Department of Organic and Macromolecular Chemistry

Research group Supramolecular Chemistry

Supramolecular Grid Structures For

Hydrogels With Self-Healing Character

Thesis submitted to obtain

the degree of Master of Science in Chemistry by

Ruben Lenaerts

Academic year 2014 - 2015

Promoter: prof. dr. Richard Hoogenboom

Supervisors: Kanykei Ryskulova and Lenny Voorhaar

Acknowledgments Writing this thesis and thereby finalizing my career as a chemistry student was not possible

without a lot of people, who all have my sincerest thanks.

I would like to thank professor Richard Hoogenboom, for giving me the chance to work in his

research group on my project of choice. I would like to thank him for always being helpful

when the project didn’t go as planned and for always making time to answer my questions

and for his help in writing my thesis.

Gratitude goes to professor Johan Winne, for giving me advice on the synthetic strategy and

making time to look over synthetic problems.

Throughout the year, I have learned a lot and all this would not be possible without my

supervisors Kanykei Ryskulova and Lenny Voorhaar. They learned me how to work in the lab

and helped me writing my thesis. I would like to thank them for always making time when I

had questions.

Furthermore I like to thank everyone in the lab who have helped me with various aspects,

Bart, Maarten and Maarten, Brynn, Joachim, Zhangyao, Gertjan, Victor, Ding-Ying, Mathias,

Glenn, Maji, Kathleen, Jos, Jan, Duchan, Bram and Vincent, who were always willing to spend

time for letting me get to know new practical and theoretical tricks and for creating a great

lab environment.

I would also like to thank all the other master students around in the lab, Mathijs, Jorne,

Lieselot, Jente and Dephine, for the fun moments we experienced.

A special thanks goes to Dries and Roald, for all their help for making my thesis better and all

the fun moments we encountered during our five years as chemistry students.

For all the chromatographic and MS analysis, my sincerest thanks goes to Ir. Jan Goeman, for

all the fast analysis and his willingness for answering my questions. For all their help with NMR

measurements, I would like to thank Tim Courtin, Dieter Buyst and Niels Geudens.

Lastly but no less sincere, I would like to thank my parents for supporting me throughout my

career as a student. Without them, I wouldn’t be able to get a master degree in science. I

would also like to thank all my friends and my sister for all their support. All of this wouldn’t

be possible without all of you.

Content 1 Theoretical Part .................................................................................................................. 1

1.1 Aim of the Thesis ......................................................................................................... 1

1.2 Supramolecular Grid structures .................................................................................. 4

1.2.1 Supramolecular Interactions by Metal-Ligand Interactions ................................ 4

1.2.2 Supramolecular Metallogrids ............................................................................... 8

1.2.3 Supramolecular Grids in Solution ....................................................................... 10

1.3 Supramolecular Hydrogels ........................................................................................ 13

1.3.1 Hydrogels ............................................................................................................ 14

1.3.2 Self-Healing Hydrogels ....................................................................................... 15

1.3.3 Applications of Self-Healing Hydrogels .............................................................. 16

1.4 Supramolecular Grids in Polymers and Gels ............................................................. 18

1.5 Supramolecular Metal-Ligand Interactions in Hydrogels .......................................... 20

1.6 Poly(2-oxazoline)s ...................................................................................................... 21

2 Results and Discussion ..................................................................................................... 25

2.1 Synthesis Route ......................................................................................................... 25

2.2 Ligand Synthesis......................................................................................................... 26

2.2.1 Synthesis of 2-Bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 ........................... 26

2.2.2 Synthesis of 2-[(Triisopropylsilyl)ethynyl]-6-(tributylstannyl)pyridine 4 ........... 31

2.2.3 Stille Coupling Reactions .................................................................................... 34

2.2.4 Synthesis of 4,6-Bis[4’-(ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 7A ................. 38

2.3 Polymer and Oligomer Synthesis ............................................................................... 40

2.3.1 Synthesis of 2-(2-(2-Methoxyethoxy)ethoxy)ethyl Azide 13 ............................. 40

2.3.2 Synthesis of Poly-α-methyl-ω-azido-(2-ethyl-2-oxazoline) ................................ 41

2.4 Cu(I)-Catalyzed Azide-Alkyne Cycloadditions ............................................................ 46

3 Conclusion and Outlook ................................................................................................... 49

4 Experimental .................................................................................................................... 53

4.1 Materials and Equipment .......................................................................................... 53

