertaken to identify the current status of intellectual property relating to the use of compounds that can be dispersed in inks or coatings to act as taggants or markers that when applied to packaging are machine readable for the purpose of separation. Of specific interest is any intellectual property that could limit or restrict the use of machine readable inks for the separation of packaging waste. The patent search encompassed the following patent search engine resources: 1. Espacenet (UK and European Patent Offices) - covering 80 million patent documents 2. US Patent & Trademark Office (USPTO) covering two million patents dating back 26 years 3. Supplier-7 Patent Search Network (GPSN) that includes Chinese patent documents from 1985 to World Intellectual Property Organisation covering 35 million patent documents. The search results were focussed on combinations of the key words; identification, marking, tracer, sorting, packaging, waste, plastic, fluorescent, dyes, ink. An examination of each of the first 500 patents extending from 2007 to 2014 reveals a very small number of recent and related patents that are described in section 3.1 of this report. Publications and literature were more difficult to search in a comprehensively manner; however, it was clear that not a great amount of information on the use of markers for the application of sorting packaging waste has been presented. Optimising the use of machine readable inks for food packaging sorting 8

11 3.1 Relevant Patents and Literature summary Work was conducted in the 1990 s as the technologies for automated sorting were being developed. Attempts to address particular limitations were made through modification of the detectors themselves and also of the packaging articles being sorted. Once successful NIRbased systems became established commercially in the early 2000 s, the requirement to modify the packaging to affect a successful sort was greatly reduced and interest and development in this area was reduced. Of specific interest was the entry of this work into the journal of Environmental Technology Vol 25 by Dr Rafi Ahmad titled A new technology for automatic identification and sorting of plastics for recycling. This article clearly describes in detail the concepts of this project, including the use of a combination of fluorescent dyes in a range of polymer materials and methods to identify and sort the packaging at commercial rates. More recently an article in Green and Sustainable Chemistry reported the possibility to identify polymers based on their auto fluorescence and doping with fluorescent dyes. The article highlighted the possible use of fluorescent dyes to identify special batches of a basic polymeric type Machine readable inks development Readable inks development involved sourcing commercially available inks and pigments, developing methods to apply to a label and design and evaluation of lighting and detection options to enable practical automated sorting using existing techniques. 4.1 Sourcing of pigments The criteria for the selection are as follows: The pigment needs to exhibit fluorescence within the red to near-infrared wavelength band (600nm-740 nm) of the spectrum. The pigment needs to have relatively high fluorescence quantum yields. The quenching of fluorescence should not reduce the signal-to-noise ratio (S/N > 2) The pigment should have low toxicity for disposal prior to recycling. The pigment should be non-reactive with the host materials. The pigment should be stable to provide adequate shelf life. Of the ten companies who were contacted for provision of pigments, three companies were able to supply pigments there were able to meet the project requirements. 6 Langhals, H., Zgela, D. and Schlücker, T. (2014) High Performance Recycling of Polymers by Means of Their Fluorescence Lifetimes. Green and Sustainable Chemistry, 4, Optimising the use of machine readable inks for food packaging sorting 9

12 Table 1: Companies investigated for the provision of pigments Supplier Pigments Comment CCL labels ER-1, IY-1 and SR-1 Organic pigments Supplier-2 SL-1 and SC-1 Organic pigments Supplier-3 NE-1, NQ-1, NQ-2, FU-1 Inorganic pigments Supplier-4 Supplier-5 Supplier-6 Supplier-7 Supplier-8 Supplier-9 Supplier-10 Unable to meet requirements Unable to meet requirements Unable to meet requirements Unable to meet requirements Laser excitation-too expensive No response No response 4.2 Bench scale test equipment and preparation of label samples An Ocean Optics Spectrometer operating within the spectral range of 350nm to 1000nm was determined to meet the performance criteria was procured and commissioned. This unit was used in combination with UV-LED lighting to rapidly evaluate different compounds, application techniques and other test variables as described below for each supplier. Use of Myer bars and a rubber pad allowed a defined thickness of lacquer to be deposited onto a PET sheet that was then cut to size and used to prepare test specimens for spectrometer testing and the preparation of labels for testing with automatic sorting. Ink and pigment concentrations were varied in samples that were drawn-down as a 10micron wet film onto 70µm PET sheet using a K1.5 bar then allowed to air dry (Figure 1). Two clear lacquers supplied by CCL labels based on nitrocellulose and vinyl were used in initial testing to compare performance of the pigment and ink samples. Figure 1: 70µm PET sheet with wet draw down of lacquer sample together with K1.5 bar Optimising the use of machine readable inks for food packaging sorting 10

13 4.3 Summary of bench scale testing results CCL Label CCL Label provided the initial pigment samples and these were used in the first round of bench scale testing. Samples of ER-1 and IY-1 pigments at 50,000ppm were mixed with both nitrocellulose and vinyl lacquers, drawn down onto PET film and evaluated. Typical spectra obtained by this technique from a number of pigment suppliers is shown with two samples showing suitable emission intensity at 621nm (Figure 2). Figure 2: Fluorescence spectra of different pigments prepared in two lacquer systems Supplier-1 The following conclusions were made: At the high loading of pigment (50,000 ppm) and a long exposure time of 100 milliseconds, the signals for the ER-1 and IY-1 pigments, were strong and readily detectable. The signals are about 1.8 times stronger in the nitrocellulose compared to that in vinyl lacquer. Although the detectable signal of IY-1 pigment was strong, it had significant overlap with the signal from the lacquer and PET substrate and was rejected as a potential candidate for the present application. It is was estimated that the ER-1 should be detectable at concentrations below 10,000ppm and at exposure times lower than 10ms. This could allow for rapid identification at low pigment concentration suited to automatic sorting. An estimation of the lower concentration limit for detection was made by developing a trend line demonstrating a relationship between the fluorescent strength and pigment concentrations. Using pigment SR-1, the trend line indicated that at concentrations down to about 5,000 ppm, the fluorescence signal could be still suitable for automated sorting units (see Figure 3 below). Optimising the use of machine readable inks for food packaging sorting 11

14 Figure 3: Concentration dependence or SR-1 dye The markers intended to identify packaging during sorting need to be lightfast throughout the packs use, and its collection for recycling through to the final sorting. Tests were undertaken with pigments SR-1 and ER-1 together with NQ-2 to establish their response when exposed to outdoor daylight for extended periods. Tests show that within about 7 days of outdoor exposure the SR-1 and NQ-2 samples had lost its fluorescence yield by almost an order of magnitude, while ER-1 was only slightly reduced (see Figure 4 below). Samples kept indoors showed no drop in performance over 6 months. Exposure to indoor lighting in a retail and domestic situation will have a lower level of UV exposure, and requires further assessment to quantify the level of impact on emission intensity. Most primary packaging is protected inside secondary packaging and has minimal direct sun exposure until it is discarded and collected through the recycling process. Also when baled for transport or storage only a small percentage of the articles are exposed at the surface of the bale. The sensitivity of the fluorescent compounds to UV light could be improved by the use of UV blockers or selecting fluorescent compounds with inherent UV stability. Figure 4: Impact of outdoor exposure on emission intensity of selected dyes. Optimising the use of machine readable inks for food packaging sorting 12

