Dylan Nagle Thesis (PDF 3MB)

Infrared Spectroscopic Investigation of the Effects of Titania Photocatalyst on the Degradation of Linear Low Density Polyethylene Film for Commercial Applications.

by Dylan John Nagle, B. App. Sci. (App. Chem.), M. App. Chem.

A thesis submitted to the School of Physical and Chemical Sciences in partial fulfilment of the requirements for the degree of

Doctor of Philosophy

Queensland University of Technology

October 2009

1 2

Do it once, do it properly; never do it again.

These words of Peter M. Fredericks are probably the greatest lesson I have learnt in my years of study to complete the PhD degree, a lesson that requires continual revisiting. I wish to acknowledge the mentoring of my supervisory team; Peter Fredericks, Llew Rintoul and Graeme George. I am grateful for what I have learned from each one, academically and personally. I also acknowledge the efforts of my family, who have offered their utmost encouragement and support. Likewise my friends and colleagues at QUT. Ultimately it was my wife Mi Jeong who carried me when life was at its most challenging, and celebrated life with me at its most rewarding. I am forever thankful.

The work presented in this thesis is, to the best of my knowledge and belief, original and my own work, except where acknowledged in the text. This material has not been submitted, either in whole or in part, for a degree at this or any other university.

Dylan John Nagle October 2009

Abstract ______8 List of Abbreviations ______10 Introduction ______11 1.1 Polymer degradation ______13 1.1.1 Thermooxidation ______14 1.1.2 Heterogeneous vs. homogenous thermooxidation kinetics______15 1.1.3 Photooxidation ______17 1.1.4 Role of hydroperoxides in polyethylene photooxidation ______20 1.1.5 Stabilisation of commercial polyethylene ______23 1.2 Prodegradants______27 1.2.1 Titanium dioxide ______30 1.2.2 Titania photocatalysis______33 1.2.3 Factors affecting titania activity in polymers ______35 1.2.4 Surface chemistry of titania ______36 1.2.5 Surface modification of titania ______37 1.2.6 Doping ______39 1.2.7 Effect of UVA vs. UVC radiation on polymer – TiO2 systems ______40 1.2.8 Summary of sections 1.1 and 1.2 ______41 1.3 Polymer degradation characterisation techniques ______42 1.3.1 Characterization of the bulk via physical tests ______43 1.3.2 Surface Characterisation______43 1.3.3 Chemical Characterisation ______44 1.3.4 Achieving high lateral resolution ______53 1.3.5 Characterisation techniques used in this thesis______57 1.4 Objectives ______58 Experimental ______63 2.1 Ciba films investigation ______63 2.2 Accelerated aging of samples______65 2.3 Mid-IR spectroscopy ______67 2.4 Imaging IR Spectroscopy______68 2.5 Synchrotron experimental ______69 2.6 Scanning electron microscopy ______72 Effect of UV pre-irradiation on the degradation of polyethylene ______73 3.1 Introduction ______73 3.2 Physical characteristics of commercial titanias and general comments ____ 73 3.2.1 Degussa P25 ______73 3.2.2 Kronos ______74 3.2.3 Huntsman Tioxide ______75 3.2.4 Sachtleben Hombitan ______76 3.2.5 Section summary ______77 3.3 Sample whitening ______78 3.4 Times to embrittlement for LLDPE film containing titania______80 3.5 IR spectral analysis – control film (undegraded)______86 3.5.1 Polyethylene absorption table______86 3.5.2 Titania absorption in the mid-infrared ______88

5 3.6 Processing agent absorptions ______88 3.7 IR spectral analysis – control film (degraded) ______90 3.7.1 OH stretc.hing region (3800-3200 cm-1) ______91 3.7.2 Carbonyl region ______91 3.7.3 Below 1500 cm-1 ______93 3.7.4 Section summary ______93 3.8 Effect of UV irradiation – control film (degraded) ______94 3.8.1 Control, weatherometer aged samples ______94 3.8.2 Control, oven aged samples______98 3.8.3 Section Summary______100 3.9 IR spectral analysis – film containing titania (degraded) ______101 3.9.1 Carbonyl region ______101 3.9.2 Fingerprint region ______104 3.9.3 Section summary ______104 3.10 LLDPE containing Degussa P25 (degraded) ______105 3.10.1 Degussa P25, weatherometer aged samples ______105 3.10.2 Section summary______109 3.10.3 Degussa P25, oven aged samples ______110 3.10.4 3% Degussa P25 samples ______111 3.10.5 Section summary______114 3.11 LLDPE containing Kronos 1002 (degraded) ______115 3.11.1 1% Kronos 1002, weatherometer aged samples,______115 3.11.2 3% Kronos 1002, weatherometer aged samples,______116 3.11.3 1% Kronos 1002, oven aged samples, ______117 3.11.4 3% Kronos 1002, oven aged samples, ______118 3.11.5 Section Summary ______118 3.12 LLDPE containing Huntsman Tioxide (degraded) ______119 3.12.1 3% Huntsman tioxide A-HR, weatherometer aged______119 3.12.2 3% Huntsman tioxide A-HRF, weatherometer aged______121 3.12.3 3% Huntsman tioxide A-HR, oven aged______122 3.12.4 3% Huntsman tioxide A-HRF, oven aged______123 3.12.5 Section summary______123 3.13 LLDPE containing Sachtleben Hombitan (degraded)______124 3.13.1 3% Sachtleben Hombitan, weatherometer aged ______124 3.13.2 3% Sachtleben Hombitan, oven aged ______126 3.13.3 Section summary______127 3.14 Discussion of the effects of titania ______128 3.15 Conclusions ______131 Multivariate Data Analysis ______135 4.1 Introduction______135 4.2 Data treatment______136 4.3 Analysis of samples subjected to oven aging______137 4.3.1 Samples without pre-irradiation______137 4.3.2 Samples with pre-irradiation ______141 4.3.3 UVA vs UVC pre-irradiation: extent of degradation information ______144 4.3.4 Section Summary______151 4.4 Weatherometer aging ______151 4.4.1 Water vapour ______152 4.4.2 UVA vs. UVC pre-irradiation ______157 4.4.3 Section summary ______157 4.5 Conclusions ______157

6 Obtaining spatial information around titania particles via a model polymer system ______159 5.1 Introduction ______159 5.2 Experimental______161 5.3 Imaging ATR/FTIR spectroscopy results______163 5.3.1 Determination of titania particle location(s)______165 5.3.2 Discussion of heterogeneous oxidation ______174 5.4 Conclusions ______178 Investigation of degradation in the mid-IR using a synchrotron light source ______181 6.1 Introduction ______181 6.2 Experimental______181 6.3 Synchrotron results and discussion______184 6.4 Conclusions ______191 Conclusions ______193 References ______199

7 Abstract

There is a need in industry for a commodity polyethylene film with controllable degradation properties that will degrade in an environmentally neutral way, for applications such as shopping bags and packaging film. Additives such as starch have been shown to accelerate the degradation of plastic films, however control of degradation is required so that the film will retain its mechanical properties during storage and use, and then degrade when no longer required. By the addition of a photocatalyst it is hoped that polymer film will breakdown with exposure to sunlight. Furthermore, it is desired that the polymer film will degrade in the dark, after a short initial exposure to sunlight.

Research has been undertaken into the photo- and thermo-oxidative degradation processes of 25 µm thick LLDPE (linear low density polyethylene) film containing titania from different manufacturers. Films were aged in a suntest or in an oven at 50 °C, and the oxidation product formation was followed using IR spectroscopy. Degussa P25, Kronos 1002, and various organic-modified and doped titanias of the types Satchleben Hombitan and Hunstsman Tioxide incorporated into LLDPE films were assessed for photoactivity. Degussa P25 was found to be the most photoactive with UVA and UVC exposure. Surface modification of titania was found to reduce photoactivity. Crystal phase is thought to be among the most important factors when assessing the photoactivity of titania as a photocatalyst for degradation. Pre-irradiation with UVA or UVC for 24 hours of the film containing 3% Degussa P25 titania prior to aging in an oven resulted in embrittlement in ca. 200 days.

The multivariate data analysis technique PCA (principal component analysis) was used as an exploratory tool to investigate the IR spectral data. Oxidation products formed in similar relative concentrations across all samples, confirming that titania was catalysing the oxidation of the LLDPE film without changing the oxidation pathway. PCA was also employed to compare rates of degradation in different films. PCA enabled the discovery of water vapour trapped inside cavities formed by oxidation by titania particles.

Imaging ATR/FTIR spectroscopy with high lateral resolution was used in a novel experiment to examine the heterogeneous nature of oxidation of a model polymer compound caused by the presence of titania particles. A model polymer containing Degussa P25 titania was solvent cast onto the internal reflection element of the imaging ATR/FTIR and the oxidation under UVC was examined over time. Sensitisation of 5 µm domains by titania resulted in areas of relatively high oxidation product concentration.

The suitability of transmission IR with a synchrotron light source to the study of polymer film oxidation was assessed as the Australian Synchrotron in Melbourne, Australia. Challenges such as interference fringes and poor signal-to- noise ratio need to be addressed before this can become a routine technique.

List of Abbreviations

ATR/FTIR Attenuated total reflectance/FTIR (spectroscopy) CB Chain breaking EDAX Energy dispersive X-ray analysis ETD Everhart-Thornley detector FTIR Fourier transform infrared (spectroscopy) FPA Focal plane array detector HALS Hindered amine light stabiliser HOMO Highest occupied molecular orbital HDPE High density polyethylene IR Infrared IRE Internal reflection element LDPE Low density polyethylene LLDPE Linear low density polyethylene LUMO Lowest unoccupied molecular orbital MCT Mercury cadmium telluride NMR Nuclear magnetic resonance (spectroscopy) PC Principal component PCA Principal component analysis PMMA Polymethyl methacrylate PVC Polyvinyl chloride QUT Queensland University of Technology RI Refractive index SEM Scanning electron microscopy S/N Signal-to-noise ratio SSD Silicon strip detector UV Ultraviolet

10 Introduction Low-end commodity plastics such as polyethylene are in high demand – 60 million metric tons were produced in 2004 worldwide1. Low Density Polyethylene (LDPE) and Linear Low Density Polyethylene (LLDPE) are the two most common forms of polyethylene other than High Density Polyethylene (HDPE), and are mostly processed as sheets and films for applications in packaging, shopping bags, agriculture, etc.. Due to our high consumption of polyethylene, the matter of disposal of used plastic has evolved as a contentious issue. As a global community we are becoming more successful at recycling unwanted plastic, but in many situations the added cost of recycling is too heavy an economic burden, and for industries such as agriculture it is wholly impractical. Currently the most common method of disposal is burying beneath soil, which coincidentally prevents the plastic from degrading due to the absence of sunlight.

In recent times scientists have sought to develop plastics with more controllable degradation properties to create an environmentally neutral film. An example is the addition of starch to polyethylene, attempting to make it biodegradable2. Unfortunately degradable additives such as starch often inhibit mechanical properties3 and, rather than achieving ‘controllable’ degradation, serve merely to accelerate the degradation process. This has a clear effect on the properties of the material in question, such as shelf life, where the polymer is already degrading before being used.

To combat these issues technology is being developed to more strictly control the degradation properties of various plastics. A successful approach has been the addition of a material that will accelerate degradation processes when exposed to sunlight. Such additives are termed ‘photosensitisers’, and exploit the radical chemistry occurring during photodegradation. Among other materials, transition metal salts in particular such as cobalt4, iron5 and nickel6 have been demonstrated to exhibit photosensitizing effects in polymeric materials.

11 A common photosensitiser is nano-particulate titania, which has been demonstrated to greatly enhance the degradation properties of various polymers when exposed to UV radiation7. Titania holds great potential as a photosensitiser for real world applications as it accelerates the degradation process, hopefully preventing a buildup of buried undegraded plastic. Once the molecular weight of a polymer has been sufficiently reduced via photooxidation, microbial or biotic degradation can proceed8.

While technology such as this is certainly a step in the right direction, the demand for plastics with a high degree of control over degradation is increasing, and thus science must look deeper to provide better degradation management. Beyond simply accelerating the degradation process, it is desirable to pre- determine the length of time a plastic film will maintain its mechanical properties, tunable to the situation required. Thus the ultimate objective of this research is to investigate a method for controlling LLDPE film lifetime, according to the application.

A method of achieving this goal will be investigated by examining the effects of pre-irradiation of LLDPE film containing titania with UV before aging in a dark environment. Titania catalyses oxidation of organic materials by absorbing UV radiation and creating radical species that are involved in the initiation step of oxidation processes9. However the concept under investigation is that of pre- irradiation, which involves the exposure of a polymer containing titania to UV irradiation in order to create reactive sites throughout the polymer matrix, which can then proceed to propagate degradation reactions which spread throughout the material, even in the absence of light, similarly to an infection spreading through a population10.

By utilising pre-irradiation technology, a measured dose of UV can be applied to a polymeric material, such as a shopping bag, in order to initiate oxidation processes. The polymer will then proceed to degrade, within a known time frame pre-determined by the strength and the time of the UV dosage.

12 In order to achieve an understanding of pre-irradiation and the effect of titania as a photosensitiser in commercially available LLDPE film, samples containing several different types of titania from different manufacturers have been exposed to UV irradiation, and then aged under accelerated conditions while being periodically monitored by mid-infrared spectroscopy. The mechanism of spread of oxidation originating at a titania particle has also been examined using infrared imaging spectroscopy, as well as high lateral resolution spectroscopy using a synchrotron radiation source.

An understanding of empirical and mechanistic effects of pre-irradiation of nano- particulate titania with UV on the photodegradation of LLDPE will be developed by analysis of the data obtained from the experimental methods outlined above. It is hoped that data will provide a greater understanding of the fundamental processes involved in titania-catalysed degradation, which can be exploited by future researchers to assist in developing technology that will allow more accurate control over the degradation of commodity plastic film.

1.1 Polymer degradation

There are seven processes by which a polymer can degrade11:

1. Thermal: the application of heat 2. Mechanical: the application of force 3. Ultrasonic: the application of sound waves 4. Hydrolytic: attack on certain functional groups along the polymer chain by water 5. Chemical: attack by corrosive chemicals or gases, such as ozone 6. Biological: attack on certain functional groups by microbes 7. Radiation: absorption of radiation at certain frequencies that induces reactions

Often, there is not just one process at work in the degradation of a polymer, and the nature of oxidation processes involved in the breakdown of a particular plastic will depend on the degradation environment of the plastic. Following the description of the goals of this project presented in the introduction, it is

13 desirable to develop a plastic film that retains its mechanical properties during its usable lifetime, and then will disintegrate into particles small enough to allow microbial action to breakdown the molecular structure of the polymer. Mechanical degradation is of lesser relevance to this study than other degradation processes, as the focus is on the breakdown of the film after disposal, by which time the mechanical properties of the film are no longer relevant. Additionally, the technology has been designed to oxidise the plastic film without requiring the application of mechanical degradation processes.

Biotic breakdown of the plastic film is important to ensure that the film is environmentally neutral; however this will not be discussed further in this thesis as it does not pertain directly to oxidative degradation. Ultrasonic and hydrolytic degradation processes are also not relevant to the degradation of waste polyethylene film for commercial applications. Chemical degradation will be discussed from the point of view of oxidation, or chemical attack by atmospheric oxygen. The degradation processes to be investigated in this thesis are termed thermooxidation (application of heat and attack by oxygen) and photooxidation (application of radiation and attack by oxygen).

1.1.1 Thermooxidation There are three principal steps involved in the oxidation of a polyolefin12:

1. Initiation: By radical generator I (initiator) 2r

r + RH rH + R By hydroperoxide ROOH R + HOO

HO ROOH RO + Scheme 1-1

2. Propagation

R + O2 ROO

ROO ++RH ROOH R

RO 2ROOH + ROO + H2O Scheme 1-2

3. Termination

2ROO ROH + O2 + R=O or ROOR +O2

ROO + R ROOR

2R R-R Scheme 1-3

RH = polymer, R• = polymer macroradical

Initiation of a polymer chain radical, or macroradical, occurs via the abstraction of a hydrogen from the carbon backbone by a radical species. Alternatively, initiation reactions can result from the cleavage of a hydroperoxide, which is itself an oxidation species. Subsequent attack by O2 on the macroradical results in the formation of a reactive hydroperoxide radical at the carbon centre. Oxidation can then spread to other polymer chains13. Radicals are inherently unstable, and will then terminate by creating hydroxyl groups, carbonyl functional groups or cross-links. It is not uncommon to see all these processes occurring simultaneously during the degradation of a polyolefin14.

1.1.2 Heterogeneous vs. homogenous thermooxidation kinetics The oxidation reactions presented in Section 1.1.1 are used in combination with chemical measurements, such as oxygen uptake, to develop models describing the kinetics of polymer degradation15,16. Ultimately, such kinetic models are developed to assist in polymer lifetime prediction studies17.

15 It has been convention to interpret polymer degradation in terms of the steady state approximation18. This is at least partially due to the use of oxygen uptake measurements, which is a bulk measurement. Oxygen uptake curves demonstrate linearity past the induction period of a polymer19, providing good correlation with the steady state approximation.

In addition to oxygen uptake measurements, kinetic information has conventionally been obtained from polymers in solution20. Thus complications such as radical mobility, polymer chain mobility, morphology, and oxygen diffusion limitation occurring in solid state systems cannot be correctly accounted for in a homogenous oxidation model10.

Chemiluminescence data was used by George and Celina13 to propose a heterogeneous oxidation model for the oxidation of polypropylene. Investigation of the oxidation of polypropylene powder at 150 °C revealed that a particle undergoing oxidation could infect a nearby stable particle. It was found that even after short oxidation times, oxidation products could be observed in localised zones, which were thought to exist around particles of residual catalyst. Furthermore, George and Celina postulated that the oxidation of polypropylene was heterogeneous even within amorphous regions of the films.

This view of localised oxidation zones on polypropylene films was used to explain the phenomenon of cracking in oxidised polypropylene. Once oxidation was initiated around a catalyst particle, an oxidation front was formed which progressed through amorphous regions, resulting in defects on a macroscopic scale. With further oxidation these linked defects formed cracks in the polymer surface. This explains why slightly oxidised polypropylene sheets demonstrated reduced tensile strength, despite only low concentrations of oxidation products. The effect of reduced tensile strength at low levels of oxidation has been well demonstrated in the literature21-23.

The concept of heterogeneous oxidation in solid state polymer films, which then leads to cracking, is fundamental to the chemistry underlying the experiments carried out in this thesis. The phenomenon of oxidation spreading through a

16 polymer in the solid state proposed by George and Celina can be exploited by the addition of chromophoric materials to enhance degradation. Similar to the spread of oxidation from catalyst residues, oxidation spreads from introduced photocatalysts to enhance degradation.

1.1.3 Photooxidation A great deal of research has been done on the photooxidation of polyolefins, and in particular polyethylene, over the last half century24-34. This section describes some fundamental photooxidation chemistry as described by recognised research leaders in this field.

Polymers containing only C-C, C-H and C-O single bonds are not expected to absorb in the UV wavelength range35. For such polymers to degrade photochemically a chromophore must be present. A chromophore might be an impurity which is chemically bonded to a polymer chain, either in the middle section, or at the end of a chain. Alternatively, a chromophore might be an impurity present as an occlusion and is not chemically bound, but is contained within the polymer matrix. A typical example is catalyst residues. Finally, a chromophore might be part of the polymer structure itself, such as double bonds, etc.. It is via these chromophores that photodegradation reactions are initiated.

The differences and similarities between thermooxidation and photooxidation have been studied for many decades. In 1954 Rugg et al.25 determined that thermooxidation of polyethylene resulted in little or no differences in the infrared absorption intensity of unsaturated moieties. Photooxidation however produced an overall increase in unsaturation, particularly in terminal vinyl group concentration, and internal double bonds. Side-chain methylene groups were found to decrease in concentration.

The degradation pathways favoured by polymers are typically reverse-analysed; information regarding the structure of oxidation products is obtained using conventional characterisation methods such as infrared spectroscopy, and from this the likely degradation pathway is deduced. It is critical, therefore, to

17 understand the relationship between degradation products and the process(es) that resulted in the products. One of the most indicative degradation products is the carbonyl group.

Carbonyl groups formed mid-chain, such as ketones, are generally the result of chain branching reactions, whereas terminal carbonyl groups, such as aldehydes, are a consequence of β–scission. Ketones can undergo reactions resulting in cleavage near the carbonyl bond via Norrish type I (resulting in two radical species) or Norrish type II (yielding a vinyl group and a ketone) reactions.

1. Norrish type I

hυ +

O O

+ CO

O Scheme 1-4

2. Norrish type II O H O hυ

O Scheme 1-5

Polymer conformation, the availability of γ-hydrogens, polymer mobility and other factors control the probability of Norrish type I and Norrish type II photoreactions. Below the glass transition (Tg) temperature the rate of formation

18 of Norrish type II depends on the ability to form the cyclic intermediate, and thus 35 the reaction is limited by the mobility of the polymer chains . Above Tg the mobility of the chains is such that the cyclic intermediate is no longer rate controlling and is kinetically similar to a polymer in solution. Below Tg the lack of chain mobility prevents separation of the Norrish type I radical species, and thus does not occur.

Allen and Edge11 describe the importance of carbonyl species in photodegradation of solid state polymers. Carbonyls are chromophores, and by absorbing UV radiation, the carbonyl oxygen can be promoted to an excited triplet state. This may be quenched by ground state molecular oxygen, resulting in a transfer of energy to the O2 molecule, giving an excited singlet oxygen. This reacts with unsaturated sites to produce hydroperoxides, according to:

3O hυ 2 1 O O* O + O2

1 O2 +

O2H Scheme 1-6

The exact significance of singlet oxygen in photooxidation of polymers such as polyethylene is still disputed. This is due to the fact that much of the experimental data comes from model system experiments, involving polymers above the Tg. Experimental evidence suggests that the above mechanism is inefficient in the absence of ketones, while others theorise various conflicting mechanisms for the above reaction to proceed. Clearly oxygen, photons, chromophores and unsaturation combine to result in oxidation; however the exact mechanism is unknown. It is possible that the many different mechanisms exist in competition with each other, and the many factors affecting oxidation such as temperature, incident radiation wavelength, presence and type of chromophores,

19 polymer chain mobility, etc.etera, determine the most likely degradation pathway.

Other species involved in photooxidation outlined by Allen and Edge are oxygen-polymer charge transfer complexes. Charge transfer complexes are used to describe an alternative pathway for the formation of hydroperoxides via attack by oxygen. Oxygen abstracts an electron from a hydrogen on the polymer backbone to generate an charge-separated complex. An intermediate of a polymer radical and hydroperoxide radical is formed, which recombine to give the final hydroperoxide. However, questions still remain regarding the efficiency of this reaction, while others argue that once an initial hydroperoxide if formed, oxygen-polymer charged transfer species are auto-catalysing35. It is likely that in processed polymers they have little practical significance compared to the effect of hydroperoxides36.

1.1.4 Role of hydroperoxides in polyethylene photooxidation In the 1980s Arnaud et al.37,38 produced some important papers regarding the photooxidation of polyethylene. It was found that most unsaturated groups formed by Norrish II reactions rapidly disappeared due to subsequent radical attack. Preferential oxidation sites were carbons in the α-position to the vinylidene. This was not true for the vinyl groups, and was believed to be due to low lability of the vinyl hydrogen. After an initial increase in the formation of vinyl groups upon exposure to UV radiation, the rate of vinyl group formation was found to parallel that of acid groups, indicating subsequent oxidation. Also, it was found that vinyl and vinylidene groups were competing for radicals during photooxidation.

In 1990 Gugumus39 suggested some novel reactions to explain the presence and relative concentrations of some degradation products of photooxidised LLDPE, as well as the lack of hydroperoxide accumulation in polyethylene when exposed to radiation. In contrast to much of the published literature, Gugumus suggested that the photolytic decomposition of hydroperoxide did not involve a radical

20 species. The proposed mechanisms involved a 6 membered transition state, as well as the evolution of water as a product.

Gugumus used these 6 membered transition state reactions to propose reactions that give vinyl, ketone and aldehyde products. Included as an example in Scheme 1-7 is the reaction between a hydroperoxide and polymeric carbon to yield a ketone.

C C C

O H O H H hν O

O HC O HC O + HC

H H H H H H

Scheme 1-7

The following year Lacoste et al.40 performed a similar study to Gugumus producing similar results; however Lacoste suggested already established mechanisms to explain the same degradation products. In order to simplify the reaction system, LLDPE samples were pre-oxidised by γ–radiation in air slightly to develop hydroperoxides, and then exposed to UV radiation in the absence of oxygen so that the degradation products of these hydroperoxides could be studied. Secondary hydroperoxides were formed and lost during 100 hours of irradiation. Carbonyl and free alcohol species increased in concentration. End carboxylic acid groups and esters also increased, along with γ–lactones. Some vinyl groups were initially lost, although a slight increase in trans-vinylene was found. Ketones were found to be created by Norrish type I and II cleavage reactions. In all cases Lacoste et al. used radical chemistry to explain the formation of oxidation products, give in Scheme 1-8.

21 ROOH heat or RO + OH light

RO + OH + RH R +H2O+ROH

R + O2 ROO

ROO + RH ROOH + R

2ROO ROH + R'C(=O)R" + O2 Scheme 1-8

It is apparent that there is not a single, elegant solution to describe the exact process of polyethylene photooxidation. Different oxidation products, in differing concentrations, result from different reaction conditions, and even manufacture of polyethylene41. The deeper one delves into the published literature, the deeper the divides in the opinion of the polymer degradation community become apparent. Conjecture and supposition regarding mechanisms are based on scientific evidence; it is the interpretation of experimental data that is likely to be debated for some time to come. It is helpful to consider the mechanism proposed by Tidjani42, proffering a simplified overview of the polyethylene photooxidation process, stemming from a widely accepted hydroperoxide intermediate.

22 H

radi c al atta c k C

OOH

hυ C OO H

H O C + OH C cage effect PH O O O O H

C C C C + P + H 2O OR C H H 2 H C OH + 2 Es ter 17 3 5cm- 1 A lc o hol 3400 cm-1

Norrish I RO or OH N orr is h II O

By following the various degradation pathways in the Tidjani degradation scheme, degradation products including esters, alcohols, acids, ketones and vinyl moieties are expected in photooxidised polyethylene. Although the exact mechanisms may not be fully agreed upon, it is clear that there is a relationship between the products, hydroperoxide intermediates and the effects of UV radiation absorption.

1.1.5 Stabilisation of commercial polyethylene As we have seen there has been a great deal of research devoted to understanding and establishing the degradation pathways of polyethylene. However for use in commercial applications, these degradation processes must be moderated for a polyolefin film to serve its intended purpose. Thus, antioxidant additives are included during processing to prolong the lifetime of polyethylene.

Antioxidants can be classified into two groups36:

• Chain breaking (CB) antioxidants, which trap radicals formed during the propagation step, and; • Preventive antioxidants, which stabilise hydroperoxide, effectively reducing the rate of initiation.

Chain breaking antioxidants are commonly added as stabilisers against thermooxidation, and are of particular importance for polyolefins due to high processing temperatures43. These antioxidants trap alkyl radicals, preventing further propagation reactions:

R + CB R-CB Scheme 1-10

A common chain breaking type antioxidant is Irganox 1010 pictured in Figure 1-1. Trapping of the alkyl radical occurs at the phenol. Irganox 1010 in particular has many industrial applications and is used by Ciba, whose films are used in this thesis. Some hindered amine stabilisers with multiple aromatic groups also provide UV stability, for example Tinuvin 327 and Chimassorb 81.

O OH

4 Figure 1-1 Irganox 1010

24 Some of the most effective preventive antioxidants are nickel dithiolate complexes, which remove hydroperoxy groups. Studies have shown that polymer hydroperoxides cannot be detected in polyethylene or polypropylene processed with nickel dithiolate complexes36. The mechanism for scavenging of hydroperoxide by the phosphate version of a dithiolate complex published by Scott in 198344 is included in Scheme 1-11.

Scheme 1-11

Another important class of hydroperoxide decomposing stabiliser is phosphite or phosphonite stabiliser45. Aryl phosphites, such as pictured in Scheme 1-12, demonstrate very efficient competition with polymer RH for chain propagating radicals. Alkyl phosphites are not used as stabilisers as the radical formed in the reduction of the phosphate radical is alkyl and will create further active radical species.

25 OOR ArO

P OAr +OOR ArO P OAr

ArO OAr O

ArO P OAr + RO

OAr

OR ArO

P OAr + RO ArO P OAr

ArO OAr

ArO

P OR + ArO ArO

ArO + OOR Inactive Products Scheme 1-12

Best stabilisation of polyethylene, and many other types of polyolefins for that matter, is achieved by combining both of these classes of stabiliser45,46. Typically, this includes high molecular mass or hindered amine light stabilisers (HALS), in combination with phosphites or phosphonites47. Thus chain breaking antioxidants are strongest during the early lifetime of the polymer, interrupting crosslinking reactions and competing with hydroperoxides, while preventive antioxidants compete with polymer chains for hydroperoxy radicals, helping to remove them from the system.

Although antioxidants provide a mechanism to prolong the lifetime of a polymer, especially by preventing degradation reactions during the melt, it is the object of this thesis to examine methods of accelerating oxidation reactions to produce a polymer with controllable degradation characteristics, as that is the ultimate goal of this work.

1.2 Prodegradants

The presence of foreign substances, such as metal ions, in a polymer matrix has a pronounced effect on the degradation of that material6,31,48-50. In 1970 May and Basharah51 listed the degradation reactions involving metal ions. These reactions were adapted from an earlier paper produced by Chalk and Smith in 195752, and are as follows:

hυ RH R + H

R + O2 ROO

ROO + RH ROOH + R

+ (n-1)+ ROOH + Mn+ ROO + H + M

ROOH + M(n-1)+ RO +MOH + n+

RO + RH ROH + R Scheme 1-13

It was found that the catalytic activity of the metals appeared to be related to their oxidation potential. The order of catalytic activity of the metals is Co > Fe > Ce = Cu > Mn = Pb > Zn > Ca. This implies that the electromotive force associated with reduction is related to the catalytic activity of the metals. Stabilisers such as metal deactivators can be added to the polymer to slow degradation. Metal deactivators, for example phenylamines, trap the metal ions, inhibiting their oxidative catalytic effect53.

There are various methods by which metals and/or metallic ions can be included in a polymeric material. The vast majority of commercially manufactured polymers contain metal ions as impurities from polymer catalysts11, and these impurities most often result in accelerated degradation of the material54. Metals can also be deliberately introduced in the form of ions or complexes as prodegradants to accelerate oxidation4-6,31,48,55,56. There are also cases where

27 metals can actually interact with the polymer system to work as oxidation retardants54, and are termed prohibitors.

In 1988 Osawa54 listed five different mechanisms by which a metallic compound can behave as a prodegradant in a polymer matrix. These reactions are an extension of the degradation reactions given by May and Bashara involving metal ions shown in Scheme 1-13.