4.2 Ligand Synthesis......................................................................................................... 55

4.2.1 2-Bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3, PdCl2(PPh3)2 Catalyst,

Purification by Column Chromatography ......................................................................... 55

4.2.2 2-Bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3, Pd(PPh3)4 Catalyst, Purification

by Column Chromatography ............................................................................................ 56

4.2.3 2-Bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3, Purification by Kugelrohr ..... 57

4.2.4 2-[(Triisopropylsilyl)ethynyl]-6-(tributylstannyl)pyridine 4 ............................... 58

4.2.5 2-[(Triisopropylsilyl)ethynyl]-6-(deuterio)pyridine Test Reaction ..................... 58

4.2.6 2-[(Triisopropylsilyl)ethynyl]-6-(tributylstannyl)pyridine 4 ............................... 59

4.2.7 Stille Coupling Diiodopyrimidine 6 ..................................................................... 59

4.2.8 4,6-Bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 6A with

PdCl2(PPh3)2 ...................................................................................................................... 60

4.2.9 4,6-Bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 6A with

Tetrakis ............................................................................................................................ 60

4.2.10 4,6-Bis[4’-(ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 7A ..................................... 61

4.3 Polymer and Oligomer Synthesis ............................................................................... 62

4.3.1 2-(2-(2-Methoxyethoxy)ethoxy)ethyl Azide ....................................................... 62

4.3.2 Poly-α-methyl-ω-azido-(2-ethyl-2-oxazoline) .................................................... 63

4.4 Cu(I)-Catalyzed Azide-Alkyne Cycloadditions ............................................................ 65

4.4.1 4,6-Bis[4’-((2-(2-(2-Methoxyethoxy)ethoxy)ethyl)-1”,2”,3”-triazol-4”-yl)pyrid-2’-

yl]-2-phenylpyrimidine 8A-TEG ........................................................................................ 65

5 Bibliography ...................................................................................................................... 67

6 Scientific Article ................................................................................................................ 75

7 Dutch Summary ................................................................................................................ 87

8 Supporting Info ................................................................................................................. 89

Abbreviations Asym Asymmetric Bu3SnCl Tributyltin chloride CuAAC Cu(I)-Catalyzed Azide-Alkyne Cycloaddition DCM Dichloromethane DMA N,N-Dimethylacetamide DMAP 4-Dimethylaminopyridine DPP 3,6-Di(2-pyridyl)pyridazine EGDMA Ethylene Glycol Dimethacrylate Et2O Diethylether EtAc Ethyl acetate EtOH Ethanol HEMA (2-Hydroxyethyl Methacrylate) MALDI-TOF Matrix Assisted Laser Desorption/Ionization Time Of Flight MeOH Methanol MeOD Methanol-d4 Mn Number Average Molecular Weight Mw Weight Average Molecular Weight NEt3 Trietylamine NMR Nuclear Magnetic Resonance PAOx Poly(2-alkyl/aryl-2-oxazoline) PEtOxn Poly(2-ethyl-2-oxazoline) with n repeating units PEtOxn-N3 Poly-α-methyl-ω-azido-(2-ethyl-2-oxazoline) with n repeating units PHEMA Poly(2-Hydroxyethyl Methacrylate) PMDETA N,N,N′,N′′,N′′-Pentamethyldiethylenetriamine PMeOx Poly(2-methyl-2-oxazoline) PMMA Polymethylmethacrylate SEC Size-Exclusion Chromatography SI Supporting Info Sym Symmetric TBAF Tetra-n-butylammonium fluoride TEG Triethyleneglycol THF Tetrahydrofuran TIPS Triisopropylsilyl

1

1 Theoretical Part

1.1 Aim of the Thesis

Hydrogels are three-dimensional solid structures that contain a substantial amount of water.

1,2,3,4 The properties they have are interesting for different biomedical applications, including

tissue engineering and drug delivery.1,5 Self-healing hydrogels are hydrogels which have the

property to self-heal after the gel has been disrupted. Self-healing is a remarkable

phenomenon observed in nature. When tissue, skin or bone is damaged, it regains its former

structure. If a hydrogel has the ability to self-heal without external stimuli like in some

biological systems it will expand its potential application even broader. This makes the gel

injectable to the body, instead of direct implantations which is done with most non-self-

healing hydrogels. The aim of this project is to synthesize self-healing hydrogels with

supramolecular crosslinkers. The key element that makes all of these promising features

possible, is the type of crosslinker that will be used for this, namely, a supramolecular

