Engineering Polymers for Organic Photovoltaics

Engineering Polymers for rganic Photovoltaics by Elijah Bultz A document submitted in partial fulfillment of the degree Masters of Applied Science. Department of Chemical Engineering and Applied Chemistry University of Toronto Copyright by Elijah Bultz 2010

Engineering Polymers for Photovoltiacs Elijah Bultz Master s of Applied Science Chemical Engineering and Applied Chemistry University of Toronto 2010 Abstract In this thesis, I have produced low polydispersity polymers of poly(chloromethylstyrene) and polyvinylphenol. Modification reactions for both polymers were attempted and were successful for only poly(chloromethylstyrene) as conditions in our experiments produced insoluble polymer salts using polyvinylphenol. A set of modified polymers were produced using a low polydisperse poly(chloromethylstyrene) as we believe that polymers with PDIs < 1.4 will be preferable for use in an eventual solar cell device. Finally a set of experiments were completed to determine the conditions to minimize the PDI of poly(chloromethylstyrene) and found that the main variable was maximizing the amount of 3-chloromethylstyrene with respect to 4-chloromethylstyrene in a polymerization. This result has lead us to believe that 3-chloromethylstyrene is less prone to transfer reactions, which in turn leads to smaller PDI values. ii

Acknowledgments I would like to thank my family and friends for support while completing this thesis iii

Table of Contents Acknowledgments... iii Table of Contents... iv List of Tables... vii List of Figures... viii List of Schemes... xi Chapter 1 Introduction and Literature Review... 1 1.1 Photovoltaics... 1 1.2 Free Radical Polymerization and Controlled Radical Polymerization... 4 1.2.1 Nitroxide Mediated Polymerization (NMP)/ Stable Free Radical Polymerization (SFRP)... 5 1.2.2 Atom Transfer Radical Polymerization (ATRP)... 8 1.2.3 Reversible Addition-Fragmentation Chain Transfer (RAFT)... 9 1.3 Multiphase Polymer Blends... 11 1.4 Polymer Blend Casting Techniques... 14 1.4.1 Drop Casting... 14 1.4.2 Spin Casting... 14 1.4.3 Doctor Blading... 15 1.4.4 Inkjet Printing... 16 1.4.5 Roll-to-roll printing techniques... 17 Chapter 2 Thesis Statement and Proposed Research... 19 Chapter 3 Addition of Phase Separating Groups to Controlled Polymers... 22 3.1 Controlled polymerization... 22 3.2 Synthesis of poly(vinylphenol)... 26 3.3 Amplification Reaction of Poly(vinylphenol)... 29 3.4 Poly(chloromethylstyrene) Modification... 32 iv

3.5 Selection of Phase Separating Groups... 39 3.6 Addition of PSG 24 - from Benzoic Acid... 45 3.7 Addition of PSG 25 - from Hexanoic Acid... 47 3.8 Addition of PSG 26 - from Trifluoroacetic Acid... 49 3.9 Addition PSG 29 - from Perflurobenzoic Acid... 50 3.10 Materials and Methods... 51 3.10.1 RAFT Chain Transfer Agent... 51 3.10.2 Controlled 4-Acetoxystyrene Polymerization Solution... 52 3.10.3 Controlled 4-Acetoxystyrene Polymerization Bulk... 52 3.10.4 Polymerization of Styrene using BST as an initiator... 52 3.10.5 Polyvinylphenol Reaction - Deacetylation... 53 3.10.6 (±)-3-Chloro-1,2-propanediol amplification reaction... 53 3.10.7 DL-1,2-Isopropylideneglycerol (Solketal) Tosylate amplification reaction... 53 3.10.8 Trityl Bromide PSG addition to polyvinylphenol... 54 3.10.9 Stearic Acid PSG addition to poly(chloromethylstryene)... 54 3.10.10 Perfluorotetradecanoic acid PSG addition to poly(chloromethylstyrene)... 54 3.10.11 Hexanoic acid PSG addition to polychloromethylstyrene... 55 3.10.12 Benzoic acid PSG addition to polychloromethylstryene... 55 3.10.13 Trifluoroacetate PSG addition to polychloromethylstyrene... 55 3.10.14 Pentafluorobenzoic acid PSG addition to polychloromethylstyrene... 56 Chapter 4 ptimizing the polydispersity of functional styrentic polymers... 57 4.1 Background & Introduction... 57 4.2 Experimental... 59 4.2.1 Equipment... 59 4.2.2 Materials and Methods... 60 2-phenyl-2-((2,2,6,6-tetramethylpiperidin-1-yl)oxy)ethyl benzoate (BST) unimer preparation.. 60 v

4.2.3 Polymerization method... 60 4.3 Results and Discussion... 63 4.4 Conclusions... 75 Chapter 5 Conclusions, Recommendations and Future Work... 76 References... 77 vi

List of Tables Table 1. PSG reaction with poly(chloromethylstyrene) and stearic acid... 36 Table 2 Molecular weight data for poly(chloromethylstyrene) substitution with perfluorotetradecanoic acid... 38 Table 3 shows the results from Blends calculations for the Flory-Huggins paramater using hexanoic acid as the base modification for polychloromethylstyrene for the polymers in Figure 19... 41 Table 4. The molecular weights for the modified poly(chloromethylstyrene)... 44 Table 5. The reaction conditions for the various polymerizations... 63 vii