15 4.3.2 Supplier-2 Fluorescence spectra were generated for SC-1 and SL-1, two pigments which were sourced from Supplier-2 and compared to that generated by SR-1. SC-1 exhibited very similar peak fluorescence intensities to SR-1 while SL-1 was much weaker (Table 2). Table 2: Fluorescence of SR-1, SL-1 and SC-1 at different concentrations Pigment SR-1 SL-1 SC-1 Conc. Ratio Peak Ratio Peak Ratio Peak Ratio ppm Intensity Intensity Intensity 12, , , , , , , , , , , , When excited by UV, SR-1 has a bright red emission while SC-1 has a bright cyan emission. Having access to a number of markers which are invisible to the naked eye yet offer different coloured emissions enables selective marking of different types of plastic packaging materials and could assist in their identification during sorting Supplier-3 The test results on the fluorescence properties of the samples from Supplier-3 (see Figure 5 below) clearly indicated that the fluorescence of these dyes were severely reduced when used with the two types of lacquer, compared to what had been previously reported in HDPE polymer matrices. NQ-2 had a good light fastness from the outdoor exposure and exhibited positive properties but it did not produce an acceptable signal even at a concentration of 50,000 ppm. Figure 5: Intensity of response from Supplier-3 in two lacquer systems Optimising the use of machine readable inks for food packaging sorting 13

16 5.0 Detectability of print and labels on recycled packaging Labels are made from different materials and are applied in different ways. Labels can be directly printed onto the package, and permanently bonded such as in-mould labels. There are also shrink, wrap-glued and pressure sensitive labels that can be completely or partially remove after disposal. The same decoration techniques are used for many types of plastic and packaging, all of which might benefit from the improved identification capabilities offered by machine readable inks. This project evaluated if packaging labels with machine readable inks could be used to identify food-grade PP as a specific example to demonstrate that the technology could be used more broadly. The ability for any label to be detected will depend on numerous factors including: Extent of coverage of the label on pack, for instance there is much greater chance of detecting a larger label or if the label extends around the pack The integrity of the label following collection, sorting and separation Presence of debris on the label which reduces excitation and strength of fluorescence Composition of in-feed material; high proportions of other plastics could hide tagged food grade PP. A semi-quantitative scheme was developed to evaluate the likelihood that a readable label surface would present to a detection system in a manner that would enable it to be detected and ejected. Figure 6 shows examples of packaging that are indicative of the four different categories identified. Waste residue and sorted PP packaging equal to 63kg was sorted to obtain 1,888 articles from two plastic recovery facilities (PRFs) that were audited to determine the percent of packaging items in each category. Photographs were taken to illustrate the audit findings and further details of the audit work are shown in Appendix 2. Figure 6: Ranking categories whether readable inks can be used in detectable labels 100% detectable Labels / Inks all round 50% detectable Labels / Inks on one side only 10% detectable Residual labels minimal print 0% detectable No label or printing Articles with all round label or printing would be detectable regardless of its orientation or condition when presented to the sorting system and were assigned a 100% ranking. Similarly articles printed only on one side have a 50:50 chance of detection depending on their orientation during sorting and so were given a 50% ranking (Figure 7). Articles with severely deteriorated or mostly detached labels or minimal print have a limited chance of being detected so were given a 10% ranking and final article without labels or printing were given a 0% ranking. Optimising the use of machine readable inks for food packaging sorting 14

17 Figure 7: Examples of PP packaging found in each of the detection groups a) 100% detectable b) 50% detectable c) 10% detecatable d) 0% detectable PP articles were hand-sorted based on the above ranking system with the weight and number of packs being recorded for each category to determine the chance of detection from the given a percent ranking. Table 3 shows the results of the audit work carried out at two UK PRFs. Visual assessments were also made at a third UK PRF which could not provide quantitative results, but anecdotally supports the same finding. Table 3: Summary of audits carried out at two UK PRFs (% weight of total). Category PRF 1 PRF 2 Average 100% detectable % detectable % detectable % detectable Total weight(kg) Number of articles 442 1,446 In summary: The results show that the PP sorted material from both PRFs were mainly either 100% or 0% detectable based on their design and the use and recovery process they had been through. Around 54% of packs had labels either all the way around the packaging or with labels/inks on one side of the container indicating which would suggest a good chance of being detected in the sorting machine. There is a small variation in the weight of recovered packaging samples and the total bale weight samples due to the presence of other film products and non-plastic waste such as cardboard and glass. The 0% category consisted mostly of fragments of plastic packaging and clear meat trays. Optimising the use of machine readable inks for food packaging sorting 15

18 Based on labels identifying the product and package type, it was estimated that the food grade PP packaging represented 80% by weight, when compared to non-food grade. 6.0 Sorting trials with commercial equipment 6.1 Initial Static Scan The hand draw down film samples of the Supplier-1 dyes IY-1 and SR-1, both at 50,000 ppm, along with 0 ppm control samples in nitrocellulose and vinyl lacquers were sent for evaluation at the Tomra laboratory using its UV and NIR detection system. Static measurements were made of all variables to establish a baseline of performance and to create a classification for each of the machine readable ink samples. With standard detection systems (using a standard halogen or a UV-C lamp as the source of lighting) no response was detected. Both ink samples were detectable using UV-A (black light) lighting, with the light source at less than 10cm from the sample. Further testing combined UV-A with standard halogen lighting improved the signal response, without negative interference. Further samples of SR-1 ink at reduced concentrations were then also supplied and tested under the same conditions. Results show that the ink was readily detectable to 6,250ppm the minimum concentration supplied. Results from both sample sets are represented by the green intensity image and the pink and purple classification images shown in Figure 8. Higher signal intensity was achieved as shown in (a) and (b) with the more intense sections of the green, grey and pink stripes, with the UV-A light at a distance of less than 5cm for both samples. Moving the UV-A lamp to >10cm caused a significant loss of signal strength and detectability indicated by the less intense sections of the green, grey and pink stripes. A strong classification was achieved at all ink concentrations 50,000ppm, 25,000ppm, 12,500ppm and 6,250ppm with the UV-A light at less than 5cm (c). Figure 8: Effect of light source distance and ink concentration. (a) Detectability varying with UV-A light at increasing distance from sample. (b) Detectability varying with UV-A light at increasing distance from sample. (c) Left to right decreasing ink concentration with UV-A light at less than 5cm. The use of an elliptical mirror to focus the lighting with the objective of increasing the effective distance between the samples and the light source from 10cm to 20cm, which would be more practical in a commercial sorting situation, did not improve results. Optimising the use of machine readable inks for food packaging sorting 16

19 6.2 Commercial simulation dynamic scan Having established a classification and solid signal response with a static experiment, the next phase of the bench scale trials was to scan the samples on a moving conveyor belt operating at commercial sorting speeds of 3m/s. The trial used UV-LED s as a lighting system to increase the distance between the light source and samples and still maintain or improve the detectability of the labels. The use of the UV-LED s increased the excitation and detection distance from 5cm to at least 23cm, providing solid and reliable detection at 3m/s with LED s operating in conjunction with the halogen lighting that is required for simultaneous NIR detection and sorting. All previously prepared concentrations of pigment from 50,000ppm to 6,250ppm were readily identified creating the option to reduce pigment concentrations further in future developments. Figure 9: UV LED array over conveyor (left) and excitation of label samples (right). Bench scale trials confirmed that the machine readable inks applied as a lacquer on a film could be detected at concentrations of 6,250ppm and at commercial sorting speeds of 3m/s. The higher intensity UV-LED lighting was required to obtain a suitable signal under practical conditions, and this could be used in conjunction with standard halogen lighting without interference with NIR sorting. 6.3 Commercial sorting trial Working with a commercial scale sorting facility at Tomra, a range of trials were undertaken with pots, tubs and trays (PTT s) to investigate the robustness of a fluorescence-based machine readable identification and sorting system: Trial 1 Impact of pot or lid type and labelling on machine readability Trial 2 Impact of pigment concentration on machine readability Trial 3 Impact of label size and distribution on machine readability Trial 4 Impact of pack transparency/opacity on machine readability Trial 5 Sorting of samples mixed with PP waste packaging Static and sorting simulation trials had demonstrated the improved identification efficiency if lighting from UV-LED s was used in addition to standard halogen lighting used for NIR detection. It became impractical to make this modification to a large sorting facility in the UK, so a 1 metre wide UV-LED array was prepared on the sorting facility at Tomra in Optimising the use of machine readable inks for food packaging sorting 17