crosslinker. A supramolecular crosslinker is a system for linking polymer chains together,

which relies on supramolecular interactions. In contrary to a covalent interaction, which binds

two atoms permanently, this non-covalent supramolecular bond is dynamic.6,7,8,9 There is an

equilibrium between the associated molecular assemblies and the individual molecules

resulting in continuous dynamic exchange between the supramolecular assembled entities

and non-assembled individual molecules. In this thesis, such a new supramolecular crosslinker

was developed. The entity that was used as supramolecular crosslinker is a supramolecular

metallogrid, where metal-ligand coordination is used as the dynamic bond. This type of

supramolecular interaction was chosen, because of the strong supramolecular bonds this

compounds forms, compared to the bonds of other supramolecular interactions. Another

advantage is that a grid structure, which has four ligands in one grid, gives the possibility of

crosslinking between different polymer chains.

2

Figure 1: Left: Location of Fe(II) complexation by compound 8. Right: the pyridine analogue of compound 8 10.

The ligand designed for this study that comprises all desired aspects, compound 8, is shown

in Figure 1. Octahedral coordinating metals, including iron(II) and zinc(II), can be coordinated

by ligand 8 in water by the free electron pairs of nitrogen, on the locations that are shown in

Figure 1. These coordinated transition metal ions have three more binding sites free which are

available for binding with the second ligand to give full metal coordination. When the other

transition metal ion also does the same and all of this goes further on, a supramolecular grid

structure can be obtained as the grid structure represents a thermodynamic minimum with

maximal supramolecular interactions and still relatively high entropy (Figure 2). It should be

kept in mind that these grid structures are formed by supramolecular interactions, meaning

in this case, they are not always present like this, but they are in a dynamic equilibrium.

Sometimes the four iron atoms and four ligands have all coordination sites occupied,

sometimes they don’t. Analogues of compound 8 can be found in literature.10 These analogues

have two pyridines instead of the two triazole rings of our compound. With Fe(II)(BF4)2, these

analogues are known to yield supramolecular grid structures. The triazole ring in compound 8

is made by a copper catalyzed azide alkyne cycloaddition. Due to this reaction, different R

groups can easily be introduced, which is not the case for the already reported pyridine

analogues.

Figure 2: Schematic representation of the supramolecular grid structure formed by compound 8 with Fe(II) as

metal ion

3

Within this thesis a stepwise approach will be followed to develop these self-healing hydrogels

as will be explained in the following manner, including the reasons why the different steps

were taken.

At first the synthesis of small molecules as R group, more specifically, triethyleneglycol will be

targeted to obtain water-soluble ligands. After the compound had been made,

supramolecular assembly of compound 8 in water will be tested with Fe(II) and Zn(II). Besides

oligomers, also polymers of different lengths will be attached to the ligand. The polymer

chosen for this task was poly(2-ethyl-2-oxazoline) (PEtOx), a polymer in which our group has

a lot of expertise, that is hydrophilic and in which it is straightforward to include an azide end-

group by endcapping of the polymerization. It will be tested if the polymeric ligands can still

form with Fe(II) and Zn(II) when polyethyloxazolines of different lengths are attached on it.

After completing all previous tasks, it will be possible to test the use of the grid structures for

making supramolecular hydrogels in water. Therefore the ligand will be reacted on one side

with a triethylene glycol oligomer and the remaining acetylene group will be attached to

polymers whit an azide functionality on both chain ends, instead of only one azide

functionality at one chain end. With this, a polymer having ligands at both chain ends will be

obtained to form a supramolecular network as shown in Figure 3. Finally, the hydrogel

properties and self-healing behavior should be evaluated.

Figure 3: Polymer chains interconnected by ligand 8 as crosslinker

The remainder of this theoretical part will introduce supramolecular interactions,

supramolecular metallogrids and self-healing hydrogels, together with applications of self-

healing hydrogels.

4

1.2 Supramolecular Grid structures

The grid structures that are designed as crosslinker for hydrogels rely on supramolecular

interactions. This part will go more in details about these supramolecular interactions. In 1.2.1,

a brief introduction will be given to supramolecular interactions, in particular metal-ligand

interactions. Also the concept of self-assembly will be addressed here. 1.2.2 will talk more in

detail about the supramolecular metallogrids. 1.2.3 shows examples of how these grid forming

ligands behave in solutions of metal ions. Important aspects of compound 8 and its metallogrid

formation will be discussed.