List of Figures Figure 1 presents the traditional p-n junction seen in IPV devices while the device architecture on the right shows a typical PV heterojunction v... 2 Figure 2 shows the energy levels of the HM and LUM levels in a typical bulk heterojunction v... 3 Figure 3 shows a typical bulk heterojunction device. iv... 4 Figure 4 shows the dynamic equilibrium between the dormant polymer (left) and the propagating polymer (right)... 5 Figure 5 indicates the main mechanism for ATRP polymerization x.... 9 Figure 6 shows the general RAFT chain transfer agent reacting (CTA) with a radical R' xv... 9 Figure 7 shows the mechanism of the RAFT controlled polymerization process... 10 Figure 8 displays possible morphologies for multiphase polymer blends from melt mixes.... 11 Figure 9 shows the general apparatus of a spin coat device... 15 Figure 10 illustrates the common techniques for printing in roll-to-roll processing. The coating units are shown in grey and the coated substrate is shown as dotted xx... 18 Figure 11 shows the process of dispersing the photoactive layer in a blend of conductive polymers with different CTGs and PSGs for each polymer and final incorporation into the Bender cell... 20 Figure 12. The Mn and PDI vs conversion plot for the polymerization of 4-acetoxystyrene at 50% wt in chlorobenzene... 27 Figure 13 shows the Mw and PDI vs time plot for the polymerization of 4-acetoxystyrene with 90% 4-acetoxystyrene and 10% chlorobenzene... 28 Figure 14 is the GPC trace of poly(4-acetoxystyrene) precipitated after a reaction time of 3h... 29 viii

Figure 15 shows the NMR for the poly(vinylphenol) modification reaction with glycidol... 31 Figure 16 GPC trace of poly(chloromethylstyrene) pre modification... 34 Figure 17 GPC trace of the poly(chloromethylstyrene) reacted with stearic acid and potassium carbonate in DMAc... 36 Figure 18 GPC trace of run 3 of the poly(chloromethylstyrene) modification reaction with perflourotetradecanoic acid... 38 Figure 19 shows the polymer structures used to detemrine the miscibility of two polymers... 40 Figure 20 shows the GPC trace of poly(chloromethylstyrene) used in subsequent modification reactions... 43 Figure 21 shows the 1H NMR for poly(chloromethylstyrene) used in subsequent modification reactions.... 44 Figure 22 shows the GPC overlay of the starting polychloromethylstyrene in black and the modified polymer 24 in red... 45 Figure 23 shows the NMR for poly(chloromethylstryene) reacted with benzoic acid to produce polymer 24... 46 Figure 24 shows the 1H NMR for polychloromethylstyrene reacted with hexanoic acid to produce polymer 25... 48 Figure 25 shows the GPC traces of the polychloromethylstyrene starting material in red and the final product after reacting with hexanoic acid to produce polymer 25 in black... 49 Figure 26 shows the NMR spectrum for a partially converted polychloromethylstyrene to polymer 26... 50 Figure 27 plots Mn vs conversion for the different polymerization runs of chloromethylstyrene.... 64 ix

Figure 28 shows the PDI versus conversion for the polymerization of chloromethylstyrene evolves... 65 Figure 29 plots M n as a function of conversion and the linearity indicates that reaction proceeds in a controlled manner... 69 Figure 30 shows the GPC traces of the polymerization 7 using the mixture of isomers... 70 Figure 31 indicates the percent mass of m-chloromethylstyrene in solution as a function of time for the polymerization of chlormethylstyrene... 70 Figure 32 shows the H1 NMR of poly(4-chloromethylstyrene)... 72 Figure 33 shows the H1 NMR of poly(3/4-chloromethylstyrene)... 73 Figure 34 shows an overlay of the GPC traces for polymeriztion of acetoxystyrene - run 9... 74 Figure 35 shows the Mn Vs Conversion Plot for the polymerization of 4-acetoxystyrene in run 9... 74 x

List of Schemes Scheme 1 shows the synthesis of a unimolecular initiator with benzoyl peroxide, styrene and TEMP (BST, 4)... 6 Scheme 2 shows the synthetic pathway for the RAFT CTA for polymerization of 4- acetoxystyrene...23 Scheme 3 illustrates the synthetic scheme for the production of thin film polymer blends from modified poly(vinylphenol) either by amplification of the phenol (left) or without amplification (right)... 25 Scheme 4 shows the tosylation reaction for solketal as a group for the amplification reaction of poly(vinylphenol)... 32 Scheme 5 illustrates the synthetic process for producing polymer blends from modified poly(chloromethystyrene).. 33 Scheme 6 shows the phase separating groups chosen from the computer modeling for synthesis. 42 Scheme 7 shows the NMP polymerization of chloromoethylstyrene with the BST unimer. 64 xi

1 Chapter 1 Introduction and Literature Review 1 1.1 Photovoltaics As the effects of climate change are becoming more easily detected by the average person and scientific data points to anthropogenic atmospheric C 2 release as the primary route cause of the change. We will therefore require carbon neutral energy sources to mitigate or even over time reverse the damage. Solar energy could provide a large potential to mitigate the future climate change issues as there is an immense amount of sunlight illuminating the earth. A 2005 study indicated that with an area of 100km x 100km and a 15% power conversion efficiency solar energy could produce enough power for American energy needs i. Photovoltaic (PV) devices directly convert solar energy into electricity and are a potential route to harvesting this immense energy from the sun. Inorganic PV devices such as silicon or mixtures of cadmium indium gallium and selenium are currently dominating the market as they can achieve overall efficiencies of 15 20% ii. The downside to these devices is the cost of manufacturing their active layer(s) is large as they are made from expensive materials which require large amounts of energy during processing to be made at high purity. (Hydro)carbon based materials are much cheaper and more abundant than the inorganic alternatives and theoretically can produce the same efficiency as inorganic devices iii. However current organic photovoltaic (PV) devices lag behind in both efficiency and lifetime. ne difference between inorganic and organic PV devices is the strength of the charge carrier bond. In the inorganic device, the holes and electrons (charge carriers) are not tightly bound together and there dissociation energy is small compared to the energy available at room temperature iv. Upon illumination, these charge carriers are photogenerated as free charges and easily travel through a built-in electric field within the device to their respective electrode as seen on the left of Figure 1 v.