20 Germany. The Tomra facility is a functional MRF that can sort up to 2 tonnes per hour of recycled plastic packaging at speeds and conditions typically found in commercial MRF s in the UK. Figure 10 shows the automated NIR sorting unit used in the trials and the purpose built double array of UV-LED lighting to enhance the detection of labels with machine readable ink. Figure 10: The experimental setup at the Tomra sorting facility in Germany. Over 500 PP pots, tubs and trays (PTT s) and lids, were sourced from RPC and modified with machine readable labels as shown in Figure 11 and Table 4. After the initial trials to quantify the detection and sorting of different configurations of machine readable labels, these prepared samples were manually flattened and mixed with 26kg of PP general packaging recovered from a UK MRF and then a sorting performance trial was conducted. Figure 11: PP containers used in the sorting trials #1 Lid, Injection White #2 Lid, Injection Yellow #3 Pot, Injection Translucent #4 Pot, Injection Translucent #5 Pot, Injection Translucent #6 Pot, Injection White #7 Pot, Injection Black #8 Tub, Thermoform Clear #9 Tray, Thermoformed White Sample preparation Self-adhesive PP sheets were prepared as described in section 4.2 with nitrocellulose lacquer and SR-1 pigment at concentrations of 12,500 ppm to 0 ppm. Sheets were then cut to size Optimising the use of machine readable inks for food packaging sorting 18

21 and adhered to the external side of the pot or lid, in most cases trying to maximise the labelled area. A number of trays and pots were also labelled with dots and lines to determine the effect of different label sizes, designs and positions on the detectability. A small proportion of pots and lids were left unlabelled for system checking purposes, as well as the 0 ppm labelled controls. The number of pots labelled for each classification and the configuration of the labels is shown in Table 4 and Table 5. Table 4: Classification of PTT samples received and prepared with machine readable labels No. #1 #2 #3 #4 #5 #6 #7 #8 #9 Format & Colour Lid, Injection White Lid, Injection Yellow Pot, Injection Translucent Pot, Injection Translucent Pot, Injection Translucent Pot, Injection White Pot, Injection Black Tub, Thermoform Clear Tray, Thermoform White Total PTT s Full label 0 ppm A No label B Full label 12,500 ppm C Full label 6,250 ppm D Full label 3,125 ppm E Dot and lines 12,500 ppm F dots 9 lines dots 3 lines Total Optimising the use of machine readable inks for food packaging sorting 19

22 Table 5: Dimension details of machine readable samples for sorting trials. No Format & Colour Label size (cm) Label area (cm 2 ) Pot weight (g) Label area (%) #1 Lid, Injection, White 10 x #2 Lid, Injection, Yellow 10 x #3 Pot, Injection, Translucent 25 x #4 Pot, Injection, Translucent 25 x #5 Pot, Injection, Translucent 25 x #6 Pot, Injection, White 15 x #7 Pot, Injection, Black 32 x #8 Tub, Thermoformed, Clear 25 x #9 Tray, Thermoformed, White 41 x These labelled PTTs were separated into their categories and analysed through five different trials that are detailed in the sections below Trial 1: Impact of package type on machine readability Selected PTT s were examined to see the effect of the type of container, and the presence of a machine readable label on the response of the automated sorting system. A known quantity of samples from classifications A (0ppm control), B (no label system check) and C (machine readable labels) which represents these variations, were then assessed for their individual response to the sorting system, the results are shown in Table 6 below. Figure 12: Unlabelled, 0ppm control and 12,500ppm pots used in Trial 1. Optimising the use of machine readable inks for food packaging sorting 20

23 Table 6: Impact of pot or lid type and classification on machine readability No. Label Area (%) A (control) 0ppm B (check) No label C 12,500ppm Detect Eject Detect Eject Detect Eject #1 50 0/5 0/5 5/5 5/5 #2 60 0/5 0/5 5/5 5/5 #3 90 0/10 0/10 6/6 6/6 #4 90 0/10 0/10 3/3 3/3 #5 90 0/10 0/10 0/10 0/10 9/9 9/9 #6 30 0/10 0/10 8/8 8/8 #7 90 0/10 0/10 6/6 6/6 #8 90 0/10 0/10 10/10 10/10 #9 60 6/6 6/6 Samples were all tested at commercial speeds of 3m/s with positive sorting, meaning that articles with the machine readable labels were identified and then ejected. The unlabelled or undetected pots would fall into the drop fraction. Results in Table 6 show a very high proficiency (58/58 or 100%) when samples were tested as a single package, with the detection and ejection of the samples with machine readable labels for all package formats while none of unlabelled (check) and 0ppm (control) samples were detected or ejected independent of package type. These results provided a high level of confidence the machine readable label technology was able to perform with high reliability at commercial speeds. PTT s in this trial had not yet been crushed and flattened, as would typically be the case with commercial baled materials. Crushing and flattening reduces the movement of cylindrical packaging on the sorting belt and allows for efficient ejection. It was necessary to ensure that the samples did not roll or move on the belt, as this movement is known to cause incorrect ejection. Close attention was paid to this aspect to minimise this issue. Optimising the use of machine readable inks for food packaging sorting 21

24 Figure 13: Fluorescent inherent in packaging and small label example. (a) Lid (top) with paper label changing appearance of MRI label compared to direct print and MRI label (bottom). (b) Inherent background blue colour fluorescence of unlabelled PP pot under UV light compared to MRI labelled sample. (c) Example of a small MRI label on larger pot that was less reliably ejected. The key discussion points which arose from Trial 1 involved lids which use paper labels that appear to contain optical brighteners, clear PP pots that have background fluorescence due to clarifying additives within the polymer itself and the effect of small labels on ejection efficiency. The additional fluorescence from paper labels and clear pots effectively changed the nature of the signal from the machine readable labels, as shown in Figure 13 and neither of which was previously known or evaluated. However both articles were still able to be identified and sorted with high efficiency. For the #7 black tubs that had smaller machine readable labels with only 30% coverage, although these were detected, ejection efficiency was reduced because the number of air jets that are activated is reduced in response to the smaller label area. In some cases this caused either misalignments or insufficient air to effectively eject the larger article over the divider. It was significant that the use of a machine-readable label provided a new and positive means of identifying and ejecting a black plastic item as this cannot be typically achieved for plastics using carbon black pigments with conventional NIR sorting equipment Trial 2: Impact of pigment concentration on machine readability Sample #5 (pot, injection, translucent, 350g) labelled C, D and E were examined to see the effect of pigment concentration and if lower concentrations of pigment could be detected. Table 7: Sorting efficiency machine readable labels with reduced concentration of pigment No. Label Area (%) C 12,500ppm D 6,250ppm E 3,125ppm Detect Eject Detect Eject Detect Eject #5 90 5/5 5/5 5/5 5/5 5/5 5/5 Five samples of each pigment concentration were tested within this trial. The results (Table 7) show that the labels were detected and ejected at 100% proficiency at all concentrations of pigment down to 3,125ppm. It was apparent that there was a reduction in signal strength at the lowest concentration and this would appear to represent a lower concentration limit for this specific fluorescent compound in this lighting configuration. This value may vary significantly with other fluorescent compounds and lighting systems and further testing would be required to determine the most effective concentration in each case. Optimising the use of machine readable inks for food packaging sorting 22