1.2.1 Supramolecular Interactions by Metal-Ligand Interactions

Supramolecular chemistry is a different kind of chemistry compared to the covalent one.

Supramolecular forces are non-covalent physical interactions that act between different

molecules. These interactions can be directional and non-directional. They include host-guest

interactions, metal-ligand binding, hydrogen bonding, electrostatic interactions, π-π stacking

and hydrophobic effects. These supramolecular interactions all have different binding

strengths and strongly depend on the interaction partners as well as the conditions. The main

difference between covalent and non-covalent bonds is that these non-covalent interactions

are dynamic. The supramolecular bond is reversible, meaning that there is an equilibrium

between the molecules present as a supramolecular assembled entity and the molecules that

are not in this associated assembly.6,7,8 The metal-ligand interactions are the most important

concerning the formation of grid-structures.

Five parameters are important concerning supramolecular interactions. One describing the

equilibrium (the association constant Ka; equation (1) and (2)), two describing the binding

dynamics namely the rate of association (ka) and the rate of dissociation (kd) and the last two

being the concentration C and temperature T.11,12,13,14 Ka is a thermodynamic parameter which

describes the equilibrium between the molecules present as a supramolecular assembled

entity and the molecules that are not in this associated assembly. With this in mind, it can be

understood that a higher association constant represents a higher extent of formation of the

supramolecular entity. This constant is a widely used method for quantitating the affinity for

a supramolecular metal-ligand complex to be formed in solution. The Ka value of metal-ligand

5

interactions are very high, although they strongly depend on the chosen metal and may range

from rather weak up to being stronger than covalent bonds.15 High supramolecular association

constants are in general difficult to achieve in aqueous solutions, which is the reason metal-

ligand interactions were chosen to act as a crosslinker for hydrogels as they are known to

remain strong in water.

The equation for Ka at a certain temperature is different for 1:1, 1:2, 2:1 etc. complexes. In a

1:1 complex, one metal ion forms a complex with one ligand and the same trend is set for 1:2

and 2:1 complexes.13 The equation for a 1:1 equilibrium, the most simple equilibrium (a), is

given in formula (1).13,16 The more general formula (2) is valid for all supramolecular metal-

ligand complexes in which m is the number of metal ions in one complex and n is the number

of the ligands in that one complex.13 Formula (1) is a special case of formula (2) in which m

and n are equal to one. As an example, the values m=1 and n=2 are filled in formula (2). A

complex consisting of more than two molecules isn’t likely to form in one step, but more in a

stepwise process. Formulas (3) and (4) are the association constants K1 and K2 for a 1:2

complex, with the equilibria (c) and (d) respectively.13 The more processes there are, the more

difficult to fit and get to the KA value.13,17

𝐾𝑎 =[𝑀𝐿𝑥+]

[𝑀𝑛𝑥+][𝐿]

(1)

𝑀𝑛𝑥+ + 𝐿 ⇋𝐾𝑎 𝑀𝐿𝑥+ + 𝑛(𝐻2𝑂) (a)

𝛽𝑚𝑛 =[𝑀𝑚𝐿𝑛]

[𝑀]𝑚[𝐿]𝑛 (2)

𝑚𝑀 + 𝑛𝐿 ⇋𝛽𝑚𝑛 𝑀𝑚𝐿𝑛 (b)

𝐾1 =[𝑀𝐿]

[𝑀][𝐿] (3)

𝐾2 =[𝑀𝐿2]

[𝐿][𝑀𝐿] (4)

𝑀 + 𝐿 ⇋𝐾1 𝑀𝐿 (c)

𝑀𝐿 + 𝐿 ⇋𝐾2 𝑀𝐿2 (d)

Four metals and four ligands participate in the formation of a [2x2] supramolecular grid. The

overall binding constant is found in formula (5).