2 Figure 1 presents the traditional p-n junction seen in IPV architecture on the right shows a typical PV heterojunction v devices while the device In an PV device the exciton (an excited hole-electron pair) binding energy is at least an order of magnitude higher than the thermal energy present at room temperature vi. To dissociate excitons produced in the active layer of an PV device into individual charge carriers an interface between the donor and acceptor layer is needed. The donor/acceptor layers are comprised of organic semiconductors with different Highest ccupied Molecular rbital (HM) and Lowest Unoccupied Molecular rbital (LUM). By definition the donor layer has the lowest ionization potential and is able to donate an electron from its HM thereby producing a radical cation. Conversely, the acceptor has the highest electron affinity and has a LUM energy level sufficient to accept an electron and form a radical anion iv. This is schematically shown in Figure 2. In this figure the donor layer also acts as the photoactive layer.

3 Figure 2 shows the energy levels of the HM and LUM levels in a typical bulk heterojunction v In the bulk heterojunction, an interpenetrating network of organic semiconductors or an organic semiconductor with an inorganic semiconductor, phase separation occurs between the hole transporting layer and the electron-transporting layer. The interface of a heterojunction is known to efficiently separate an exciton into a hole and electron if the interface is reached in time before decay of the exciton to the ground state or other energetically degenerating processes can occur vii. The diffusion length of a typical exciton is estimated to be 10nm viii. Maximizing the interfacial area of the heterojunction between the donor/acceptor (or hole and electron transporting layers respectively) is required to ensure that as many excitons reach the interface before recombination occurs. Also controlling the domain size of each particular layer in the heterojunction is required to ensure that a diffusing electron can reach the interface. In summary the ideal case for an PV will be a heterojunction whereby the two domains produce a bicontinuous interpenetrating network with a maximized interfacial area and as a consequence small domain sizes thus maximizing the chance of exciton splitting and charge transport the respective electrodes. An example of a bulk heterojunction device is shown in Figure 3 with the donor and acceptor layers in different colours.

4 Figure 3 shows a typical bulk heterojunction device. iv 1.2 Free Radical Polymerization and Controlled Radical Polymerization Free radical polymerization (FRP) is a means of producing high molecular weight macromolecules (polymers) by using a process by which free radicals react with monomers containing vinyl (ethylenic) groups. There are four main reactions that occur in free radical polymerization: initiation, propagation, transfer and termination. Usually in a FRP reaction a thermosensitive initiator is used which degrades into a single or pair of free radicals. These radicals then react with the vinyl bond of a monomer producing a new sp 3 C-C bond with the radical positioned at the last monomer added. This process then repeats itself in a process known as propagation and occurs until a transfer or termination reaction occurs. In FRP chain transfer occurs when the propagating polymer radical abstracts a atom from another molecule leaving a terminated chain (non propagating chain) and a newly formed radical that can itself begin to propagate. Two radicals on different propagating polymer chains can react with each other to produce a new C-C bond and a dead polymer chain in a termination process known as combination. These termination and transfer mechanism cause FRP polymers to exhibit large polydispersities (molecular weight distributions) because these transfer and termination reactions, while they can occur to polymers at any given molecular weight, generally result in a doubling of the molecular weight.

5 Controlled free radical polymerization (CFRP) is a technique used to produce polymers of low polydispersity by substantially lowering the rate of both transfer and termination reactions. The three main methods of controlled are described below. 1.2.1 Nitroxide Mediated Polymerization (NMP)/ Stable Free Radical Polymerization (SFRP) The first reported method for controlled free radical polymerization was by Georges et al. in 1993 ix. This method uses a stable free radical such as 2,2,6,6-tetramethyl-1-piperidynyl-N-oxy or TEMP, a nitroxide, as a capping agent for a growing polymer. In NMP there is equilibrium between dormant, capped polymers and propagating polymers as shown in Figure 4. Figure 4 shows the dynamic equilibrium between the dormant polymer (left) and the propagating polymer (right) x At reaction temperatures (typically around 120ºC) the equilibrium lies far to the left towards the dormant state. The right side of the equilibrium does not exist for sufficiently long periods of time such that termination and transfer processes are largely avoided while propagation occurs at a reduced rate. To maintain control of the polymerization (i.e. ensure that the polydispersity stays low) the nitroxide should not be able to abstract an atom from another molecule and should also not be able to initiate a polymerization with another monomer. The nitroxide is present in the reaction at the same or greater concentration as initiator to ensure that all propagating polymers

6 are capped (i.e. molar ratio to initiator is typically 1). Also the reaction temperatures must be kept low enough such that the monomers do not auto-polymerize. A NMP polymerization also does not need to necessarily be initiated by a conventional initiator. A molecule that contains a single repeat unit of the polymer capped with a nitroxide group is known as a unimolecular initiator or unimer. The concept was first developed and studied by Hawker xi. With a unimer, the initiator to nitroxide ratio is by definition kept constant at 1. A synthetic route for a unimer, 4 formed of benzoyl peroxide, 2, styrene, 1, and TEMP, 3, is shown in Scheme 1. + + N N 1 2 3-4 Scheme 1 shows the synthesis of a unimolecular initiator with benzoyl peroxide, styrene and TEMP (BST, 4) The kinetics for a NMP process can not be modeled by a simple rate law as in traditional free radical polymerization due to the excess of polymer chains in the dormant state and less than one percent in the active propagating state at a given time. This rate law therefore becomes a set of differential equations, which has been described by Veregin et al xii. Their assumptions exclude the effect of initiation because the temperature is set well above the 10h half-life of the given initiator so all the initiator has decomposed very early into the polymerization leading to the equilibrium and propagation equation shown below.