25 Figure 14: Different concentrations of MRI on label. C (left) 12,500ppm, D (middle) 6,250ppm and E (right) 3,125ppm. The impact of reduced pigment concentration on the fluorescence of the test samples can be seen in Figure 14 above. A less intense visual colour is noted in sample E (right) Trial 3: Impact of label size on machine readability Sample #5 (pot, injection, translucent, 350g) marked F was examined to see the effect of the label shape and size. The F samples were labelled with stripes or dots as indicated in Table 8 below; 9 samples with label stripes of 3 lines (5cmx0.5cm) at equal distance around the outside of surface. 9 samples of dots with 3 circles (2cm diameter) adhered at equal distances around the outside surface. Table 8: The effect of label design with florescent stripes and dots. No. Total Label Area (%) F - Stripes 12,500ppm F - Dots 12,500ppm Detect Eject Detect Eject #5 6-9% 5/9 5/9 7/9 5/9 The stripes and dots have similar limitations to the #7 tubs in Trial 1, in that they were mostly detected, but not ejected as efficiently because the smaller label area activated only air jets of the same proportion as the label size. When this was either misaligned with the whole package or had insufficient force to lift the whole package over the divider bar, the samples falls into the drop fraction, rather than be sorted into the correct bin. Figure 15 illustrates the reduced label area of stripes and dots used in Trial 3. Optimising the use of machine readable inks for food packaging sorting 23

26 Figure 15: Dots and stripes labels used to show the effect within the sorting system. For accurate and efficient ejecting of samples the air jets in most types of automated sorting equipment are activated in direct proportion to the size of the detected sample. So large articles will activate a greater number of air jets for an extended time, and small samples will activate a reduced number of jets for a shorter time as determined by the detected surface area. Striped labels will be detected as a long and narrow or a wide and short article to the detector and in both cases fewer air jets would be activated, reducing the probability of correct ejection over the divider bar causing it to fall into the drop fraction. Dots also activate only a reduced number of air jets compared to the package size and regardless of orientation, again reducing the probability of being correctly ejected, and causing them to fall into the drop fraction. The results from Trial 3 are an important factor and highlight that the machine readable labels need to be proportional to the size of the packaging to optimise the sorting efficiency Trial 4: Impact of pack opacity on machine readability The potential to detect the machine readable labels through translucent containers was examined using injection moulded translucent pots #3 and thermoformed translucent pots #8. Labels were positioned on the external base of the pot to avoid interference from the existing labels in the container walls, and then they were placed upright on the sorting belt so illumination and detection could only occur through the PP polymer in the base. Table 9: The effect of container opacity on machine readability. C 12,500ppm No. Label Area (%) Detected Ejected #3 20 5/5 2/5 #8 20 5/5 5/5 Optimising the use of machine readable inks for food packaging sorting 24

27 Figure 16: Thermoformed tub (left) and Injection pot (right) with MRI labels on their base. The #3 injection moulded pot had a background fluorescence in the polymer that made it look like it had optical brighteners in it. For both samples all 5 pots were detected, but only 2 of the #3 pots were ejected. This was probably because of the relatively small size of the label relative to the containers size, as discussed in Trial 3, and also the higher weight of the injection moulded #3 pot making it more difficult to eject with reduced air jets compared to the #8 lighter weight thermoformed tub Trial 5: Sorting of samples mixed with PP waste packaging Labelled samples from Table 5 were manually crushed to simulate the real world situation where packaging is compressed into bales for extended periods of time, before being sorted at MRF facilities or Plastic Recovery Facilities (PRF). As a result almost all packaging articles recovered for recycling are flattened as can be seen in Figure 17 below. Figure 17: Samples used in the Trial 5 sorting. Crushed PTT s ready to be put on the sorting line. Mixed PTT s after the sorting process. All crushed PTT s from categories C, D, E and F (Table 5) were mixed with 26.1 kg of PP packaging waste collected at PRF 2 as shown in Table 10 and then fed onto the sorting line via the ballistic separator, to provide a consistent and even spread of materials onto the sorting belt. Sorting results that are presented below in Table 11 show a high level of detection of the machine readable labels. Again ejection was impacted by label size as described in Trial 1 and 3 as well as the statistical presentation of the label to the detector and the rolling of the large pots that did not stay flattened after the manual crushing. As with earlier trials the MRI samples were positively sorted (ejected) and the PP waste packaging was the drop fraction. A second recovery pass was then conducted with the drop Optimising the use of machine readable inks for food packaging sorting 25

28 fraction, to minimise the impact of the few pots that were rolling on the belt and were detected but not ejected. Table 10: Composition of mixed material for sorting % Number % Material kg (weight) of parts (number) Sorted PP packaging waste % 1,685 85% MRI labelled and crushed pots % % Total ,986 Figure 18: PP waste and marked containers used in Trial 6. PP waste and marked PTT s on conveyor belt before sorting Ejected fraction of marked PTT s after sorting Table 11: First and Recovery pass sorting efficiency of mixed material. 1 st Pass Recovery (2 nd ) Pass Labelled PTT s input 4.9kg 1.29kg Total eject fraction 3.9kg 1.0kg Contamination (PP waste) 0.3kg 0.09kg Labelled PTTs ejected 3.6kg 0.95kg Total drop fraction 26.6kg 25.6kg Concentration of MRI input 16% 5% Yield of MRI samples 74% 74% Purity of MRI samples 93% 92% Labelled PTTs left in system after 1 st pass 1.29kg Optimising the use of machine readable inks for food packaging sorting 26

29 Some MRI labelled and waste PP articles were lost in the system, either dropping through the ballistic separator or otherwise being held up on the conveyor line, so that fraction totals do not equal 100% of the starting weight. Despite being flattened, some rolling and movement of the articles on the sorting belt was observed, causing ejection errors and a reduced yield that had an impact on the final purity of the sort. Table 12: Overall sort efficiency and yield after addition of recovery pass. Machine readable labelled samples Total input of MRI labelled containers Total output of MRI labelled containers 4.9kg 4.55kg Yield average of first and second pass 74% Total Yield (two sequential passes) 93% Total average Purity 93% The sorting results (summarised in Table 12) showed that using the mix of MRI labelled containers the yield in both sorting passes was 74% at two different concentrations (16% and 5%) of fluorescent labels. The yield is influenced by the statistical presentation of some labels on the lids, the dots on the pots and the inadequate flattening of some of the pots. The average purity of the sort fractions in both passes was 93%. Regardless of these minor trial issues, the machine readable labelled samples were able to be detected at levels close to 100% and sorted with high efficiency (close to 100% with ideal presentation to the detector) to provide a high purity stream (average 93%) which confirms that this technology would be able to be used to sort food grade and non-food grade plastic packaging articles, as well as other difficult to sort packaging materials Sorting trial conclusions Trials 1-5 at the MRF facility have confirmed that packaging articles with machine readable labels can be sorted under commercial conditions with yields of at least 74% and purity of 93% in a single pass. It is standard practice in PRF s to apply a second sorting pass on the target material, and as such this would be acceptable for sorting food contact plastic packaging. A second sorting pass will normally be used to further improve purity. Two passes of sorting at 93% purity will result in a final purity of 99.5%, which would exceed the current EFSA target of 99% purity for food grade polymer purity (as specified for recycled HDPE). An audit of the sorted fractions in Trial 5 enabled a further theoretical assessment of the sorting performance. This was based on the isolation of the MRI samples that were previously found to be difficult to sort, such as the samples with small labels (#7), dots and stripes (#5) and packages that rolled. If these types of articles are removed from the calculation, then the yield and purity of the sort is further improved. It is estimated that yield would be improved to 98% and purity improved slightly to 95% with the removal of PP waste that was contaminated with MRI ink from the test sample. The trials have highlighted the need for the label size or marked area to be comparable to that of the package for efficient sorting. Also that some packaging materials have existing background fluorescence activity that may need to be considered. Further optimisation is needed in terms of design to integrate with existing labelling or a separate application of the lacquer or a coating maybe needed. Some of the limitations of manually preparing the labelled samples impacted on the final results, however these issues are not expected to be of concern for a commercial processes. Optimising the use of machine readable inks for food packaging sorting 27