1. Catalytic decomposition of hydroperoxides Metal ions can react with hydroperoxides to produce free radicals, according to the following reactions:

ROOH + Mn+ RO + M(n+1)+ + OH-

(n+1)+ n+ + ROOH + M ROO + M + H Scheme 1-14

The reactions are in reverse order to those given by May and Basharah in Scheme 1-13. However the process of metal catalysis and product formation can be summarised by:

Mn+/M(n+1)+ 2ROOH RO + ROO + H2O Scheme 1-15

2. Direct reaction with the substrate This results in the production of free radicals:

RH + MX2 R + MX + HX

RH + MX R + M + HX Scheme 1-16

3. Activation of oxygen Transition metals may interact with oxygen to produce a charge transfer complex, which can then create hydroperoxy radicals which react with the polymer:

n+ (n+1)+ M + O2 M + O2

+ O2 + H HO2 Scheme 1-17

4. Decomposition of a metallic compound Energy can initiate the decomposition of a metallic compound to produce a free radical, which can then go on to react with the polymer:

hυ MX M + X

RH + X R + HX Scheme 1-18

5. Photo-sensitising action An electron in the metal’s outer shell may be promoted from the ground state to an excited state by the absorption of radiation. Subsequent transfer of energy to the polymer molecule upon relaxation induces a radical.

hυ M M*

M* + RH M + RH*

RH* R + H Scheme 1-19

The last reaction type involves semi-conductors, and is often called semiconductor photocatalysis. A common semiconductor incorporated into polymeric systems is titanium dioxide.

1.2.1 Titanium dioxide Titanium dioxide is commonly used in polymer manufacture as a pigment, and made up about 60% of global pigment production in 200257. Titanium dioxide

(TiO2), or titania, exists in three different crystal lattice structures: rutile, anatase and brookite. Brookite is not commonly used due to its poor stability, and therefore the considerable majority of discussion found in the literature regarding the photoactivity of titania involves either anatase or rutile. Rutile is the most thermodynamically stable of these forms. Microparticle TiO2 powder is suitable for use as a white pigment due to its high refractive index and lack of absorption in the visible range of the spectrum between 380 nm and 700 nm wavelength.

Rutile TiO2 has a refractive index of 2.7, slightly higher than anatase at 2.55.

Anatase and rutile have numerous structural and functional differences. Commercially available anatase is typically less than 50 nm in size with the particles possessing a band gap of 3.2 eV, corresponding to a UV wavelength of 387 nm58. The adsorptive affinity of anatase for organic compounds is higher than that of rutile, and anatase exhibits lower rates of recombination in comparison to rutile. In contrast, the thermodynamically stable rutile phase generally contains particles larger than 200 nm with a smaller band-gap of 3.0 eV. The excitation wavelengths extend into the visible spectrum at 410 nm. Despite this, anatase is generally regarded as the more photochemically active phase, due to the combined effect of lower rates of recombination and higher surface adsorptive capacity59.

The different crystal faces of rutile and anatase titania influence the chemistry occurring at the surface of a titania particle9. The most thermally stable crystal face of rutile TiO2 is (110), depicted in Figure 1-2a. Anatase has two stable

30 surfaces, (101) and (001), of which (001) is the most common and is given in Figure 1-3a. The (100) face is less common in nanoparticles. Oxygen deficiencies in the rutile (110) of titania provide reaction sites for redox chemistry such as water cleavage and oxygen adsorption.

It is the photoactivity of titania nanoparticles that is a desired property when using titania as a prodegradant in polymeric materials. The reactions relating to the photochemistry of titania and its photocatalytic properties have been an increasing area of interest for some time56,60.

32 1.2.2 Titania photocatalysis Although titania had been well known as a white pigment in paint due to its reflective properties, it was not until the first half of the 20th century that research was first conducted into the phenomenon of paint chalking in sunlight9. Chalking is the appearance of white powder on the surface of paint, so named for its similarity with chalk. It was recognised that oxidation and reduction reactions were occurring simultaneously.

In their review of semiconductor photocatalysis Mills and Le Hunte49 group the terms ‘photocatalysis’, ‘photoinduced reaction’, ‘photoactivated reaction’ and ‘photosensitisation’, and define them as “a process by which a photochemical alteration occurs in one chemical species as a result of the initial absorption of radiation by another chemical species called the photosensitiser”.

If a semiconductor absorbs light of energy greater than the ∆E of the bandgap (Figure 1-4), an electron (e-) can be promoted from the valence band (HOMO, or Highest Occupied Molecular Orbital) to the conduction band (LUMO, or Lowest Unoccupied Molecular Orbital), creating a hole (h+) in the valence band. This electron-hole pair is termed an ‘exciton’. There are several possible outcomes of such a reaction, with simple recombination of the e- and h+ being the most common.

33 Atomic orbitals Molecule Cluster Q-size particle Semiconductor N=1 N=2 N=10 N=2000 N>>2000

LUMO

Energy

∆E ∆E ∆E ∆E

HOMO

Figure 1-4 The energy required to excite electrons from the ground state (HOMO) of a semi-conductor to the excited state (LUMO) decreases with increasing number of units N.

According to Scheme 1-20 and Scheme 1-21, if an electron acceptor or donor such as oxygen, water, hydrogen peroxide or organic molecule is present, the electron-hole pair may form a radical species instead of recombining:

Acceptor:

e+ O2 O2

e + OH H2O2 + OH

e + R + H RH Scheme 1-20

Donor:

h +OO2 2

+ OH h H2O + H

h + RH R + H Scheme 1-21

These reduction and oxidation reactions provide the radical species that can initiate degradation reactions in polymers. Holes are the primary oxidising

34 species in photocatalytic reactions9, reacting with water to produce hydroxyl radicals, and with organic molecules to produce carbon-centred radicals. Oxygen is an important species in these reactions, not only by accepting electrons to create superoxy radicals, but also by assisting charge separation, resulting in carbon-centred radical formation61.

1.2.3 Factors affecting titania activity in polymers Research into the photoactivity of titania in recent decades has uncovered aspects of titania chemistry that influence its activity in polymeric materials49,54,56,62. Among other factors, these include particle size63, crystalline structure64, phase composition65-67, surface area68,69, nature and concentration of lattice defects70, surface hydroxyl groups71-73, and impurities74,75.

Possibly the single biggest factor affecting the activity of titania is particle size. This is due to several reasons. The larger a titania particle, the relatively less surface area is available for reaction with oxygen, reducing its effectiveness. Also, larger particle size makes it easier to achieve hole – electron recombination.

Revisiting Figure 1-4, it can be seen that the more TiO2 molecules in a particle, the closer the gap between the HOMO and the LUMO. Thus ideally, to maximize the photoactive effect of titania in polymer systems for the purpose of degradation, it is more desirable to have nanoscale particles with good dispersion.

The different effects of micro- and nano-sized titania particles on the degradation of cumene as a model for polymer degradation was investigated by Allen et al. and reported in two papers76,77. A clear difference was noted between the activity of the two, with nano-sized particles considered to be the more active. Furthermore, nanoparticles were found to influence the degradation of a material much earlier, starting at the manufacture of the polymer.

Another detrimental effect of large particle size on the photocatalytic properties of titania is due to whitening; larger particles reflect UV light, which is demonstrated by titanias’ use as a stabilizer when used with microscale particle size.

In 2001 Cho and Choi78 established the difference between photolytic and photocatalytic degradation of PVC containing Degussa P25 TiO2. SEM images revealed that photocatalytic degradation occurred more rapidly in an area localized around the TiO2 particles, and photolytic degradation occurred at evenly distributed centers throughout the PVC matrix. Dispersion of the titania particles was a problem, and micrometer-sized agglomerates were reported to reduce significantly the photosensitizing effect of TiO2.

Particle size in a polymeric system is determined by the manufacture of the titania powder, and by agglomeration of the titania once it is mixed with molten polymer. The propensity of titania to agglomerate in polymers is well recognized, and researchers have looked to modify the surface of titania particles to achieve good particle separation, and thus small particle size once the polymer product is finalised.

1.2.4 Surface chemistry of titania The ease of molecule adsorption onto the surface of titania catalyst particles has a significant effect on the activity of titania. The surface of titania particles is highly heterogeneous, with anatase containing Lewis acid sites and several different forms of hydroxyl groups79. Anatase also has larger crystal faces than rutile, and is generally considered more suitable as a catalyst as it demonstrates higher adsorptivity80. In addition to hydroxyl groups on the surface of titania particles, defects must be present to trap oxygen to allow catalytic degradation of organic molecules81.

Degussa P25 titania is one of the most efficient and extensively used commercial photocatalysts available due to its high surface area, high photoactivity and minimal impurities81. The high activity of Degussa P25 is thought to be due to a more positive conduction band potential in rutile compared to anatase. This allows photogenerated electrons to pass from anatase to rutile, preventing recombination within the anatase. The removal of electrons by rutile is similar to the action of dopants, discussed below. Degussa P25 titania is formed as

36 nanoparticles, and thus has much greater surface area than most other forms of titania. This allows for significantly increased molecule adsorption rates.

The importance of water molecules trapped onto the surface of the titania particles is still open for debate82. Although it has been established that hydroxyl radicals form the main oxidising species during titania photocatalysis, there are several suggested pathways to creating such a species, which may or may not necessitate water. The evidence in the literature appears incomplete in either case, and more study is required before the role of water can be properly determined.

1.2.5 Surface modification of titania Unmodified nano-titania does not demonstrate good dispersion characteristics in polymeric systems due to titania-polymer interactions9. To improve the dispersion of titania nanoparticles researchers have modified the surface of titania by grafting polymers with better polymer miscibility properties onto the particle surface83-87.

This effectiveness of this approach is determined by the polarity of the functional groups present on the polymer chain; the grafting of a short-chain polymer with polar groups onto the surface of titania nanoparticles allows for improved particle dispersion in a polymer with polar functional groups87. An example of a silicon grafting agent ‘WD-70’ is given in Figure 1-5, used by Zan et al.85 in 2004 to improve the dispersion of nano-titania in polystyrene. Dispersion of the titania was reported to be successful, and the polystyrene with grafted TiO2 degraded at a faster rate than pure polystyrene. The titania particles used in this case were laboratory-prepared by Zan et al. with a size range of 70-100nm.

Figure 1-5 The structure of WD-70 which is grafted onto the surface of TiO2 particles by Zan et al.84 to improve titania dispersity in polystyrene.

Although there is some literature describing the grafting of short-chain polymers to provide electrostatic interactions with a polymer such as those mentioned, there is considerably less literature on surface modified titania for application as a photocatalyst in polyethylene.. The lack of scientific literature is presumably due to the difficulty of identifying an appropriate polymer for use as a grafting agent. The absence of functional groups in polyethylene chains disqualifies the use of grafting agents such as that presented in Figure 1-5.

In 1992 Allen et al.7 conducted a study on low density polyethylene (LDPE) films containing nine different types of titanium dioxide pigments (coated and uncoated, rutile and anatase titania). The nature of the coating was not divulged, except to say that it was organic in nature. It is presumed that the coating was designed to improve dispersion. Thermooxidative and photooxidative degradation was compared, and it was found that all the titania pigments acted as photosensitisers, with uncoated anatase and uncoated fine crystal rutile types being the most active. The significance of this result lies in that nature of the titanium dioxide. Even though the titania was pigment grade, manufactured to behave as a photostabiliser in other applications, it behaved as a prodegradant in LDPE. Additionally, surface modification to promote better dispersion had a negative effect on the photocatalytic properties of the titania particles.

It was also found that during thermooxidation, the rate of carbonyl formation as followed by IR techniques was less dependent on the nature of the particle as the temperature was increased. It was found that the role of titania particles was dependant on crystal size and structure, as well as the nature of surface treatment. The photocatalytic activity of Allen’s titania particles increased with decreasing particle size.

In 2006 Zhao et al.88 reported enhanced photooxidation of polyethylene containing 0.1 - 1% TiO2. Degussa P25 (a mixture of 75% anatase and 25% rutile titania) was used. The titania did not appear to be modified, and unfortunately the size of any agglomerated particles was not reported. SEM images showed creation of cavities in the polymer following irradiation, which were attributed to the evolution of small volatiles.

1.2.6 Doping Doping of titania involves the addition of an element or compound to titania which can enhance the quantum efficiency of electron-hole separation. Elements used in doping depend on the application, and are often used to enhance the photoactivity of titania under visible light89-99.

Vanadium doped titania photocatalysts 98,100-104 are a good example of the mechanism of doped titania photocatalysis. Martin et al.98 reported on the mechanism of quantum sized vanadium doped titania nanoparticles in the oxidation of the dye 4-chlorophenol. The 1-5 nm sized particles synthesized in the laboratory aggregated to 50 µm size particles; however each crystallite was reported to be electronically isolated.

H - O O O O O OH

V V V

O O O O O O

Titania particle surface Figure 1-6 Mechanism for charge separation of vanadium doped titania.

Figure 1-6 shows the mechanism given by Martin et al.98 for charge separation + by V(V) (VO2 ) doped titania. V(V) attached to the surface of a titania particle can abstract a hydrogen from an organic molecule. When applied to polymer degradation, a carbon centered radical formed by hydrogen abstraction can lead to hydroperoxide formation, according to mechanisms discussed earlier.

In this thesis the photocatalytic effect of titania doped with antinomy is reported. There is apparently no literature discussing such a system, however it is assumed that antinomy is present to assist in charge-carrier separation similarly to the example of vanadium given here.

1.2.7 Effect of UVA vs. UVC radiation on polymer – TiO2 systems Allen has contributed greatly to the knowledge of the activity of titania in polymeric materials7,11,62,76,77,105-113. However, in 1996 Allen and Katami110 found an unusual result when they conducted a comparison of aging conditions on the degradation of linear low density polyethylene films containing titania. Using a narrow band 365 nm radiation source all types of titania studied, except heavily coated rutile particles, acted as photosensitisers. However, with narrow band 254 nm irradiation all types of titania acted as UV screeners and stabilised the polymer. This was explained by the penetration depth of the higher-energy wavelength light. The 365 nm radiation was said to be absorbed closer to the surface of the titania particles. However, the 254 nm exhibited "deep crystal lattice penetration", and the surface functional groups of the titania particles were not activated. It was concluded that the thermal and photo generation of active carriers on the surface of pigment particles strongly influences the photoactivity of titania particles.

Although Allen has reported several times the photostabilising effect of titania under 254 nm radiation76,77,108,110,114, it appears that there has been limited supporting experimental evidence. Only two papers77,108 show different experiments that support the theory of deep crystal lattice penetration. Seeing that many different factors, such as particle size, temperature, etc., affect the role that titania plays in polymer degradation, it is reasonable to suggest that more experimental evidence is necessary to corroborate this theory.

In 2006 Zan et al.115 developed a low density polyethylene film incorporating titania nanoparticles via a melt-blending technique. Degussa P25 titania was employed, and an irradiation source of 254 nm was chosen. The result of

40 photocatalytic degradation was investigated via several established analytical techniques including IR and SEM imaging. The titania was found to act as a photosensitiser, and the polymer weight loss following degradation was found to be much higher than for pure polyethylene degradation.

This result can be considered significant, as Allen had found that titania acted as a mild photostabiliser under 254 nm radiation. This was attributed to the depth of penetration of incident radiation into the crystal lattice. However, Zan et al. found that degradation occurred at 254 nm faster than in sunlight, which was accelerated compared to pure polyethylene. Both researchers used similar materials under comparable conditions. The difference, however, was the titania used. Zan et al. used Degussa P25, a nano sized particle, while Allen used a laboratory prepared experimental grade titania particle.

As stated by Allen et al.105 and Mills and Hunte49, the manufacture history plays a large a role in the photoactivity of the titania particles. It is possible that the particle size of Allen’s titania affected the degradation kinetics to a greater extent than the wavelength of the incident radiation. In all likelihood it may have been a combination of both factors, as the 254 nm radiation penetrated deeper than 300+ nm light into the crystal lattice structure, and the larger particle size further restricted the activation of surface groups.

1.2.8 Summary of sections 1.1 and 1.2 The importance of polyethylene as a commodity plastic has resulted in the thermooxidation and photooxidation degradation pathways being well studied for many decades. Polyethylene degradation is driven by radical reactions, eventually giving rise to chain scission, crosslinking of the polymer chains, and the development of oxygenated functional groups.

It has been revealed that degradation of polymers in the solid state is not a homogenous process. Oxidation occurs at points in the polymer, such as metal catalyst residues, and spreads to other areas of the bulk, eventually resulting in cracks and the loss of mechanical properties. This phenomenon can be exploited

41 by the addition of prodegradant materials to enhance the oxidation process. Among potential prodegradant materials titania has achieved noteworthy status, being both highly photoactive and economically viable for commercial applications.

Producing a polyethylene film with controllable degradation properties is not as straightforward as merely adding titania powder to a melt. This is due to the many factors affecting the activity of titania in polymers, including particle aggregation, particle size, surface properties, crystal phase, etc.. Researchers have attempted to enhance the activity of titania by surface modification, and addition of dopants to improve the separation of electrons and holes.

Although the activity of titania under UVA irradiation is well characterised, there is some contention regarding the effect of UVC on a polymer/titania composite material, especially where titania particles are between pigment and nano-sized. Additionally, although these materials have been well studied under conditions of constant irradiation, it is unknown whether degradation reactions initiated by titania with UV irradiation will continue to occur rapidly in the dark, such as might be expected in the lifecycle of waste plastic packaging, or shopping bags.

1.3 Polymer degradation characterisation techniques

There are various approaches employed by polymer researchers to explore, visualize and quantify physical and chemical aspects of polymer degradation occurring in the solid state. The applicability of a characterization technique to a given problem is determined by the nature of the information that is sought. In order to determine degradation pathways for, say, the design of new antioxidants, FTIR and NMR provide detailed information regarding molecular structure of oxidation products. Alternatively tensile strength testing is better suited to quantifying the physical effects of oxidation on the mechanical properties of a polymer film.

Areas of interest can be broadly grouped into three categories; physical, or bulk, effects; surface characteristics; and molecular structural information. A

42 combination of these aspects of polymer degradation provides an overall description of the entire degradation process.

1.3.1 Characterization of the bulk via physical tests

1.3.1.1 Elongation at break Although bulk testing is more commonly employed by commercial assessors in industry116, it is also a powerful tool for researchers in polymer degradation, due to its ability to detect early signs of degradation117. Elongation at break tests involve measuring the strain required to break a piece of film of predetermined dimensions. Madfa et al.118 demonstrated the applicability of elongation at break testing to LDPE films that had been exposed to natural weathering. Tests showed that cross-linking induced by radiation absorption became significant after just one week of weathering. The materials used in this experiment were commercially produced; however they wholly degraded after just 4 months of weathering.

Roy et al.55,119-123 employs elongation at break tests among several other characterization techniques when investigating accelerated polyethylene degradation. His results demonstrate not only decreased elongation at break at early stages of oxidation, but also that increasing the concentration of prodegradant (typically cobalt stearate) have a negative effect on the mechanical properties.

1.3.2 Surface Characterisation For solid state polymers exposed to oxidative environments such as natural weathering or accelerated conditions it is expected that degradation will occur initially at the surface124. There are numerous methods in existence for examining the surface of degraded polymers, of which the most common technique is scanning electron microscopy (SEM).

SEM provides physical information from the surface of degraded polymers, such the appearance of cracks and voids125,126, and the influence of domains in

43 polymer blends 127. It is particularly useful in characterising the surface of degraded polymers containing transition metal photocatalysts due to its inherent image contrast between metals and organic compounds. Thus agglomeration and distribution of transition metal photocatalysts128 and the immediate environment surrounding these photocatalysts can be clearly observed.

Zhao et al.88 provided some excellent SEM images demonstrating the appearance of cavities around titania particles in polyethylene following UV irradiation (Figure 1-7). In these images the titania particles are difficult to detect, however as demonstrated later in this thesis backscattered images show the location of titania particles quite clearly.

One of the potential drawbacks to SEM imaging is the loss of sample, as the sample must be coated with carbon or gold prior to examination. However the technique does not require a large sample, and as the sample is already degraded, sample loss does not generally constitute a significant issue when characterizing the surface of degraded polyolefins with SEM.

1.3.3 Chemical Characterisation

1.3.3.1 Oxygen uptake As stated earlier, oxygen uptake measurements are used to determine kinetic information regarding degradation reactions. Zeynalov and Allen112,113,129-131 used oxygen uptake measurements to examine the effects of antioxidants and

44 prodegradants on the degradation kinetics polymers, using the model compound cumene. Oxygen uptake measurements were used to obtain information such as the rate of radical scavenging by the antioxidants, nature of the rate dependence on the concentration of inhibitors and the activity of different phases of titania nanoparticles.

Although oxygen uptake measurements are useful for investigation of the kinetics of oxidative degradation, other techniques provide more accurate chemical information regarding the nature of degradation products, which can then be used to determine degradation pathways.

1.3.3.2 Nuclear Magnetic Resonance Nuclear Magnetic Resonance (NMR) allows the polymer chemist to investigate the properties of a degraded polymer sample on a molecular scale, yielding information regarding carbon hybridization132, chain scission and cross-linking phenomena133,134, polymer chain mobility135-137 and morphology138. NMR can be used at high field frequency to detect specific molecular changes, and at low power can detect degradation related changes in the bulk sample139.

NMR is not as well represented in the literature as other characterisation tools, such as FTIR, for polyolefin degradation. There are several possible reasons for this, including the lack of spatial information, the relative difficulty in obtaining solid state NMR spectra compared with other characterisation techniques, and the nature of the information obtained. Additionally, NMR is a destructive technique, making it difficult to obtain information from bulky samples.

1.3.3.3 Mid-Infrared Spectroscopic Techniques Vibrational spectroscopic techniques have been used for many decades in the characterisation of polymers, and the characterisation of polymer degradation products. The most important of these techniques has been mid-Infrared (IR) spectroscopy, for its ease of application to many polymeric materials, and the type and quality of information obtained. Mid-infrared instruments usually

45 examine the wavelength range from 2.5 µm to ~17 µm (4000 cm-1 to ~600 cm-1). Of IR techniques, transmission has been in use the longest.

Transmission IR Many journals and books have been written discussing the application of transmission FTIR for characterisation of polymeric materials, and of degradation related products. In the 1950s Rugg et al. investigated polyethylene structure24 and oxidation products25 using IR spectroscopy. Much of the work performed by Rugg et al. is still used in the literature today, demonstrating the reliability and reproducibility of transmission IR. Transmission IR can also be used to quantitatively examine oxidation products of polyolefins140, although in more recent times emission FTIR has also been demonstrated to be appropriate to the task141.

Transmission IR can pose some challenges in the study of thin films. If the film in question is too thick, then over-absorption can occur, distorting the spectrum by cutting off the top of the CH2 absorption peaks of a polyolefin. At an absorbance value of 2 or higher less than 1% transmission occurs and detectors become increasing unreliable. However, when examining oxidation products this is not usually a problem, as over-absorption can actually be used to enlarge weak absorptions that might otherwise be difficult to observe clearly. This technique can be used to examine the carbonyl and fingerprint regions at early stages of oxidation. If over absorption is unwanted it can often be overcome by cutting off a thinner section of the polymer for examination.

More challenging than over-absorption of a film is the issue of interference fringes. An interference fringe is described as a sinusoidal intensity variation due to interference of radiation that undergoes multiple reflection between two flat and parallel surfaces142. Figure 1-8 shows an IR transmission spectrum with an interference fringe (upper plot). It can be seen in Figure 1-9 that light reflected at the lower n1/n2 interface may reflect again inside the polymer film on the opposite interface, and upon passing through the sample will interfere constructively and destructively in a sinusoidal fashion (depending on the

46 wavelength) with the transmitted beam, resulting in the superimposition of a sine wave on the final spectrum.

Figure 1-8 Avoidance of interference fringes by using polarised light at the Brewster angle143.

Incident IR beam

n2 (air)

Reflection at interface n1 (polymer film)

n2 (air)

Sinusoidally interfering waves Figure 1-9 Schematic demonstrating the interference fringe phenomenon, which causes the superimposition of a sinusoidal wave over a transmission IR spectrum.

For interference fringes to occur in the IR spectra of polymer films the surfaces must be flat, parallel, and spaced the correct distance apart (film thickness between 5 µm and 2.5 mm)144. Although the intensity and position of the polymer absorption bands remain unchanged in the spectrum, interference fringes can complicate the spectrum by making small peaks difficult to interpret as they can lie of the shoulder of a sine wave. Additionally, spectral comparison methods such as overlapping, spectral subtraction and curve fitting become very difficult to achieve reliably. When investigating early stages of oxidation with IR transmission methods interference fringes can pose more than a mere nuisance.

There are some methods to alleviate the problem of interference fringes in transmission spectra. One of the most effective methods is use of the Brewster angle, suggested by Harrick143 in 1976. This involves orientation of the polymer film at the Brewster angle with respect to the incident light. A demonstration of this is included in Figure 1-8, where a transmission spectrum of polyester film has had interference fringes avoided by use of the Brewster angle and polarised light.

This method would be expected to be less effective in studying commercial polyethylene films due to crystal orientation of the polyethylene chains during film blowing145, which would yield spectral results dependant on film orientation during sampling146,147. Other methods for reducing fringes include scratching the surface of the polymer with steel wool to create a rough surface, and clamping the film between two mid-IR transparent windows144. However neither of these techniques is applicable to the study of degraded polyolefin films.

The spacing of peaks in interference fringes can be used to determine the thickness of the film under investigation, if the refractive index of the material is known144. The thickness is calculated by counting the number of waves over a wavenumber range, according to:

∆n t = 2(ν2−ν1)η Equation 1 Where: t = thickness ∆n = number of waves in spectral range

ν2−ν1 = spectral range (wavenumbers) η = refractive index of film

ATR/FTIR Attenuated Total Reflectance FTIR (ATR/FTIR) measures the near surface layer144, and has demonstrated suitability to the investigation of surface degradation of polyolefin films148-150. A comparison of the vibrational spectroscopic techniques transmission IR, emission IR and ATR/FTIR by Delor et al.151 concluded that ATR/FTIR is a reliable method for the study of the evolution of degradation products of elastomers by infrared spectroscopy.

ATR/FTIR is an internal reflection technique144 that uses the optical principle of light passing through a medium of high refractive index internally reflecting when impinging on a surface of lower refractive index at an angle less than the

‘critical angle’. The critical angle, θc, can be described as the threshold angle below which light will internally reflect at a boundary between two media, and is given by:

η2 sin θc = η1 Equation 2 Where: η2 = lower refractive index

η1 = higher refractive index

In ATR/FTIR internal reflection is achieved by placing a crystal that is transparent to mid-IR, and possessing a high refractive index, in optical contact with the sample (Figure 1-10). When light is internally reflected at the crystal/sample boundary, evanescent mid-IR waves can be absorbed by the

49 samples molecules. The resulting signal is Fourier transformed to produce an infrared spectrum. ATR/FTIR crystals can be multi-bounce or single bounce, depending on the size of sampling area sought, and the strength of the signal required.

Incident mid-IR Exiting mid-IR

Internal ref lection element (IRE)

Sample Figure 1-10 Schematic of multi-bounce ATR/FTIR.

As ATR/FTIR is a semi-surface technique, it is essential to know the depth of sample being described by the spectrum obtained. The depth of penetration of light is dependant the relative difference in refractive index between the IRE and the sample, and on the wavelength of the radiation. As a mid-infrared spectrum is obtained over a range of wavelengths, the depth of penetration will vary over the spectrum, according to the Harrick equation:

λ dp = 2πn (sin2 θ - n2 )1/2 1 21

(12)

Where: dp = depth of penetration λ = wavelength (nm)

n1 = refractive index of IRE

n21 = ratio of refractive index of sample/objective θ = angle of incidence

50 This raises some important factors that must be considered when viewing an ATR/FTIR spectrum. Firstly, higher energy, shorter wavelength light (high- wavenumber end of the spectrum) penetrates less into the sample, giving a weaker absorbance than the low wavenumber end of the spectrum. Software used for the manipulation of spectra usually includes an ATR correction formula that will balance this disparity in absorption intensity caused by the difference in penetration depth with wavenumber. Also, the depth of penetration is dependent on the refractive index of the material and the IRE. Naturally, the refractive index of the sample cannot be changed, however by using an IRE with a relatively high refractive index (germanium for example has a refractive index of 4 and is transparent to mid-IR), the depth of penetration can be lowered significantly151,152.

In this thesis the low wavenumber end of the spectrum (down from approximately 1800 cm-1) is the region of greatest interest, as it contains chemical information relating to degradation. ATR correction formulas change the scaling by decreasing the relative absorption strength of bands in this region, which is undesirable in this instance as this the region contains the most relevant information. Especially at low levels of oxidation, ATR correction would cause already difficult to detect changes to be more difficult to observe. Thus spectra obtained using ATR/FTIR methods in this thesis have not undergone an ATR correction, but are analysed in all cases as the raw data obtained when the spectra were acquired.

1.3.3.4 Polyethylene absorptions in the mid-IR Different types of polyethylene (high density, low density, branched, etc.) display different absorption bands in the mid-IR153. Transmission IR spectra of thick polyethylene specimens will exhibit over-absorption of the main C-H vibrations, allowing weaker, skeletal vibrations to be more easily identified.

High density polyethylene (HDPE) contains less –CH3 groups than other forms of polyethylene, and thus has a relatively weaker symmetrical bend C-CH3 absorption at 1378 cm-1. Additionally, HDPE has a higher vinyl unsaturation

51 content, absorbing at 910 and 990 cm-1. In contrast low density polyethylene (LDPE) and in particular linear low density polyethylene (LLDPE) have a much -1 higher –CH3 content, and thus a stronger C–CH3 at 1378 cm . Due to the higher numbers of short chain branches in LLDPE, there is a high content of vinylidene (pendant methylene, -1 >CH=CH2), absorbing at 888 cm .

Crystallinity of the polymer also affects the mid-IR spectrum. A band at 1303 cm-1 increases with increasing amorphous content, while sharp absorptions at 1175 and 1050 cm-1 increasing with higher crystalline content.

1.3.3.5 Depth Profiling by ATR/FTIR Transmission IR experiments can be performed by coupling an IR microscope to an FTIR spectrometer154. Similarly, micro-ATR/FTIR experiments can be carried out by using an ATR objective on the microscope155. An aperture can be adjusted on a micro-ATR/FTIR spectrometer to measure an area on the sample smaller than the size of the contact surface. Thus micro-ATR/FTIR allows for improved lateral resolution, but at the cost of signal/noise ratio, as there is less reflected light received by the detector.

Do et al.152 performed an experiment to deduce an optimum aperture size for the measurement of some carbon-filled polymeric materials. It was found that the minimum aperture setting, which allowed for a best possible compromise between signal/noise ratio and collection time, was 40 µm. The authors152 then add that the actual spatial resolution being achieved when using micro- ATR/FTIR can be determined by dividing the aperture size by the refractive index of the crystal. In this case it was found to be around 12 µm, which is approximately the diffraction limit of infrared radiation.

Line-mapping by micro-ATR/FTIR was demonstrated to produce spectra that could be used to obtain oxidation profiles of a cross-sectioned surface of a polymer sample156. These profiles evidenced higher levels of degradation towards the exposed edge of the rubber than compared with the centre. The

52 authors152 concluded that micro-ATR/FTIR was a technique suitable to the study of polymeric materials, and good quality, low noise spectra could be obtained using a silicon IRE in just a few minutes.

1.3.4 Achieving high lateral resolution With respect to the heterogeneous degradation processes occurring in solid polymeric materials (Section 1.1.2), there is a clear advantage to be able to view degradation processes with respect to the dimensions of the material in question. Additionally, the inclusion of natural (crystalline or amorphous regions) and introduced (prodegradants, inhibitors) heterogeneities within polyolefins heighten the need for spatial information. Lateral (commonly referred to as spatial) resolution refers to the smallest distance at which two objects can be distinguished, and recent developments in IR technology has seen lateral resolution reduced to below the wavelength of IR light in an air medium. Imaging ATR/FTIR is a technique that has achieved considerable success in increasing the spatial resolution available to IR spectroscopists.