7 P i + P i + T KL k L p M P i i+ 1 Where P is the growing polymer, T is the nitroxide, L is the reversibly capped polymer, M is the monomer and i is denotes the length of the polymer. For controlled and living free radical polymerizations P << L and overall this leads to two differential equations to describe the propagation above. d[ Pi ] = k p[ M ]([ Pi dt d[ Li ] k dt L k [ T ][ P ] i 1 ] [ P ]) k L i [ T ][ P] [ Li ] [ L] L k [ T ][ P ] + i L [ T ][ P] [ Li ] [ L] These two differential equations ignore termination and transfer reactions which for a well controlled polymerization should be minimized and these equations have been shown to provide a good fit for well controlled polymerization in the aforementioned reference. Here k p refers to the rate constant for the propagation reaction, k L is the rate constant for the deactivation reaction to the dormant stage and [T][P]/[L] is K L, or the equilibrium constant for the nitroxide capping reaction. Here P i is proportional to M n as M n should increase linearly with the change of conversion. More complete kinetic studies have been completed using a combination of 21 differential equations to solve variables for M n and M w which include possible side reactions and termination xiii. Another positive aspect of nitroxide mediated polymerizations is the absence of the gel effect, also known as the Tronsdorff effect. This effect occurs in traditional free radical polymerizations. At approximately 30% conversion, the viscosity of the reaction mixture becomes large enough that the termination reactions slow and propagation continues. leading to

8 an autoacceleration of the reaction, which in turn leads to an increase of temperature due to the exothermic behavior of radical polymerizations and a run-away reaction, which leads to problems for the scale up for these polymerization. Using NMP, it has experimentally been shown that this effect is not present as an isothermal reaction can be kept for up until high conversion values xiv. 1.2.2 Atom Transfer Radical Polymerization (ATRP) In the atom transfer radical polymerization (ATRP) process, as in NMP, there is a capping agent and a dynamic equilibrium which prefers the dormant state over the active state. In ATRP the catalyst used is a combination of transition metal (M n ) which can expand its coordination sphere and increase its oxidation number x, a ligand (L) for complexing the transition metal, a counter ion, and a alkylhalogen which is required for producing the free radical initiation during which the halogen is added to the transition metal complex. The transition metal of choice is copper (Cu) and the halide of choice is bromine (Br). In ATRP the alkyl halide acts as the initiator and the halide is caps the dormant polymer as well. The alkyl halide in the presence of the transition metal complex, M n, cleaves the R-X bond producing a free radical species that propagates in the presence of monomer and produces an oxidized transition metal species complex, M n+1 XL shown in Figure 5. At reaction temperatures, the equilibrium lies far to the left towards the dormant species, as in NMP, thereby reducing the possibility of termination and transfer products during the polymerization. The ATRP equilibrium constant (k ATRP =k act /k deact ) can be altered by the ligand/ initiating system allowing for an optimization of the polymerization.

9 Figure 5 indicates the main mechanism for ATRP polymerization x. 1.2.3 Reversible Addition-Fragmentation Chain Transfer (RAFT) The reversible addition-fragmentation chain transfer (RAFT) process uses reversible chain transfer to ensure that control of the polymerization occurs. This is done with a chain transfer agent (CTA) also known as a RAFT agent. The general RAFT agent is shown in Figure 6 with a carbon sulfur double bond and a carbon sulfur single bond with leaving group that can reinitiate a polymerization (R). The Z group on the RAFT agent is used to stabilize the radical form of agent. Figure 6 shows the general RAFT chain transfer agent reacting (CTA) with a radical R' xv

10 Depending on the reactivity of the monomer different RAFT agents will be required. RAFT agents can be dithioesters, dithiocarbamates, xanthates or trithiocarbonates all having different reactivities due to their difference in the chemical nature of the Z group (Figure 6) and depending on the stability/energy of a growing polymer requires a different Z group to allow for good control of the polymerization to occur. Initiation of a RAFT polymerization is normally initiated with traditional azo initiators. At the heart of the RAFT process is the (CTA), which an initiating radical or propagating polymer radical attacks the C=S bond of the RAFT agent forming a radical adduct with two propagating chains (2 or 4 in Figure 7). This adduct is unstable and either chain can cleave reforming a stable RAFT CTA with a single polymer chain and a radical that continues to propagate Post-polymerization polymers made by RAFT polymerization can be used as macro-raft agents allowing for the easy production of block copolymers. The RAFT agent is very versatile as it can be designed to work with a large number of monomers, solvent systems and reaction conditions compared to the other controlled polymerization techniques xv Figure 7 shows the mechanism of the RAFT controlled polymerization process

11 1.3 Multiphase Polymer Blends Multiphase polymer blends occur when two or more immiscible polymers are mixed together. These polymer blends have been identified as the most versatile and economical method to produce new high performance materials xvi. The two main parameters that control the properties of the blend are the control of the interface and control of the morphology. When referring to multiphase polymer blends, the morphology refers to the distribution of one polymer component phase with respect to the others. There are different morphologies that can be produced depending on the properties of the polymers in the blend with respect to each other as seen in Figure 8. Figure 8 displays possible morphologies for multiphase polymer blends from melt mixes. Some of the commonly produced morphologies seen in multiphase polymer blends are shown above in Figure 8. Although these are morphologies produced from melt mixtures, similar morphologies can be produced through solution casting. Droplets in a matrix occur when there is one component in the continuous phase while the other component is dispersed in that phase.