30 It was necessary to conduct the sorting trial at a smaller MRF facility due to the requirement for an additional lighting source, of UV LED s. It was not possible to build the new light source and retrofit a machine in a large UK MRF without causing a major disruption to the MRF operations. Despite the reduced trial size there is full confidence in the ability to scale up the making of MRI labelled packaging samples using existing commercial processes and the addition of UV LED lighting source to retrofit existing sorting equipment. Significant additional testing and equipment design is required before full commercialisation is possible, however modification and retrofitting to existing equipment, with UV LEDs of this type is expected to be feasible for recent model equipment. The LED s provide a long operational life at relatively low operating cost while capital costs for modifications of this type are still speculative at this early stage. 7.0 Evaluation of cost implications The attributes of a label-based marking system have been summarised previously as: Identifiable at commercial sorting speeds; Visible from all sides of the container; Robust and remain intact throughout the recycling process; Marker must be destroyed or removed during the recycling process; Marker must not interfere with branding; and Low cost with labels typically < 0.75pence per label 7. The various cost elements that need to be considered include those relating to the cost of the label and pigment, its method of application to a pack, costs associated with additional sorting, and costs to modify existing sorting equipment. The costs of developing a label are considered variable due to size and coverage as well as the final pigment type and concentration. A surface coating over the entire label has been utilised in this study, in practice another form may be used. Some labels already use a varnish layer into which a fluorescent or other machine readable ink could be added, so there would be minimal additional cost. Other label formats may require the varnish layer to be added if the pigment cannot be otherwise incorporated in the existing design. It is expected that the labels will be able to be applied in the same way, and minimal if any additional cost is expected in that aspect unless the design requires additional labelling to improve the probabilities of detection and sorting. The cost of additional sorting is expected to be compensated by the additional value and yield of packaging that a MRF or PRF would obtain from materials. Although this value may be modest initially, it would increase as detectable labels for a given application gained in volume. The study has shown that optimal sorting results are achieved by using specific UV emitting LED s as an illumination source, in addition to the standard halogen lighting used for NIRoptical sorting. The UV LED system is compatible with halogen lighting and NIR sorting they are relatively inexpensive to purchase and operate and to retrofit to existing equipment. These arrays use much less power than mercury vapour lamps and last ten times as long however they need to be given a power supply and provided with adequate cooling. It would be necessary to optimise the design and integrate this array in a commercial system. 7 Sorting plastics for food use WRAP summary report. Axion Consulting 2012 Optimising the use of machine readable inks for food packaging sorting 28

31 7.1 Estimation of incremental pigment costs and requirements The incremental costs of pigment used to create labels with areas of 100cm 2 are considered here, with a machine readable pigment concentration of 12,500ppm as these were representative of the range of labels used in the current trials. Manually prepared labels used 0.15ml of pigment/lacquer/solvent stock solution to cover a 10cmx10cm area using 1.875mg of pigment. Using pigment prices of /kg provided by the suppliers, the pigment costs were estimated at pence per label (details in Appendix 2). With this non-optimised system, additional pigment costs are estimated to be 5-15% of the 0.75 pence price of an existing applied label. Table 13: Estimation of pigment costs and volumes Pigment concentration (ppm) 12,500 3,000 1,000 Pigment laid down (mg) Pigment 187/kg (pence/label) Pigment 575/kg (pence/label) Total Pigment required* (tonne) * Recent data 8 estimates that 143,000 tonnes of food grade PP are placed on the UK market. Assuming a weight of 15g per pack, the UK uses 9.53 billion PP food packs per year. If each pack was to carry a 10cm x 10cm tagged label carrying mg of pigment (12,500ppm), then the pigment requirements are estimated at 17.9 tonnes per year. Trials conducted suggest that pigment concentrations much lower than 12,500pm were readily detectable and it is very likely that lower concentrations may be used in a commercial system that would reduce costs and pigment volumes to 25% of those used for most of the trials. Table 13 estimates the reduced cost and tonnage of material required if pigment levels were able to be reduced to 3,000ppm as was successfully detected in the trial, and at 1,000ppm, which was successfully evaluated in bench scale testing. Machine readable labels add minimal cost and can be readily available for commercial use and eventual sorting of food-grade plastic packaging. Using the lower pigment cost of 187/kg at 12,500ppm, costs would be an additional 3,342,625 per year. With a pigment application of 1,000ppm, pigment use is reduced to 1.43 tonnes per year and carries an additional cost of 267,410 or just under 0.4% the estimated cost of an applied label priced at 0.75pence. Using the higher pigment cost of 575/kg the cost and the higher 12,500ppm concentration that was used in trials, costs increase to pence/label (details in Appendix 2). 7.2 Sorting equipment modification The first static trials with commercial sorting equipment established that the existing halogen lighting did not provide adequate excitation of the machine readable inks, for a commercial process. The bench top experiments had utilised UV LED s for screening compounds so a UV LED array was selected as the most cost effective technique for the MRF trial. A water cooled 8 UK market compositional data of polypropylene packaging. WRAP July 2012 Optimising the use of machine readable inks for food packaging sorting 29

32 double array across the 1 metre wide sorting belts was made and this proved to very effective in trials as shown in Section 6 of this report. After the trials it has been confirmed that retrofitting a UV LED light source to existing recent model equipment is practical, however further design and optimisation is required to better integrate with these existing systems. Costs for upgrades are difficult to estimate at this early stage, and is dependent on several factors, including the selection of compounds, label size and existing equipment configuration. New equipment could also be produced with the UV LED light source fully integrated. While any additional capital cost is significant for an existing MRF and PRF operators, it is not expected to be prohibitive, given the potential for additional value that could be achieved from a food grade sorted plastic or other new product streams. 7.3 Sorting Protocols for Fluorescent markers In this project the markers have been added individually however they can be added in multiple combinations to generate a more informative coding as has been shown previously by Ahmad and co-workers 9. When this is done then multiple signals are emitted at the different wavelengths related to the specific pigments. This means that in principle if only four fluorescent pigments were used then there would be fifteen possible combinations of signals (given by the formula 2 n -1). Six markers would provide 63 possible combinations. The principle is shown in Figures 19 and 20. Figure 19: The spectra of two tracers A and B shown individually and in combination 9 S Rafi Ahmad, A new technology for automatic identification and sorting of plastics for recycling, Environmental Technology, 25, , 2004 Optimising the use of machine readable inks for food packaging sorting 30

33 Figure 20: How four fluorescent markers could be used to code package classification Number of doppants (n d ) for n number of tracers: n d = 2 n -1 Detector Signal Identification by Binary Code TYPE A Fluorescence Yes B C D E No F Illumination G H Tracer Doped Plastic Product Yes I J K L Yes M N O This means that the use of multiple markers can be used to sub-categorise the package in ways that have not been previously possible. Clearly it can be applied for the case of identifying food grade materials within a polymer type such as PP, HDPE and PET once the package has been categorised by the NIR polymer sorting process as per the current practice. In addition, special categories could be created to actively include or exclude packages during sorting. For example in HDPE, one category could be un-pigmented or pigmented milk bottles to ensure that only these products are widely recycled back into milk bottles. The food grade PP stream could be split into homopolymer and copolymer fractions for optimal recycling into high performance products. Another category could be allocated to packages that are used for household chemicals, which could ensure that these packages are actively excluded from food grade applications. Black plastics of each major polymer type could be also be given specific categories. In the recovery of PET bottles, many bottles are sleeved and are ejected as non-pet. A marker combination category in the sleeve could be programmed to indicate that a specific bottle could be safely added to the clear PET bottle bunker. Equally another marker combination could indicate that the sleeved bottle was pigmented and then safely directed into the coloured PET fraction. A further category in PET, is water bottles that contain AA (acetaldehyde) absorber so that they are not mixed with clear carbonated bottles. Other categories could be reserved for multilayer products in all polymer types. Bioplastics of the various types could be sorted individually or generically. The concept could also be applied to sort plastics containing high value pigments such as pigments that don t absorb infra-red radiation (as used in CPET) or specific chemicals such as BPA (bis phenol A), flame retardants, plasticisers or other materials of interest or concern. The use of four non-overlapping fluorescing markers would allow the development of fifteen subcategories that at this stage would seems to provide a lot of scope for splitting the recovered packages into discrete recycling streams. More categories would be available if additional markers were introduced to the sorting protocol. Optimising the use of machine readable inks for food packaging sorting 31