1.3.4.1 Imaging ATR/FTIR There are two primary methods of image acquisition that are of particular interest to polymer chemists157. When performing projection imaging, an area of interest is selected and uniformly illuminated by a broad beam. Reflected or transmitted radiation from the specimen is directed back via a system of optics to an arrayed set of detectors, known as a focal plane array detector. Scanning imaging involves moving the sample or detector such that different areas on the specimen surface are sampled in a raster pattern. In both methods, the signal is received by the detector and reconstructed to give chemical and spatial information.

Imaging detectors are typically constructed of an m x n array of detectors. An optical signal impinges on the detector related to a point on the object, and if the radiation is sufficiently large it will produce a current, I(t), that flows through a load resistor, RL, and produces a voltage, v(t), according to:

53 v(t) = I(t)RL = [S(t) * h(t)]RRL Equation 3

Where: S(t) = intensity envelope of the optical signal h(t) = impulse response function of the detector R = responsivity (ratio of photosignal to radiation power incident on detector) specified in units of amperes per watt * = denotes convolution

Ideally, detectors should have a high signal-to-noise ratio (S/N), and a high spectral response. S/N is often limited by the sensitivity of the detector, which can be described in terms of the current produced per unit of incident radiation. Signal received by the detector is considered noise if it does not originate at the conjugate object point. Spectral response refers to the range of wavenumbers over which the detector will produce useful information. It is often necessary to sacrifice spectral response in order to increase the S/N of a focal plane array detector. MCT (HgCdTe) detectors are commonly used in FTIR instruments to improve the S/N.

MCT detectors offer greater sensitivity when using ATR/FTIR. FPA detectors consist of a 2-dimensional square array of MCT detectors, usually in the order of 64 x 64, or 128 x 128 detectors. One of the drawbacks of using such a sensitive photovoltaic detector is its ease of saturation – large amounts of incident radiation quickly overwhelm the detector. Furthermore, in order to achieve maximum spectral response with low levels of incident radiation, it must be used at very low temperatures. This problem is resolved by operating the detector with liquid nitrogen cooling.

High spatial resolution is important in imaging studies in order to examine as small an area as possible. As spectroscopic measurements are determined using photons, spatial coherence of the photons limits the spatial resolution. Spatial coherence is distance below which the interference of the harmonic signal is constructive. This is inversely related to the wavelength, according to:

K = 2π/λ Equation 4

Where: K = wave vector λ = wavelength

In imaging spectrometers, pixel resolution plays a role in determining the overall system resolution. One pixel measures a fixed width, and pixel resolution is given as the length of an image in one direction divided by the number of pixels in that direction.

In recent times, the use of µ-ATR/FTIR imaging has introduced some exciting improvements to the achievable spatial resolution of IR imaging spectrometers158-161. Chan and Kazarian published new findings in 2003, reporting the achievement of spatial resolution of 3-4 µm using µ-ATR/FTIR with a Ge IRE. This is a further improvement on a spatial resolution of 8 µm achieved by Sommer et al.162 in 2001.

Chan and Kazarian used stringent criteria when determining lateral resolution159,162. A chemically heterogeneous polymer sample, consisting of poly(methyl methacrylate) (PMMA) patterned using electron beam lithography and fixed on a silicon wafer, was examined using µ-ATR/FTIR. Two different areas (in this case, clean PMMA vs lithographed PMMA) were considered to be resolved when the spectra showed a 5-95% absorbance profile as a function of distance, i.e. when the spectra changed from showing 5% of one component to 95%. This was achieved with a dimension limit of 4 µm. The change in wavelength of light when passing through media of high refractive index is responsible for such an achievement. The relationship between the refractive index and wavelength of light is given as:

55 λ n = 0 λ n Equation 5 Where: n = refractive index

λ0 = wavelength of light in a vacuum

λn = wavelength of light in a medium of refractive index n

The wavelength and velocity of light both decrease with increasing refractive index of the medium..The shorter wavelength of light when applying µ- ATR/FTIR techniques with a Ge IRE (RI = 4.0) allows for much greater spatial resolution than using, say, transmission IR, where the light passes through a medium of air. The application of this principle allowed Chan and Kazarian159 to obtain spatial resolution higher than had previously been reported.

1.3.4.2 Synchrotron radiation source While a conventional FTIR microspectroscopy cannot use an aperture smaller than approximately 20 µm, due mainly to S/N restrictions, a synchrotron light source is powerful enough to achieve a reasonable signal with a much smaller aperture. A synchrotron light source is some 300 times brighter than a light source on a conventional IR spectrometer163. However, this translates to an improvement of brightness of 3 orders of magnitude when light is projected through a pinhole of 10 µm diameter, resulting in greatly improved S/N for MCT detectors.

Despite the significant attenuation of the light source, spatial resolution of 6 µm has been reported in the literature164. Many of the applications of synchrotron FTIR studies are biological in nature, examining organic materials such as tissues165 and fungi166, where improved spatial resolution has been beneficial, and imaging ATR/FTIR was not practical.

Utilisation of a synchrotron light source for the investigation of polymer degradation has not been popular with research scientists. Although little information can be seen in the literature, an article has been by published by

56 Wetzel and Carter167 who examined the degradation of acrylic polymer automotive coatings using a synchrotron light source in 1998. The coating was cross sectioned and spectra obtained in 1 µm steps in transmission mode. Although this paper demonstrated the applicability of microspectroscopy with a synchrotron light source to obtain spectral information with high spatial information it is uncertain whether information was obtained that could not have been obtained via imaging ATR/FTIR.

Despite the improved spatial resolution of FTIR microspectroscopy using a synchrotron light source over conventional FTIR microspectroscopy, the spatial resolution obtainable is still slightly inferior to that of an imaging ATR/FTIR system. However, as not all samples are suitable to examination by imaging ATR/FTIR, there is a range of applications for both methods.

1.3.5 Characterisation techniques used in this thesis There is a broad range of characterisation techniques that have some relevance to the investigation of the degradation of polyethylene film containing titania. However, some methods provide more pertinent data than others. In particular, IR spectroscopy has a long history in this field, and the information regarding degradation-related functional groups has been well studied. Furthermore IR spectroscopy allows the researcher to link the products with a degradation pathway.

Other chemical characterisation techniques are less useful for this study. Raman spectroscopy provides better lateral resolution, however it is less sensitive to oxygen containing functional groups compared to IR spectroscopy. Solid-state NMR requires destruction of the sample, and the spectra are more difficult to interpret with a broad range of oxidation products in low concentrations. X-ray photoelectron spectrsocopy (XPS) examines the surface layer of the sample, whereas when examining the films used in this thesis information was sought from deeper than the surface of the material. ATR/FTIR penetrates up to 3 microns into the sample, which examines that part of the sample with the highest concentration of oxidation products. Oxygen uptake measurements provide

57 information regarding the rate of oxidation; however this has been achievable using IR spectroscopy. SEM and back-scattered SEM images provide information regarding the dispersion of titania particles and the physical environment around the particles, and was used for the purpose of characterisation.

There are also mechanical methods of determining the extent of degradation of polyethylene film, such as stress-at-break measurements. However, the film was not required to degrade until it had lost its useful mechanical properties, but was required to break up into fine particles to allow for microbiotic action to render the polymer environmentally neutral. Thus all films were degraded until embrittlement, such that they could no longer be held in their sample holders. In this situation it was decided that mechanical measurements would not present information pertinent to the investigation, and so were not conducted.

It was decided therefore that IR spectroscopy in various forms would be the main characterisation tool in this study principally because of its non-destructive nature as well as the high level of molecular structural information that it provides.

1.4 Objectives

There is a strong demand from industry to develop an environmentally neutral commodity plastic film with controllable degradation qualities for applications such as shopping bags, packaging, agricultural film, etc.168. LLDPE is a natural choice for such a material as it has proven applications in these fields, and its degradation chemistry has been well studied in the literature23,26,27,50,56,120,169.

The cost of the final product is an important consideration, as this research is being performed with a view to potential commercial applications. With this in mind LLDPE film blown by Ciba containing commercially available types of titania from different manufacturers has been chosen as the subject material. Although there is published research regarding the effect of titania on the degradation of polyethylene88,107,115,170,171, these studies concern mainly

58 polyethylene and/or titania manufactured in a research science laboratory, and therefore these results may not be directly transferable to commercial applications. The research undertaken in this thesis will be directly applicable to the development of environmentally neutral films.

Titania particle size has been demonstrated to strongly affect the photosensitivity of polymer-titania composite systems63. Although nanoparticles of titania are more photoactive then pigment grade titania110, they have a greater tendency to agglomerate9,63, reducing their effectiveness. The surface of titania particles has been modified by researchers to encourage better dispersion83,84,86,87, however, surface modification not only reduces the photoactivity87, but also increases the cost of a commercial material.

To date there has not been a published study on the effects of UV irradiation or thermal oxidation of the commercial films used in this thesis. For films used in agriculture, it is important to know how long these films can be expected to last when exposed to sunlight. Additionally titania from different manufacturers has been used in the films, and it would be useful to establish a relative order of photoactivity of these titanias in polyethylene. Therefore examining the degradation under simulated solar irradiation will be performed extensively to determine the activity of titania.

Although the topic is debated among prominent polymer degradation research scientists39,40, the oxidative degradation of polyethylene is most likely radical based35, whereby the formation of carbon centred radicals leads to hydroperoxides, resulting in degradation products. Therefore by controlling the radical population in a polyolefin one can control, to a greater or lesser extent, the rate of degradation. This principle has been used in the development of radical scavenging antioxidants36,43, used to prolong the lifetime of a polyolefin.

When a titania particle absorbs UV radiation an electron/hole pair is created9,49. If they do not recombine, it is possible for these species to migrate to a nearby organic molecule, creating a carbon centred radical9,49. In a polyolefin such a radical species can then react with oxygen to form a hydroperoxide macroradical,

59 which can subsequently infect the polymer10. From such infection sites in a polymer degradation can be seen to spread throughout the bulk13.

The main objective of this thesis is to exploit this phenomenon to control the degradation of polyethylene-titania composite material, even in the dark. Radical species will be initiated in polyethylene-titania film via exposure to UV pre- irradiation before being subjected to accelerated degradation conditions. It is anticipated that oxidation will spread from these sites of infection, resulting in degradation of the bulk polymer film. The concept of pre-irradiation with UV to control subsequent degradation of polyolefin film in the dark is a novel approach, not yet seen in scientific literature. In order for the final film to have a viable commercial application, it is desired that the polymer film should break down completely in the dark following pre-irradiation after approximately 6 months.

Although a good deal is known about the degradation of LLDPE, there is little information regarding the chemistry of the environment surrounding a titania particle during degradation. If more fundamental knowledge regarding the degradation pathways and chemical species surrounding the particle was available it is hoped that this could lead to refining the composite material to create one with strictly controllable properties. State of the art IR techniques allowing improved lateral resolution will be assessed to determine their applicability to the study of the degradation around titania particles, especially in the early stages of oxidation.

The parameters to be studied include pre-irradiation wavelength, exposure time, and subsequent degradation conditions, which need to be understood in order to further develop the technology. For example there are reports of titania irradiated with UVC acting as a stabiliser110, while others have found it has a photosensitising effect115. Such issues need to be clarified to determine the optimum conditions for pre-irradiation. Also the difference in activity between commercially available grades of titania needs to be investigated, along with the effects of surface modification and doping.

60 This technology would have a clear application for commodities such as shopping bag film, whereby careful pre-irradiation of a polyethylene-titania composite film prior to use could result in shopping bags that will degrade in the dark. Packaging film could be pre-irradiated so that once the expected useable lifetime of the film has expired, it can be disposed of, and will degrade even without further exposure to sunlight. It is hoped in the near future we will be using films that degrade controllably according to the demands of the application in an environmentally neutral manner.

61 62 Experimental

2.1 Ciba films investigation

10 samples of polyethylene film were obtained from Ciba AG (Postfach Schwarzwaldallee 215, Basel, Switzerland). The sheets ranged between 22 and 27 µm in thickness and were composed of Dowlex LLDPE. Dowlex is described on the Dow Plastics company website172:

“Next Generation DOWLEX* NG 5056 E and 5056.01 E are ethylene octene-1 copolymers specifically designed for use in blown film applications requiring the finished film to show high impact strength and exceptional optical properties, as well as good retention of properties at low temperatures. Typical applications of use include lamination, bag-in-box liners and form-fill-seal packaging of frozen vegetables and liquid foods. DOWLEX NG 5056.01 E is the slip and anti-block version of the resin.” Dowlex 5056 has a melt index of 1.1 and a density of 0.919.

Dowlex 5056 contains the synergistic stabilisers Irgafos 168 (phosphate), and hindered phenolic type stabilisers Irganox 10176, Irganox 1010 and Irganox 1330. The stabilisers are present in concentrations of less than 0.5% w/w.

Nine of the polyethylene samples contained titania, and one control without titania. A list describing the titania in the LLDPE film was supplied by Ciba to accompany the samples and is in a modified version is provided in Table 1.

Table 1 List of types of titania in LLDPE obtained from Ciba.

Titania Titania Description Average Loading Particle Size (%) (nm) Degussa P25 approx 75% anatase, 25% rutile, no surface modification 25-35 1 Degussa P25 approx 75% anatase, 25% rutile, no surface modification 25-35 3 Kronos 1002 100% anatase, apparently no surface modification 20-200 1 Kronos 1002 100% anatase, apparently no surface modification 20-200 3 Huntsman Organic coating; tioxide A-HR micronised 100% anatase, water dispersible 150 3 Huntsman Organic coating; tioxide A-HRF micronised 100% anatase, dispersible in organic systems n/a 3 Sachtleben Anatase microcrystal Hombitan LW-S- with an antimony-doped U crystal lattice 30 3 Sachtleben Organic coating on Hombitan LW-S- anatase microcrystal U-HD

30 3 Sachtleben Organic coating on Hombitan LW-S- anatase microcrystal 12 31 3 Sachtleben Aluminium and organic Hombitan LC-S coating on anatase microcrystal 32 3 Control nil n/a 0

64 2.2 Accelerated aging of samples

The samples were subjected to a multilevel factorial design experiment set up to concurrently investigate several variables. The phenomena under investigation were the effects of:

• Pre-irradiation of the samples with UVC or UVA light prior to aging • Pre-irradiation exposure times of 0 s, 60 s, 3 hrs or 24 hrs • Subsequent aging in either a weatherometer or an oven

This created 176 unique samples (see Table 2- Table 12, pg 75), which were arranged in random order to minimize systematic errors. Squares of polyethylene were cut from the sheets and held inside 35 mm photographic slide mounts to create individual samples (left). As the number of samples that could be processed at one time was limited by the available space in the weatherometer, the samples were

processed in batches of 25. Once the samples aged in the Figure 2-1 LLDPE film in projector slide weatherometer had all achieved embrittlement, the next sample holder batch of samples was processed. Samples that were aged in the oven were added to the same oven on different shelves.

Before samples were subjected to accelerated aging conditions in some cases they were pre-irradiated. Pre-irradiation is a key concept in this thesis, and involved exposing the films to a measured dose of UV irradiation prior to accelerated aging. UVA pre-irradiation was conducted in a Q-UV aging cabinet, incorporating a battery of eight 40 W Q-UV-A lamps, at a distance of 5 cm from the samples (dose rate ~1,200 W/m2). The peak emission was at 340 nm with a cut-off at 295 nm. All pre-irradiation was conducted at ambient atmospheric conditions.

UVC pre-irradiation was conducted using 2 x 60 W low-pressure mercury vapour lamps with single wavelength 254 nm emission, purchased from Heraeus.

65 The system power was approximately 50 W/m2 at the irradiation platform, including radiation from a parabolic reflector for collection of stray UV light. The spectral emission of a low-pressure mercury vapour lamp is a line spectrum with approximately 90 % of its output at 254 nm.

Weatherometer aging was conducted in a Heraeus Suntest CPS+ Weatherometer™ device operating at an irradiation level of approximately 765 W/m2 at the plane of the samples. Temperature ranged between 35 and 45 ºC, with ambient humidity. Air was drawn from outside the weatherometer by an internal fan and blown over the samples. The weatherometer was set to a cycle of 72 hours irradiation time, after which the samples were removed from the weatherometer and ATR/FTIR spectra were obtained. Non-embrittled samples were returned to the weatherometer subsequent to measurement for another 3 day cycle. Embrittled samples were removed from the population.

Oven aging was conducted in a Contherm Digital Series Oven™ set to 50 ºC under atmospheric conditions. The oven was opened weekly, allowing fresh air inside. Oven aged samples were removed weekly or fortnightly for the first 3 months, and then less frequently for the remainder of the aging time for ATR/FTIR analysis. The samples were returned to the oven immediately following analysis.

The temperature of 50 ºC was chosen for accelerated thermooxidation, which is 10 - 20 ºC lower than others reporting in the literature for similar films4,173,174. The choice of temperature reflected the attempt to best mimic natural conditions within the time available. Additionally, is has been reported that with increasing temperature the nature of titania particles (crystal phase, surface modification, particle size, etc.) has lesser influence on the rate of oxidation7. As the effects of different forms of titania were under investigation, lower temperatures were more appropriate.

Samples were considered to have embrittled when they either fractured and developed tears during the aging process, or when the film was easily punctured and the material tore easily under the application of light pressure from a

66 relatively blunt object. All weatherometer aged samples were aged to embrittlement, while all oven aged samples were aged for a minimum of 200 days, and in some cases up to 400 or more.

2.3 Mid-IR spectroscopy

ATR/FTIR was performed on a Nicolet Nexus 870™ spectrometer using a Smart Endurance macro diamond ATR crystal. 64 scans were co-added at 4 cm-1 resolution. Spectra were manipulated using Grams32 AI software. The spectra were not ATR corrected.

The carbonyl index of the samples was recorded for each spectrum. The carbonyl index was obtained by measuring the ratio of the area under the carbonyl peak -1 -1 (between 1705 cm to 1735 cm ) to the area under the CH2 deformation peak (1460 cm-1 to 1475 cm-1).

Multivariate analysis was performed using Solo™, a standalone version of the PLS-toolbox add-on designed for Matlab by Eigenvector. Outliers were identified by abnormal Hotelling T2 or Q residuals. In all cases spectra identified as outliers were examined first to determine that they were true outliers (abnormal baseline, additive absorptions, etc.). If an abnormality was found, the spectrum was labeled an outlier and not included in the analysis. If the spectrum appeared ‘normal’ in all other respects, it was generally included. In general very few apparent outliers were identified as outliers.

The PCA program in Solo™ was used for all multivariate analysis work. The spectra were normalized by setting the area to the same value (Area = 1), followed by mean centering. The PCA model was cross-validated by a ‘leave one out’ method. Typically 3-5 PCs were used to create a model. Data handling is described in more detail in Section 4.2.

67 2.4 Imaging IR Spectroscopy

Data were collected using a Varian FT-IR imaging system. The system consists of a rapid scan Varian 3100 FT-IR spectrometer, a Varian 600 UMA FT-IR microscope equipped with an ATR objective, and a 32 x 32 liquid nitrogen cooled mercury cadmium telluride (MCT) focal plane array detector. A germanium slide-on crystal was used in the ATR objective. Spectra were processed and images created using Varian Resolution Pro software.

Topas® was used as the polymeric material for solvent casting degradation experiments covered in Chapter 5. Topas is an ethylene/norbornene copolymer (Figure 2-2), which is easier to dissolve in solvents than polyethyelene.

x y Figure 2-2 Molecular structure of Topas®.

Five grams of Topas was dissolved in 50 ml of cyclohexane on a hot plate stirrer set to 40 ºC and left to dissolve overnight with stirring. Droplets of 2.5 µL of the polymer/cyclohexane solutions were pipetted onto glass slides in triplicate and the cyclohexane evaporated off under a fume hood. A glass slide supported control droplets, not containing titania. A second glass slide supported Topas that had Degussa P25 titania powder mixed with the Topas solution at approximately 5% loading. The third slide supported droplets onto which Degussa P25 titania powder had been dusted over the surface before the solvent was evaporated in the fume hood.

The droplets on the slide were irradiated with UVC described in Section 2.2. ATR/FTIR spectra were obtained for the droplets on the 4 slides hourly over 4 hours. It was found that the slide supporting Topas with titania deposited on the surface demonstrated the greatest carbonyl intensity, and it was concluded that this method would be used for the following step of the imaging ATR/FTIR experimental.

Figure 2-3 ATR Ge crystal slide assembly.

Using the same solution of 5 g Topas in 50 ml cyclohexane, 2.5 µl was deposited onto the IRE surface of the slide-out germanium ATR/FTIR crystal objective (Figure 2-3). Degussa P25 titania was deposited onto the wet surface, and the slide-out crystal assembly with the solvent-cast polymer was left to dry under a heated vacuum overnight. The assembly was then irradiated with UVC, and images collected every hour, up till 8 hours of irradiation. Spectra were acquired at 16 cm-1 resolution with 1064 scans co-added, over a range of 4000–850 cm-1

2.5 Synchrotron experimental

Spectra were acquired at the Australian Synchrotron (Clayton, Victoria) on the IR beamline175,176 using a Bruker V80v spectrometer with Bruker Hyperion 2000 infrared microscope in micro-transmission mode. An aperture of 10 x 10 µm was used. Spectra were examined using OPUS 6.5 software.

A Perspex box was built around the microscope stage allow purging with nitrogen gas (Figure 2-4). The box had a door at the front to access the stage. Once the doors had been closed they were left shut for 10 minutes to allow for adequate purging.

69 Synchrotron light source Perspex purge box

Stage controls Doors Figure 2-4 Bruker FTIR microscope with Perspex purge box at the Australian Synchrotron.

UVA from an Omnicure® 2000 high pressure 2000 W mercury lamp emitting at 300-500 nm (unfiltered) with a flexible fiber optic cable was used inside the Perspex box by threading the fiber optic cable through a similarly sized hole in the Perspex. This was clamped in place 4 cm above the sample, at an approximately 70 ºC angle from the horizontal, with the power level at 100. (Previous proof of concept experiments not published in this thesis had adequately demonstrated that this positioning and light strength provided enough UVA in atmospheric conditions to degrade a titania containing LLDPE sample such that a carbonyl absorption could be observed in the IR spectrum after about 10-15 minutes of irradiation).

The material investigated was a polyethylene film blown by members of the CRC-P project at QUT. It was comprised of LLDPE obtained from DOW Chemicals, with 1% polyisobutylene (for processability and to improve tackiness) and 3% Degussa P25 titania. The film was 15 µm thick and clear. A piece of LLDPE was cut from the sheet, and placed over a metal slide containing a hole, to allow transmission experiments

70 Fiber optic

UVA

Film Hole in plate

Sampled area Tape Metal plate

Figure 2-5 Schematic representing the experimental set up for synchrotron transmission IR experiments.

The LLDPE film was mapped in micro-transmission mode in a 2 x 3 pattern, with an aperture of 10 x 10 µm. Two hundred and fifty six scans were co-added for each spectrum, at a spectral resolution of 4 cm-1. A background was taken via the hole in the metal plate to the side of the film edge before each spectrum.

UVA fibre optic Microscope magnification objective

Upper cassegrain

Perspex box extension for stage movement

Sample on metal sample holder

Microscope stage Figure 2-6 Photograph taken at the Australian Synchrotron showing the IR microscope and stage. The stage is currently in position for map acquisition, and is moved to bring the sample under the UVA probe when irradiating.

After taking a background the stage was moved to a predetermined position using the microscope controlling software, and the doors to the Perspex box were opened for 3 minutes to allow air into the box. The film was then exposed to 2 minutes of irradiation, with the doors open. The UVA lamp was switched off, the doors were closed, and 10 minutes was allowed for the nitrogen purge to remove most of the air inside the box. A 2x3 map was acquired, which took approximately 10 minutes. This process was repeated until the film had undergone a total of 30 minutes of irradiation. Each map took a total of 30 minutes to acquire, which includes the time taken for purging, sample irradiation, etc. 30 minutes of irradiation resulted in a total experiment time of 8 hours.

2.6 Scanning electron microscopy

SEM images were acquired on a FEI Quanta 200 Environmental SEM equipped with an Everhart-Thornley detector (ETD), while backscattered electron images were acquired using a silicon strip detector (SSD). Elemental microanalysis was conducted using energy dispersive x-ray analysis (EDAX) and all samples were coated in carbon films.

72 Effect of UV pre-irradiation on the degradation of polyethylene

3.1 Introduction

As stated in Chapter 1, the aim of this project is to utilize UV pre-irradiation to enhance and control the degradation of polyethylene, such that it will degrade in the dark. Various types of titania from different manufactures have been incorporated into polyethylene film, and have been subjected to a range of degradation conditions. Factors influencing the photoactivity of titania particles such as size distribution, agglomeration, crystal phase, modification, etc. have been investigated with a view to understanding the effects of these factors on the degradation pathway of polyethylene.

This chapter scrutinizes the large amount of data obtained during the investigation, and examines and compares the photoactivity of different titania particles

3.2 Physical characteristics of commercial titanias and general comments

3.2.1 Degussa P25 Polyethylene film containing 1% Degussa P25 was translucent, although slightly hazy. With several layers of thickness the film appeared opaque and glossy white, with a distinct sheen. If the film was held up to the light numerous small imperfections were seen included in the sheet. These imperfections were approximately 200 microns or smaller in diameter and appeared evenly distributed.

This film lasted only 6 to 12 days in the weatherometer. All samples turned entirely white very early, and disintegrated into fine particles upon embrittlement. Rubbing the embrittled material between fingertips resulted in a

73 crumbly, flaky collection of small (sub-millimetre) particles of approximately even size.

The sample with 3% Degussa P25 loading was also translucent, although much less so than the 1% sample. Several layers thickness of the film showed the plastic was white. UV degradation experiments produced similar results to the 1% loading. All samples reached embrittlement in 6 to 12 days, although they tended to embrittle earlier than the 1% sample. Again the samples whitened and became opaque very early.

Backscattered SEM images (Figure 3-1) show a particle size in the film of up to, and in some cases exceeding, 5 µm. The Degussa P25 titania appears agglomerated and poorly dispersed. This poor dispersion in the polyethylene film is possibly due to the lack of surface modification of the titania particles.

Figure 3-1 Backscattered SEM images of film containing 3% (left) and 1% (right) Degussa P25. (50 µm size scale).

3.2.2 Kronos The Kronos 1% loading film was semi-transparent white, with a fine dispersion of particles. The 3% loading film had a similar, albeit less transparent, appearance to the 1% film. A size dispersion of particles could be seen similar to the 1% loading film (Figure 3-2).

74 The Kronos 1% loading film did not undergo a significant colour change with exposure to UV. Embrittlement typically took between 20 and 30 days. The 3% films on the other hand whitened considerably and heterogeneously. The 3% loading material tended to flake when fully embrittled.

Figure 3-2 Backscattered SEM images of film containing 1% (left) and 3% (right) Kronos Titania. (5 0 µm size scale).

3.2.3 Huntsman Tioxide The two Huntsman Tioxide samples were very white when observing a single film and this became opaque with 4 sheets thickness. The plastic film has a very glossy look and feel. It demonstrated longer embrittlement times than the Degussa P25 samples. The film containing water dispersible organic-coated particles (A-HR) had a wide range of embrittlement times, varying from 3 to 21 days. Organic dispersible (A-HRF) titania in LLDPE showed more consistency in aging times, ranging from 9 to 18 days to embrittlement. The Huntsman Tioxide samples whitened heterogeneously with exposure to UV.

Backscattered SEM images (Figure 3-3) show good particle dispersion, and smaller particle size than the Degussa P25 samples, although there is still some agglomeration, and many particles around 1 µm in size.

Figure 3-3 Backscattered SEM images of film containing 3% A-HR (left) and A-HRF (right) Huntsman Tioxide. (20 µm size scale on the left and 50 µm on the right.)

3.2.4 Sachtleben Hombitan The Sachtleben Hombitan films were all white, and 4 sheets thickness resulted in opacity. The particles appeared to be evenly dispersed, as no imperfections could be seen in the film when held up to the light. A tinge of brown could be seen in the folded antimony-doped and organic coated films, perhaps due to the nature of the modifiers on the titania. The organic and aluminium coated titania film was very white and had a very high gloss. No brown tinge could be seen in this film.

The Sachtleben Hombitan films demonstrated the longest times to embrittlement of all films in the weatherometer, generally between 12 and 30 days, with some exceptions. The Sachtleben Hombitan LW-S-12 sample exhibited the greatest degree of whiteness when subjected to UV radiation, whilst the aluminium and organic coated particles remained relatively clear until embrittlement. The aluminium and organic coated titania samples tended to tear when embrittled, rather than become flaky as the whitened samples containing Degussa P25 titania. These samples took up to 45 days to embrittle.

Sachtleben Hombitan titania appeared well distributed in the polyethylene films (Figure 3-4), with a particle size of under 1 µm. The organic coated particles demonstrated the best dispersion.

Figure 3-4 Backscattered SEM images of film containing 3% LW-S-U (upper left), LW-S-U-HD (upper right), LW-S-12 (lower left) LC-S (lower right) Sachtleben Hombitan titania. (20 µm size scale).

3.2.5 Section summary All of the films were translucent, varying slightly in colour between white and off-white. Higher titania loadings resulted in less transparency. The LLDPE film containing Degussa P25 titania showed the poorest particle distribution, as well as significant agglomeration of particles. Titania with organic coatings demonstrated improved particle dispersal and a narrow particle size distribution.

77 LLDPE films that degraded rapidly in the weatherometer became white, and disintegrated into small flaky particles when embrittled. LLDPE films that took longer to degrade in the weatherometer tended to remain translucent, and tore at embrittlement. The difference in degradation results was due to the photoactivity of the titania, and the cause of sample whitening is discussed in the following section.

3.3 Sample whitening

The phenomenon of whitening was examined by SEM. An image taken of a whitened, degraded polyethylene film containing 1% Degussa P25 titania and the corresponding backscattered image is shown in Figure 3-5.

Figure 3-5 SEM image of polyethylene containing 1% Degussa P25 titania particles following photodegradation. The image on the right is the backscattered image. Note the appearance of dark areas in the backscattered image around the titania particles, indicating an absence of material at these locations.

The titania particles clearly show up as white dots in the backscattered image. It can be seen that there are dark areas in close proximity to many of these particles, often in the shape of a ‘tail’, or a ‘wormhole’, which represents an absence of material. The effect of titania creating holes in polymeric material under the application of UV radiation with subsequent whitening has been demonstrated many times in the literature 77,78,88,177-179.

78 It has been postulated that the whitening of materials containing titania can be explained by the presence of cavities78. Cavities created by titania-catalysed photodegradation, such as those observed in Figure 3-5, scatter visible light, producing a whitening effect. The scattering of light by pores in a material is utilised by the coatings industry to produce white coatings180, and the pore size must be one half the wavelength of incident light to achieve maximum 181,182 scattering . Thus to scatter visible light (λrange = 400 nm – 750 nm), a pore diameter of between approximately 0. 2 µm and 0.4 µm is required. A cavity created by titania particles, such as captured by the SEM images in Figure 3-5, is therefore capable of scattering visible light such that the LLDPE film appeared white.