12 In melt-mixing droplets are broken up threads and smaller droplets which can coalesce into larger droplets. The competition between the coalescing and breaking up determines the particle sizes of this component. As stated above, large-scale phase separation and larger domain sizes can also be seen in solution cast polymer blends when poorer solvents and slow evaporation occurs. Droplet-in-droplet morphologies will generally be formed in binary melt mixes when the concentrations are near the phase inversion region (the region where one component changes from the disperse phase to the continuous phase) or by controlling the time of mixing. These are known as composite droplets and can be formed in solution cast polymer blends as well. Likely the most important blend morphology for polymer PV devices is the co-continuous or bicontinuous morphology. In this morphology there is not one polymer component dispersed in the continuous phase. Both components are continuous and interpenetrating producing better electrical properties due to the presence of complete conductive pathways to the electrodes available only for this particular morphology. In contrast the droplet morphology it is possible that the interfacial area is high but because one component is dispersed in the other it is possible to traps where that droplet does not have access to the particular electrode. Determination of the morphology produced is explained below. There are two main driving forces that determine the final morphology of a polymer blend cast from solution: thermodynamic forces and kinetic effects. The thermodynamic properties include the solubility of the polymers in a given solvent, the Flory-Huggins parameter, χ, which describes energy required for interdispersing a polymer into a solvent and the ratio between the constituents in solution. The kinetic effects apply during the formation of a thin film and include the evaporation rate, the rate of any crystallization processes and any post casting annealing processes conducted. The solvent effect has both kinetic and thermodynamic aspects to the bulk morphology of a thin film. The solvent choice has been shown to greatly affect the overall efficiency of a polymer solar cell where there is excellent solubility of both components compared to a poor solvent. Poorer solvation of a component will lead to larger domain sizes as full dissolution and complete mixing within solution is less likely to occur when compared to a better solvent xvii. With respect

13 to kinetics, the solvent evaporation rate can have a significant effect on domain size. When spincasting is used to produce thin film polymer blends, the domain size is generally much smaller than when slower solvent evaporation occurs as in drop casting xviii. These factors are known to have an effect in melt mixing as well. Coalescence of droplets occurs when cooling of the melt occurs too slowly leading to much larger domain sizes. This case normally occurs in roll-to-roll printing and inkjet printing where the evaporation rate is normally much smaller than spin casting. Another variable controlling the blend morphology is the relative concentration of the two components. In melt mixes the change from droplet-in-matrix form to droplet-in-droplet morphology to co-continuous is seen as one component increases with respect to the other. The droplet-in-matrix form will form in the opposite direction if the second component is much smaller than the first Thermal annealing is another way to alter the morphology of a polymer blend. If a component of the blend can crystallize then the addition of heat can allow the intermolecular forces to crystallize small domains of the polymer if the temperature is raised about the glass transition temperature and domains within the blend can reorganize. Annealing also allows aid in the stability of the resulting blend when it is subjected to higher temperatures post processing. Thermal annealing can also cause negative effects including a dewetting at the surface of the film on the coated surface leading to a decrease in performance and adhesion. The relaxation of polymers during the annealing process can also decrease the sharpness of the interface as well. Analysis of the bulk morphology of polymer blends usually consists of varying microscopy techniques including transmission electron microscopy (TEM), scanning electron microscopy (SEM), scanning probe microscopy (SPM), and atomic force microscopy (AFM). The main use for TEM is the study of lateral morphological changes while SEM and SPM are both used to study the surface topography of a thin film blend. However, these analytical techniques only give

14 a two dimensional view of the thin film. Some research has been dedicated towards analyzing the thin film of a polymer blend in three dimensions. ne such method involves using a sample containing a deuterated component coupled with time-of-flight secondary ion mass spectroscopy (TF-SIMS) as a means to analyze the film xix. To determine crystallinity within a sample a variety of x-ray diffraction techniques can be used. 1.4 Polymer Blend Casting Techniques Polymer blends can be produced by varying techniques. As polymers have very low to nonexistent vapor pressures the best way to process them is through solution casting techniques. The two main classes of solution processes are coating/casting from solution and printing from solution. In printing, an ink can be transferred from a stamp to a substrate by reverse action or directly printing (for example with inkjet). In the case of printing there is usually an opportunity for pixilation or patterning of the coating solution onto the substrate. While the process of coating or casting is different in that the solution is transferred directly to the substrate by pouring and there is no opportunity for pixilation or patterning xx. The solution casting methods most commonly used are described below. 1.4.1 Drop Casting Drop casting is the simplest way to produce a thin film. In this coating method a horizontal surface is required and the polymer solution is dropped or poured on and the solvent is evaporated. This method provides no major control with respect to film thickness. 1.4.2 Spin Casting

15 Spin casting is the most common technique in academic laboratories for producing thin film polymer blends. In this technique, like in drop casting, the polymer solution is dropped onto a spinning substrate as seen in Figure 9 [xx]. The thickness of the film produced, d, is governed by this equation: α d = kω Here, ω is the angular velocity and k and α are empirical constants for given solvent, polymer and substrate. The thickness and morphology and surface topography of films produced from this method have been found to be very reproducible xxi. There are two major disadvantages with spin casting: the first is there is a significant amount of waste produced as much of the ink is lost through radial spraying and secondly spin casting is in no way scalable to larger substrates or commercial production. Because spin coating is normally done on a small scale the waste is not an issue, although for industrial purposes spin casting is not a viable option. Figure 9 shows the general apparatus of a spin coat device 1.4.3 Doctor Blading

16 Doctor blading is a technique where the a sharp blade is placed at a fixed distance above the substrate surface. The polymer solution is placed in front of the blade that moves across the substrate, coating the film with a thin polymer solution. The film thickness, d, is empirically determined using the given formula: d 1 c = g 2 ρ where g is the distance between the blade and the substrate, c is the concentration of the solids in solution and ρ is the density of the final film. The main advantage over spin coating is that material waste can be kept to a minimum, although this requires optimization of the process although the solvent evaporation rate in doctor blading is generally lower than spin coating, which can lead to larger scale phase separation. 1.4.4 Inkjet Printing Inkjet printing works in a similar fashion to those used for the home ink jet printer. The printing head is made to be resistant to organic solvent. This method has the ability to produce pixilation at high resolution and requires no master copy commonly required with other printing techniques. However, printing speed is generally slower than other printing methods. The film thickness is described by the equation below where N d is the number of droplets per area V d is the volume of the droplets and c is the concentration of the solids in solution and ρ is the density of the final film. The drops can be formed by mechanical compression or the ink (polymer solution) is heating and electrostatically charged and propelled towards the substrate. Inkjet printing of pigments (crystals) dispersions is much more difficult than with solutions. With this technique a mixture of more than one solvent is commonly used with one or more having a high volatility to other low volatility. This solvent pair allows for both film setting and final film formation. d = N d V d c ρ