34 In order to exploit this new sorting capacity a key requirement would be to establish a protocol for unique and standardised set of markers for the specific categories and to ensure that only these approved markers are used within the regional and potentially trans-regional market place. 8.0 Standardisation protocol for machine readable ink marker technology The introduction of new technology that involves the use of specific (fluorescent) signals for the sorting of packaging will require coordination between the whole supply chain once the technology has been well defined by the initial development of commercial machinery and label technology. Within Europe there are a number of organisations that would be involved in the coordination of a system of markers so that each fraction that is separated is based on consistent use of materials to avoid any confusion in the ultimate step of recycling. The European organisations that would be consulted would be EUPR Plastics Recyclers Europe - representing over 110 recycling companies EPRO European Association of Plastics Recycling and Recovery Organisations representing 19 organisations in 14 European countries. Petcore Europe - a pan European organisation of specialists from the PET Value Chain. It develops efficient solutions for the sustainable management of PET recycling. EFBW - The European Federation of Bottled Water (EFBW) represents the interests of bottled water producers, bottlers and companies at a European and international level. In UK the organisations that could be consulted would include industry organisations brand owners and retailers. Within these categories the following organisations could be consulted; BPF - British Plastics Federation - is the leading trade association for the UK Plastic Industry, with over 450 members and 1,200 affiliated members. BPF Recycling Group - this is now a stand-alone membership group within the British Plastics Federation representing the recycling organisations in UK. WRAP- is the leading organisation that has coordinated similar integration activities as enablers of change in many sectors of the packaging supply chain. RECOUP - Recoup is the UK's leading authority on plastics waste and resource management, providing expertise and guidance to a wide range of clients across the plastics supply, use and disposal chain. There is an existing EU FP7 research project coordinated by Petcore called POLYMARK 10, focussing on the recovery of PET bottles using the concept of adding markers to the bottles. Its objectives are similar to those of this project but focussed on PET bottle applications. Discussions have been held with Petcore and they have expressed support and encouragement for cooperation once this project has been completed. More recently an 10 Optimising the use of machine readable inks for food packaging sorting 32

35 Innovate UK project (REFLEX 11 ) also started looking at marking techniques to identify and separate post consumer flexible plastic packaging. The list of organisations is not exhaustive and more organisations can be sensibly added to the both the UK and EU consultation process. It is clear that it may take a significant amount of time to gain coordination and finally consensus between the many groups. 9.0 Conclusions The results of the patent and IP search, while not exhaustive, has indicated that it is unlikely that there is any existing IP that would prevent or otherwise restrict the use of machine readable inks for use in sorting recycled plastic packaging. There has been IP activity in the development of spectral taggants or machine readable markers of several types, however these are mostly focussed on security, tamper evidence and forgery prevention. This study used existing commercial compounds, and has not developed new compounds that could be the subject of new IP. Previous publications have provided insight to a number of specific techniques for detection and sorting of waste packaging using machine readable inks or other spectral taggants. These do not appear to restrict the development, but also limits any potential for new IP to be registered based on the sorting techniques described in this report. The search for suitable commercially available compounds, identified a small number of options that had sufficient emission intensity in the target wavelength range with the available light source. Organic dyes were preferred for lower cost and long term stability, while inorganic compounds tended to have a higher cost per kg, but higher intensity so that they could be used at lower concentrations. Optimisation of cost / concentration / response and stability requires further work and screening of additional compounds. The sorting trial used an inorganic compound that was commercially available and emitted in the visible and NIR spectrum when excited under a UV light source so it was identifiable using detectors of the type in use on existing NIR sorting equipment. Based on the audit of commercial of PP waste indicating that 55% detectability of existing packages together with potential yields of 98% from the sorting trials, it can be estimated that of the 143,000 tonne of PP food packaging, 77,077 tonnes could be recovered each year in the UK based on the recovery of all of the postconsumer PP packaging. This would increase dramatically if label design and application was modified to better accommodate identification and sorting requirements. The material cost for the additional labelling are from to pence / label dependent on pigment costs and concentrations. This may be able to be applied as part of the existing labelling at very little further cost, however there will be some situations where additional or new labels designs would be required to provide an effective recovery during sorting. These additional costs are difficult to estimate, however it is expected that existing commercial process can be used for their application. A customised UV light source to excite the fluorescent compounds was required in the sorting trial to achieve adequate identification. This made it impractical to conduct the sorting trial at a large commercial MRF in the UK, however using the smaller 1 tonne/ hour trial MRF, is was possible to conduct a trials under commercial sorting conditions of 3m/s and a mixture with recycled PP packaging sourced from PRF-2 in the UK. Trial results show a high yield of 74% with a purity of the 93% of the machine readable labelled samples. Yield could be improved by using a recovery loop on the reject stream and 11 REFLEX - Optimising the use of machine readable inks for food packaging sorting 33

36 purity improved further by a 2 nd pass on the target stream, as is commonly used in existing commercial MRF s and PRF s. These excellent sorting results provide a high level of confidence in the possibility to recover a significant amount of target materials such as HDPE, PET and PP plastic packaging for food applications at purities sufficient for use back into high value applications. Trials indicated that results are independent of colour or transparency, including black, and package type. The trials focused on PP packaging as an example to show the potential to identify and separate PP used in food applications from PP packaging used for non-food applications. It is clear that by using a printed label or surface lacquer the same technique could be utilised on a wide range of plastic packaging formats to provide a unique marker that would enable identification and separation of selected articles to a targeted stream. Investigation is required to determine real life performance of the machine readable inks through the supply chain. Stability of the active compound is required for only a few days, or perhaps several months under a wide range of environmental conditions. Ideally the label or marking should be retained with the packaging until after sorting after which it would be mechanically removed or washed off during further processing. A removable label as opposed to an in-mould or direct printed label is seen as the preferred application technique as this can be more easily removed from the pack after sorting during recycling. Existing wash processes for food applications have been shown to remove a large amount of ink from labels even if the label remains on the packaging, such as IML and also from direct printed packaging articles. 12 This suggests that ink removal will not be an obstacle to the introduction of machine readable labels in these cases. Labels that are not easily removed and that trap the ink under the label substrate may be problematic and this would need to be assessed as part of the labelling protocol. In recent years new selfpeeling adhesive labels have become available on the market and this technology could potentially be used in conjunction with the machine readable labels. It is important that the machine readable marking is eliminated from the packaging during the recycling process, to minimise contamination into food grade applications. Existing food grade processing equipment and processes are designed to remove existing labels and printing inks to prevent contamination. As part of the process to determine a suitable format for machine readable labels and inks it will be necessary to conduct further specific trials to assess the performance of the existing processes to effectively remove the machine readable inks or lacquers. Consistent with testing for printed ink residues, it will be necessary to show that residues from machine readable inks do not contaminate the recycled plastic for food applications or for identification and resorting in closed loop applications. Also it is preferable to remove the label so that a food contact item that may pass through to a non-food grade recycling process is not later incorrectly identified as food contact on its next trip through the recycling loop because of residual food grade label. Recycling applications will need to be evaluated to ensure that any machine readable labels intended to be sorted to a food application stream but that remain in non-food applications are effectively removed or otherwise rendered inactive to prevent a non-food grade article being incorrectly identified as food grade, giving a false positive. The deactivation may occur from the existing processes of intensive washing and melt compounding, effectively thermally degrading the fluorescent additive. An additional influence of the melting stage in recycling would be to dilute fluorescent dyes to lower concentrations making the signal strength from any residues much weaker. Often these recycled materials are processed into opaque coloured materials that would not provide a significant emission, however the process will require validation as part of any further machine readable ink development. 12 Further analysis of decontaminated recycled polypropylene (rpp)- WRAP report 2012 Nextek Ltd Optimising the use of machine readable inks for food packaging sorting 34