Whitening of the LLDPE film provides an indication of the relative photoactive strengths of the various types of titania used in this thesis. Samples containing modified titania, such as the Satchleben Hombitan organic coated titanias, did not whiten, and took longer to degraded. These films appeared similar to the control sample at embrittlement, remaining quite clear. Conversely LLDPE film containing Degussa P25 embrittled quickly and whitened.

The cavities formed by titania affect the physical properties of the film. The loss of material, as evidenced by the wormholes in Figure 3-5, combined with rapid and extensive oxidation of the LLDPE chains, causes the film to fall apart with exposure to UV irradiation. This is evidenced by the homogeneous flaking of the film at embrittlement, rather than the shrinking and tearing of the control film and those containing less active titania.

79 3.4 Times to embrittlement for LLDPE film containing titania

The following tables describe the experiments performed for each LLDPE film, and the time taken for each sample to achieve embrittlement. The first two columns indicate whether the samples were aged in the oven or suntest, the next two columns indicated what wavelength of pre-irradiation UV was used, and the subsequent columns indicated the length of pre-irradiation (a time of 0 sec indicates that the sample was not subjected to pre-irradiation). The last column denotes the numbers of days aged until embrittlement was achieved, and a ‘+’ after the number indicates that the sample had still not achieved embrittlement at the conclusion of the experiment.

Thus, as an example, the 3rd line in Table 2 shows that the LLDPE film containing 1% Degussa P25 was pre-irradiated with UVA for 3 hours and aged in the oven, and embrittlement was not achieved in over 330 days of aging.

The experiments were not performed in the order presented in the tables, but were randomised instead to assist in reducing drift and sample carry over errors (Section 2.2).

Table 2 Data for the LLDPE film containing 1.00% TiO2 Degussa P25 showing sample aging details and days taken to reach embrittlement.

1.00% TiO2 Degussa P25 Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 370+ 400+ 330+ 200+ 270+ 400+ 392 24 6 12 6 9 12

80 12 9 3

Table 3 Data for the LLDPE film containing 3.00% TiO2 Degussa P25 showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Degussa P25 Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 375 330+ 330+ 181 330+ 381 371 229 12 6 6 9 9 9 6 6

Table 4 Data for the LLDPE film containing 1.00% TiO2 Kronos 1002 showing sample aging details and days taken to reach embrittlement.

1.00% TiO2 Kronos 1002 Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 400+ 330+ 330+ 370+ 330+ 400+ 330+ 330 27 27 21 27 30 36 24 3

Table 5 Data for the LLDPE film containing 3.00% TiO2 Kronos 1002showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Kronos 1002 Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 370+ 470+ 370+ 330+ 400+ 370+ 200+ 270+ 24 24 18 15 21 24 15 9

Table 6 Data for the LLDPE film containing 3.00% TiO2 Huntsman Tioxide A-HR showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Huntsman Tioxide A-HR Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 200+ 270+ 415+ 370+ 200+ 470+ 200+ 233 15 15 12 15 15 21 12 3

Table 7 Data for the LLDPE film containing 3.00% TiO2 Huntsman Tioxide A-HRF showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Huntsman Tioxide A-HRF Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 200+ 270+ 330+ 270+ 470+ 270+ 270+ 371 15 15 15 18 12 15 9 9

Table 8 Data for the LLDPE film containing 3.00% TiO2 Sachtleben Hombitan LW-S-U showing sample aging details and days taken to reach embrittlement. '+' indicates the sample had not reached embrittlement when the experiment had ended.

3.00% TiO2 Sachtleben Hombitan LW-S-U Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 330+ 370+ 470+ 270+ 270+ 415+ 330+ 327 21 21 30 30 24 32 9 12

Table 9 Data for the LLDPE film containing 3.00% TiO2 Sachtleben Hombitan LW-S-U- HD showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Sachtleben Hombitan LW-S-U-HD Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 430+ 430+ 415+ 370+ 200+ 430+ 270+ 330 27 24 27 27 30 24 27 12

Table 10 Data for the LLDPE film containing 3.00% TiO2 Sachtleben Hombitan LW-S-12 showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Sachtleben Hombitan LW-S-12 Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 200+ 430+ 330+ 200+ 470+ 470+ 200+ 73 18 15 21 18 15 12 12 9

Table 11 Data for the LLDPE film containing 3.00% TiO2 Sachtleben Hombitan LC-S showing sample aging details and days taken to reach embrittlement.

3.00% TiO2 Sachtleben Hombitan LC-S Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 430+ 470+ 430+ 330+ 200+ 370+ 470+ 392 39 30 27 27 45 33 39 21

Table 12 Data for the control sample showing sample aging details and days taken to reach embrittlement. Control Days to Oven Suntest UVA UVC 0 sec 60 sec 3 hr 24 hr emb. 266+ 266+ 266+ 266+ 266+ 266+ 266+ 266+ 66 66 60 57 66 66 48 24

These tables show that the samples have been aged in the oven for considerable periods of time; up to 470 days for some samples. As stated in the objectives (Section 1.4) it is desirable that the films embrittle completely in the dark after

85 about 6 months of oven aging following pre-irradiation, and thus all samples have undergone a minimum of 200 days spent aging in a dark environment at 50 °C. All samples tested in the weatherometer were aged to embrittlement.

Two samples have come close to the target of embrittlement in the oven by 6 months; both samples are LLDPE containing 3% Degussa P25 titania (see Table 3), and were pre-irradiated with 24 hours of UVA (181 days) and UVC (229 days) respectively. All other samples exposed to UV pre-irradiation and aged in the oven took over 200 days to embrittle. It is probable that for a commercial application, 24 hours of UV pre-irradiation is too long to be practical; however from the point of view of exploring new technologies, it is important to recognise that these samples provide evidence of the success of pre-irradiation as a concept, and a starting platform for the continued research into this new technology. An in-depth analysis of these, and all samples in the pre-irradiation experiment, is provided in the following sections.

3.5 IR spectral analysis – control film (undegraded)

3.5.1 Polyethylene absorption table The infrared characterisation of polyethylene is well known, and the early work published by Rugg et al. in the 1950’s 24,25 is still referred to 169,183,184. A complete description of straight chain alkane vibrations from C3H8 through to n- 185,186 C19H40 was provided by Snyder and Schachtschneider . The differences in absorption in the mid-IR of different types of polyethylene, and the effects of crystalline and amorphous regions were discussed in Section 1.3.3.4. An ATR/FTIR spectrum of the control film used in this thesis is provided in Figure 3-6, and a table assigning the absorptions is provided in Table 13.

86 0.5

0.4 ce n

a 0.3 rb so

Ab 0.2

0.1

0.0

4000 3500 3000 2500 2000 1500 1000 Wavenumbers (cm-1)

Figure 3-6 ATR/FTIR spectrum of the LLDPE control film. This film does not contain titania.

Table 13 Mid-infrared absorption table for Dowlex 5056 G polyethylene. Absorption (cm- Appearance Assignment 24,25,186,187 1)

~2953 Weak, shoulder -CH3 antisymmetric stretc.h

2914 Very strong -CH2 antisymmetric stretc.h

2847 Very strong -CH2 symmetric stretc.h

2690-2630 Weak, broad collection Various CH2C and CH3C of small bands structural features 188, possible overtones

1471, 1463 Strong, doublet -CH2 deformation

~1445 weak, shoulder -CH3 antisymmetric bend 1212, 1195, Weak, sharp bands Methyl bending vibrations

729, 718 Strong, sharp CH2 rocking modes

3.5.2 Titania absorption in the mid-infrared

0.5 Kronos film Control film 0.4 Subraction result n o i

t 0.3 rp so

Ab 0.2

0.1

0.0

4000 3500 3000 2500 2000 1500 1000 Wavenumbers (cm-1)

Figure 3-7 Spectral subtraction result (Black) of control film (Magenta) subtracted from 1.00% Kronos film (Blue). Spectra are to scale. The spectra have been offset.

Anatase and Rutile titania absorb mid infrared light below ~850 cm-1 189. The strength of this absorption increases with increasing titania concentration. The presence of titania does not appear to otherwise significantly affect the infrared spectrum of polyethylene, which can be seen in Figure 3-7 where the subtraction result demonstrates the additional titania absorption in the Kronos film, while the

CH2 absorption peaks are largely similar aside from some minor broadening in the CH stretch region.

3.6 Processing agent absorptions

Absorptions occur in the spectra of the Ciba films that cannot be assigned to polyethylene, and are representative of an organic material that has been added to

88 the LLDPE film. Spectral subtraction was used to extract a spectrum of this material from the polyethylene (Figure 3-8), and possible band assignments are provided in Table 14.

0.030

0.025

0.020 ance

rb 0.015 o s b A 0.010

0.005

0.000

1750 1500 1250 1000 750 Wavenumbers (cm-1)

Figure 3-8 Subtraction spectrum showing additive absorption peaks.

Table 14 Mid-infrared absorption table for processing agent present on the surface of polyethylene Absorption (cm-1) Appearance Assignment 24,25,186,187

1490, 1460 Strong, doublet -CH2 deformation

1398, 1361 Medium, doublet t-butyl –(CH3)3

1212, 1195 Medium, doublet t-butyl –(CH3)3

1081 Medium, sharp isopropyl –(CH3)2 (?) 906 Weak, sharp Vinyl (?) 854 Weak, sharp Vinyl (?) 776 Weak, sharp t-butyl symmetric skeletal stretc.h

719 Strong -CH2 rock

89 There are no absorptions in the spectrum above 1500 cm-1, which excludes the possibility of aromatic or carbonyl functional groups on the compound. Additionally, this spectrum only appears on one side of the polymer film, and is on the surface. It is likely that it is some kind of aliphatic processing agent, possibly a film blowing, anti-static or anti-slip agent. Similar compounds are known to be used in some film blowing processes190. The spectrum of this additive looks similar to polyisoprene or polybutylene, although some absorptions are not so closely matched that a positive identification can be made.

There is a great deal of processing agents used by different manufacturers for different purposes, making positive identification extremely difficult. Soxhlet extraction techniques were performed using a variety of solvents in attempts to isolate this additive, without success.

The focus of this thesis is to obtain results that will be directly transferable into ‘real world’ applications. The materials subjected to study are commercially available ones, and contain additives and processing agents such as these. In the context of shopping bags and other applications, it reasonable to expect that the side subjected to sunlight cannot be controlled, and the effect of pre-irradiation and aging processes must be investigated in a manner that variables such as additive concentrations, processing agents on the surface, machine direction of the blown film, etc., are randomised. Thus the side of the film was not deliberately taken into consideration when aging the film. The effect of the processing agent on degradation is investigated in Chapter 4.

3.7 IR spectral analysis – control film (degraded)

Oxidative degradation of the polyethylene film resulted in significant changes in the infrared absorption spectrum (Figure 3-9). In all cases an increase in absorption occurred in the carbonyl region (1850-1650 cm-1), indicating the presence of degradation products including some kind of oxygenated functional group. A broad increase in absorption was noticed below 1300 cm-1, with some specific peaks that can be related to esters and unsaturation. The OH stretc.hing region displays evidence of carboxylic acid type OH formation.

0.3 Control Control 66 days weatherometer Subtraction result

0.2 ance rb o s

b 0.1 A

0.0

4000 3500 3000 2500 2000 1500 1000 Wavenumbers (cm-1)

Figure 3-9 ATR/FTIR spectral subtraction result (Blue) of unaged control sample (Black) subtracted from 66 days weatherometer aged control sample (Red). Spectra have been offset.

3.7.1 OH stretc.hing region (3800-3200 cm-1) Absorption at the high wavenumber end of the spectrum indicating the presence of alcohol oxidation products is expectedly weak using ATR/FTIR spectroscopy144. For reasons discussed in Section 1.3.3.3, the spectra have not undergone ATR correction, and thus the high wavenumber end of the spectrum does not display full strength absorptions. However subtraction spectra such as that shown in Figure 3-9 indicate that some alcohol functional groups are present in small quantities.

3.7.2 Carbonyl region Examination of the carbonyl region of the spectrum of degraded polyethylene shows multiple, overlapping bands. Curve fitting of this region has been

91 performed to extract peak positions. It can be seen in Table 15 that more assignments have been made than there have been bands fitted. The carbonyl region of these degraded polymers is in most cases fairly broad and undefined, and addition of further bands to the curve-fitting calculations does not result in a more accurate fit. As was seen in Section 1.1.3 the carbonyl region of degraded polymers, such as polyethylene, shows the absorptions of many different kinds of carbonyl-containing degradation products. In a curve fitting exercise it is useful to assign the major bands, whilst being aware that multiple absorptions are likely to be hidden under the same peak. The results are shown in Figure 3-10, and the peak assignments are given in Table 15.

1800 1700 1600 Wavenumbers (cm-1) Figure 3-10 Carbonyl region of Figure 3-9.

Table 15 Curve fitting results of the carbonyl region for the control film Absorption (cm-1) Assignment 184,187 1785 Lactones, anhydrides, peracids 1763 Peresters, anhydrides 1733 Esters and aldehydes 1710 Ketones and carboxylic acid

1641 (sharp) RCH=CH2 1639 Carboxylates

92 3.7.3 Below 1500 cm-1 Several changes occur in the absorption spectrum of polyethylene below 1500 cm-1 with oxidation (Figure 3-11). A feature that is common to all polyethylene films studied was the broad absorption increase from approximately 1400 cm-1 to 600 cm-1. This absorption increase is likely to be a complex combination of signatures arising from absorptions such as C-O stretc.hes of esters, anhydrides, carboxylates, etc.. An increase in absorption intensity is in this region is common to all degraded samples, both those containing titania and those without. The strong absorption at 1180 cm-1 is assigned to an ester C-O stretc.hing vibration187.

0.030

0.025

0.020

0.015 ance rb

o 0.010 s b A 0.005

0.000

-0.005

1250 1000 750 Wavenumbers (cm-1)

Figure 3-11 Expansion of subtraction result in Figure 3-9 below 1500 cm-1.

3.7.4 Section summary The LLDPE used in this study oxidises to produce absorption peaks in the infrared spectrum that agree with exhaustive studies published in the literature over decades. A weak, broad absorption above 3200 cm-1 indicates the presence

93 of alcohols. Investigation of the carbonyl region between 1900 and 1650 cm-1 shows that esters and acids form a large proportion of degradation products, with some anhydride and lactone absorptions at higher wavenumbers. The area below 1400 cm-1 has a very broad increase in absorption, indicating the presence of various oxygenated degradation products, with a strong absorption at 1180 cm-1 assigned to an ester C-O stretc.h. An aliphatic processing agent is present on one side of the film, which appears to contain a high concentration of tertiary methyl groups.

3.8 Effect of UV irradiation – control film (degraded)

Degradation experiments in the weatherometer and in the oven were performed according to the details provide in Section 2.2. Section 3.8 covers the control film, and will examine the development of the carbonyl index, drawing comparisons between the films. The carbonyl index was acquired according to the method described in Section 2.3.

3.8.1 Control, weatherometer aged samples This section will examine the effect that different times of exposure to UV irradiation had on the control sample aged in the weatherometer. UVC radiation will be investigated first, followed by UVA radiation.

94 3.8.1.1 Effect of pre-irradiation – UVC

0.30

0.25

0.20 index l 0.15

0 secs Carbony 0.10 60 secs 3 hrs 0.05 24 hrs

0.00 0 10203040506070 Days aged in weatherometer

Figure 3-12 Carbonyl index plots of the control film pre-irradiated with UVC for 0 secs, 60 secs, 3 hours and 24 hours and subsequent exposure in the weatherometer. Note the earlier time to embrittlement combined with lower carbonyl index at embrittlement of the 24 hour exposed sample. Second order polynomial trend lines have been fitted to depict a trend, and do not imply reaction mechanism or theory.

The carbonyl index plots in Figure 3-12 show the effects of pre-irradiating LLDPE with UVC, and subsequent weatherometer aging. Two key points can be recognised from observation of the plots in Figure 3-12.

Firstly, significant UVC exposure shortens the lifetime of the polymer. The carbonyl index plot of the sample exposed to 60 secs of UVC is virtually indistinguishable from the untreated sample, while the samples exposed to 3 hours and 24 hours of UVC demonstrate accelerated degradation. In this case, 3 hours of pre-irradiation causes a reduction of the time to embrittlement of the polymer by one third, and 24 hours pre-irradiation reduced this by a further one third. Embrittlement is defined in Section 2.2.

Secondly, the slopes of the carbonyl index plots in Figure 3-12 imply that oxidation is occurring more rapidly in the samples with longer UVC exposure times. If the slopes had been similar across all samples, with only an offset on the y-axis to demonstrate an increased concentration of oxidation products formed during UVC treatment, it could be concluded that the pre-irradiation had merely started degradation earlier. However, the steeper slopes indicate a faster rate of reaction, which is attributed to the polymer samples with longer UVC exposure times being more susceptible to further photooxidation in the weatherometer than those samples with lesser UVC pre-treatment.

The carbonyl index plots clearly demonstrate that UV irradiation increases the rate of degradation. Furthermore, larger doses of UVC irradiation appear to have had a significant effect on the rate of oxidation product formation. The spectra of the 3 hour and 24 hour UVC pre-irradiated samples prior to weatherometer aging show an absorption at 1641 cm-1, which is assigned to a vinylic absorption (Table 15). In addition to the creation of unsaturated polymer chains, the films have shrunk in the sample holder, indicating extensive crosslinking35. Crosslinked polymers have a higher concentration of hydrogens tertiary to backbone carbons, which are more susceptible to hydroperoxide attack11.

Section 1.1.2 covered the effect of oxidation occurring initially in defective sites in polymer, and spreading from there to the polymer bulk. High doses of UV radiation, and in this case particularly UVC, appear to have created reactive sites via double bonds and crosslinks, which has resulted in a faster rate of degradation of the bulk polymer. This is evidenced by the steep slope of the carbonyl plots of the UVC pre-irradiated samples. Thus ‘weakening’ of the polymer via UV irradiation has resulted in shorter embrittlement times.

96 3.8.1.2 Effect of pre-irradiation – UVA

0.30

0.25

x 0.20 de in l

y 0.15 on

rb 0 secs

Ca 0.10 60 secs 3 hrs 24 hrs 0.05

0.00 0 10203040506070 Days aged in weatherometer

Figure 3-13 Carbonyl index plots of the control film pre-irradiated for 0 secs, 60 secs, 3 hrs and 24 hours with UVA radiation. Second order polynomial trend lines have been added.

In contrast to their UVC pre-irradiated counterparts, the carbonyl index plots of the UVA pre-treated samples in Figure 3-13 do not demonstrate significant differences. For all UVA pre-irradiated samples, there is little change in the carbonyl region following pre-irradiation. Close examination of the carbonyl absorption region of the 24 hour pre-irradiated sample prior to weatherometer aging does show some absorption; however this is barely above the signal noise.

Despite the similar carbonyl index plots, UVA pre-irradiation affected the outcome of film degradation. Increasing the length of pre-irradiation resulted in shorter embrittlement times, as samples pre-irradiated for 0 secs, 60 secs, 3 hours and 24 hours embrittled in 66, 66, 60 and 57 days in the weatherometer respectively (Table 12).

97 3.8.1.3 Section summary Pre-irradiation of LLDPE film under UVC or UVA before aging in a weatherometer decreases the time taken to achieve embrittlement. The film must be irradiated for a significant period of time to produce an effect. Irradiation results in reactive sites in the polymer matrix that are susceptible to hydroperoxide formation, increasing the rate of degradation. Although oxidation occurs more quickly, the heavily irradiated samples show less oxidation product formation at embrittlement, suggesting that other degradation mechanisms are occurring.

Pre-irradiation with UVC has a more significant impact than with UVA on the time taken to embrittlement of the control sample. This is most likely due to the higher energy of UVC, which results in the formation of more reactive sites. The degrading effect of UVC is reflected by the faster embrittlement times, and lower carbonyl index at embrittlement of the UVC pre-irradiated samples.

3.8.2 Control, oven aged samples This section will examine the effect that different times of exposure to UV irradiation have on the control sample aged in the oven at 50 °C. UVC pre- radiated samples will be investigated first, followed by UVA pre-radiated samples.

98 3.8.2.1 Effect of pre-irradiation – UVC

0.6

0.5

0 secs 0.4 60 secs 3 hrs dex 24 hrs

l in 0.3

0.2 Carbony

0.1

0.0 0 50 100 150 200 250 300 Days aged in oven

Figure 3-14 Carbonyl index plots for UVC pre-irradiated control samples aged in the oven. Second order polynomial trend lines have been added. Error bars showing standard deviation have been added to this figure and Figure 3-15; however they have been omitted from the other figures for clarity

Pre-irradiation of the control film with UVC has strongly affected the degradation of the polymer. Figure 3-14 shows little difference in the carbonyl index of the non-irradiated and the 60 sec irradiated films.

99 3.8.2.2 Effect of pre-irradiation – UVA

0.06 0 secs 60 secs 3 hrs 24 hrs

x 0.04 de in l y on rb

Ca 0.02

0.00 0 50 100 150 200 250 300 Days aged in oven

Figure 3-15 Carbonyl Index plots for UVA pre-irradiated control samples aged in the oven. Second order polynomial trend lines have been added. An outlier has been removed from 3 hr sample.

Similarly to those results seen for the weatherometer aged control sample, UVA pre-irradiation did not have as large an impact on the rate of oxidation as UVC pre-treatment (Figure 3-15). Even the sample with 24 hours exposure to UVA did not degrade significantly until after 100 days in the oven. Note that the carbonyl index scale on the y-axis shows that the LLDPE control film samples are still in the early stages of oxidation.

3.8.3 Section Summary Oven aged control samples (LLDPE film without titania) take much longer to degrade than weatherometer aged control samples. None of the samples have achieved embrittlement in the oven, despite aging for over 266 days. Pre-

100 irradiation with UVC results in more significant carbonyl product formation than UVA with similar irradiation times.

3.9 IR spectral analysis – film containing titania (degraded)

Films containing titania degraded in the weatherometer and in the oven considerably faster than the control. Not all films containing titania behaved the same, with some films achieving embrittlement much faster than others. The following sections will examine the effect of titania on the IR spectra, and compare the degradation of LLDPE film containing titania from different manufacturers.

3.9.1 Carbonyl region An interesting feature of the degradation of this film is the carbonyl absorption at embrittlement. Figure 3-16 shows spectra of 1% Degussa P25 film aged in the oven and weatherometer compared with the control aged in the weatherometer.

101 Control in weatherometer 0.02 1% Degussa P25 oven 1% Degussa P25 weatherometer ce n Absorba

0.01

1800 1750 1700 1650 1600 Wavenumbers (cm-1)

Figure 3-16 Comparison of control film without pre-irradiation embrittled in the weatherometer, 1% Degussa P25 without pre-irradiation after 330 days in the oven, and 1% Degussa P25 without pre-irradiation embrittled after 12 days in the weatherometer. Both the 1% Degussa P25 containing samples had embrittled, while the control sample had not. Spectra are to scale.

It is immediately evident that the carbonyl region of the weatherometer aged film containing 1% Degussa P25 titania at embrittlement has not developed to the extent of the control film at the end of its lifetime in the weatherometer. This implies that there are less oxygenated degradation products present in the material at the point of failure in the 1% Degussa P25 film weatherometer aged, than in the control film, and therefore degradation processes other than those resulting in the formation of products containing carbonyl species are combining to result in embrittlement. The ‘wormhole’-like cavities shown in the SEM images in Figure 3-5 demonstrate the loss of material due to irradiation of titania particles. It was shown that this effect was strongest in the materials containing Degussa P25, implying that these particles are most active in degrading the polymer matrix.

102 The material that has been destroyed by the titania particles has in all likelihood 191 been converted directly to volatiles including CO2 and H2O . The likelihood of this is substantiated by multivariate statistical analysis presented in Section 4.4.1. This at least partially explains both the clear absence of material evident in the SEM images, and the lack of carbonyl absorption strength in the infrared spectra. The titania is contributing to the faster embrittlement times when the LLDPE is subjected to extensive UV irradiation, by forming cavities which could weaken the polymer matrix.

The magenta spectrum in Figure 3-16 is of the LLDPE film containing 1% Degussa P25 and aged in the oven for 330 days (it had not embrittled after this time). The film had not undergone UV pre-irradiation, and does not show a strong absorption at 1732 cm-1, an absorption present in both the weatherometer aged control and 1% Degussa P25 films. This supports the degradation scheme proposed by Tidjani, given in Scheme 1-9. According to this scheme, oxygen centred radicals formed by the absorption of radiation by hydroperoxides give rise to ester oxidation products (among other products). Thus in samples exposed to significant amounts of UV (such as the weatherometer aged samples) can be expected to have ester products in higher concentrations. Alternatively, hydroperoxides that decompose by heat produce a carbon centred radical, which is more likely to result in acid oxidation products. This explains the relatively lower ester concentration in the samples aged in the oven compared to those aged in the weatherometer.

Further to the differences in absorption of mid-IR caused by the different aging conditions (oven vs. weatherometer), it is worth noting the similarity of the carbonyl region of the control sample and LLDPE film containing 1% Degussa P25. Despite a difference in overall absorption intensity, as discussed in earlier paragraphs, the shape of the absorption peaks occurring in this region are very similar. This indicates similar relative concentrations of degradation products, with different absolute concentrations. It is apparent from these observations that the titania is speeding up the process of embrittlement, however the oxidation pathway of the material is not changing.

103 3.9.2 Fingerprint region Corresponding with the growth of an absorption in the carbonyl region with increasing oxidation, a peak formed in the infrared spectra at 1178 cm-1. This is most likely to be the C-O stretc.h of an ester peak. Also, there is a broad increase in absorption in this region (Figure 3-17).

0 days 3 days 0.01 6 days on i Absorpt

0.00 1400 1350 1300 1250 1200 1150 1100 1050 1000 950 900 Wavenumbers (cm-1)

Figure 3-17 IR Spectra of 1% Degussa P25 without pre-irradiation aged for 0 days, 3 days, and 6 days in the weatherometer. Note the increase appearance of an ester C-O absorption at 1180 cm-1, combined with genearally higher absorption in this region. The absorption at 1080 cm-1 is the processing agent.

3.9.3 Section summary The presence of photoactive titania caused the LLDPE film containing 1% Degussa P25 to embrittle after just a few days in the weatherometer. Degradation could be followed in the IR spectrum by an increase in carbonyl absorption and a broad increase in absorption of the fingerprint region.

104 The carbonyl region of the IR spectra of samples aged in the weatherometer containing 1% Degussa P25 and the control show very similar relative absorption intensities, implying that there are similar relative amounts of oxidation products. However the control sample had a much stronger carbonyl absorption overall, indicating a greater extent of oxidation at embrittlement. The lack of oxidation product build-up in the weatherometer aged LLDPE film containing titania is probably due to the effects of other degradation processes, such as weakening of the film due to cavities.

Oven aged samples showed very little ester absorption intensity in the IR spectrum, confirming the applicability of the Tidjani degradation scheme to these LLDPE film samples.

3.10 LLDPE containing Degussa P25 (degraded)

3.10.1 Degussa P25, weatherometer aged samples All samples containing Degussa P25 titania degraded very quickly in the weatherometer. This made it difficult to notice any significant effects of pre- irradiation on the embrittlement times of these samples.

Figure 3-18 shows the effect of different exposure times of UVC on the carbonyl absorption of the polymer containing 1% Degussa P25. Sixty seconds exposure did not appear to result in significant changes to this region. Three hours exposure shows a gain in absorption of carbonyl containing species such as ketones, esters and acids. Twenty-four hours exposure resulted in a substantial carbonyl absorption, as well as an unsaturation peak at 1641 cm-1. The formation of unsaturation to varying degrees occurred in samples exposed to all forms of radiation, including pre irradiation with UVA and UVC, and the weatherometer. However, the absorption band was sharpest and relatively most intense in samples exposed to large doses of UVC.

105

3.10.1.1 Effect of pre-irradiation – UVC

0.006 0 secs 60 secs 0.005 3 hrs 24 hrs 0.004 on i 0.003

0.002 Absorpt

0.001

0.000

-0.001

1850 1800 1750 1700 1650 1600 Wavenumbers (cm-1)

Figure 3-18 Comparison of the carbonyl region of LLDPE film containing 1% Degussa P25 after pre-irradiation with UVC for 24 hours (Blue), 3 hours (Green), 60 seconds (Red) and 0 seconds (Black).

The 1% Degussa P25 film showed some differences to the control film when aged in the weatherometer. Figure 3-19 gives the carbonyl index plots for the weatherometer aged samples.

106 0.24 0 secs 0.22 60 secs 0.20 3 hrs 24 hrs 0.18 0.16

dex 0.14

l in 0.12

bony 0.10

Car 0.08 0.06 0.04 0.02 0.00 02468101214 Days aged in weatherometer

Figure 3-19 Carbonyl index plots for UVC pre-irradiated LLDPE film containing 1% Degussa P25 titania and aged in the weatherometer. Order polynomial trend lines have been added.

The effect of titania can be seen in these carbonyl plots when compared with those of the control sample given in Figure 3-12. The samples have achieved embrittlement much earlier than the control sample, and pre-irradiation with UVC has reduced that time even further. Additionally, the carbonyl index at embrittlement is much lower than the control samples. These observations imply quicker degradation rates, with less oxygenated functional groups at embrittlement.

The lower concentration of oxygenated degradation products at embrittlement suggests that other degradation processes are having a significant effect. IR spectra of these materials exhibit a more intense unsaturation absorption band than the control sample. Furthermore, as discussed in Section 3.9.1 the polymer appears to have been ‘burned away’ by titania, and the cavities left have weakened the material, hastening the embrittlement.

107 3.10.1.2 Effect of pre-irradiation – UVA The samples pre-treated with UVA radiation showed similar trends to those treated with UVC. Figure 3-20 shows the carbonyl plots for these samples.

0 secs 0.10 60 secs 3 hrs 0.09 24 hrs

0.08

0.07

0.06 dex in

l 0.05 y n 0.04 rbo

Ca 0.03

0.02

0.01

0.00 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Days aged in weatherometer

Figure 3-20 Carbonyl index plots for UVA pre-irradiated LLDPE film containing 1% Degussa P25 titania and aged in the weatherometer. Second order polynomial trend lines have been added.

Samples pre-irradiated with UVA embrittled by 12 days, which is comparable to the UVC irradiated samples. The carbonyl index of the samples at embrittlement is similar for all samples irradiated with UVA, with the exception of the sample irradiated for 24 hours. However, it was noted that the sample actually appeared embrittled after 6 days, and by 9 days the polyethylene film had been nearly completely destroyed. The most significant difference between the samples irradiated with UVA or UVC and aged in the weatherometer is the carbonyl index of the samples at embrittlement of the 24 hour pre-irradiated samples. The

108 higher carbonyl index at embrittlement of the 24 hour UVC irradiated sample indicates a higher concentration of oxidation products in this sample.

Considering that the sample entered the weatherometer at a higher starting carbonyl index, it is clear that significant oxidation had taken place during the 24 hours of UVC irradiation. In fact, the starting carbonyl index of 0.9 is comparable to the carbonyl index of samples at embrittlement after several days of aging in the weatherometer. This indicates that although the sample had undergone significant oxidation under UVC irradiation the sample had not embrittled, that is to say chain scission reactions had not proceeded to the extent that the film began to fall apart. The sample irradiated with 3 hours of UVC shows a similar trend.

3.10.2 Section summary Degussa P25 titania greatly reduces the time taken to embrittlement in the weatherometer compared to the control film, resulting in some cases in a 10-fold increase in the degradation rate. Pre-irradiation with UV did not have a great impact on embrittlement times, with the exception of the sample irradiated for 24 hours with UVC. This sample demonstrated higher carbonyl index at the start of aging and at embrittlement, indicating a higher concentration of oxidation products, but not a greater extent of chain scission reactions.