17 1.4.5 Roll-to-roll printing techniques Roll-to-roll processing techniques (Figure 10) have traditionally been used for printing newspapers, books and magazines commercially as a high volume of material can be printed in a short amount of time. This technique will eventually be used to mass produce organic photovoltaics so the future cost is minimized. In roll-to-roll processing the substrate is very long sheet that is rolled up. The substrate is unwound and which can be patterned and printed through a variety of techniques. In roll-to-roll processing the analogous technique to doctor blading would be knife-over-edge coating where a polymer solution would be kept introduced to the end of the blade while the roll is moving underneath. Slot-die coating is used when thin stripes of material need to be printed on a Gravure printing allows for 2D patterning by roll-to-roll printing. Gravure printing uses a roll with an engraved pattern that is partially immersed in a bath of polymer solution. A knife blade removes excess solution on this roll and the pattern is transferred to the moving substrate that moves between this coating roll and a support roll. Less complicated techniques to print polymer solutions include meniscus printing where an solution is added to a roll and transferred to the substrate in larger areas. ther processes can also take place as the sheet is unwound including heating, curing or annealing. nce complete, the substrate is rolled up for later manufacturing processes.

18 Figure 10 illustrates the common techniques for printing in roll-to-roll processing. The coating units are shown in grey and the coated substrate is shown as dotted xx

19 Chapter 2 Thesis Statement and Proposed Research 2 The solar cell architecture being developed in the Bender lab is a variation on the bulk heterojunction solar cell. In this device photon absorption has been separated from hole and electron transport. The photon absorption material is made of either dye-doped-organic nanoparticles or engineered organic crystals while the conductive polymers will have semiconductor pendant groups rather than using a polymer with a conjugated and conductive backbone. The polymers used will be solution processable, that is to say they will be soluble in a particular solvent or solvent mixture while the photon absorbing dye doped nanoparticles or engineered crystals will be insoluble but dispersible in that solvent or solvent mixture. The dispersion is cast onto a substrate and the solvent(s) are evaporated leaving a multiphase polymer blend with dispersed light-absorbing particles as illustrated in Figure 11 The nanoscale morphology of the phase separated blend will dictate the interfacial area between the hole and electron transporting layers. The area should be maximized so that exciton dissociation, the first step in electrical charge generation, can easily occur. Also the layers should be co-continuous to allow for an easy pathway for charge transport to the respective electrode. Photovoltaic devices are generally produced from solution casting so that film thickness can be kept thin. The solvent is then evaporated usually at an elevated temperature leaving the thin film polymer blend. The goal of my thesis is therefore to explore the polymer chemistry and post polymerization modification chemistry which might lead to a pair of polymers derived from the same base polymer which when solution cast together in a thin-film would phase separated in discrete domains.

20 PSG N n n m H R R' PSG PSG PAM - nonsoluble N R = PSG R' = CTG PSG PSG PSG PSG PSG PSG Printing PSG' PSG' In Solution Solvent Evaporation PSG' PSG' PSG' PSG' PSG' PSG' Figure 11 shows the process of dispersing the photoactive layer in a blend of conductive polymers with different CTGs and PSGs for each polymer and final incorporation into the Bender cell. The initial goal of my project is to produce a base polymer which is styrenic in nature and amenable to post polymerization transformation/chemistry. Initial targets will be poly(4- acetoxystyrene) and poly(chloromethylstyrene). Secondly, I will explore the post polymerization chemistry in order to ascertain a preference between poly(4-acetoxystyrene) and poly(chloromethylstyrene) as based polymers. Initially I will target the combination of fluorocarbon and hydrocarbon fatty chains introduced post polymerization. These differences in properties should ultimately ensure that the polymers are immiscible and phase separate in the solid state. I will also use in silico modeling to prove this pair is a suitable non-miscible pair. A system of this sort of blend from a single source has not been attempted or reported in the literature even without the conductive pendant groups and therefore provides a potential new

21 route towards developing materials for organic photovoltaics (PVs) which is novel on multiple levels. This method allows for the independent control of the electronic properties and the physical properties in contrast to other systems where each is dependent (sometimes inversely dependent) on the other.

22 Chapter 3 Addition of Phase Separating Groups to Controlled Polymers 3 3.1 Controlled polymerization As stated above, the goal of my research is to produce phase separated polymer blends by postpolymerization functionalizing of a base polymer which is to be produced by radical polymerization. Large polydispersities are undesirable for my purposes as outlined above. I will there use controlled radical polymerization techniques to minimize the polydispersity of my base polymer. As outlined in the preceding chapter, there are three methods for producing controlled polymers with narrow polydispersities and each have there own strengths and weaknesses. ATRP was not chosen for this work due to the use of copper salts that may have required an extensive workup post polymerization to remove any leftover salts. Such a workup might be necessary as the presence of copper in organic electronic materials is known to negatively affect the use of the polymer as a functional material. In the literature RAFT has been shown to be very versatile xxii although the syntheses were often reported as facile, they were often difficult to reproduce and the workup was equally as difficult.