37 Further development work is required to identify additional suitable compounds to improve the cost and performance of the machine readable inks. Additional work is also required to ensure the application to the packaging is optimised for sorting and recovery either by the use of a surface coating or lacquer on existing labelling as was used in the trials in this study, or as a separate application if label size is inadequate, is prone to premature detachment from the package or is otherwise unsuitable. Further coordination on the development of a standardised protocol system for marker technology for sorting plastic packaging in UK and Europe is recommended via the POLYMARK and REFLEX project and the other EU-based industry and governmental organisations Recommendations for further work It is recommended that further work be completed to optimise the performance and cost effectiveness of a label based fluorescent machine readable ink technology. Further development is required to optimise the active florescent compounds being used. Ideally a low cost and/or higher performance materials could be identified. Evaluate stability of the compound in the supply chain to ensure that it remains effective but does not otherwise have any negative impact throughout the supply chain, through recovery and final sorting. Assess the ability of the MRI labels and compounds to be effectively removed during the cleaning and decontamination process, particularly for direct print and in-mould labels that tends to remain with the packaging. Conduct a large scale trial with brand owners, converters, retailers, and recyclers to identify any limitations for full commercialisation once preliminaries are optimised. Conduct a full feasibility (business case) study on the costs and benefits for the supply chain based on requirements and outcomes of the large scale trial. Identify and evaluate other target materials that could be better recovered for closed loop recycling by using MRI labels. As an example black plastic packages with a clear MRI label were successfully identified by this technique. Additionally different packages that require positive or negative differentiation could be sorted after a primary polymer sort such as multilayer bottles and trays, milk bottles, bio-plastics or bottles used for household chemicals and garden products that contain harmful contents. Develop a comprehensive standard set of criteria for sorting using MRI technology so that sorting into generic food grade types including PET, HDPE and PP and other target materials can be conducted uniformly across the UK and European packaging and recycling sector. Optimising the use of machine readable inks for food packaging sorting 35

38 Appendix 1 Patent and Literature search Directly relevant Patents and Literature. A good amount of work was conducted in the 1990 s as the technologies for automated sorting were being developed. Attempts to address particular limitations were made through modification of the detectors themselves and also of the packaging articles being sorted. Once successful NIR-based systems became established commercially in the early 2000 s, the requirement to modify the packaging to affect a successful sort was greatly reduced and interest and development in this area was reduced. The earliest patent entitled, Identification of materials and products, by Dowty Seals Ltd (GB) UK Pat. No. EP , was published on 25 Jan The patent claimed that when different fluorescent materials are added to a manufacturing base of different polymeric materials, these can be identified from their characteristic fluorescence signatures. A range of suitable fluorescent materials were identified for use in transparent polymer materials. The following year Bayer AG (DE) filed a patent under the name, Method for the identification of plastics, US patent # Claims were based on tagging plastics with a fluorescent marker having different fluorescence life time (duration) under fast UV pulse excitation and identification of characteristic emission wavelengths of the dyes incorporated in specified polymers. Kyung-Tae Han et al. filed a patent application DE A1 entitled, Sorting and separation of plastic waste and into its component types by marking with a suitable dye during manufacturing and combined measurement of fluorescent wavelength and lifetime. Claims were similar to those of Bayer AG that the plastic is marked during the course of its production with a dye, the colour and specific fluorescence lifetime of which are measured to allow separation of plastic waste mixtures. Research on the same topic was conducted at The Cranfield University (then Royal Military College of Science) under the leadership of Dr Sheikh Rafi Ahmad and following the successful basic research he advertised for interested organizations within the EU countries to join in a near-market R&D programme with EU funding under the BRITE-EURAM programme. The same authors further developed the idea which has undergone industrial validation, from which the concept and the industrial trial results have been published widely. Of specific interest was the entry of this work into the journal of Environmental Technology Vol 25 titled A new technology for automatic identification and sorting of plastics for recycling. This article clearly describes in detail the concepts of this project, including the use of a combination of fluorescent dyes in a range of polymer materials and methods to identify and sort the packaging at commercial rates. Machine readable inks Searching using the key words machine and readable resulted in a very broad response for a range of techniques including barcodes, sound coding, radio frequency (RFID) and many other methods that are employed as security, tracking, communication and other verification applications. In this broad field there are many thousands of patents however they are not directly relevant to this project. Including the search term ink provided results that incorporated mostly the design of digital printing cartridges for authentications of important documents such as passport, bank notes etc. and of designer items such as perfume bottles, special medicine, bottles containing poison etc. No patent on tracer ink sprayed on labels or coatings of plastic Optimising the use of machine readable inks for food packaging sorting 36

39 packaging has been identified. This area of IP may have some relevance to the sorting of plastic waste as a method of application to packaging or labels. Information on some of the patents granted over the last decade or so is presented below. Patent #US B1 and # WO A1: Printer cartridge including machine readable ink. Assignee: Silverbrook Research Pty ltd. Inventors: Kia Silverbrook et al. Filing date 23 May, Patent #US B1 and US B1: Machine readable water based red fluorescent ink compositions. Assignee: Pitney Bowes Inc. filing date: Inventors: Judith A. Auslander, et al. Filing date:3 April, 2000 and 2 Feb Patent #US A1: Printed, machine-readable code, document provided with a code of this type and a method for producing said code and document. Assignee: George Depta. Filing date: 4 Oct Other related patents activity In the broader field of markers, taggants, identification and sorting for a wide range of applications there has been a good amount of ongoing R&D and patent activity. While not having a direct impact on the objectives and intended technologies of this project, it is important to be aware of this activity and the nature of these related patents in order to better understand the position and novelty of the current project activities. Below are listed just a few of the most interesting related patent titles and the abstracts from many hundreds that were reviewed. Empire Technology, titled Tags for sorting of plastics in January 2014 Methods and systems for identifying and sorting of different plastics rely on the differences in the surface polarity of the various plastics. A marking system may be used which provides markers that adhere specifically to only one type of plastic, and the adhering markers may then allow for a positive identification of the plastic. This type of system may be used for the identification and separation of polylactic acid plastics from various other plastics. Polysecure GmbH, title Material having a marker and method for sorting a mixture of materials March 2013, with an abstract reading The present invention relates to a material, preferably a plastic material, containing foreign matter, characterised in that the material further contains fluorescent compounds, and to a method for separating a mixture of materials. The patent relates to the incorporation of a fluorescent marker to identify PVC window frames for recycling. Barowski et al, title Waste recycling system using tagged, bar coded or other distinctively marked containers, method of recycling, and container device September The invention refers to a recycling system using uniquely marked bags, preferably electronically tagged, for maintaining separation and custody tracking of targeted waste from collection to recycling. Paper cup stock are cited as a high value waste stream which if collected separately can be recycled into food grade paperboard. Conclusion of Patent and literature review This project has attempted to utilise existing compounds, application methods and sorting techniques to minimise barriers to commercialisation. In this regard the development of novel compounds, application methods and sorting technologies is secondary and this may limit the possibility of developing novel solutions. Searches of patent abstracts carried out using the above search engines and key words revealed six relevant patents; Optimising the use of machine readable inks for food packaging sorting 37