109 3.10.3 Degussa P25, oven aged samples

3.10.3.1 Effect of pre-irradiation – UVC

0.30 0 secs 60 secs 0.25 3 hrs 24 hrs

0.20 x e ind

l 0.15 ony rb

a 0.10 C

0.05

0.00 0 50 100 150 200 250 300 350 400 450 Days aged in oven

Figure 3-21 Carbonyl index plots for UVC pre-irradiated LLDPE film containing 1% Degussa P25 titania and aged in the oven. Second order polynomial trend lines have been added.

The samples pre-irradiated for 24 hours and 3 hours reached embrittlement in the oven at 24 days and 392 days respectively. It appears as though the 24 hour irradiated sample was already close to embrittlement before oven aging. This is confirmed by comparison with the carbonyl index plot of the weatherometer aged sample shown in Figure 3-19, where the 24 hour UVC treated polymer embrittled after just 3 days.

110 3.10.3.2 Effect of pre-irradiation – UVA

0 secs 0.10 60 secs 3 hrs 24 hrs 0.08 x

de 0.06 in l y on

rb 0.04 Ca

0.02

0.00 0 50 100 150 200 250 300 350 400 Days aged in oven

Figure 3-22 Carbonyl index plots for UVA pre-irradiated LLDPE film containing 1% Degussa P25 titania and aged in the oven. Second order polynomial trend lines have been added.

Figure 3-22 shows that pre-irradiation with UVA has been much less effective than UVC in accelerating degradation in the oven. Pre-irradiation for 60 secs with UVA has almost no effect on the rate of degradation of the polymer. Even those samples exposed to higher doses of UVA are not strongly deviating from the curve for the non-irradiated sample, and after nearly 400 days of oven aging these samples only show a relatively moderate carbonyl index.

3.10.4 3% Degussa P25 samples Increasing the titania loading from 1% to 3% affected the degradation of the polyethylene film to an extent. The weatherometer aged samples generally embrittled earlier, taking around 6-9 days. Furthermore, in some cases the carbonyl index actually starts to drop away, indicating that some of the

111 oxygenated functional products have degraded even further to form volatiles such as CO2, H2O and small organic molecules.

0.34 0.32 0.30 0.28 0.26 0.24 0.22 0.20 0.18 0 secs UVC 0.16 l index 0.14 60 secs UVC 0.12 3 hrs UVC 0.10 24 hrs UVC

Carbony 0.08 0 secs UVA 0.06 60 secs UVA 0.04 3 hrs UVA 0.02 0.00 24 hrs UVA -0.02 -0.04 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Days aged in weatherometer

Figure 3-23 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Degussa P25 titania and aged in the weatherometer. Second order polynomial trend lines have been added. The data for samples aged for 60 secs UVC and 24 hours UVC contain outliers.

With the exception of the 24 hr pre-irradiated sample, the samples pre-treated with UVC and aged in the oven did not demonstrate significantly different lifetimes (Figure 3-23). Figure 3-24 shows that the carbonyl index of these samples is not greatly different at embrittlement. The lifetime of the 24 hr pre- irradiated sample is shortened by about 150 days; however the slope of the carbonyl index plot is quite similar to the lesser pre-irradiated samples.

112 0.40

0.35

0.30

x 0.25 e ind l 0.20 C6 y B bon r 0.15 C7 a

C C14 0.10 C11 C12 0.05 C9 C13 0.00 0 50 100 150 200 250 300 350 400 Days aged in oven

Figure 3-24 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Degussa P25 titania and aged in the oven. Linear or second order polynomial trend lines have been added.

The samples pre-irradiated under UVA and oven aged produced some different results. The sample exposed to 24 hours of UVA actually embrittled after just 180 days of aging, and at a very high carbonyl index of over 0.35. Figure 3-24 shows that these samples have a much higher carbonyl index then their 1% Degussa P25 counterparts after a similar length of time spent in the oven. Exposure to 3 hrs or less UVA did not seem to greatly affect the degradation of this material.

The higher carbonyl index of the oven aged samples at embrittlement, or after long periods of time in the oven, indicate that there is more oxidation occurring in these samples compared with the weatherometer aged samples. The creation of reactive sites in the polymer films due to pre-irradiation by UV proceed to be oxidised further in the oven, and can eventually result in embrittlement. By contrast, weatherometer aged samples embrittle earlier and at lower carbonyl

113 indexes, demonstrating the greater effect of non-oxidation related processes, and the burning of the material by titania.

In addition, the higher carbonyl index of the 24 hour UVA and UVC pre- irradiated samples was much more pronounced in the 3% Degussa P25 samples then the 1% Degussa P25 samples. This phenomenon was not found in the weatherometer aged samples. It suggests that at higher concentration of Degussa P25, the wavelength of pre-irradiation becomes less relevant, as there is enough titania to form a sufficient quantity of reactive sites that can induce more rapid polymer aging.

The reactions provided in Scheme 1-21 describe how titania can produce carbon centered radicals. Macroradicals can then proceed to crosslink the material (Section 1.1.1). The information contained in the carbonyl plots seen thus far indicate that this is occurring in the LLDPE, specifically that the titania is absorbing UV irradiation to form a charge separated species, which is giving rise to macroradicals, resulting in degradation processes such as crosslinking.

3.10.5 Section summary Pre-irradiation of polymer film containing 1% Degussa P25 titania with UV light results in faster degradation than non-pre-irradiated samples. UVC has a much more significant effect on this material than UVA. Despite the faster degradation, the polymers are still taking a long time to embrittle in the oven – a film exposed to 3 hours of UVC took over 1 year to embrittle.

The 3% Degussa P25 film behaves differently to the 1% film, with the samples in the weatherometer tending to embrittle earlier. Three hours or less dosage with UVC or UVA did not have a significant effect on the degradation of these films, although 24 hours of pre-irradiation greatly shortened the lifetime in both cases. Titania is having a significant effect on the polymer by introducing reactive sites in the polymer chains, which proceed to react in a dark environment. With increased concentrations of Degussa P25, the wavelength of pre-irradiation light

114 becomes less relevant, provided that the polymer has been irradiated for a sufficiently long period of time.

3.11 LLDPE containing Kronos 1002 (degraded)

The films containing Kronos 1002 titania degraded much more slowly than the Degussa films. Overall the titania appeared to be much less active. Only heavy doses of irradiation affected the degradation outcomes.

3.11.1 1% Kronos 1002, weatherometer aged samples,

0.16

0.14

0.12

0.10

l index 0.08 0 secs UVC 60 secs UVC 0.06 3 hrs UVC

Carbony 24 hrs UVC 0.04 0 secs UVA 60 secs UVA 0.02 3 hrs UVA 24 hrs UVA 0.00 0 5 10 15 20 25 30 35 40 Days aged in weatherometer

Figure 3-25 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 1% Kronos titania and aged in the weatherometer. Polynomial trend lines have been added.

Figure 3-25 shows the carbonyl index plots for all LLDPE films containing 1% Kronos titania and aged in the weatherometer, UVA and UVC pre-irradiated combined. The films generally degraded in about half of the time of the control sample, showing that titania has some effect on the degradation.

115 3.11.2 3% Kronos 1002, weatherometer aged samples,

0 secs UVC 60 secs UVC 0.18 3 hrs UVC 24 hrs UVC 0.16 0 secs UVA 60 secs UVA 0.14 3 hrs UVA 0.12 24 hrs UVA

0.10 l index

0.08 bony

Car 0.06

0.04

0.02

0.00 0 5 10 15 20 25 30 Days aged in weatherometer

Figure 3-26 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Kronos titania and aged in the weatherometer. Second order polynomial trend lines have been added.

Increasing the concentration of Kronos 1002 titania from 1% to 3% resulted in shorter embrittlement times in the weatherometer. Overall, the samples degraded between 15 and 25 days, and the carbonyl index at embrittlement was higher than the 1% samples. Pre-irradiation had less of an effect on the 3% loading film.

116 3.11.3 1% Kronos 1002, oven aged samples,

0 secs UVC 60 secs UVC 3 hrs UVC 0.5 24 hrs UVC 0 secs UVA 60 secs UVA 0.4 3 hrs UVA 24 hrs UVA x

de 0.3 l in ony

rb 0.2 Ca

0.1

0.0 0 100 200 300 400 Days aged in oven

Figure 3-27 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 1% Kronos titania and aged in the oven. Second order polynomial trend lines have been added.

The lack of impact that pre-irradiation had on these samples is show in Figure 3-27. As has been the trend with all samples, the 24 hr UVC pre-irradiated sample showed a significantly different carbonyl index plot, although even this sample took 330 days to embrittle in the oven. None of the other samples had achieved embrittlement at the conclusion of the experiment.

117 3.11.4 3% Kronos 1002, oven aged samples,

0 secs UVC 0.45 60 secs UVC 3 hrs UVC 0.40 24 hrs UVC 0 secs UVA 0.35 60 secs UVA 3 hrs UVA 0.30 24 hrs UVA

0.25 l index 0.20 bony 0.15 Car

0.10

0.05

0.00 0 100 200 300 400 Days aged in oven

Figure 3-28 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Kronos titania and aged in the oven. Second order polynomial trend lines have been added.

The carbonyl index plots of the oven aged 3% loading Kronos 1002 film look very similar to those of the 1% film show in Figure 3-27. Apart from the 24 hr UVC pre-irradiated sample, none of the films had achieved embrittlement.

3.11.5 Section Summary The Kronos 1002 titania appears to be relatively inactive in polyethylene film with respect to UV treatment. Pre-irradiation has not had a significant impact, except for the sample treated for 24 hrs with UVC. The strong photosensitising effect of titania when irradiated with large doses of UVC it’s a common trend in all samples studied. The Kronos samples took a considerable time to age in the weatherometer, and the oven aged samples also degraded slowly. The carbonyl index plots of the oven aged samples are similar to those of the control, indicating that pre-irradiation did not have a strong effect on the outcome of

118 oxidation. Increasing the loading from 1% to 3% resulted in somewhat shortened lifetimes in the weatherometer, although there was little effect on the oven aged samples.

3.12 LLDPE containing Huntsman Tioxide (degraded)

The films containing Huntsman Tioxide titania were reasonably sensitive to UV, degrading in about 15 days in the weatherometer. UV pre-irradiation also affected the rate of degradation of film aged in the oven.

3.12.1 3% Huntsman tioxide A-HR, weatherometer aged This Huntsman tioxide film contained titania that was 100% anatase, and water dispersible. It was more active than the 100% anatase Kronos films. Figure 3-29 shows the weatherometer aging results. Films tended to break down after about 15 days. Pre-irradiation did not have a significant impact on carbonyl index, with the exception of the 24 hour UVC pre-irradiated sample.

119 0 secs UVC 60 secs UVC 0.16 3 hrs UVC 24 hrs UVC 0.14 0 secs UVA 60 secs UVA 0.12 3 hrs UVA 24 hrs UVA 0.10

l index 0.08 bony 0.06 Car

0.04

0.02

0.00 0 2 4 6 8 10 12 14 16 18 20 22 Days aged in weatherometer

Figure 3-29 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Huntsman Tioxide A-HR and aged in the weatherometer. Second order polynomial trend lines have been added.

The sample exposed to 60 secs of UVC gave an apparently anomalous result, taking significantly longer to embrittle. There is no obvious reason for this, and it is postulated that this inconsistency was due to a factor not controlled in this experiment, such as perhaps variable film thickness, higher antioxidant concentration, heterogeneous titania dispersion or human error. When working with real-world samples, such as these films are, anomalous or unusual results can appear quite regularly.

120 3.12.2 3% Huntsman tioxide A-HRF, weatherometer aged

0 secs UVC 60 secs UVC 0.18 3 hrs UVC 24 hrs UVC 0.16 0 secs UVA 0.14 60 secs UVA 3 hrs UVA 0.12 24 hrs UVA x de 0.10 in l y

on 0.08 rb

Ca 0.06

0.04

0.02

0.00 0 2 4 6 8 10 12 14 16 18 20 Days aged in weatherometer

Figure 3-30 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Huntsman Tioxide A-HRF and aged in the weatherometer. Second order polynomial trend lines have been added.

The Huntsman tioxide A-HRF samples behaved similarly to the A-HR equivalent in the weatherometer. The samples pre-irradiated with UVA did not degrade faster than the untreated samples, and the 24 hour UVA treated sample took the longest time to achieve embrittlement, with lower carbonyl index measurements. UVC pre-irradiation did not lead to significantly different carbonyl plots to UVA pre-irradiation, a difference some other films have shown.

121 3.12.3 3% Huntsman tioxide A-HR, oven aged

0 secs UVC 60 secs UVC 0.30 3 hrs UVC 0.28 24 hrs UVC 0.26 0 secs UVA 0.24 60 secs UVA 0.22 3 hrs UVA 0.20 24 hrs UVA x 0.18 de 0.16 l in 0.14 ony 0.12 rb 0.10 Ca 0.08 0.06 0.04 0.02 0.00 0 100 200 300 400 Days aged in oven

Figure 3-31 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Huntsman Tioxide A-HR and aged in the oven. Second order polynomial trend lines have been added.

Pre-irradiation of LLDPE containing 3% Huntsman tioxide A-HR with UVA or UVC did not have a strong effect on the outcome of oven aging, with the exception of the sample pre-irradiated for 24 hours of UVC. Again, it can be seen that this treatment causes earlier embrittlement times, and a higher carbonyl index at embrittlement, however the sample still took 250 days to embrittle.

122 3.12.4 3% Huntsman tioxide A-HRF, oven aged

0.30 0 secs UVC 0.28 60 secs UVC 0.26 3 hrs UVC 0.24 24 hrs UVC 0.22 0 secs UVA 0.20 60 secs UVA 0.18 dex 3 hrs UVA n 0.16

l i 24 hrs UVA 0.14 0.12 0.10 Carbony 0.08 0.06 0.04 0.02 0.00 0 100 200 300 400 Days aged in oven

Figure 3-32 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Huntsman Tioxide A-HRF and aged in the oven. Second order polynomial trend lines have been added.

As with the weatherometer aged samples, pre-irradiation does not appear to have a significant impact on the degradation of the polyethylene film containing 3% Huntsman tioxide A-HRF and aged in the oven. The exception to this is the 24 hr UVC pre-irradiated sample.

3.12.5 Section summary The polyethylene films containing Huntsman tioxide titania demonstrated increased sensitivity to UV radiation than the control sample, although not as much as the Degussa P25 films. The samples degraded in about 15 days in the weatherometer, and did not demonstrate significant differences according to UV pre-treatment.

Pre-irradiation had little effect on the oven aging of these films. The A-HR material showed increased carbonyl index measurements during oven aging with pre-irradiation, but this effect could not be observed in the A-HRF material. The

123 material did not always behave consistently, possibly due to manufacture disparities such as titania distribution and film thickness.

3.13 LLDPE containing Sachtleben Hombitan (degraded)

Generally, the Sachtleben Hombitan films were among the least responsive to UV treatment of all the films containing titania studied. Weatherometer aged samples often took over 20 days to embrittle, and up to 45 days. Only those samples pre-irradiated with 24 hrs of UVC had achieved embrittlement in the oven. Of the 4 films containing Sachtleben Hombitan titania, the LLDPE film containing LW-S-12 (organic coating on anatase microcrystal) was the most active.

3.13.1 3% Sachtleben Hombitan, weatherometer aged Pre-irradiation had very little effect on samples aged in the weatherometer. Figure 3-33 shows the carbonyl index plots for the film containing antinomy doped titania particles. Higher doses of UV pre-treatment resulted in slightly shortened lifetimes in the weatherometer.

124 0.18

0.16

0.14

0.12 x e 0.10 ind l y 0.08 0 secs UVC 60 secs UVC 0.06 3 hrs UVC Carbon 24 hrs UVC 0.04 0 secs UVA 60 secs UVA 0.02 3 hrs UVA 24 hrs UVA 0.00 0 5 10 15 20 25 30 35 Days aged in weatherometer

Figure 3-33 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Sachtleben Hombitan LW-S-U titania and aged in the weatherometer. Second order polynomial trend lines have been added.

The LW-S-U film (Figure 3-33) demonstrated longer embrittlement times than the LW-S-12 film (Figure 3-34). However neither samples showed a significant response to pre-irradiation.

125

0 secs UVC 60 secs UVC 0.18 3 hrs UVC 24 hrs UVC 0.16 0 secs UVA 60 secs UVA 0.14 3 hrs UVA 0.12 24 hrs UVA x de 0.10 l in

ony 0.08 rb

Ca 0.06

0.04

0.02

0.00 0 2 4 6 8 10 12 14 16 18 20 22 Days aged in weatherometer

Figure 3-34 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Sachtleben Hombitan LW-S-12 titania and aged in the weatherometer. Second order polynomial trend lines have been added.

3.13.2 3% Sachtleben Hombitan, oven aged The oven aged samples proved to be similar to the results seen so far. Pre- irradiation with 24 hours of UVC shortened the embrittlement time, however even these samples generally took over 300 days to achieve embrittlement.

126 0 secs UVC 0.35 60 secs UVC 3 hrs UVC 0.30 24 hrs UVC 0 secs UVA 0.25 60 secs UVA 3 hrs UVA 24 hrs UVA

ndex 0.20 i l y n 0.15 Carbo 0.10

0.05

0.00 0 50 100 150 200 250 300 350 400 450 Days aged in oven

Figure 3-35 Carbonyl index plots for UVC and UVA pre-irradiated LLDPE film containing 3% Sachtleben Hombitan LC-S titania and aged in the weatherometer. Second order polynomial trend lines have been added.

Figure 3-35 shows the carbonyl index plots of oven aged samples. Samples which received minimal pre-irradiation appear to be degrading very slowly, and extrapolation from the plots indicates they would take years to embrittle. Heavy pre-irradiation still resulted in an embrittlement time of just less than 400 days.

The LW-S-12 film gave improved results, with the 24 hr UVC pre-treated sample achieving embrittlement in 75 days. The remainder of the samples of this film appear to be following the general pattern. For example, the sample pre- irradiated for 60 secs with UVC does not appear close to embrittlement after over 450 days in the oven.

3.13.3 Section summary The Sachtleben Hombitan films proved resistant to UV pre-irradiation effects. They took longer than other films to embrittle in the weatherometer, and after over 400 days spent in the oven many films still appear intact, with low carbonyl

127 indexes. The most active of these films was the LW-S-12 film; however this film was considerably less responsive to UV treatment than the polyethylene film containing Degussa P25 titania.

3.14 Discussion of the effects of titania

The results provided in this chapter have demonstrated the effect of UV pre- irradiation of LLDPE film, with and without titania, on the relative rates of degradation. UV pre-irradiation decreased the time taken to embrittle, for both weatherometer and oven aged samples. SEM images showed that the Degussa P25 titania exhibited poor dispersion, and a tendency to agglomerate into large particles. Other types of titania were much more evenly distributed, and had a narrower size distribution also.

All LLDPE samples containing titania degraded faster than the control sample, although oven aged samples were difficult to gauge precisely as mostly they had not degraded after over 200 days spent aging. Thus all modified and unmodified titanias used in this study exhibit prodegradant qualities. Of these types of titania, Degussa P25 demonstrated the most reactivity, and samples containing Degussa P25 degraded in ca. 200 days in the dark.

The effect of rutile acting as a dopant to assist in electron-hole separation was discussed in Section 1.2.4. The results from this large study demonstrate the relative strength of Degussa P25 as a photocatalyst, and therefore there is little doubt as to the importance of electron-hole separation when identifying a strong photocatalyst. Additionally, the Degussa P25 titania was not surface modified, and thus there exists a large amount of surface area for oxygen adsorption. The factors affecting titania photoactivity were listed in Section 1.2.3, namely particle size, crystalline structure, phase composition, surface area, nature and concentration of lattice defects, surface hydroxyl groups, and impurities. Degussa P25 exhibits small particle size, high surface area (and thus high availability of surface defects to oxygen), surface hydroxyl groups and low impurities.

128 Kronos titania does not contain a mixed crystal phase, being 100% anatase. Kronos exhibited the lowest photoactivity, and it is thought that this was due to the higher chance of electron-hole recombination. The combination of electron- hole recombination and the lower surface area, leading to reduced availability of surface defects, resulted in Kronos displaying the lowest photoactivity of the titanias investigated in this study.

The Huntsman Tioxide samples were coated with an organic modifier to improve dispersibility. However these coatings appear to have lowered the available surface area, reducing the photoactive potential; likewise for the Satchleben Hombitan coated samples, which also showed poor photoactivity. The antimony- doped titania performed poorly; however this could be due the doping effect of antimony, which has not been reported elsewhere in the literature. Overall, these results indicate that the available surface of titania, particle size (regardless of agglomeration when distributed in an organic phase) and the crystal phase are the dominant factors when considering the potential of titania as a photocatalyst.

Increasing the titania loading from 1% to 3% had little significant affect on the relative rates of degradation of the LLDPE films. It is likely that this has arisen due to agglomeration of the particles, reducing the available surface area. This is supported by the SEM images presented in Section 3.2, which show larger titania particle sizes in the 3% loading samples compared to the 1% loadings for Degussa P25 and Kronos films.

The data presented in this chapter demonstrate the effectiveness of pre-irradiation to accelerate degradation in LLDPE film, even in the dark. By exposing the sample containing 3% Degussa P25 to 24 hours of UVA or UVC, embrittlement was achieved in approximately 200 days. This result indicates that a significant concentration of highly photoactive titania, pre-irradiation with a large dose of UV radiation and higher temperatures are required to greatly accelerate the oxidation of LLDPE film. The continuing challenge for research into this technology will be likely to be focussed on methods to reduce the loading of titania, and improving particle dispersion in addition to utilising a smaller particle size with a narrower distribution. Additionally, it would be desirable to

129 reduce the intensity of UV pre-irradiation by increasing the photosensitivity of the film.

It can be seen from the carbonyl index plots of the control sample aged in the weatherometer that UVC pre-irradiation not only reduces the time taken to embrittlement, but the carbonyl absorption is not as intense also. This effect was also found in LLDPE containing titania. However, this was reversed for the oven aged samples – samples that had undergone 24 hours of UVC irradiation degraded with a much higher carbonyl index.

This phenomenon is explained when considering the effect of UV irradiation on polyethylene. As was written in Section 1.1.3, UV irradiation induces crosslinking and unsaturation. Titania photoreactions also produce macroradicals that can result in similar products (Scheme 1-20, Scheme 1-21). Oxidation reactions are initiated at these reactive sites that have now become sensitive to oxidation, and proceed to spread throughout the bulk (Section 1.1.2).

The more active the titania is, the more reactive sites are produced. In the case of weatherometer aging, degradation processes not showing an oxygenated functional group signature (e.g. crosslinking and cavity formation) dominate the degradation process, reducing the carbonyl index at embrittlement. However for oven aged samples, the reactive sites are attacked by oxygen to produce oxidation products containing a carbonyl functional group. This is in agreement with the literature discussed in Section 1.1.4.

The similarity of the relative carbonyl absorption intensities in all samples strongly indicates that the reaction pathway is not affected by titania. Hence titania is considered to be catalysing degradation of the film by increasing the number of radicals available for reaction. Photosensitisation products such as crosslinks and unsaturation are in higher concentration, resulting in a more rapid rate of oxidation.

Allen had described a stabilising effect of pigment-grade titania particles on organic material when irradiated with UVC (Section 1.2.7). This effect was not

130 found in the course of these experiments, including samples containing pigment grade titania, modified titania or unmodified nano-titania. It is possible that this effect was unique to the sample set that was investigated in that experiment; however those results should not be extrapolated into other systems. It is concluded from the experiments conducted in this thesis that titania has a sensitising, not stabilising, effect on polyolefins irradiated with UVC.

3.15 Conclusions

The activity of the different titanias as a prodegradant in polyethylene is summarised in Table 16.

131

Table 16 Summary of titania activity.

Titania Oven Weatherometer Effect of pre-

Most lifetime lifetime (max) irradiation active Degussa P25 ~200 days 12 days Greatly reduced lifetime in weatherometer and oven with large doses

Huntsman >300 days 15 days Oven samples Tioxide A- showed higher deg. HR/F rate Least Active Sachtleben >300 days 18 days Oven samples Hombitan showed higher deg. LW-S-12 rate

Sachtleben >300 days 30 days Oven samples Hombitan showed higher deg. LW-SU/HD, rate LC-S

Kronos 1002 >300 days 30 days Reduced effect

The carbonyl index plots and comparisons of times taken to embrittlement have revealed that Degussa P25 is the most photoactive titania in LLDPE. Furthermore, the most important titania characteristics that determine the photoactivity of a titania particle are available surface area for oxygen adsorption and crystal phase. Titania particles are thought to create photosensitised regions, which result in faster rates of degradation. Pre-irradiation with 24 hours of UV can result in a plastic film that will degrade in around 200 days in the dark at

132 50 °C. This result demonstrates potential for technology involving pre- irradiation.

In order to develop the technology required to create a LLDPE film with more accurately controllable degradation properties, more information regarding the chemistry of the degradation processes occurring is required. During the course of these experiments a great deal of spectra data have been acquired. These data have been subjected to statistical analysis in order to glean as much information as possible from this unique data set.

133 134 Multivariate Data Analysis

4.1 Introduction

In Chapter 3 it was seen that a large body of data had been collected comparing the degradation of LLDPE containing nano-titania particles from different manufacturers, under different degradation environments, and with different kinds of pre-treatment. While techniques such as spectral subtraction and carbonyl index plots can elucidate some information regarding degradation processes, standard ‘data mining’ methods such as these struggle to cope with the size and complexity of a data set like the one in this study.

Multivariate data analysis (also known as Chemometrics when specifically applied to chemical data) allows the user to highlight aspects of data that are varying in relation to each other. It is especially applicable to large data sets, such as a series of measurements taken over time, and has the potential to extract information that is difficult to see with the unaided eye.

This is not to suggest that multivariate data analysis techniques can see something that is not there. Rather, they allow data to be presented in such as way that only relevant information is examined. In the analogy of ‘mining’ data for information, chemometrics is an excavator and sorter combined.

It is hoped that the spectral data collected on the photooxidation of LLDPE containing titania holds information that may be exploited to produce a polymer with controllable degradation properties. A chemometric investigation is the most likely method of data mining to discover this information, and expand the knowledge base regarding the degradation of polyethylene photosensitised with titania.

135 4.2 Data treatment

Principal Component Analysis (PCA) is a multivariate data analysis technique that decomposes data into one or more components. These components describe the variations that are happening in the data set. The first component, Principal Component 1 (PC1) describes the most significant variance, PC2 describes the second most significant variance, and so forth. Eventually, a PC and all subsequent PCs will describe mostly noise. PCs from this point are not considered significant.

The data must be pretreated in order to remove artifacts such as signal strength variations, baseline differences, or other variations that can affect the outcome of a PCA model. Firstly, the data are normalised, by giving the total area of each spectrum the same value (arbitrarily set to 1). This removes differences in signal strength due to artifacts such as poor contact with the internal reflection element when performing ATR/FTIR, possible light scattering or slight variations in refractive index, etc.. Secondly, the data are mean centered, which creates an average spectrum, and the variables (absorption at a given wavenumber) are described in terms of negative or positive variance from the mean. If the data are not mean-centered, the first PC does not describe variance, but actually shows an average spectrum, and describes how far other spectra are from this average. As our interest is not the mean, but variations around the mean, the data are mean- centered first.

Data that do not contain information relevant to the investigation were removed. The spectral ranges under investigation varied between models, however in all cases data above 1900 cm-1 were removed prior to any statistical analysis, as at low levels of oxidation relevant information is difficult to distinguish from signal noise.

A PCA calculation on a data set is termed a model, and the model needs to be validated with external data to determine its reliability. Cross-validation methods were used, in which one spectrum (a data subset) is removed from the data set (test set) to leave the data set minus the subset (model building set), and the

136 model was recalculated. The spectrum removed was then predicted by the model. Reiterations were performed until every spectrum had been left out. Two cross- validation methods were used according to the number of spectra; if there were less than 20 spectra, then the leave-one-out method was used. In the case of 20 or more spectra, a “venetian blind” model was used, in which regularly spaced spectra (eg 3rd, 4th, 5th etc) from the test set was used in the model building set while the remaining spectra were used as a validation set. Venetian blind cross validation uses less computing power and less time to calculate the model, as leave-one-out cross validation is inappropriate for large data sets due to the complexity of the calculation.

4.3 Analysis of samples subjected to oven aging

4.3.1 Samples without pre-irradiation It is necessary to investigate any effects of titania upon degradation processes in polyethylene in the absence of irradiation in order to distinguish these phenomena from photoactivated effects. The sample containing 3% Degussa P25 demonstrated much higher carbonyl intensity than the control after similar times of aging in the oven at 50 ºC without either sample being subjected to any irradiation (Section 3.10.4). This implied that titania is increasing the rate of oxidation, even in the absence of light.

PCA was used to make a comparison between the control sample, and the sample containing 3% Degussa P25 titania. Both samples were oven aged and did not undergo any UV pre-irradiation.

137 -3 x 10 5 17 54 4

3 35 244 ) 2

1% 00 150 9 . 1 33 18 80 327286 3 (5 24 217 C 0 115 P 94 on

s -1 e r

o 177 07 375 c

S -2 136 61 -3 266 180 -4 225 -5 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 Scores on PC 1 (75.91%)

Figure 4-1 Scores plot for control film (green star) and film containing 3% Degussa P25 (red triangle). Both films were aged in the oven and were not pre-irradiated. The numbers next to each point represent the days spent aging in the oven.

0.25 )

91% 0.2 3 (5.

C 0.15 P on

ngs 0.1 Loadi

), 0.05 91% 0 75. 1 ( C -0.05 on P

ngs -0.1 di Loa -0.15 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 Wavenumber

Figure 4-2 Loadings plots for PC1 (blue) and PC3 (green) from Figure 4-1.

138 Figure 4-1 shows a plot of the scores for PC1 vs. PC3. PC1 describes the general trend of carbonyl absorption development with time, and the inverted peaks of the processing agent spectrum demonstrate that the processing agent is actually disappearing from the spectrum. (An absorption assignment table for the processing agent is provided in Section 3.6). Not only is the processing agent disappearing with more time spent aging in the oven, but the PCA model gives some indication as to whether the processing agent is being oxidised, or simply being lost as a volatile compound.

Comparing the relative scores of the control sample after around 200 days of aging with that of the sample containing 3% Degussa P25 titania after a similar length of aging, it is seen that they have very different scores on PC1. The loadings plot of PC1 is describing variances occurring mostly to the processing agent spectrum. The processing agent absorptions at 1080 cm-1 and around 1375 cm-1 are both being lost as the carbonyl absorption increases in intensity. After 200 days spent aging in the oven, the control samples still have a negative score on PC1, and therefore still contain some amount of processing agent. The heavily oxidised LLDPE film containing 3% Degussa P25 titania, however loses its absorptions due to the processing agent after a similar period of time spent in the oven. Thus we can see that the processing agent is being lost from the more heavily oxidised sample, and it is inferred that this represents an oxidation of the processing agent.