23 Br Mg THF, dry MgBr CS 2 S S-MgBr K 2 C 3 KI/I 2 /H 2 S S S S 5 6 7 8 8 + CN N N CN 2 S NC S 9 10 NC 10 S S n 9 Scheme 2 shows the synthetic pathway for the RAFT CTA for polymerization of 4- acetoxystyrene The RAFT chain transfer agent, 10, shown in Scheme 2 was designed with the aim of controlling the polymerization of a styrenic monomer shown at the bottom of the figure. To synthesize the RAFT CTA, a Gringard reagent, 6, was produced with 1-bromonaphthalene, 5, in the presence of magnesium metal in anhydrous THF to which carbon disulfide was added to produce a the red dithio acid salt, 7. Titration of this acid with I 2, a reducing agent, produced the disulfide, 8, compound required to produce the final RAFT CTA. 1,1 -azobis(cyclohexanecarbonitrile), 9, the azo compound, decomposes at high temperatures to release nitrogen and 2 cyclohexanecarbonitrile radicals, which react with the disulfide and add on as the Z leaving

24 group on the RAFT CTA. 9, which has high decomposition temperature, was used over other azo initiators due to restrictions on importing them into the country. We were unable to purify the intermediates in this multi-step synthesis, which lead to a final reaction mixture of a crude RAFT CTA. We attempted purification techniques, which included large silica chromatography columns, run with solvents mixtures as the mobile phase to separate the impurities from the RAFT CTA. All attempts resulted in RAFT CTAs less than 90% pure by HPLC. As the purification of this RAFT agent was clearly tedious and exceedingly difficult, RAFT controlled polymerizations were not going to be pursued. Nitroxide mediated polymerization (NMP) uses a radical initiator as in RAFT controlled polymerizations but uses a stable free radical to cap the propagating polymer. Unlike RAFT polymerizations, all the molecules required to produce controlled polymers in NMP are purchasable at high purity such as TEMP, the nitroxide stable free radical and the initiator benzoyl peroxide. Polymers are generally easily purified of the low molecular weight byproducts, which are soluble in a range of solvents while a high molecular weight polymer generally does not. NMP is also known for controlling styrenic polymerizations, that is keeping the polydispersity low. For these reasons nitroxide mediated polymerization was chosen as the method for producing the polymer backbone.

25 BP TEMP N i NH 4 H/IPA N i n' N m ' 11 n N m 12 Cl H H H 14 15 K 2 C 3 p yrid ine H (1) Ts 16 K 2C 3 (2) 13 P SG N HTC PS G E TC PS G P SG PS G ET C N i H + /H 2 H H "Amplified"Polymer, 17 n' N m ' P SG' HTC P SG =, -- (CH 2 ) 17 CH 3 Scheme 3 illustrates the synthetic scheme for the production of thin film polymer blends from modified poly(vinylphenol) either by amplification of the phenol (left) or without amplification (right) The choice in monomer is also of vital importance to the end goal of phase separated blends by modification of a controlled polymer its need to undergo chemical derivatization. 4- Acetoxystyrene, 11, was chosen first as its polymer, 12, is a precursor to poly(vinylphenol), 13, which contains a single phenolic group per repeat unit which can be modified postpolymerization by various means. We initially proposed a scheme by which the single phenolic group would be amplify to two hydroxyl groups with the addition of a glycerol group to each repeating unit, 17. Ideally this would have doubled the number of reactive functional groups on the polymer, allowing for a higher concentration of phase separating groups or charge transfer groups. The amplification reactions attempted, explained in detail below, were unsuccessful as a polymer salt was formed and an irreversible precipitation occurred. Additionally we found that chemical derivatization of poly(vinylphenol) itself was problematic for the same reason.

26 Therefore 4-chloromethylstyrene, 20, was chosen as an alternative monomer to produce polymers, 21, for post-polymerization modification by displacement of the chloride atom with nucleophiles such nucleophile can be generated from both phase separating groups and conductive groups. The reaction scheme for poly(chloromethylstyrene), 21, is shown below in Scheme 5. 3.2 Synthesis of poly(vinylphenol) 4-Acetoxystyrene, 11, was polymerized to poly(4-acetoxystyrene), 12, by a number of reaction conditions as a precursor to poly(vinylphenol) in radical polymerization. This route must be taken because the polymerization of vinylphenol would not progress, as phenols are known inhibitors of radicals, thus the most facile route to produce poly(vinylphenol) is the deacetylation of poly(4-acetoxystyrene) in alkali reaction conditions. We first polymerized 4-acetoxystyrene in a 50% solution in benzene with the azo initiator 2,2'- azobis(2-methylbutyronitrile). The resulting polymer reacted until the solution was too viscous for the stir bar to spin and was of a high molecular weigh (>100,000 Da) with a bimodal weight distribution. This polymer was used to optimize the deacetylation reaction to make poly(vinylphenol). The first attempt at nitroxide-mediated polymerization of 4-acetoxystyrene using TEMP and BP run in a 50% wt solution in chlorobenzene. Figure 15 shows that the polymerization of 4- acetoxystyrene is not easily kept under control with the polydispersity is increasing over time. In these conditions it is likely that a higher TEMP:BP ratio is required to keep a very tight molecular weight distribution as more nitroxide present in solution would increase the possibility of a reversibly capped polymer rather than a termination product being formed. This increase in

27 TEMP may lead to a slower propagation rate. However, in place of testing this hypothesis, we proceeded to scope out the remaining procedure. 12000 10000 y = 116.08x + 1151.1 R 2 = 0.9774 2.5 2 Mn (Da) 8000 6000 4000 1.5 1 PDI 2000 0.5 0 0 10 20 30 40 50 60 70 80 90 100 Conversion (%) 0 Mn PDI Linear (Mn) Figure 12. The Mn and PDI vs conversion plot for the polymerization of 4-acetoxystyrene at 50% wt in chlorobenzene. Different reaction conditions, (90% 4-acetoxystyrene, 10% chlorobenzene) were used in a kinetic study of the 4-acetoxystyrene polymerization to determine if a better polymer could be synthesized to meet our requirements of low polydispersity (PDI < 1.4), following the procedure of Dollin et al. xxiii. This experiment had a molar TEMP/BP radical ratio greater than one so there is more TEMP present then growing polymer chains to inhibit termination products. Figure 13 shows the molecular weight and polydispersity both increasing over time to unacceptably high values and again losing control so it was decided to terminate the reaction at 3h in a subsequent reaction by precipitating the polymer to ensure that it was of an acceptably low polydispersity. The GPC trace of this final polymer that would go to modification is shown in Figure 14. This polymer has a Mw of approximately 20,000 Da and a polydispersity of 1.34. A deviation, or tailing, from a Gaussian curve is seen on the right side of the peak, which