40 List of specifically relevant patents Patent No DE (A1) WO (A1) GB (A) US (A1) US (A1) US Patent Title Sorting and separation of plastic waste and into its component types by marking with a suitable dye during manufacture and combined measurement of fluorescent wavelength and lifetime. Material having a marker, and method for sorting a mixture of materials. Fluorescent tracers. Invisible-fluorescent identification tags for material. Waste recycling system using tagged, bar coded or other distinctively marked containers, method of recycling, and container device. Organic solvent based ink for invisible marking/identification. Patent GB (A1) in the above list was authored by principal investigators within the current project team, Rafi Ahmad and Edwin Billiet. The only paper of specific interest that was identified is the entry in the Environmental Technology by Ahmad et al 13 and the subsequent publication in Materials Recycling Week, 13 th February This paper discusses mixing the fluorescent markers with the plastic prior to fabrication rather than being applied as a label or coating post fabrication as intended by this project, but it does highlight the potential to sort different grades of similar polymers, such as food grade and non-food grade polypropylene packaging using this technology. Review of the existing patent and literature has shown that is that there is a great amount of IP in the area of the compounds that might be used as spectral markers, also in their use in a range of applications, however for the purpose of identification and sorting of plastic waste there are relatively few directly relevant patents. Similarly with the literature review, very few publications that provide specific reference to the sorting of plastic waste were located. The content and disclosure of these few patents and publications indicates that if this project is successful it has good prospects of being able to operate freely in the market place. In addition, specific novel compounds as well as application or identification techniques would need to be developed in order to have sufficient novelty that might itself be patentable. The aim of this project will be to utilise existing materials and processes to improve the probability and speed with which industry might adopt and commercialise the project outcomes. 13 S Rafi Ahmad, A new technology for automatic identification and sorting of plastics for recycling, Environmental Technology, 25, , 2004 Optimising the use of machine readable inks for food packaging sorting 38

41 Appendix 2 Audit PP articles in the waste stream Aim To gather data on how easily packaging that used machine readable ink on the labels would be detected and sorted. Rationale: Pots, tubs and trays (PTTs) that could end up in landfill, including polyethylene terephthalate (PET), polyethylene (PE) and polypropylene (PP) were targeted to assess the likelihood of being detected if readable inks were incorporated within the label. A semi-quantitative scheme was developed to evaluate the likelihood that a readable label-surface would present itself to a detection system (Table 1). The weight and number of packs was recorded for each label category to determine the percentage chance of detection. Photographs were taken of representative packs to provide visual evidence to support the audit findings. Categories used for assessing whether readable inks can be used in detectable labels. Chance of being Detected (%) Types of Labels 100 Labels/inks on both sides 50 Label/inks on one side 10 hardly any labels/inks left 0 no labels i) Trial at PRF-1 PTTs that end up as waste residue (Figure 1) in the CLR plant were analysed and audited to ascertain the extent of labelling and have the chance of being detected. Waste residue bale at PRF-1 Optimising the use of machine readable inks for food packaging sorting 39

42 100% chance of detection: Examples of PTTs with labelling all the way round the pack or on both sides where 100% of packs would be detected are shown below. Optimising the use of machine readable inks for food packaging sorting 40

43 50% chance of detection: Examples of PTTs with labelling on one side of the container indicating a 50% chance of being detected are shown below. 10% chance of detection: Examples of labels/inks that have mostly come off the container and detection would be most unlikely are shown below. 0% chance of detection: Examples of packs without labels indicating detection is not possible. Optimising the use of machine readable inks for food packaging sorting 41

44 PRF-1: Audit data A sample of two bales was separated into the four categories and weighed to determine the percentage of each fraction Summary of audits 1-5 from PTT s at PRF-1 Category Weight (kg) Fraction (%) 100% detectable % detectable % detectable % detectable Total Weight (kg) 9.14 Concluding Remarks The results show that from the PRF-1 audit the packaging materials were mostly either 100% or 0% detectable based on the process they had been through. The weight of each sample did not match the total category weight because there was a lot of other film products and other waste materials in each sample. The percentage of PP was approximately 20-25% of each audit The 0% category consisted of broken bits of plastic packaging and clear meat trays. ii) Trial at MRF Due to an unexpected machine shut down at the time of the audit, the research team were unable to access sorted material which prevented detailed gravimetric analysis of separate packaging samples. As such, only a limited visual audit of the likelihood of label readability could be undertaken from the side of the bales, however packaging labels similar to PRF-1 was observed, indicating commonality at different sorting facilities. iii) Trial of sorted PP packaging from PRF-2 To obtained material for future sorting trials a third facility was also audited by analysing 80 Kg of sorted PP packaging. 80 kg of sorted PP from PRF-2, separated, weighed and counted. Optimising the use of machine readable inks for food packaging sorting 42

45 100% chance of detection: Examples of PTTs with labelling all the way round the pack or on both sides where 100% of packs would be detected are shown below. Optimising the use of machine readable inks for food packaging sorting 43

46 50% chance of detection: Examples of PTTs with labelling on one side of the container indicating a 50% chance of being detected are shown below 10% chance of detection: Examples of labels/inks that have mostly come off the container and detection would be most unlikely are shown below. Optimising the use of machine readable inks for food packaging sorting 44

47 0% chance of detection: Examples of packs without labels indicating detection is not possible Optimising the use of machine readable inks for food packaging sorting 45

48 Summary of audits 1-7 from PP sorted material from PRF-2 Category Weight (kg) Fraction (%) 100% detectable % detectable % detectable % detectable Total Weight (kg) Summary of audits carried at PRF-1 and PRF-2 (% weight of total) Category PRF-1 PRF-2 Average 100% detectable % detectable % detectable % detectable Optimising the use of machine readable inks for food packaging sorting 46

49 Appendix 3: Cost evaluation summary Incremental cost of pigment to make label measuring 10cm x 10cm together with estimation of UK pigment demands based on 14.3 billion PP packs. 1) Concentration of pigment in stock ink solution is 12,500ppm or 12,500mg per 1000ml 2) Therefore, each ml of stock ink contains 12,500/1000 = 12.5mg of pigment 3) If we need 1ml of stock ink to make a label measuring 20cm x 10 cm, then the label will contain 12.5 mg pigment 4) based on a price of pence per mg, then pigment costs for label are 12.5x0.0187= pence per label 5) Experiments demonstrated that 0.3ml of stock ink were needed to cover a label measuring 20x10cm label laying down 3.75mg of pigment with costs estimated at 3.75x0.0187= pence per label To cover 20x10cm label Vol used (ml) Pigment laid down (mg) Pigment costs (pence) To cover 10cm x 10 cm label requires 0.15ml stock ink (using range of concentrations) 12, Pigment laid down (mg) Pigment laid down (g) Pigment costs (pence) Pigment required for 14.3 billion packs (g) 17,875,000 8,580,000 4,290,000 2,145,000 1,430, , ,000 Pigment required for 14.3 billion packs (kg) Pigment required for 14.3 billion packs (tonnes) Pigment concentration (ppm) 12, Pigment laid down (mg) Pigment laid down (g) Pigment costs 187/kg Pigment costs 575/kg Pigment concentration (ppm) 12, Pigment required for 14.3 billion packs (g) 17,875,000 8,580,000 4,290,000 2,145,000 1,430, , ,000 Pigment required for 14.3 billion packs (kg) 17,875 8,580 4,290 2,145 1, Pigment required for 14.3 billion packs (tonnes) Optimising the use of machine readable inks for food packaging sorting 47

50