PC3 appears to be describing the difference in degradation processes between that of the processing agent and the polymer. Consider that PC1 tells us that as the carbonyl absorption increases, the absorption peaks of the processing agent decrease. However, PC3 shows us that the methyl absorptions at 1390 cm-1 and 1361 cm-1 of the processing agent are actually increasing with carbonyl absorption. In addition, there is a new absorption at 1100 cm-1 that decreases with increasing carbonyl, and an absorption band at 1280 cm-1 which is also decreasing with increasing carbonyl. Both of these bands are indicative of a low molecular weight ester C-O stretc.h, such as formate or acetate ester.

139 Due to the difficulties in determining the exact nature of the processing agent, interpretation of these data is not straightforward. However, it appears that the processing agent has degraded to form a low molecular weight oxidation product, which may have then further oxidised to form volatile compounds which do not appear in the spectra. The processing agent has disappeared from the spectra of the heavily oxidised sample containing titania faster than from the control sample. The data do not only yield information about the processing agent, however, but also about the LLDPE.

The most interesting information to come from this investigation is the lack of separation between the control and the sample containing titania. Processing agent chemistry aside, the two materials appear to be degrading by the same process. This has been established by the calculation of many PCA models of different systems comparing non pre-irradiated, oven aged LLDPE samples. Rather than present pages of examples showing a lack of separation according to degradation chemistry between the control and samples containing titania, an example of the degradation chemistry of the processing agent has been provided here to demonstrate the information obtained by application of PCA.

When examining scores plots for separation, one is looking for grouping of the data. If groups of data can be identified, then the loadings plots will reveal on what basis they are separated. In this instance PC1 shows significant separation of the control and titania-containing sample, however PC1 is describing the general extent of degradation, and shows that the titania-containing samples are simply more degraded. By comparing the value of the PC1 score and the degradation time, it is clear that the effect of the titania is to speed up the degradation process considerably, however the lack of separation on the PC3 axis tends to indicate that the degradation processes are similar.

PC2 contained information not relevant to degradation chemistry, but showed fluctuations in the strength of the additive absorption and did not separate the data, while PC3 did not separate the control sample or titania containing sample either. It is apparent then that in the absence of irradiation, the control sample is

140 degrading in a similar manner to the titania-containing sample, albeit more slowly.

4.3.2 Samples with pre-irradiation Multivariate statistical analysis of the spectral data obtained from the oven aging of the UV pre-irradiated samples containing 3% Degussa P25 revealed interesting chemistry regarding the effect of UV irradiation. A comparison was made using the carbonyl region of the spectra of the samples that were non- irradiated, 3 hours of UVC pre-irradiated, and 24 hours of UVA irradiated. (24 hours of UVC pre-irradiation was found to be too harsh to make accurate comparisons. 3 hours of UVC pre-irradiation has a more similar effect to 24 hours of UVA pre-irradiation).

0.06

282 123 371 0.04 244 327 95 286 375 330 ) 217

% 0.02 3 4 150 1.

9 241 ( 162 1 140

C 0

P 181 54 70 on 115 80 12 180 199 es

r 35 70 98 o -0.02

c 24

S 61 32 17 -0.04 24

07 -0.06 00 5 10 15 20 25 30 Sample

Figure 4-3 Scores plot for PC1 comparing 3% Degussa P25 containing film non-irradiated (red triangle), 24 hours UVA pre- irradiated (green star) and 3 hours UVC pre- irradiated (black circle). Samples were aged in an oven. The label ‘Sample’ on the x-axis refers to the spectrum number in the series of spectra.

141 0.35

0.3

0.25 ) %

3 0.2 4 1. 9 ( 0.15 1 C P 0.1 on gs n i 0.05 ad Lo 0

-0.05

-0.1 1680 1700 1720 1740 1760 1780 1800 1820 1840 1860 1880 Wavenumber

Figure 4-4 PC1 for Figure 4-3. This PC describes the general variances seen in the spectra during the process of degradation. .

PC1 shows the trend of increased carbonyl absorption with time spent in the oven. This PC describes over 90% of the total variance in the carbonyl region. PC2 and PC3 describe much less, and are discussed below.

142 -3 x 10 8 24 95 6 61 123

4 98 199 70

) 2 07 162 12 % 24 241 140 4 330 282 181 .8 0 1 371

( 375

3 00

C 180 -2 327 17 P n 244 o

s -4 217 e r o

c 150

S -6 54

80 -8

-10 35

-12 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 Scores on PC 2 (4.97%)

Figure 4-5 Scores plot for PC2 and PC3 comparing 3% Degussa P25 containing film non- irradiated (red triangle), 24 hours UVA pre-irradiated (green star) and 3 hours UVC pre- irradiated (black circle).

0.25 ) %

7 0.2 .8 1 (

3 0.15 C P n

o 0.1 s g n i d

a 0.05 o L , )

% 0 9 .3 5 ( -0.05 2 C P n -0.1 o s g n i

d -0.15 a o L -0.2 1680 1700 1720 1740 1760 1780 1800 1820 1840 1860 1880 Wavenumber

Figure 4-6 PC2 (blue) and PC3 (green) for Figure 4-5.

143 The loadings plots show that PC2 describes the absence of the carbonyl absorption at 1713 cm-1 and the presence of a carbonyl absorption at 1732 cm-1. Referring to Section 3.7.2, these are assigned to ketones/carboxylic acids, and esters/ketones respectively. PC3 shows another ester absorption at higher wavenumbers, around 1740-45 cm-1, and also some absorptions around 1860 cm- 1. These latter absorptions are probably due to less common carbonyl containing degradation products involving cyclic structures, anhydrides, lactones and peracids

Figure 4-5 shows that the samples subjected to pre-irradiation score more highly on PC3 than do the non-irradiated samples. Thus we know that these samples contain more of the lesser absorbing oxygenated functions such as anhydrides, lactones, etc. which have absorptions above 1800 cm-1. The samples pre- irradiated with UVA score more highly on PC2, suggesting less carbonyl at 1713 cm-1. The other samples seem to be spread over this PC.

The trends suggested by this analysis are that pre-irradiation of samples containing titania results in slightly higher ester, lactone and anhydride concentrations. This confirms Tidjani’s degradation pathway schematic in Section 1.1.4. In this diagram we see that degradation involving UV results in various products containing carbonyl functionality which can absorb at higher wavenumbers, while the thermal aging route to the left side of the diagram gives rise to acids, which typically absorb around 1713 cm-1, as seen in this multivariate statistical analysis.

4.3.3 UVA vs UVC pre-irradiation: extent of degradation PCA information can be used to obtain plots describing the extent of degradation similar to the carbonyl plots in Chapter 3. An extent of degradation plot using multivariate statistical information has distinct advantages over a conventional plot obtained by plotting the area under the carbonyl absorption. When using a chemometric approach, a much wider spectral range can be used, and selected absorptions that do not contain relevant information can be omitted. This is

144 compared to the carbonyl index plots obtained by measuring the area under the carbonyl peak, such as those presented in Chapter 3.

Additionally, if there is more than one influence leading to variances in a spectrum over time, then the PC that describes the variances occurring only from degradation can be separated from the rest of the spectral information and used to obtain an extent of degradation plot. This is relevant when studying real world materials, where we have already seen the effect of processing agent on the spectrum of the materials being investigated. Finally, a PCA calculation requires only a few minutes to set up, and seconds to calculate. Compare this with conventional carbonyl plots, where depending on the number of spectra, it takes hours of measuring and calculation to obtain the area under the carbonyl peak and plot the results.

Spectral data obtained from the 3% Schatleben Hombitan LW-S-U-HD (anatase crystal with organic coating) titania in LLDPE film aged in an oven at 50 ºC was subjected to a PCA investigation. Two samples were used: one was pre-irradiated with 24 hours of UVC, and the other pre-irradiated with 24 hours of UVA. The PCA results are presented below.

145 0.015 282 241 C096330 0.01 95 162 199

24 61 ) 0.005 00 123 12 93% 79.

( 0 1 C 371 -0.005 on P 330

es 241 r 282

o 199

c 123 162

S -0.01 61 24 -0.015 12

-0.02 2 4 6 8 10 12 14 16 18 20 22 Sample

Figure 4-7 PC1 scores for 3% Schatleben Hombitan LW-S-U-HD titania in LLDPE film aged in an oven at 50 ºC. 24 hours UVA pre-irradiated (red triangle) vs 24 hours UVC pre- irradiated (black circle).

0.2

0.15

0.1 )

93% 0.05 (79. 1

C 0 on P -0.05 ngs

Loadi -0.1

-0.15

-0.2 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 Wavenumber

Figure 4-8 PC1 loadings plot from Figure 4-7.

146 PC1 describes variances occurring with oxidation: the growth of the carbonyl peak, and loss of methyl deformation modes. The data are separated due to the size of the initial carbonyl absorption following pre-irradiation: the UVC pre- irradiated sample entered the oven with a much stronger absorption

Taking the values for the scores values and plotting them against days aged in the oven provides a curve describing the extent of degradation (Figure 4-9).

0.016

0.014

0.012

0.01

e 0.008 r o 0.006 1 sc C

P 0.004

0.002

0 0 50 100 150 200 250 300 350 400 -0.002

-0.004 Days aged in oven

Figure 4-9 PC1 scores against time taken from Figure 4-7. The sample is 3% Schatleben Hombitan LW-S-U-HD titania in LLDPE film, and the black points represent 24 hours UVC pre-irradiated, and the red points 24 hours UVA pre-irradiated. Polynomial trendlines have been fitted. The scores have been offset to allow a better comparison.

It can be seen from the extent of degradation plot in Figure 4-9 that the two trendlines are nearly parallel. This is a clear indication that the rates of degradation in the oven of these samples are the same, regardless of the type of irradiation that they were subjected to initially. A carbonyl index plot taken by measuring the area under the carbonyl absorption and plotted against time is presented in Figure 4-10.

147 0.5

0.4

0.3 x e d in l

y 0.2 n o b r a

C 0.1

0 0 50 100 150 200 250 300 350

-0.1 Days aged in oven

Figure 4-10 Carbonyl index plot for 3% Schatleben Hombitan LW-S-U-HD titania in LLDPE film aged in an oven at 50 ºC. 24 hours UVA pre-irradiated (red) vs 24 hours UVC pre-irradiated (black). Polynomial trendlines have been fitted.

It is evident that the plots presented in Figure 4-9 and Figure 4-10 are significantly different. As the plots derived from PCA contain a great deal more spectral data, and the loadings plots demonstrate that these data are directly related to degradation, it is likely that Figure 4-9 is the more accurate one. The UVC pre-irradiated sample had embrittled after 370 days of aging, whereas the UVA pre-irradiated sample was still intact by this time. It is probable that plotting only the carbonyl area does not reveal all of the degradation related variances such as unsaturation loss or gain and crosslinking, and as the sample reaches embrittlement the carbonyl intensity will start to drop off due to the evolution of small volatile degradation products (Section 1.1.4).

A comparison between PCA derived data and carbonyl index information to graphically represent the extent of degradation is provided in the following plots. For this example, PCA has been performed on the same range in the carbonyl region (ca. 1705 – 1735 cm-1) as the range used for the carbonyl index measurements. The PE film contained 1% Kronos titania and was aged in the oven. Trendlines have been fitted to the data, and the r2 values are reported in the plots.

148

0.12

0.10

0.08

0.06 0 secs (r2=0.28) PC 0.04 24 hrs UVC (r2=0.86) 60 secs UVA (r2=0.04) 0.02 ore on c S 0.00

-0.02

-0.04

0 100 200 300 400 Days aged in oven

Figure 4-11 Extent of degradation plot for selected samples containing 1% Kronos titania and aged in the oven. The carbonyl region selected for PCA analysis corresponded to that used for carbonyl index calculations.

149

0.5

0.4

2 0.3 0 secs (r =0.91) 24 hrs UVC (r2=0.95) l index 60 secs UVA (r2=0.95) 0.2 Carbony 0.1

0.0

0 100 200 300 400 Days aged in oven

Figure 4-12 Carbonyl index plot for selected samples containing 1% Kronos titania and aged in the oven. Considering that a similar region of the IR spectrum of the degraded materials was examined for both plots, it would be expected that the plots should look similar. Indeed, the plots are very comparable, indicating that the two methods are presenting the same information in different ways. This result was found to be repeatable in other series of data, and demonstrates that data obtained via PCA for extent of degradation plots are reliable and can be used to show the progression of oxidation in a series of data.

The most significant difference between the two types of plots is the closeness of fit. The r2 values strongly suggest that the carbonyl index data are more robust, especially at low levels of oxidation. This is reflected in the plots shown in Figure 4-9 and Figure 4-10, where the carbonyl index data also show better fit.

It appears that despite the advantage of being able to examine a broader range of spectral data when employing PCA to determine the extent of degradation, carbonyl index derived plots provide better accuracy. Therefore one should

150 consider the region(s) of spectra that ought to be considered, and the accuracy of the data required, when deciding which plot is the most suitable for a given task.

4.3.4 Section Summary PCA is a useful tool for the exploration of large amounts of spectral data. It can sometimes highlight very subtle differences, and the ability to choose certain areas of the spectra for investigation is a considerable advantage over conventional techniques. Scores and loadings plots provide an easily interpretable method of data representation.

UV pre-irradiation of LLDPE film containing photoactive titania with subsequent oven aging has produced some differences in the resulting degradation products. UV irradiation has formed more complex degradation structures, including anhydrides, esters and possibly some cyclic oxygenated functions. These are found only in very small concentrations, and most of the aging products give absorptions typical of acid and ketone carbonyls. This is in close agreement with Tidjani’s degradation pathway schematic.

Information regarding the extent of degradation can be obtained from PCA data plotted against time. There are several advantages in this method, including a broader range of spectral data that can be analysed, and the speed of calculation. However, a comparison of carbonyl index plots and PCA derived plots over the same spectral range reveals that the carbonyl index data are more robust with regards to closeness of fit.

4.4 Weatherometer aging

In Section 4.3 it was seen that PCA can be used to extract data regarding the distribution of degradation products, and to obtain information relating to the rate of degradation similar to carbonyl index plots. This section will apply PCA as an exploratory tool to mine IR spectral data obtained from samples aged in the weatherometer.

151 4.4.1 Water vapour Investigation of the data obtained from the weatherometer aging of LLDPE film containing Degussa P25 and Huntsman titanias (these are listed as the most active titanias in Table 16, Section 3.14) revealed the presence of water vapour in the spectra.

-3 x 10 6

15 4

39 12

) 06 03 36 21 00 % 2 09

6 30 8 .

0 51

4 ( 48

C 0 18 06 66 57 24 03 12 15 on P 54 ores

c -2 63 S 60 33 36 -4 27

-6 03 -0.04 -0.02 0 0.02 0.04 0.06 0.08 Scores on PC 1 (91.16%)

Figure 4-13 Scores for PC1 and PC4 for the LLDPE film containing 3% Huntsman A-HR titania (100% anatase, water dispersible; red triangle) and control sample (black circle). Both samples were aged in the weatherometer and were not pre-irradiated

152 0.12

0.1

0.08 ) 0.06 86% . 0 (

4 0.04 C P n 0.02 o s g n 0 oadi L -0.02

-0.04

-0.06

1300 1400 1500 1600 1700 1800 Wavenumber

Figure 4-14 Loadings plot for PC4 from Figure 4-13.

The distribution of samples on PC4 (Figure 4-13) shows a grouping of the samples containing titania. Examination of the loadings plot of PC4 in Figure 4-14 reveals a ‘spectrum’ that contains signature absorptions (small, sharp absorptions sitting above the noise) of a water vapour spectrum. The titania containing samples scored highly on this PC, indicating that there is relatively more of this spectrum contained in their spectra. The presence of water vapour is not unique to this sample: it can be found in most of the spectra of degraded LLDPE containing active titania types. It should be noted that PC4 concerns less than 1% of the total variance in the spectra between 1900 cm-1 and 1200 cm-1, and demonstrates the effectiveness of PCA in detecting small phenomena in complex spectra, and its ability to highlight these differences and present them in a visually comprehensible manner.

To the best of the author’s knowledge, water vapour has not been found trapped in plastic film containing photoactive prodegradants before, and no literature discussing such a phenomenon could be found. Water vapour was not found in any of the oven aged samples, nor was it found in the weatherometer aged

153 samples containing titania types that were relatively inactive, such as those manufactured by Kronos or Satchleben Hombitan.

Section 1.1.1 listed the reactions involved in oxidation of a polyolefin. The initiation and propagation steps are reviewed here:

Initiation: By radical generator I (initiator) 2r

r + RH rH + R By hydroperoxide ROOH R + HOO

HO ROOH RO + Scheme 4-1

Propagation

R + O2 ROO

ROO ++RH ROOH R

RO 2ROOH + ROO + H2O Scheme 4-2

Comparing these with the reactions involving titania;

Acceptor:

e+ O2 O2

e + OH H2O2 + OH

e + R + H RH Scheme 4-3

154

Donor:

h +OO2 2

+ OH h H2O + H

h + RH R + H Scheme 4-4 it can be seen that the water vapour likely plays an important role in the degradation process.

It was mentioned in Section 1.2.2 that exciton holes (h+) play an important role in titania-catalysed degradation. Thus absorption of UV light by the titania particle produces a hole, which gives rise to a macro radical and a hydrogen radical. The macro radical reacts with oxygen in the propagation step to produce a hydroperoxide, which among other possible reactions can combine with another hydroperoxide to produce a water molecule (3rd step of Scheme 4.4-2). It is conceivable that this water molecule can then react further with exciton holes to produce hydroxyl radicals, which may then act as initiators for new degradation reactions. These reactions are summarised in Scheme 4-5 below.

155 h + RH R + H

I (initiator) 2r

r + RH rH + R

R + O2 ROO

ROO ++RH ROOH R

2ROOH RO + ROO + H2O

+ OH h H2O + H Scheme 4-5

According to this reaction pathway, to produce a water molecule two hydroperoxide molecules must combine. Therefore hydroperoxides would need to be in a relatively high concentration for such recombination to occur. It is expected that the more active a titania is in producing radicals, the more hydroperoxides are formed, and the higher the likelihood of two hydroperoxide molecules combining to produce a water molecule.

The cavities produced by titania vigorously oxidising the polymeric material (Section 3.2) also play a role. The water vapour was noticed in the most oxidised materials, and these materials also whitened, which was attributed to the presence of cavities (Section 3.3) and confirmed by SEM. These cavities contain water vapour in high enough concentrations to be detectable in the infrared spectra via PCA. As mentioned earlier it is possible that the water re-enters the reaction cycle to produce more degradation initiating molecules, however as the water vapour is apparently collecting inside the cavities, it is more probable that the exciton holes preferentially react with carbon chains rather than with water.

156 4.4.2 UVA vs. UVC pre-irradiation PCA analysis of samples pre-irradiated with UVA or with UVC and aged in the weatherometer showed no separation on principal components. Pre-irradiation did not significantly enhance degradation of the samples, and the degradation- related information in the IR spectra was found to closely represent the information found in the carbonyl plots. In many cases, and especially in the more heavily pre-irradiated samples and those samples with the most photoactive titania types, there were insufficient data points, (i.e. insufficient number of spectra) to form a credible PCA calculation. This is due to the short times to embrittlement, and therefore in some cases only 2 or 3 spectra were acquired.

4.4.3 Section summary LLDPE film containing photoactive titania and aged in the weatherometer formed cavities caused by the degradation of material around the titania particles. Water vapour collected in these cavities, and is detectable in the infrared spectra. It is likely that this water is created by the combination of two hydroperoxide moieties, and it is also possible that it re-enters the reaction pathway to create more oxidation initiating species. Pre-irradiation with UVA or UVC has little effect of the degradation outcome of LLDPE film containing titania and aged in the weatherometer, with the samples unable to be distinguished by PCA.

4.5 Conclusions

Multivariate data mining techniques have been used to explore large amounts of spectral data, providing an advantage over conventional techniques not just in time saved to perform an analysis, but representation of the data in visual ways that enhance relevant areas of variance. Principal component analysis has been used to effectively explore the IR spectral data collected from LLDPE film containing titania and subjected to different forms of pre-treatment and aging conditions.

Pre-irradiation with UV was found to promote degradation, decreasing the time taken for the LLDPE to degrade in the oven. Some differences, such as higher

157 ester concentrations, were found in the spectra of those samples that underwent more significant periods of pre-irradiation and contained active titania types. In general, however, the titania served to enhance the degradation process without greatly changing the types of degradation products formed. The experimental evidence here supported Tidjani’s degradation pathway schematic presented in 1.1.3.

PCA information was also used to create plots showing relative rates of degradation. Using PCA scores of a degradation-related PC has some clear advantages over using conventional carbonyl index for such plots, including the ability to select only particular or relevant areas of the spectrum to be analysed, and the speed of calculation. However, a comparison of carbonyl index plots and PCA derived plots over the same spectral range revealed that the carbonyl index data have a closer agreement, and fitted trendlines possessed a superior closeness of fit.

A potentially important discovery found through application of PCA was that of water vapour, which was established to reside in the cavities in the LLDPE film produced by titania when subjected to aging in the weatherometer. The water molecules are likely to be produced by the combination of two hydroperoxides. It is possible that water molecules may then re-enter the reaction pathway by reaction with titania-generated exciton holes to produce more oxidation initiation species. The relevance of water is titania photoreactions is discussed in Section 1.2.4.

Furthermore it has been established that titania does not change the degradation pathway, however, by sensitising regions of the polymer when subjected to UV light it behaves as a photocatalyst, increasing the overall rate of degradation. This is in agreement with the observations discussed in Chapter 3. It is therefore relevant to examine the changes occurring around titania particles in order to detect reactive regions, and examine any regions of faster rates of degradation. Recent changes in IR technology have enabled scientists to reach beyond the traditional limits of spatial resolution, allowing the investigation of heterogeneous oxidation via mid-IR spectroscopy.

158 Obtaining spatial information around titania particles via a model polymer system

5.1 Introduction

To this point the focus of the thesis has involved the investigation of data acquired by taking single-point mid-IR spectra of the degraded LLDPE films containing various titania types from different manufacturers. By characterising the oxidation products and comparing the relative rates of degradation, it has been possible to determine the effects of pre-irradiation on the degradation processes of bulk LLDPE containing a prodegradant.

The purpose of pre-irradiation was to initiate oxidation reactions, which would then propagate further oxidation throughout the bulk (see Section 1.1.2). Considering the success of pre-irradiation in shortening the lifetimes of LLDPE film aged in an oven, it is likely that the film degraded heterogeneously, as shown by the SEM images which depicted the heterogeneous distribution of titania particles. Higher concentrations of oxidation products are expected in the close vicinity of the photoactive titania nanoparticle centres.

There would be obvious advantages to obtain chemical information regarding the heterogeneous degradation processes occurring around the titania particles and propagation of oxidation from particles into the bulk, to assist in the development of a polymer film that will degrade quickly and controllably. Information such as the rate of spreading, optimum distance between particles, optimum particle size, etc. is obtainable if one can monitor the spatial progress of degradation throughout the bulk.

Until now it has not been possible to obtain chemical data following heterogeneous oxidation around prodegradant particles in the infrared. Although chemiluminescence has been used to demonstrate heterogeneous oxidation10,13,192, this technique does not provide the wealth of information regarding the development of oxidation products, and thus the ability to trace the oxidation

159 pathway, that is available in the mid-IR. Previously oxidation data has been acquired from single points (such as the data presented so far in this thesis), and in order to obtain spatial information line mapping with ATR/FTIR has been employed.152

Difficulties arise when seeking to monitor the development of oxidation around a selected prodegradant particle using ATR/FTIR (transmission-mode IR spectroscopy is generally unsuitable due to sample thickness, interference fringes in films and poor spatial resolution as discussed in Section 1.3.3). ATR/FTIR requires the sample to be in optical contact with IRE, which is most commonly achieved by the application of controlled pressure. However the IRE cannot remain in constant contact with the sample during oxidation, as the sample must be subjected to accelerated oxidation conditions between spectra acquisition.

This introduces two hindrances to the study of oxidation. Firstly, the identical spot on the sample surface must be re-located between accelerated oxidation to study the environment surrounding the same titania particle, which can be difficult for some samples. Secondly, ATR measurements on an identical location require repeated application of pressure to obtain spectra. Mechanical stresses have been demonstrated to affect photochemical degradation rates of polymers 193, and it is therefore reasonable to suggest that repeated ATR contact will affect degradation processes, resulting in an incorrect model of the degradation pathway.

The imaging ATR/FTIR study of the oxidation of a model aliphatic polymer presented in this chapter addresses both of these problems, as discussed in further paragraphs. Additionally the spatial information obtained via imaging ATR/FTIR (Section 1.3.3.3) is ideally suited to the study of heterogeneous oxidation, conditional to the size of the heterogeneous domains under investigation.

The size of heterogeneous domains largely determines the suitability of a particular technique to obtain spatially resolved chemical information. Referring to the SEM images of Degussa P25 titania in LLDPE (Section 3.2.1, Figure 3-1), some titania particles are up to 5 µm in diameter, and it is therefore reasonable to

160 suggest that a minimum lateral resolution of 5 µm is required in order to observe chemical changes occurring around a titania particle for this study.

Imaging ATR/FTIR with an IRE of high refractive index can provide lateral resolution of up to 3 – 4 µm159. Theoretical aspects of these methods have been discussed in the introduction (see Section 1.3.4). Furthermore, the ability to obtain hundreds of spectra in one image (Section 1.3.4.1), and the relative accessibility of the instrumentation (compared to, for example, a synchrotron light source) promises potential for the investigation of heterogeneous oxidation around titania centres.

The novel concept presented in this thesis involves the solvent casting of an aliphatic model polymer directly onto the IRE surface. The method, described in detail in the following section, circumvents the need to re-locate the identical location on the polymer surface, and ATR pressure is not applied as the material is in good optical contact from the start. Contingent upon a prodegradant particle being within the imaging area on the IRE surface, heterogeneous oxidation in a sensitised region around a particle can be monitored in real time, without the need for removal of the sample from the IRE surface.

5.2 Experimental

The investigation of degradation around a titania particle was performed using an imaging ATR/FTIR spectrometer at Queensland University of Technology (QUT). Imaging ATR/FTIR is a mid-IR spectroscopic technique that collects spectral information in a spatial context, to a lateral resolution of up to 4 µm159. Factors affecting the lateral resolution capability, and other aspects of imaging ATR/FTIR, have been discussed in Section 1.3.4.1. The imaging ATR/FTIR spectrometer used for the research presented in this thesis employed a 32 x 32 Focal Plane Array (FPA) detector, resulting in images containing 32 x 32 pixels. Each pixel represents the spectrum of a specific part of the sample imaged onto a particular MCT detector element in the FPA. Images are created in a number of different ways including band area or intensity, ratios of band areas or intensities, and more complex methods such as principal component analysis. In this case

161 images were constructed by ratioing the area under the carbonyl absorption to the

CH2 deformation absorption, and the pixels were assigned a colour on an arbitrary colour scale according to the numerical value of the ratio result, with red showing high values and blue showing low values.

LLDPE is unsuitable for these oxidation studies as it is largely insoluble at room temperature, and the likelihood exists of oxidation reactions occurring at the high temperatures required to dissolve polyethylene. The experiment is designed to investigate the heterogeneous oxidation of polymeric materials, and is not restricted to polyethylene. The suitability of several model aliphatic polymers (for example polyisobutylene, polypropylene) was investigated, and it was experimentally determined that Topas® (an aliphatic polymer containing a norbornene moiety (Figure 5-1)) was most appropriate.

x y Figure 5-1 Molecular structure of Topas®.

Topas was dissolved in cyclohexane and solvent cast onto the IRE surface shown in Figure 5-2. Degussa P25 titania was deposited onto the surface and the whole assembly was exposed to a total of 8 hours of UVC according the experimental procedure recorded in Section 2.4. Images were recorded hourly, and they are presented in Figure 5-3.

162 Ge Internal Ref lection Element

Figure 5-2 ATR/FTIR objective assembly.

5.3 Imaging ATR/FTIR spectroscopy results

The images presented in Figure 5-3 immediately begin to show an increase in the amount of light blue (indicating an increase in carbonyl absorption) after 1 hour of irradiation, and after 5 hours there appears to be a significant concentration of oxidation products that register 0.05 on the carbonyl index scale. By 6 hours the degradation begins to occur more rapidly, and after 8 hours of irradiation the oxidation is quite extensive.

163 Figure 5-3 Images taken of Topas containing TiO2 and irradiated with UVC. Each figure is labelled with the cumulative irradiation time. The numbers on the x and y axes represent the number of pixels. Each pixel is 1.2 µm in width. The pixels have been smoothed using Varian software.

164

The oxidation process occurring is clearly heterogeneous, as Figure 5-3-I contains pixels ranging in colour from green to red, covering 0.05 to 0.15 on the carbonyl index. Importantly, the heterogeneity is occurring with a domain size of around 5-10 pixels, or 6-12 µm, across (see Figure 5-4 below). These domains of more rapid oxidation are thought to be the photosensitised regions, caused by titania particles as discussed in Section (3.14).

These images demonstrate the applicability of imaging ATR/FTIR to the heterogeneous investigation of the oxidation of polymers. For imaging ATR/FTIR to be suitable, the domain size of the heterogeneity would need to cover at least 3 pixels, translating to approximately 4 µm in diameter. Comparison with the size of some of the larger titania agglomerates shown in the SEM images of LLDPE containing Degussa P25 (Figure 3-1), this lateral resolution would be adequate for the detection of titania particles and oxidation in the surrounding polymer.

Domains of higher oxidation product concentration

Domain of lower oxidation product concentration

Figure 5-4 Image taken from Figure 5-3I, illustrating heterogeneous domains.

5.3.1 Determination of titania particle location(s) The current FPA technology prevents direct detection of titania particles on the surface of the IRE, as the spectral range does not reach below 900 cm-1, which is higher than the Ti-O absorption in the mid-IR. An ATR/FTIR spectrum of the Degussa P25 titania used in this experiment is included in Figure 5-5, showing

165 the strong Ti-O absorption below 800 cm-1. The exclusion of this region from the FPA spectral range forces reliance on the OH stretc.hing absorption above 3000 cm-1 and bending vibration at 1635 cm-1 in order to detect titania directly.

0.2

0.1 sorbance b A

0.0 3500 3000 2500 2000 1500 1000 Wavenumbers (cm-1)

Figure 5-5 ATR/FTIR spectrum of Degussa P25 powder. The O-H stretc.hing absorption (3600 – 3000 cm-1) and bending absorption (1635 cm-1) are due to hydroxyl groups on the surface of the titania particles.

Other factors in this experiment affect the ability to detect titania. Importantly, the likely location of titania particles must be considered with respect to the depth of penetration of the IR light into the polymer film. Furthermore the surface of the film being measured is not the surface in direct contact with oxygen or under direct UV irradiation, which may also affect the spectra.

166 Topas film Imaged area Titania particle

190 nm IRE

Figure 5-6 Schematic showing Topas film with titania cast onto IRE surface. The depth of penetration of ATR/FTIR is shown by the dashed black line. The titania particle locations are hypothetical.

Figure 5-6 is a schematic representing the Topas film containing titania particles cast onto the IRE surface. The Harrick equation presented in Section 1.3.3 can be used to determine the depth of penetration (dp) of the IR radiation into the Topas film, which at 3500 cm-1 is;

2857 nm dp = 2 2 1/2 2π x4(sin 45 - (1.5/4) )

= 190 nm with an ATR angle of incidence of 45 °, and a refractive index of 1.5 for Topas194. A penetration depth of 190 nm is not a well defined cut-off point, as the strength of the signal from the evanescent waves weakens exponentially as they move through the sample. However it provides an approximate depth to which IR analysis is performed, and hence in order to observe hydroxyl stretc.hing absorptions from a titania particle the particle ought to be within approximately 190 nm of the IRE surface. This is represented by the schematic in Figure 5-6, with a titania particle slightly impinging on the edge of the imaged area. .