28 indicates that there are some dead chains present. This process was not optimized and this polymer was used in subsequent reactions. 40000 1.8 Mw (Da) 35000 30000 25000 20000 15000 10000 5000 Mw PDI 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 PDI 0 0 1 2 3 4 5 6 Time (h) 0 Figure 13 shows the Mw and PDI vs time plot for the polymerization of 4-acetoxystyrene with 90% 4-acetoxystyrene and 10% chlorobenzene

29 Figure 14 is the GPC trace of poly(4-acetoxystyrene) precipitated after a reaction time of 3h Poly(vinylphenol) was synthesized by deacetylation of the poly(4-acetoxystyrene) ammonium hydroxide and isopropyl alcohol (Scheme 3) so further modifications would be possible. The reaction was monitored by FT-IR and deemed complete once the 1715cm -1 peak was no longer present corresponding to the carbonyl C= bond in the ester on the acetyl group indicating that the deacetylation reaction was complete. This product was used in subsequent modification reactions. 3.3 Amplification Reaction of Poly(vinylphenol)

30 The amplification of the phenol group with glycerol derivatives to a repeat unit with two hydroxyl groups would double the number of reactive sites on the polymer for further modification of either phase separating groups or charge transfer groups (for either hole or electron transfer). The rationale behind amplifying the reactive sites per repeat unit is to reduce the relative strength of the styrenic π-π interactions of the repeat unit. If the π-π interactions outweigh the repulsive interactions of the phase separating groups then the polymers may become miscible and phase separation may not occur. The reaction scheme shows three potential synthetic routes for amplifying poly(vinylphenol) using either 1) Chlorohydrin, 14, (a chlorinated derivative of glycerol) in the presence of a base 2) glycidol, 15, in the presence of an amine base or 3) a tosylated solketal, 16, in the presence of a base. In the first route, the carbonate base, shown in the reaction scheme, act to deprotonate the polymer phenol, which then substitutes the chlorine on the chlorohydrin producing an amplified repeat unit, and KCl salt as a precipitate. This reaction was run overnight in dimethyl acetemide (DMAc) and conversion was monitored by the chlorohydrin/ DMAc ratio by GC. After one night of reaction the chlorhydrin peak had disappeared and a precipitate had formed in the reaction mixture. This precipitate was polymer and was not soluble in any solvent and exhibited only marginal swelling in dimethyl sulfoxide (DMS) and indication that cross-linking had occurred or an insoluble polymer salt was formed. Due to the unsuccessful processing of this polymer, this synthetic route was stopped. The second synthetic route for amplification involved using the epoxide, glycidol, in the presence of triethylamine to add to the phenol repeat unit. The reaction proceeds with nucleophilic phenolic oxygen attacks the carbon on the epoxide, opening the epoxide ring forming and ether bond with the phenolic hydrogen adding to the epoxide oxygen in step. In this reaction no polymer salts are formed and ideally no side reactions would occur, but epoxides are very reactive chemicals and the opening of one epoxide can initiate the ring opening of other epoxides leading to the formation polyether compounds. This reaction was run in DMAc with excess glycidol per repeat unit of phenol and according to the GC results all glycidol had reacted indicating that polyether compounds were produced rather than the goal of the single glycerol

31 addition. This was confirmed 13 C NMR shown in Figure 15. In the 70 80 ppm region there are multiple peaks corresponding to the ether linked carbons. This indicates that polyether linkages were occurring and the amplified polymer product was not produced successfully. Figure 15 shows the NMR for the poly(vinylphenol) modification reaction with glycidol The third route attempted used the tosylate of solketal, a glycerol with two hydroxyls protected by an acetal. This acetal is sensitive to acidic conditions, which deprotects the hydroxyl groups and produces isopropyl alcohol. It was believed that this reaction scheme would be successful, as the protected hydroxyl groups would inhibit any side reactions. To form leaving group on the free hydroxyl group of solketal a tosylation reaction is used to form a tosylate leaving group which can react with the phenol in basic environments, where the acetal is stable. The tosylation reaction is shown in Scheme 4 and this product was used in the amplification reaction with poly(vinylphenol) with a carbonate base to deprotonate the phenol in DMAc. After reacting

32 overnight, similar results to the previous reactions using carbonate base occurred, a precipitate that was not soluble in the workup was produced likely due to an insoluble polymer salt formation. H TsCl Pyridine, DCM S 19 16 Scheme 4 shows the tosylation reaction for solketal as a group for the amplification reaction of poly(vinylphenol) After multiple unsuccessful routes to amplifying poly(vinylphenol) we decided to not amplify the polymer and attempt the addition of phase separating groups to determine the viability of using poly(vinylphenol) as a polymer for modification. Trimethylphenyl bromide was chosen as the phase separating group shown in Scheme 3. Solvent mixtures of DMAc/Toluene and N- Methyl-2-pyrrolidone (NMP)/ Toluene were used with cesium carbonate as the base to deprotonate the phenol. In all cases insoluble precipitates were formed even after an acidic workup to reprotonate and resolubilize the polymer. At this point it was concluded that the polymer salts being formed and this was the most likely cause of insolubility and another polymer was chosen to modify where the salt would not form on the polymer. 3.4 Poly(chloromethylstyrene) Modification

33 BP i N TEM P Cl 20 21 Cl n m N n' m' N PS G' HTC PSG ETC P SG = ( CH 2 ) 16 CH 3 PSG ' = (CF 2 ) 12 CF 3 22 23 Scheme 5 illustrates the synthetic process for producing polymer blends from modified poly(chloromethystyrene) Chloromethylstyrene, 20, like 4-acetoxystyrene has a functional group that can undergo post polymerization modification but unlike a phenol, the chlorine on the chloromethyl group on the repeat unit is displaced in a substitution reaction and the salt exists on the modifying group rather than the polymer.