The images in Figure 5-3 show the earliest signs of oxidation in the upper left corner. Any correlation between sensitisation by titania and the relatively rapid oxidation occurring at this location can be substantiated by a relatively higher

167 OH stretc.hing absorption strength in this corner of the image acquired before oxidation UV irradiation had commenced.

The poor signal-to-noise ratio of these spectra largely prohibited the use of conventional spectral exploratory tools such as spectral subtraction and area comparison, and so PCA was used to analyse the spectra obtained in the image shown in Figure 5-3-A, i.e. Topas containing Degussa P25 titania prior to UV irradiation. It was hoped that due to OH absorptions on the surface of the titania which are visible in the mid-IR spectra, any separation in the data according to an absorption intensity difference in this region might indicate the presence of titania.

0.15 0hr414 0hr931 0hr905 0hr365 0hr561 0.1 0hr22 0h7r932 0hr504 0hr708 0hr5 430hr503 0hr486 0hr994 0hr929 0hr36 0h2 r807 0hr7 450hr524 0hr933 0hr1022 0hr9 540hr470 0hr318 0hr312 0hr267 0hr937 ) 0.05 0hr843 0hr901 0hr452 0hr6 740hr965 0hr463 0hr393

33% 0hr838 0hr309 0hr805 0hr1023

0. 0hr1021

1 0hr8 730hr831 0 0hr775 0hr1009 0hr922 0hr77 70hr1018 2 ( 0hr874 C 0hr927 0hr1015 0hr763 0hr707

P 0hr923 0hr400 0hr244 0hr990 -0.05 0hr314 0hr489 0hr732 on 0hr430 0hr5 200hr796 0hr830 0hr254 0hr457

es 0hr773

r 0hr706 0hr545

o 0hr454 0hr616 c 0hr744 0hr590 S -0.1 0hr986 0hr3 860hr812 0hr387 0hr974 0hr100 0hr442 0hr370 0hr97 0hr7 0hr99 69 0hr268 0hr427 -0.15 0hr419

-0.2 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 Scores on PC 1 (74.80%)

Figure 5-7 Scores plot for PCs 1 and 2 based on the OH stretc.h region of the spectra obtained in the image from Figure 5-3-A. The spectra are numbered from 1 to 1024, starting from the lower left corner of the image, and increasing sequentially from left to right. Every 32nd spectrum begins a new row above the previous row. The hr prefix refers to the time of UV irradiation. All spectra have been obtained from the image of the sample prior to irradiation, hence the 0hr prefix before all sample labels.

168 Figure 5-7 shows the scores plot for PCs 1 and 2 based on the OH stretc.hing region of the spectra. Figure 5-8 and Figure 5-9 are the corresponding loadings plots, accounting for 75% and 10% of the data variation respectively. Following the description of the spectra labelling method provided in the caption to Figure 5-8, the spectra in the upper left corner of the image are labelled higher than 800 in the series, and belong to the first 5 in every row of 32 spectra. Thus some separation of these spectra from the other spectra is the image is sought, which might indicate the possible location of surface OH groups on a titania particle.

Figure 5-7 shows that there is separation of the data on PC1, corresponding to the location of the spectra in the image. Thus it would appear that PC1 is describing some systematic artefact present in the spectra due the imaging technique, and not the presence of titania. Thus this PC was discounted from the investigation, and PC2 was explored, which was shaped more in accordance with a likely OH absorption.

0.13

0.128

) 0.126 80% 74. ( 0.124 1 C P

on 0.122 ngs

0.12 Loadi

0.118

0.116 3100 3150 3200 3250 3300 3350 3400 3450 3500 3550 Wavenumber

Figure 5-8 Loadings plot for PC1 from Figure 5-7.

169 0.2

0.15

0.1 ) 33% .

0 0.05 1 ( 2

C 0 P n o -0.05 ngs

Loadi -0.1

-0.15

-0.2 3100 3150 3200 3250 3300 3350 3400 3450 3500 3550 Wavenumber

Figure 5-9 Loadings plot for PC2 from Figure 5-7.

The scores plot of PC2 in Figure 5-7 contains 1024 spectra, resulting in a complicated plot from which it is difficult to observe any clear discrimination of the data. It is expected that if PC2 is describing an OH stretc.hing absorption signature to a titania particle, and assuming photosensitisation of the upper left corner of the imaged area by titania, then the cumulative scores in this area of the image might be higher than another area of the image were oxidation was slower to occur.

Two 5 x 5 pixel matrices were selected from the image of the Topas film prior to irradiation, the first from the upper left corner, and the second from an area demonstrating a slower oxidation rate. These areas are pictured in Figure 5-10.

170 1

Figure 5-10 Image of Topas on IRE prior to irradiation. The boxed area labelled 1 represents a region that oxidised quickly, compared to the boxed area labelled 2, which oxidised more slowly.

The PC2 scores for each pixel was determined and summed to determine the ‘total score’ of the 5 x 5 pixel area. Figure 5-11 and Figure 5-12 show the spectra number corresponding to these pixels, and the score of each pixel on PC2, with a summed total.

171 Column 123 456 Row

32 993 994 995 996 997

31 961 962 963 964 965

30 929 930 931 932 933 29 897 898 899 900 901

28 865 866 867 868 869

123 456

32 -0.024 0.095 0.076 0.014 -0.03

31 0.075 0.015 0.004 -0.005 0.040

30 0.089 0.041 0.129 0.110 0.078

29 0.047 0.087 0.078 0.059 0.043

28 0.001 -0.01 0.01 0.002 -0.004

Figure 5-11 5 x 5 pixel selection from Box 1 in Figure 5-10 showing the spectrum number (left) and the scores on PC2 for each pixel (right).

Sum of scores in each pixel for PC2 in Box 1 = 1.03

172 Column 16 17 18 19 20 21 Row

10 304 305 306 307 308

9 272 273 274 275 276

8 240 241 242 243 244 7 208 209 210 211 212

6 176 177 178 179 180

16 17 18 19 20 21

10 0.006 0.007 0.073 0.049 0.037

9 0.012 -0.040 -0.017 -0.045 0.083

8 0.020 0.025 -0.042 -0.034 -0.041

7 -0.027 0.006 0 -0.012 -0.019

6 0 0.003 -0.010 -0.007 0.013

Figure 5-12 5 x 5 pixel selection from Box 2 in Figure 5-10 showing the spectrum number (left) and the scores on PC2 for each pixel (right).

Sum of scores for each pixel for PC2 in Box 2 = 0.013

It can be seen that area represented by Box 1 in Figure 5-10 has a much high cumulative score on PC2 than the area in Box 2. PC2 appears representative of OH stretc.hing absorption, and it is concluded that there is a strong likelihood that the area of the film corresponding to the upper left corner of the image contains a titania particle(s), determined by detection of the OH stretc.hing absorption, which has resulted in sensitisation to oxidation of the Topas film.

While this result demonstrates the probability that the images presented in Figure 5-3 are showing the heterogeneous degradation of polymer film sensitised to

173 oxidation around titania particles, relative rates of oxidation can be examined by comparing the carbonyl index values of the spectra acquired over the course of UV irradiation.

5.3.2 Discussion of heterogeneous oxidation In order to examine the images to determine the spread of oxidation, carbonyl index values were calculated for the spectra that fell across a line map drawn through the 4th pixel of each row, as shown in Figure 5-13. These values have been plotted against their corresponding row number, and displayed in Figure 5-14. Some line maps have been omitted for clarity.

Figure 5-13 Cross section of image to plot carbonyl index.

174

Figure 5-14 Carbonyl index line maps for the 4th spectra in each row. Only selected line maps are shown for clarity.

The images presented in Figure 5-3 indicate that where the line map has been acquired the carbonyl index is quite variable, and this is represented by the significant point-to-point variations in the line maps of Figure 5-14. Overall there is a trend towards higher values at the right side of the plot, correlating with the upper left corner of the images. This corresponds with the observation of the domain of higher carbonyl intensity found in the images, thought to be around titania-sensitised regions.

The relative rates of oxidation can be obtained by plotting the carbonyl index for each pixel against time. Points corresponding to regions of high (row 29) and low (row 7) carbonyl index have been plotted, along with the average of all points for comparisons sake. Figure 5-15.

175

Figure 5-15 Rates of degradation for every 5th pixel and average of all pixels.

Figure 5-15 reveals the photosensitising effect of titania on Topas. There is a clear induction period during which little or no carbonyl moiety containing degradation products are formed, however the length of this induction period is 1 hour longer at point corresponding to row 7. At row 29, where it was previously shown that there is a high probability of a titania particle in the immediate environment, the induction period only lasts 2 hours. Once the induction period has ended the polymer tends to degrade at a rate apparently independent of titania photosensitisation effects.

The adduced evidence strongly indicates that titania is present in the imaged location of the solvent-cast Topas polymer, and that the titania has had a photosensitising effect, resulting in heterogeneous degradation and more rapid oxidation of the polymer surrounding the titania. This experiment is considered a significant step forward for the study of heterogeneously oxidising polymers using infrared spectroscopy, for the following reasons:

• It provides an advantage over other spatially-resolved techniques such as chemiluminescence imaging by supplying chemical information contained within the mid-IR spectra. This can used to trace the degradation pathway of materials, examine different oxidation products

176 and identify domains of varying sensitivity to oxidation that might correspond to different concentrations of components in a polymer blend.

• Polymer oxidation can be studied in real time, and as data are collected from the same area of polymer, the effect of impurities such as catalyst residues can be monitored.

• Samples are not subjected to pressure from ATR/FTIR techniques, removing the influence of mechanical stress on the oxidation process. Additionally, the same area of the polymer does not need to be re-located.

• Imaging ATR/FTIR with an IRE of high refractive index allows for spatial resolution of around 4 µm, which is a large improvement on physical limitations of transmission spectra through a medium of air. This also allows for the collection of a large amount of data on a small sampling area, dependent on the number of MCT elements in the FPA.

There are some technical issues regarding polymeric materials that need to be addressed before this technique can be used for routine analysis of heterogeneously degrading polymers. These include:

• Polymers need to be dissolved in solvents at room temperature to avoid initiation of oxidation reactions at elevated temperatures. These solvents are also required to be volatile and non-aggressive to the bond holding the IRE into the ATR assembly. For certain polymers, such as polyethylene, finding the appropriate solvent could prove to be quite difficult.

• Certain polymers have a tendency to shrink due to crosslinking when oxidising, and lift off the IRE surface. Additionally, polymers tend to lift off the IRE surface during solvent volatilisation.

• It is difficult to determine the imaging location on the IRE surface. This might affect heterogeneous polymers with a low concentration of a

177 secondary component, such as catalyst particles, as there is a reduced likelihood that imaged area will contain a particle.

• There is no clear method for determining the thickness of the polymer film once it has been solvent cast.

Imaging ATR/FTIR is still a new field of infrared spectroscopy, and the instrumentation is under continual research and development. Some areas of instrumentation that require improvement include:

• Poor signal-to-noise ratio of the FPA

• FPA spectral range cut-ff at 900 cm-1, which prevents the identification of signature absorptions of some materials and oxidation products.

• Varying baselines, anomalous dispersion and bad pixels can result in two spectra of the same material appearing dissimilar, with dissimilar absorption intensities.

• Attenuation of signal, resulting in the need for low spectral resolution and a high number of scans.

5.4 Conclusions

A novel experiment in which a model polymer system containing titania was photooxidised and imaged in real time to demonstrate the heterogeneous development of oxidised domains. It is thought that these domains of higher carbonyl product concentration are likely due to the photosensitisation effect of titania particles in the immediate vicinity. The existence of such domains implies that the titania is not homogeneously distributed throughout the polymer.

PCA was used to demonstrate the existence of a region of greater OH absorption in the upper left corner of the images, corresponding to where the carbonyl

178 concentration was most intense. The Degussa P25 titania mid-IR spectrum contains OH signature intensities, present as functional groups on the surface of titania. It is considered likely that titania in this region contributed to the OH absorption in the images, and the titania has sensitised regions of the Topas to photooxidation. This was supported by the greater overall gain of carbonyl absorption intensity in the spectra acquired from the region thought to contain titania.

An induction period was found to prelude more rapid acceleration of oxidation rate, which demonstrated the photosensitising effect of titania. Areas that were likely to contain titania oxidised rapidly after 2 hour of UV exposure, compared to 3 hours of regions further from titania particles.

The novel technique presented represents a significant advance in imaging ATR/FTIR spectroscopy, and has acquired previously unobtainable data in real- time. In particular the ability to acquire spatially-resolved chemical data without the need to force ATR/FTIR contact or re-locate a position on the sample surface is advantageous. While it holds great potential for the study of heterogeneous systems, there are a number of technical and instrumental issues that need to be addressed before this can become a routine technique.

179

180 Investigation of degradation in the mid-IR using a synchrotron light source

6.1 Introduction

The advantages of high lateral resolution in imaging or mapping spectroscopy were highlighted in the previous chapter (Section 5.35.1), where the evolution of oxidation product formation was observed in a domain 5 µm in diameter. IR spectroscopy with a synchrotron light source is another technique that has shown potential to achieve high lateral resolution195, and hence is expected to have some usefulness in the investigation of the spread of oxidation from titania catalyst particles at the very early stages of degradation.

This study has examined the suitability of IR spectroscopy with a synchrotron light source to investigate the titania-photocatalysed degradation of polyethylene film, with a view to the acquisition of data with a lateral component at the earliest stage of oxidation.

6.2 Experimental

The experimental procedure was described in Section 2.5; however some further explanation of the sample choice is required. A LLDPE film blown by members of the project at QUT was used instead of the films produced by Ciba discussed in Chapters 3 and 4. Among other advantages, this was because the purpose of the investigation in this case was not primarily to assess the suitability of the film for commercial applications, but to assess the suitability of transmission IR with a synchrotron light source to examine the early stages of polymer degradation. The film produced by QUT was clear, and the titania was better dispersed than the Ciba produced films. Additionally, at 15 µm thickness the QUT produced film was 10 µm thinner than the Ciba produced film, allowing better oxygen permeability.

The advantage of using a synchrotron light source over a conventional source for this type of investigation is the capacity to achieve diffraction limited lateral resolution. On laboratory bench top systems with an internal glowbar source this

181 would result in too great a reduction in signal strength; however with the high brightness and high degree of focus of synchrotron sourced light, even the small fraction of light that manages to pass through the aperture is sufficient to obtain quality spectra. Thus, theoretically, an aperture can be set to provide a beam size of 3 µm x 3 µm at the sample surface, although lateral resolution will be larger than this because of diffraction effects26.

The Bruker Hyperion 2000 microscope at the Australian Synchrotron had two single-point MCT detector options. One detector provided a greater spectral range (3800 cm-1 to 550 cm-1) at the cost of signal-to-noise ratio. This detector was chosen over the second option, which provided improved signal-to-noise ratio at the cost of spectral range, and was effective only to 750 cm-1. It was hoped that titania would be detectable in the mid-IR, which would allow degradation information to be related to the location of titania particles. As titania absorption starts at 750 cm-1 and continues to lower wavenumbers, the detector providing a more suitable spectral range was selected.

During the course of performing experiments it was found that the signal-to- noise ratio was too poor to allow an extremely small aperture size. Eventually it was established that a 10 µm x 10 µm aperture with a spectral resolution of 4 cm- 1 and 256 scans provided the best compromise between lateral resolution, noise, spectral resolution, titania absorption and acquisition time.

IR maps were obtained during the process of sample irradiation. These maps were 2 contiguous steps down by 3 contiguous steps across to provide a total of 6 pixels (Figure 6-1). All stage movements were automated, and the stage was taken back to the same place for each measurement. When the sample was to be irradiated with UVA for a 2 minute exposure, the stage was brought across to position the sample under the UV probe (See Section 2.5, Figure 2-6). Subsequent to irradiation the stage was moved back to the sampling position and a map was acquired. Maps were obtained from the same place to achieve two

182 goals: to examine any progress of heterogeneous oxidation, and to minimize any variation in the interference fringe for PCA analysis.

10 µm

10 µm M1 M2 M3 LLDPE film

M6 M5 M4

Figure 6-1 Schematic showing map pattern for LLDPE acquired in micro-transmission mid-IR mode at the Australian Synchrotron. The pixels in the map were acquired sequentially, from map point 1 (M1) to point 6 (M6). Pixels were contiguous 10 µm x 10 µm squares.

Interference fringes, discussed in Section 1.3.3.3, proved to be difficult to eradicate. Various methods were employed, such as having the film on an angle during data acquisition, and cutting the film at an angle to sample the cut surface. The method that met with the most success was to place a small piece of film in optical contact with a KBR slide; however the film did not remain in contact for longer than one or two minutes. None of these techniques were successful in reliably removing the interference fringe, and ultimately it was decided to continue with data acquisition. By revisiting the same point on the polymer film surface for mapping, it was hoped that as long as the thickness of the film did not change during irradiation, the interference fringe pattern at each point in the map should be identical. From the point of view of assessing the suitability of these techniques for future studies, it was of interest to investigate whether a chemometric analysis of the data would be able to overcome changes in the spectra due to variations in the interference fringe pattern.

183 6.3 Synchrotron results and discussion

Figure 6-2 shows a typical spectrum acquired at the Australian Synchrotron of LLDPE containing 3% Degussa P25 titania, obtained using the experimental procedure described in Section 2.5. The sinusoidal baseline is characteristic of an interference fringe. Titania does not absorb strongly in this spectrum, and the low wavenumber end of the spectrum suffers from poor signal-to-noise ratio, making titania detection quite difficult.

3.5

3.0

2.5

2.0

1.5 Absorbance

1.0

0.5

0.0

4000 3500 3000 2500 2000 1500 1000 500 Wavenumbers (cm-1)

Figure 6-2 A mid-IR spectrum of LLDPE containing 3% Degussa P25 titania obtained using a Bruker Hyperion 2000 microscope with an MCT detector with a synchrotron source, a 10 µm x 10 µm aperture, 4 cm-1 spectral resolution and 256 scans. Some data points are missing, presumbably due to conversion from a Bruker format to a Grams32 AI compatible. As spectra in the Bruker format cannot be read or manipulated by other software, it was necessary to convert to a more suitable format.

PCA proved to be a valuable tool to address challenges created by poor signal-to- noise ratio and interference fringes when analysing the carbonyl and fingerprint regions. As mentioned earlier, the detector with a broader detection range at the

184 cost of signal-to-noise ratio was employed for data acquisition. Subjection of the data to PCA resulted in a greatly improved signal-to-noise ratio of the spectra, which is key to an investigation of degradation around a titania particle as the absorption of oxidation products in early stages of degradation is likely to be weak, making it difficult to distinguish from noise.

PCA was particularly effective in assisting with interference fringes. Firstly, it should be pointed out that the interference fringes remain in the loadings plots after PCA analysis, which suggests that there are some changes occurring to the sinusoidal pattern. This is probably due to changes in the thickness of the LLDPE film as it begins to crosslink during UV irradiation. Despite the lingering presence of the fringes, improvement in the signal-to-noise ratio and the highlighting of absorptions that are changing in the series of spectra allows absorptions to stand out clearly above the interference fringe. This makes any small changes occurring in the data much more accessible to investigation.

The benefit of using PCA to data mine mid-IR spectra of this nature was clearly demonstrated when analysing the data obtained in the experiment described Section 6.2. The PCA result of the region below 1900 cm-1 from the first map point is provided in the following figures. PC1 appears to describe systematic changes as it steadily decreases with increasing irradiation time, while subsequent PCs describe noise.

To help interpretation of the figures, the reader is reminded of the schematic presented in Figure 6-1 which describes the order of pixels in the map, starting at M1. In the following figures the number of minutes the sample had undergone irradiation at the time the map was acquired is provided by a number following an underscore. Thus as an example M1_18 represents the spectrum acquired from the first pixel in the map, by the time the film had undergone 18 minutes of UV exposure. The sample number in the following scores plots refers to the sequence in the series of spectra collected during 30 minutes of irradiation.

185 -3 x 10 8 M1_22

6 M1_26 M1_28 M1_18 M1_30 4 M1_24 M1_20

) 2 M1_12 % M1_16 8 9 . 0 M1_0 M1_14 (87 1

C -2 P on -4 M1_2 M1_6 ores c -6 S

-8 M1_4 -10 M1_8 M1_10 -12 2 4 6 8 10 12 14 16 Sample Number

Figure 6-3 Scores plot for PC1 of the first map point. Each point represents a spectrum acquired during the course of the experiment. The sample number on the x-axis represents the sequential number of the spectrum.

0.3

0.25

) 0.2 % 98 7. 8 0.15 1 ( C

on P 0.1 s g n i

oad 0.05 L

-0.05 600 800 1000 1200 1400 1600 1800 Wavenumber

Figure 6-4 Loadings plot for PC1 from Figure 6-3.

186

It is not immediately clear what PC1 is describing in this case. Certainly the interference fringe is visible, and the scores plot could be describing some systematic change. It does not appear to be degradation related, as the PC scores drop steadily for 10 minutes of irradiation, then begin to increase again until 30 minutes of irradiation. It is possible that the PC is describing changes related to interference fringes, CH absorption variations, or some other artefact.

Due to the unsuitability of PCA over this broad spectral region, only the carbonyl region has been investigated for the 6 pixels in the map, and the results provided below. The data have been over fitted to show PCs 2 and 3, which appear quite noisy. PC1 is blue, PC2 is green, and PC3 is red.

-3 x 10 8 0.25

) M1_2 5% 0

. 0.2

1 6

3 ( M1_6 C M1_0 0.15 P , ) ) 4 M1_4 15% 01% . .

9 0.1 6 M1_10 M1_12 8

2 M1_8 ( 2 ( 1 C C

P M1_26 0.05 P ,

) M1_14 M1_22 M1_18 M1_28 0 M1_20 M1_30 on M1_16 M1_24 15% 0 ngs 89.

-2 adi 1 ( Lo C -0.05

on P -4 es -0.1 or c S -6 2 4 6 8 10 12 14 16 -0.15 Sample Number 1700 1720 1740 1760 1780 1800 Wavenumber M2

-3 x 10 10 0.3 ) M2_2 5% 7 . 8 0.25 1 ( 3

C 6 M2_6 0.2 P

) ), 4 M2_0 M2_4 0.15 1% 68% 9

M2_10 7. 8 (6. 2 M2_8 M2_12 0.1 1 ( 2 C C M2_30 P P , M2_14 n ) 0 M2_16 M2_24 M2_28 0.05 M2_1 M28 _20 M2_26 o % M2_22 68 ngs i . -2 0 Load 1 (87

C -4 -0.05 P

on -0.1 -6 ores c

S -0.15 -8 1710 1720 1730 1740 1750 1760 1770 1780 1790 1800 2 4 6 8 10 12 14 16 Wavenumber Sample Number

187 M3

-3 x 10 8 0.3 ) 83% . M3_2 0.25 2 6 3 (

C 0.2 P , ) ) 4 M3_4 M3_6 M3_10 0.15 17% 34% 7. 4. 7 1 2 M3_0 M3_8 0.1 1 ( 2 ( C C M3_14 M3_30 P

M3_12 P

M3_16 , n

) M3_18 M3_28 0.05 0 M3_20 M3_24 M3_26 o 17% ngs M3_22 i 0 77. -2 Load 1 (

C -0.05

on P -4 -0.1 es or c S -6 -0.15 2 4 6 8 10 12 14 16 1710 1720 1730 1740 1750 1760 1770 1780 1790 1800 Sample Number Variable

-3 x 10 4 M4_12 0.3 M4_10 3 M4_6 0.25 ) M4_2 ) %

M4_0 2 0.2 12%

M4_4 1 2 34. M4_8 M4_16 34. M4_22 ( 0.15 2 ( M4_18 2 C 1 C P

P , , 0.1 ) )

M4_20 % 1 3 31% 0 . M4_14 0.05 3 53. 5

M4_24 ( 1 1 ( -1 0 C C M4_28 -0.05 on P on P

-2 M4_26 s es ng or di c M4_30 -0.1 S

-3 Loa -0.15

-4 -0.2 2 4 6 8 10 12 14 16 1710 1720 1730 1740 1750 1760 1770 1780 1790 Sample Number Wavenumber M5

-3 x 10 12

) 0.25 M5_2 86% 10 . 1 0.2

3 ( 8 C

P M5_6 , ) 6 0.15 M5_10 ) % 44%

4 37 16. M5_4 0.1

M5_12 79. ( 2 ( C 2 M5_8 1 P C

, M5_30 0.05 P ) M5_14 M5_16 M5_28 0 M5_26 M5_18 M5_22 on 37% M5_20 M5_24 s 0 ng 79. -2 adi 1 ( Lo C -4 -0.05 on P

es -6 -0.1 or c S -8 2 4 6 8 10 12 14 -0.15 Sample Number 1700 1710 1720 1730 1740 1750 1760 1770 1780 1790 Wavenumber M6

188 -3 x 10 12 0.25 ) M6_2 69%

. 10 0.2 2

3 ( M6_6

C 8 0.15 P , ) ) 6 % 0.1 21 00% M6_10 80. 14. ( 4 0.05 1

2 ( M6_4 C C P P

, M6_8 ) 2 0 M6_12 on M6_26 s

21% M6_14 M6_18 M6_22 ng 0 M6_16 M6_20 M6_24 M6_28 -0.05 80. adi Lo

1 ( M6_30 C -2 -0.1 on P

es -4 -0.15 or c S -6 -0.2 2 4 6 8 10 12 14 1710 1720 1730 1740 1750 1760 1770 1780 1790 1800 Sample Number Wavenumber Figure 6-5 Scores plots (left) and loadings plots (right) for each pixel in a 3 x 2 map (see Figure 6-1) of LLDPE containing 3% Degussa P25 titania irradiated with UVA. Spectra were recorded every 2 minutes for 30 minutes total irradiation time. Each point in the scores plots represents a spectrum, and the label describes the pixel number and length of UV exposure of the film in minutes. The sample number on the x-axis represents the sequential number of the spectrum.

There is a clear trend in the scores and loadings plots presented in Figure 6-5. PC1 seems to be the only PC describing a systematic change in the data with time, starting at a high score which diminishes with time. PCs 2 and 3 mostly describe noise, with the possible exception of the 4th map point. And each PC1 loadings plot shows similarly positioned absorption peaks.

It is unlikely that the PCA investigation in this instance is detecting any degradation related changes in the spectra. Inspection of the loadings plots shows upward pointing absorptions at 1718 cm-1, 1735 cm-1, 1750 cm-1 and 1773 cm-1. The high starting scores continuing to low finishing scores indicate that these absorptions are actually disappearing with time. Closer inspection of these loadings plots reveals that these absorptions are describing water vapour, which in this case is in the surrounding air.

The likelihood of water vapour absorptions present in the spectra was confirmed by the regularly high value of the score of the data in all map pixels collected after the 10th minute of irradiation. It had been noted while conducting the experiment that the 3 minute nitrogen purge before data collection was accidentally omitted, and therefore more water vapour is seen as this point. There

189 are several possible reasons why water vapour was seen to be decreasing over the course of the experiment.

The experiment was conducted between the hours of 11pm and 7am. Until 11pm there had been 2 persons around the instrument, however following the commencement of the experiment there was only 1, reducing the source of atmospheric water. Additionally, after initially monitoring the experiment in the early stages, the person conducting the experiment would leave the room while waiting for the purge and data collection.

The late hour at which the experiment was performed could also have had an influence, as there were less people in the vicinity, and less doors being opened to the outside. The Australian Synchrotron has an air purging system which helps to keep the air at low humidity, and while there was less human traffic this may have help it to work more efficiently, contributing to lower water vapour.

Regardless of the water vapour content, there does not appear to be any signs of oxidation in the spectra. This was not expected, as proof of concept experiments conducted prior to this one had found the appearance of a carbonyl after 10-15 minutes of irradiation with the same source. However in the optimisation experiments the spectra were measured using diamond ATR/FTIR, which is a surface sensitive technique. In the experiment discussed here, spectra were collected in transmission and hence measured much more the bulk of the material, decreasing the relative concentration of any degradation products.

Additionally, despite the best efforts to allow oxygen back into the purge box described in Section 2.5, there would have been a constant pressure of purge gas away from the sampling area, preventing atmospheric oxygen from circulating back into the box. Any lack of oxygen would naturally result in a decrease in oxidation product formation.

190 6.4 Conclusions

Micro-transmission mid-IR spectroscopy with a synchrotron light source allows for improved spatial resolution at the sample surface, down to a possible 3 µm. With the view to correlate oxidation-related absorptions to distance from a titania particle, an MCT detector providing a spectral range broad enough to encompass titania absorption was selected. This incurred some penalties in data collection, however. Primarily, the necessary 5 µm or better lateral resolution could not be achieved due to a poor signal-to-noise ratio; the uppermost lateral resolution that could be attained was 10 µm. Furthermore the signal-to-noise ratio required 256 scans at 4 cm-1 spectral resolution, which becomes time consuming when acquiring spectra in a 6 pixel map every two minutes. Notwithstanding the long acquisition time, the spectra retained a significant level of noise, particularly at the low wavenumber end of the spectrum.

There were other challenges faced when analysing a film in the mid-IR at the Australian Synchrotron. The instrument possessed a Perspex purge box with doors opening at the front (see Figure 2-4). This made measurements acquired over time very cumbersome, as the environment inside the purge box requires up to 10 minutes or more to remove absorptions from external water and carbon dioxide when taking sensitive measurements. Additionally, the purge source is nitrogen gas, which must be replaced with dry air in order to conduct oxidation experiments.

Interference fringes could not be removed from the spectra, despite using different methods such as having the film at an angle, and placing the film in optical contact with a medium transparent in the mid-IR. Despite the difficulties caused by interference fringes when performing data manipulation such as spectral subtraction, the sensitivity of PCA analysis surpassed the problem by highlighting absorptions that are changing with time. PCA analysis also demonstrated an ability to largely remove noise considerations, and was able to clearly detect the variation of water vapour in the air during the course of the experiment, despite allowing for 10 minutes of purging time inside the Perspex box.

191

Future studies of the oxidation of LLDPE film could be successful if the purge gas can be replaced with dry air or oxygen. However due to limitations of the detector it is difficult to detect titania particles, which remains a challenge if information regarding degradation changes around a titania particle is desired.

192 Conclusions

The primary aim of the work reported in this thesis has been to exploit the nature of titania photocatalysis in order to advance the technology that will lead to the development of a commercial plastic film with controllable degradation properties, even in the dark. To this end the concept of pre-irradiation has been thoroughly examined using commercially available titania from different manufacturers, blown in a LLDPE film by a well known and respected chemical production company, Ciba. Pre-irradiation is a novel concept, and involves the exposure of the film containing titania photosensitiser to UV irradiation in order to initiate oxidation reactions prior to aging in the dark.

Nine different samples of 25 µm thick LLDPE film, containing 1-3% loadings of titania including Degussa P25, Hunstman Tioxide, Satchleben Hombitan and Kronos were subject to investigation. The degradation of the samples was followed by mid-IR spectroscopy to determine the effects of pre-irradiation wavelength (UVA vs. UVC), length of pre-irradiation time, and aging conditions (accelerated aging in the oven at 50 °C, and suntest aging).

SEM images showed that the Degussa P25 titania tended to agglomerate into particles up to several micron across, while modified titanias exhibited much better size and particle distribution. When exposed to UV irradiation some of the samples turned white – this occurred only in the samples containing photoactive titania, and the whiteness was found to be the result of light scattering caused by the titania particles completely destroying the surrounding polymer to form a cavity the shape of a ‘wormhole’.

ATR/FTIR spectra of degraded polymer film demonstrated that samples exposed to UV irradiation developed a higher concentration of oxidation products containing ester moieties, along with various other products that absorb at higher wavenumbers such as lactones and anhydrides. Oven aged samples however tended to form acids, confirming the degradation pathways proposed by Tidjani42. Importantly, it was found that although titania accelerates the

193 degradation of LLDPE film, the degradation products, and relative concentrations of these products, is not affected by titania. It is concluded that titania behaves as a catalyst by providing radicals for degradation to occur, without changing the degradation pathway.

The LLDPE film containing 3% Degussa P25 titania and pre-irradiated for 24 hours with UVA embrittled in approximately 200 days in the oven. This is a significant reduction in the lifetime of the polymer compared to the control sample pre-irradiated for 24 hours with UVA, whose carbonyl index plot did not show signs of significant oxidation occurring after 260 days in the oven. The reduction in the lifetime of the LLDPE film containing titania after pre- irradiation, in conjunction with the photocatalytic nature of titania that accelerates degradation without changing the degradation pathway, has been accepted as strong evidence that pre-irradiation is a successful concept.

Some dissimilarity was observed between pre-irradiation with UVA and UVC, whereby samples exposed to UVC often degraded more rapidly than those exposed to UVA. This was attributed solely to the higher energy of UVC, which resulted in more aggressive degradation of the LLDPE prior to aging. The degradation products and hence pathways were found to be similar, however, and in all situations UVA and UVC irradiation was found to accelerate aging.

This has an impact on the research performed by Allen 7 and co-workers. Allen found that pigment grade titania has a photostabilising effect when subjected to UVC irradiation. SEM images of the titania used in this study illustrated that it had agglomerated into pigment grade sized particles, however at no point was the LLDPE film stabilised by titania when exposed to UVC. It is concluded that the results found by Allen et al. were applicable only to that unique data set, and titania acts as a photosensitiser when subjected to UVC irradiation.

Titania surface modification was found to play a more important role in reducing the photoactive potential of the titania than particle size and aggregation, probably due to the lack of available sites for oxygen trapping. Degussa P25 titania was demonstrated to be significantly more photoactive than the other

194 forms of titania. This was attributed to the crystal phase, whereby the minority rutile fraction was considered to be acting as a dopant, assisting in electron/hole separation. The high photoactivity of Degussa P25 titania despite significant agglomeration and lack of surface modification implies that titania activity is dependent on the efficiency of electron/hole separation.

Increasing the loading from 1% to 3% had a moderate effect of increasing the rate of degradation. Low doses of UVA and UVC pre-irradiation did not greatly affect the rate of degradation, and several hours at least of pre-irradiation was required in order to achieve embrittlement significantly faster. It is likely that higher rates of degradation when increasing the loading was not seen due to the tendency of the particles to aggregate, reducing the available surface area. This was supported by SEM evidence.

The order of photoactivity of titania in the LLDPE films was determined to be Degussa P25 >> Huntsman Tioxide > Satchleben Hombitan > Kronos.

Over the course of the pre-irradiation experiment thousands of mid-IR spectra had been collected to form a broad, comprehensive data set. This data was subjected to the multivariate data analysis technique PCA, which was found to not only handle the large amount of data quickly and efficiently, but was also easily tailored to investigate various aspects of the degradation.

PCA confirmed that titania acted as a photocatalyst by increasing the rate of reaction, without altering the degradation pathway. It was used to provide an alternative measure of oxidation by plotting the scores of a series of spectra on a principal component describing degradation against time, to result in a plot describing the rate of degradation. This has the advantage over conventional carbonyl index plots of including much more spectral data in the analysis, providing a plot describing the rate of change in the whole spectrum rather than just the carbonyl region. The data did not show the closeness of fit of carbonyl index derived data.

195 The ability of PCA to detect very slight changes in a spectrum was demonstrated by the detection of water vapour in the ‘wormholes’ caused by titania completely oxidising the LLDPE. The existence of water vapour in these cavities was previously unknown, and it is possible that it plays some secondary role in the degradation processes of the LLDPE.

Heterogeneous oxidation was investigated in a novel experiment by imaging ATR/FTIR, and domains of more rapid carbonyl product formation were discovered. For the first time spectral information with a lateral component was collected in real time, and revealed the existence of localised domains of increased oxidation rate. It was thought that these regions corresponded to the location of titania particles. Although titania could be directly observed in the IR spectrum obtained using and FPA, PCA was used to demonstrate that regions of high OH concentration corresponded to sensitised domains. The OH signature was attributed to functional groups present on the surface of the titania particles. Despite the potential of imaging ATR/FTIR for the study of heterogeneously degrading polymers, some challenges need to be addressed before it can become a more routine technique.

IR transmission spectroscopy with a synchrotron light source did not prove as useful as hoped for the investigation of the early stages of oxidation production formation in a lateral context. Despite various issues complicating the acquisition of data, it was found that interference fringe issues and poor signal-to-noise could be somewhat compensated for by subjection of the data to PCA. Once again PCA demonstrated its usefulness in the analysis of mid-IR spectral data sets by overcoming interference fringe issues, and by graphical representation of small changes in the spectra, such as the drop in ambient humidity overnight. For future studies it is likely however that imaging ATR/FTIR techniques will supply more useful information.

This thesis has demonstrated that UV pre-irradiation of an LLDPE film with photoactive titania particles can result degradation of the film in the dark in a greatly accelerated time frame. It follows that the concept of pre-irradiation holds potential for the development of a plastic film with controllable degradation

196 properties for commercial outcomes. The way forward for this technology is to increase the photosensitivity of the LLDPE film, and to continue to expand the knowledge of the degradation chemistry occurring around titania nano-particles, in order to optimise the tunablity of the product for various applications.

197 198 References

1. www.plasticsinfo.co.za. 2004. 2. N/A. (Personal Products Co., USA). Application: BE: 1975, p 11 pp. 3. Willett, J. L.; Westhoff, R. P., Annual Technical Conference - Society of Plastics Engineers 52nd, 1572 1994. 4. Roy, P. K.; Surekha, P.; Rajagopal, C.; Chatterjee, S. N.; Choudhary, V., Polymer Degradation and Stability 91, 1791 2006. 5. Sekine, Y.; Fujimoto, K., Journal of Material Cycles and Waste Management 5, 107 2003. 6. Kaczmarek, H.; Sionkowska, A.; Kaminska, A.; Kowalonek, J.; Swiatek, M.; Szalla, A., Polymer Degradation and Stability 73, 437 2001. 7. Allen, N. S.; Khatami, H.; Thompson, F., European Polymer Journal 28, 817 1992. 8. Arnaud, R.; Dabin, P.; Lemaire, J.; Al-Malaika, S.; Chohan, S.; Coker, M.; Scott, G.; Fauve, A.; Maaroufi, A., Polymer Degradation and Stability 46, 211 1994. 9. Fujishima, A.; Zhang, X.; Tryk, D. A., Surface Science Reports 63, 515 2008. 10. George, G. A.; Celina, M., Environmental Science and Pollution Control Series 21, 277 2000. 11. Allen, N. S.; Edge, M., Fundamentals of Polymer Degradation and Stabilization; Springer, 1992. 12. Denisov, E. T. In Environmental Science and Pollution Control Series; Amin, M. B.; Hamid, S. H.; Maadhah, A. G., Eds.; Marcel Dekker: NY, 2000. 13. Celina, M.; George, G. A., Polymer Degradation and Stability 40, 323 1993. 14. Bandyopadhyay, P. K.; Shaw, M. T.; Weiss, R. A., :Polymer-Plastics Technology and Engineering 24, 187 1985. 15. James C. W. Chien, E. J. V. H. J.: 1968, p 381. 16. Gijsman, P.; Hennekens, J.; Vincent, J., Polymer Degradation and Stability 39, 271 1993. 17. Colin, X.; Fayolle, B.; Audouin, L.; Verdu, J., International Journal of Chemical Kinetics 38, 666 2006. 18. Gugumus, F., Polymer Degradation and Stability 52, 145 1996. 19. Scheirs, J.; Bigger, S. W.; Billingham, N. C., Polymer Testing 14, 211 1995. 20. Lawrence, J. B.; Weir, N. A., Journal of Applied Polymer Science 18, 1821 1974. 21. Gupta, A. P.; Saroop, U. K.; Gupta, V., Journal of Applied Polymer Science 106, 917 2007. 22. Tidjani, A.; Dasilva, A. O.; Fanton, E.; Arnaud, R., Journal of Macromolecular Science, Pure and Applied Chemistry A36, 633 1999. 23. Rymarz, G.; Klecan, T.; Sobocinska, A., Polimery (Warsaw, Poland) 38, 588 1993.

199 24. Rugg, F. M.; Smith, J. J.; Wartman, L. H., Journal of Polymer Science 11, 1 1953. 25. Rugg, F. M.; Smith, J. J.; Bacon, R. C., Journal of Polymer Science 13, 535 1954. 26. Beachell, H. C.; Tarbet, G. W., Journal of Polymer Science 45, 451 1960. 27. Beachell, H. C.; Nemphos, S. P., Journal of Polymer Science 21, 113 1956. 28. Lacoste, J.; Singh, R. P.; Arnaud, R.; Lemaire, J., Recent Trends Chem. React. Eng., [Proc. Int. Chem. React. Eng. Conf.] 1, 312 1987. 29. Cruz-Pinto, J. J. C.; Carvalho, M. E. S.; Ferreira, J. F. A., Angewandte Makromolekulare Chemie 216, 113 1994. 30. Tyler, D. R., Journal of Macromolecular Science, Polymer Reviews C44, 351 2004. 31. David, C.; Trojan, M.; Daro, A.; Demarteau, W., Polymer Degradation and Stability 37, 233 1992. 32. George, G. A., Mater. Forum FIELD Full Journal Title:Materials Forum 19, 145 1995. 33. Fanton, E.; Pellereau, B.; Arnaud, R.; Lemaire, Polymer Photochemistry 6, 437 1985. 34. Peeling, J.; Clark, D. T., Journal of Polymer Science, Polymer Chemistry Edition 21, 2047 1983. 35. Rabek, J. F., Polymer Photodegradation: Mechanisms and experimental methods Springer, 1996. 36. Al-Malaika, S.; Scott, G., Degradation and Stability of Polyolefins, 283 1983. 37. Arnaud, R.; Moisan, J. Y.; Lemaire, J., Macromolecules 17, 332 1984. 38. Fanton, E.; Pellereau, B.; Arnaud, R.; Lemaire, J., Polymer Photochemistry 6, 437 1985. 39. Gugumus, F., Polymer Degradation and Stability 27, 19 1990. 40. Lacoste, J.; Carlsson, D. J.; Falicki, S.; Wiles, D. M., Polymer Degradation and Stability 34, 309 1991. 41. Gijsman, P.; Sampers, J., Polymer Degradation and Stability 58, 55 1997. 42. Tidjani, A., Polymer Degradation and Stability 68, 465 2000. 43. Al-Malaika, S.; Scott, G., Degradation and Stability of Polyolefins, 247 1983. 44. Scott, G., Pure and Applied Chemistry 55, 1615 1983. 45. Habicher, W. D.; Bauer, I., Environmental Science and Pollution Control Series 21, 81 2000. 46. Gugumus, F., Polymer Degradation and Stability 50, 101 1995. 47. Gugumus, F. L., Environmental Science and Pollution Control Series 21, 1 2000. 48. Freitas, A. R.; Vidotti, G. J.; Rubira, A. F.; Muniz, E. C., Polymer Degradation and Stability 87, 425 2005. 49. Mills, A.; Le Hunte, S., Journal of Photochemistry and Photobiology A: Chemistry 108, 1 1997. 50. Barr-Kumarakulasinghe, S. A., Polymer 35, 995 1994. 51. May, W. R.; Bsharah, L., Industrial & Engineering Chemistry Product Research and Development 9, 73 1970. 52. Chalk, A. J.; Smith, J. F., Transactions of the Faraday Society 53, 1214 1957.

200 53. Dabald, A. R., Rubber Age (New York) 71, 621 1952. 54. Osawa, Z., Polymer Degradation and Stability 20, 203 1988. 55. Roy, P. K.; Surekha, P.; Rajagopal, C.; Chatterjee, S. N.; Choudhary, V., Polymer Degradation and Stability 90, 577 2005. 56. Sipinen, A. J.; Rutherford, D. R., Journal of Environmental Polymer Degradation 1, 193 1993. 57. Winkler, J., Titanium Dioxide; Vincentz Network: Hannover, 2003. 58. Mills; Andrew; Hill; George; Crow; Matthew; Hodgen; Stephanie, Thick titania films for semiconductor photocatalysis; Springer: Heidelberg, 2005. 59. Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer, M. C., Journal of Physical Chemistry B 107, 4545 2003. 60. Ivanova, S. R.; Minsker, K. S.; Zaikov, G. E., Oxidation Communications 25, 325 2002. 61. Thompson, T. L.; Yates, J. T., Jr., Topics in Catalysis 35, 197 2005. 62. Allen, N. S., Polymer Degradation and Stability 2, 155 1980. 63. Zhang, Z.; Wang, C.-C.; Zakaria, R.; Ying, J. Y., Journal of Physical Chemistry B 102, 10871 1998. 64. Bickley, R. I.; Gonzalez-Carreno, T.; Lees, J. S.; Palmisano, L.; Tilley, R. J. D., Journal of Solid State Chemistry 92, 178 1991. 65. Linsebigler, A. L.; Lu, G.; Yates, J. T., Jr., Chemical Reviews (Washington, D. C.) 95, 735 1995. 66. Watson, S. S.; Beydoun, D.; Scott, J. A.; Amal, R., Chemical Engineering Journal (Amsterdam, Netherlands) 95, 213 2003. 67. Lee, S.-K.; Robertson, P. K. J.; Mills, A.; McStay, D.; Elliott, N.; McPhail, D., Applied Catalysis, B: Environmental 44, 173 2003. 68. Ohtani, B.; Zhang, S.; Handa, J.; Kajiwara, H.; Nishimoto, S.; Kagiya, T., Journal of Photochemistry and Photobiology, A: Chemistry 64, 223 1992. 69. Dagan, G.; Tomkiewicz, M., Journal of Physical Chemistry 97, 12651 1993. 70. Heller, A.; Degani, Y.; Johnson, D. W., Jr.; Gallagher, P. K., Proceedings - Electrochemical Society 88-14, 23 1988. 71. Yang, Q.; Xie, C.; Xu, Z.; Gao, Z.; Li, Z.; Wang, D.; Du, Y., Journal of Molecular Catalysis A: Chemical 239, 144 2005. 72. Beydoun, D.; Amal, R.; Low, G. K. C.; McEvoy, S., Journal of Physical Chemistry B 104, 4387 2000. 73. Sato, S., Langmuir 4, 1156 1988. 74. Thompson, T. L.; Yates, J. T., Jr., Chemical Reviews (Washington, DC, United States) 106, 4428 2006. 75. Luo, Z.; Gao, Q., Journal of Photochemistry and Photobiology, A: Chemistry 63, 367 1992. 76. Allen, N. S.; Edge, M.; Ortega, A.; Sandoval, G.; Liauw, C. M.; Verran, J.; Stratton, J.; McIntyre, R. B., Polymer Degradation and Stability 85, 927 2004. 77. Allen, N. S.; Edge, M.; Sandoval, G.; Ortega, A.; Liauw, C. M.; Stratton, J.; McIntyre, R. B., Polymer Degradation and Stability 76, 305 2002. 78. Cho, S.; Choi, W., Journal of Photochemistry and Photobiology A: Chemistry 143, 221 2001. 79. Hadjiivanov, K. I.; Klissurski, D. G., Chemical Society Reviews 25, 61 1996.

201 80. Kontos, A. I.; Arabatzis, I. M.; Tsoukleris, D. S.; Kontos, A. G.; Bernard, M. C.; Petrakis, D. E.; Falaras, P., Catalysis Today 101, 275 2005. 81. Piscopo, A.; Robert, D.; Weber, J. V., Journal of Photochemistry and Photobiology, A: Chemistry 139, 253 2001. 82. Fox, M. A., Solar Energy Materials and Solar Cells 38, 381 1995. 83. Tang, Q.; Lin, J.; Wu, Z.; Wu, J.; Huang, M.; Yang, Y., European Polymer Journal 43, 2214 2007. 84. Zan, L.; Wang, S.; Fa, W.; Hu, Y.; Tian, L.; Deng, K., Polymer 47, 8155 2006. 85. Zan, L.; Tian, L.; Liu, Z.; Peng, Z., Applied Catalysis A: General 264, 237 2004. 86. Farrokhpay, S.; Morris, G. E.; Fornasiero, D.; Self, P., Colloids and Surfaces A: Physicochemical and Engineering Aspects 253, 183 2005. 87. Ju-Nam, Y.; Lead, J. R., Science of The Total Environment 400, 396 2008. 88. Zhao, X. u.; Li, Z.; Chen, Y.; Shi, L.; Zhu, Y., Journal of Molecular Catalysis A: Chemical 268, 101 2007. 89. Cheng, P.; Li, W.; Zhou, T.; Jin, Y.; Gu, M., Journal of Photochemistry and Photobiology, A: Chemistry 168, 97 2004. 90. Wu, J. C. S.; Chen, C.-H., Journal of Photochemistry and Photobiology, A: Chemistry 163, 509 2004. 91. Iketani, K.; Sun, R.-D.; Toki, M.; Hirota, K.; Yamaguchi, O., Materials Science & Engineering, B: Solid-State Materials for Advanced Technology B108, 187 2004. 92. Sonawane, R. S.; Kale, B. B.; Dongare, M. K., Materials Chemistry and Physics 85, 52 2004. 93. Gole, J. L.; Stout, J. D.; Burda, C.; Lou, Y.; Chen, X., Journal of Physical Chemistry B 108, 1230 2004. 94. Klosek, S.; Raftery, D., Journal of Physical Chemistry B 105, 2815 2001. 95. Jeon, M. S.; Yoon, W. S.; Joo, H.; Lee, T. K.; Lee, H., Applied Surface Science 165, 209 2000. 96. Iwasaki, M.; Hara, M.; Kawada, H.; Tada, H.; Ito, S., Journal of Colloid and Interface Science 224, 202 2000. 97. Subrahmanyam, M.; Rao, K. V. S.; Babu, M. R.; Srinivas, B., Indian Journal of Environmental Protection 18, 266 1998. 98. Martin, S. T.; Morrison, C. L.; Hoffmann, M. R., Journal of Physical Chemistry 98, 13695 1994. 99. Bickley, R. I.; Gonzalez-Carreno, T.; Gonzalez-Elipe, A. R.; Munuera, G.; Palmisano, L., Journal of the Chemical Society, Faraday Transactions 90, 2257 1994. 100. Zhao, Z.; Liu, Q., Catalysis Letters 124, 111 2008. 101. Yang, X.; Cao, C.; Hohn, K.; Erickson, L.; Maghirang, R.; Hamal, D.; Klabunde, K., Journal of Catalysis 252, 296 2007. 102. Kubacka, A.; Fuerte, A.; Martinez-Arias, A.; Fernandez-Garcia, M., Applied Catalysis, B: Environmental 74, 26 2007. 103. Mohamed, M. M.; Al-Esaimi, M. M., Journal of Molecular Catalysis A: Chemical 255, 53 2006. 104. Serpone, N.; Lawless, D.; Disdier, J.; Herrmann, J.-M., Langmuir 10, 643 1994.

202 105. Allen, N. S., Degradation and Stabilisation of Polyolefins; Applied Science Publishers: New York, 1983. 106. Allen, N. S.; Chirinis-Padron, A.; Henman, T. J., Polymer Degradation and Stability 13, 31 1985. 107. Allen, N. S.; Edge, M.; Corrales, T.; Catalina, F., Polymer Degradation and Stability 61, 139 1998. 108. Allen, N. S.; Edge, M.; Corrales, T.; Childs, A.; Liauw, C. M.; Catalina, F.; Peinado, C.; Minihan, A.; Aldcroft, D., Polymer Degradation and Stability 61, 183 1998. 109. Allen, N. S.; Edge, M.; Holdsworth, D.; Rahman, A.; Catalina, F.; Fontan, E.; Escalona, A. M.; Sibon, F. F., Polymer Degradation and Stability 67, 57 2000. 110. Allen, N. S.; Katami, H., Polymer Degradation and Stability 52, 311 1996. 111. Corrales, T.; Peinado, C.; Allen, N. S.; Edge, M.; Sandoval, G.; Catalina, F., Journal of Photochemistry and Photobiology A: Chemistry 156, 151 2003. 112. Zeynalov, E. B.; Allen, N. S., Polymer Degradation and Stability 86, 115 2004. 113. Zeynalov, E. B.; Allen, N. S., Polymer Degradation and Stability 91, 931 2006. 114. Allen, N. S.; Barcelona, A.; Edge, M.; Wilkinson, A.; Merchan, C. G.; Ruiz Santa Quiteria, V., Polymer Degradation and Stability 86, 11 2004. 115. Zan, L.; Fa, W.; Wang, S., Environmental Science and Technology 40, 1681 2006. 116. Pospísil, J.; Pilar, J.; Billingham, N. C.; Marek, A.; Horák, Z.; Nespurek, S., Polymer Degradation and Stability 91, 417 2006. 117. Parris, D. R., Proceedings of International Wire and Cable Symposium 38th, 105 1989. 118. Al-Madfa, H.; Mohamed, Z.; Kassem, M. E., Polymer Degradation and Stability 62, 105 1998. 119. Roy, P. K.; Titus, S.; Surekha, P.; Tulsi, E.; Deshmukh, C.; Rajagopal, C., Polymer Degradation and Stability 93, 1917 2008. 120. Roy, P. K.; Surekha, P.; Rajagopal, C.; Choudhary, V., Journal of Applied Polymer Science 108, 2726 2008. 121. Roy, P. K.; Surekha, P.; Rajagopal, C.; Choudhary, V., eXPRESS Polymer Letters 1, 208 2007. 122. Roy, P. K.; Surekha, P.; Rajagopal, C.; Chatterjee, S. N.; Choudhary, V., Polymer Degradation and Stability 92, 1151 2007. 123. Roy, P. K.; Surekha, P.; Rajagopal, C.; Choudhary, V., Journal of Applied Polymer Science 103, 3758 2007. 124. Nagle, D. J.; Celina, M.; Rintoul, L.; Fredericks, P. M., Polymer Degradation and Stability 92, 1446 2007. 125. Suarez, J. C. M.; Mano, E. B., Polymer Degradation and Stability 72, 217 2001. 126. Wang, S.; Yu, J.; Yu, J., Journal of Polymers and the Environment 14, 65 2006. 127. Davis, G., Materials Characterization 51, 147 2003. 128. Zhao, G.; Tian, Q.; Liu, Q.; Han, G., Surface and Coatings Technology 198, 55 2005.

203 129. Zeynalov, E. B.; Allen, N. S.; Calvet, N. L.; Stratton, J., Dyes and Pigments 75, 315 2007. 130. Zeynalov, E. B.; Allen, N. S., Polymer Degradation and Stability 91, 3390 2006. 131. Zeynalov, E. B.; Allen, N. S., Polymer Degradation and Stability 85, 847 2004. 132. Dick, C. M.; Liggat, J. J.; Snape, C. E., Polymer Degradation and Stability 74, 397 2001. 133. Maxwell, R. S.; Chambers, D.; Balazs, B.; Cohenour, R.; Sung, W., Polymer Degradation and Stability 82, 193 2003. 134. Palmas, P.; Colsenet, R.; Lemarié, L.; Sebban, M., Polymer 44, 4889 2003. 135. Assink, R. A.; Celina, M.; Minier, L. M., Journal of Applied Polymer Science 86, 3636 2002. 136. Assink, R. A.; Gillen, K. T.; Sanderson, B., Polymer 43, 1349 2002. 137. Gillen, K. T.; Assink, R. A.; Bernstein, R., Polymer Degradation and Stability 84, 419 2004. 138. Assink, R. A.; Celina, M.; Gillen, K. T.; Clough, R. L.; Alam, T. M., Polymer Degradation and Stability 73, 355 2001. 139. Somers, A. E.; Bastow, T. J.; Burgar, M. I.; Forsyth, M.; Hill, A. J., Polymer Degradation and Stability 70, 31 2000. 140. Stivala, S. S., Journal of Polymer Science, Part A: General Papers 3, 4299 1965. 141. George, G. A.; Celina, M.; Vassallo, A. M.; Cole-Clarke, P. A., Polymer Degradation and Stability 48, 199 1995. 142. Bertie, J. E. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002. 143. Harrick, N. J., Applied Spectroscopy 31, 548 1977. 144. Hannah, R. W. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002. 145. Chambon, F.; Srinivas, S., Journal of Plastic Film & Sheeting 15, 297 1999. 146. Ajji, A.; Zhang, X.; Elkoun, S., :Annual Technical Conference - Society of Plastics Engineers 62nd, 2045 2004. 147. Noda, I.; Dowrey, A. E.; Marcott, C., Modern Polymer Spectroscopy, 1 1999. 148. Choi, H. S.; Shikova, T. G.; Titov, V. A.; Rybkin, V. V., Journal of Colloid and Interface Science 300, 640 2006. 149. Kaczmarek, H.; Oldak, D.; Malanowski, P.; Chaberska, H., Polymer Degradation and Stability 88, 189 2005. 150. Stark, N. M.; Matuana, L. M., Polymer Degradation and Stability 86, 1 2004. 151. Delor, F.; Barrois-Oudin, N.; Duteurtre, X.; Cardinet, C.; Lemaire, J.; Lacoste, J., Polymer Degradation and Stability 62, 395 1998. 152. Do, T.-T.; Celina, M.; Fredericks, P. M., Polymer Degradation and Stability 77, 417 2002. 153. Chalmers, J. M.; Hannah, R. W.; Mayo, D. W. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002.

204 154. Sommer, A. J. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002. 155. Fitzpatrick, J.; Reffner, J. A. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002. 156. Fredericks, P. M. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002. 157. Koenig, J. L., Microspectroscopic Imaging of Polymers American Chemical Society: New York, 1998. 158. Chan, K. L. A.; Hammond, S. V.; Kazarian, S. G., Analytical Chemistry 75, 2140 2003. 159. Chan, K. L. A.; Kazarian, S. G., Applied Spectroscopy 57, 381 2003. 160. Chan, K. L. A.; Kazarian, S. G.; Mavraki, A.; Williams, D. R., Applied Spectroscopy 59, 149 2005. 161. van der Weerd, J.; Kazarian, S. G., Applied Spectroscopy 58, 1413 2004. 162. Sommer, A. J.; Tisinger, L. G.; Marcott, C.; Story, G. M., Applied Spectroscopy 55, 252 2001. 163. Williams, G. P. In Handbook of Vibrational Spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; John Wiley and Sons Ltd: Chichester, 2002. 164. Jilkine, K.; Gough, K. M.; Julian, R.; Kaminskyj, S. G. W., Journal of Inorganic Biochemistry 102, 540 2008. 165. Miller, L. M.; Dumas, P., Biochimica et Biophysica Acta (BBA) - Biomembranes 1758, 846 2006. 166. Kaminskyj, S.; Jilkine, K.; Szeghalmi, A.; Gough, K., FEMS Microbiology Letters 284, 1 2008. 167. Wetzel, D. L.; Carter, R. O., III, AIP Conference Proceedings 430, 567 1998. 168. Feuilloley, P.; Cesar, G.; Benguigui, L.; Grohens, Y.; Pillin, I.; Bewa, H.; Lefaux, S.; Jamal, M., Journal of Polymers and the Environment 13, 349 2005. 169. Prasad, A.; Mowery, D., Annual Technical Conference - Society of Plastics Engineers 55th, 2310 1997. 170. Jin, C.; Christensen, P. A.; Egerton, T. A.; Lawson, E. J.; White, J. R., Polymer Degradation and Stability 91, 1086 2006. 171. Zhiyong, Y.; Mielczarski, E.; Mielczarski, J.; Laub, D.; Buffat, P.; Klehm, U.; Albers, P.; Lee, K.; Kulik, A.; Kiwi-Minsker, L.; Renken, A.; Kiwi, J., Water Research 41, 862 2007. 172. http://www.resinex.be/Datasheets/Dowlex/DOWL_NG5056E.pdf. 2007. 173. Sharma, N.; Chang, L. P.; Chu, Y. L.; Ismail, H.; Ishiaku, U. S.; Mohd Ishak, Z. A., Polymer Degradation and Stability 71, 381 2001. 174. Tarantili, P. A.; Kiose, V., Journal of Applied Polymer Science 109, 674 2008. 175. Creagh, D.; Tobin, M.; Broadbent, A.; McKinlay, J., AIP Conference Proceedings 879, 615 2007. 176. Creagh, D.; McKinlay, J.; Dumas, P., Vibrational Spectroscopy 41, 213 2006. 177. Eren, T.; Okte, A. N., Journal of Applied Polymer Science 105, 1426 2007. 178. Nakayama, N.; Hayashi, T., Composites Part A: Applied Science and Manufacturing 38, 1996 2007.

205 179. Fa, W.; Zan, L.; Gong, C.; Zhong, J.; Deng, K., Applied Catalysis B: Environmental 79, 216 2008. 180. Rennel, C.; Rigdahl, M., Colloid and Polymer Science 272, 1111 1994. 181. Belyi, V. N.; Malevich, V. L.; Shraiber, Y.; Khilo, N. A., Journal of Applied Spectroscopy 75, 199 2008. 182. Chambon, E.; Florentin, E.; Torchynska, T.; Gonzalez-Hernandez, J.; Vorobiev, Y., Microelectronics Journal 36, 514 2005. 183. Cran, M. J.; Bigger, S. W., Applied Spectroscopy 57, 928 2003. 184. Salvalaggio, M.; Bagatin, R.; Fornaroli, M.; Fanutti, S.; Palmery, S.; Battistel, E., Polymer Degradation and Stability 91, 2775 2006. 185. Schachtschneider, J. H.; Snyder, R. G., Spectrochimica Acta 19, 117 1963. 186. Snyder, R. G.; Schachtschneider, J. H., Spectrochimica Acta 19, 85 1963. 187. Lin-Vien, D.; Colthup, B. N.; Fateley, W. G.; Grasselli, J. G., Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press: London, 1991. 188. Lawson, E. E.; Edwards, H. G. M.; Johnson, A. F., Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 51, 2057 1995. 189. Farmer, V. C.; Editor, Mineralogical Society Monograph 4: The Infrared Spectra of Minerals, 1974. 190. Hummel, D. O., Atlas of Plastics Additives: Analysis by Spectrometric Methods, by Dietrich O. Hummel, 2003. 191. Yang, R.; Liu, Y.; Yu, J.; Wang, K., Polymer Degradation and Stability 91, 1651 2006. 192. Blakey, I.; Billingham, N.; George, G. A., Polymer Degradation and Stability 92, 2102 2007. 193. Tyler, D. R., Journal of Macromolecular Science, Polymer Reviews C44, 351 2004. 194. Yokoyama, S.; (Tomoegawa Paper Co., Ltd., Japan). JP: 2009, p 12pp. 195. Levenson, E.; Lerch, P.; Martin, M. C., Journal of Synchrotron Radiation 15, 323 2008.

206