𝛽44 =[𝐹𝑒4𝐿4𝑛

8𝑥+]

[𝐹𝑒2+]4[𝐿]4 (5)

6

Experimentally, the Ka values at certain temperatures can be determined by measuring a

change in physical properties upon complexation, which can be done with methods like 1H

NMR spectroscopy, in which the chemical resonance is measured, and UV absorption, where

the absorption of the moieties is measured. ITC, isothermal titration calorimetry, is another

commonly used technique. ITC is, besides for measuring Ka, also commonly used for measuring

the stoichiometry of the formed supramolecular moieties. Other thermodynamic values,

including ΔG, ΔH and ΔS are also obtained with ITC.13,14,16,18

The degree of association at a given temperature,11 also called the degree of complexation,13

is dependent on both the value of Ka and on the concentration C of the supramolecular

moieties. This degree of association is proportional to [(KaC)1/2] 11. When talking about

hydrogels, the crosslink density in a supramolecular hydrogel can be estimated with this

equation.11

The supramolecular interactions of the [2x2] grid of our interest are dynamic, whereby the

individual moieties constantly associate to the supramolecular complex and constantly

dissociate back to their individual moieties. The rate at which this happens can be quantified

with ka and kd. The ratio of ka and kd leads back to Ka as shown in formula (6) and equation (e).

In a supramolecular hydrogel, a supramolecular crosslink will always fluctuate between an

active (associated supramolecular complex) and a dissociated crosslink (dissociated

supramolecular complex). 11,19 Different techniques exist to determine the kinetics of the

dynamic supramolecular exchange. The technique used depends on the time frame of the

dynamic process.6 For more information on how kinetic studies can be used to draw

mechanistic information, Bohn (2014)6 provides an interesting review.

𝐾𝑎 =𝑘𝑎

𝑘𝑑 (6)

𝑀 + 𝐿 ⇌𝑘𝑑𝑘𝑎 𝑀𝐿 (e)

Another important supramolecular concept that is important to discuss for these metal-ligand

interactions, is so-called self-assembly. Whitesides and Grzybowski defined self-assembly as

“the autonomous organization of components into patterns or structures without human

intervention.”20 The obtained supramolecular entity is formed spontaneously and depends on

external factors like the pH and temperature of the solution, the process is reversible and the

7

assembly formed in the end is the thermodynamically most stable entity. A supramolecular

assembly can be kinetically inert and kinetically labile with as difference between these two is

that a labile coordination complex has an exchange of the ligands and metals in the solution,

whereby an inert coordination complex isn’t capable to do so. An inert coordination assembly

has a high activation energy for exchanging ligands. This inert coordination is kinetically

trapped and does not necessarily go to its most thermodynamically stable supramolecular

assembly.9 The grid structure of this thesis on the other hand, with coordination by Fe(II) or

Zn(II), is proposed to be kinetically labile and should be able to go to the thermodynamically

most stable compound, due to a relatively low activation energy. It should be noted that the

grid is not formed at once when the reagents are added, but a lot of intermediates are formed

before reaching the thermodynamic minimum of the supramolecular grid.9

In Figure 4, two examples of metal-mediated supramolecular self-assembled structures are

shown. An equilibrium between a supramolecular square and triangle (Figure 4; left) is

observed when adding Pd(NO3)2, diethylamine and the ligands shown in Figure 4 in solution.21

At higher concentrations, the amount of supramolecular triangles compared to squares

increases while decreasing the concentration increases the formation of supramolecular

squares compared to supramolecular triangles due to the terms of the thermodynamic

equilibrium. The square has a lower enthalpy compared to the triangle, while the triangle is

favored over the square when considering the entropy.22,21 Another example of a metal-

mediated supramolecular self-assembled structure is the supramolecular grid (Figure 5; right),

the structure of our interest. This self-assembled entity can be formed by the combination of

a transition metal and a ligand, for example Cu(I) with 3,6-di(2-pyridyl)pyridazine, also called

DPP (Figure 5; left), which is a widely used ligand for supramolecular grids.23

8

Figure 4:Top: Equilibrium between a supramolecular square and triangle; Bottom: a supramolecular grid

structure. Reprinted from ref. 24

1.2.2 Supramolecular Metallogrids

Jean-Marie Lehn and Jack Harrowfield define that metallogrids ‘are oligonuclear metal ion

complexes in which the array of metal ions is essentially planar and each metal ion can be

considered to define a point in a square or rectangular structure.’25 To achieve such grid

structures, some design considerations need to be made. Besides the coordination number of

the metal, one also has to keep in mind the ligand’s coordination site. This coordination site

has to be perpendicular to the plane of the ligand at the metal centers.26 This coordination

setting at the site of the metal centers controls that a straight, stiff extension from one ligand

to a bigger ligand system will spontaneously give a two dimensional grid network, featured by

the metal ions lying in one plane.26 In practice, bidentate and tridentate subunits for the ligand

can be used in combination with tetrahedral, octahedral and sometimes bipyrimidal

coordinating metals. By imagining the structures, it can be understood that bidentate ligands

need a tetrahedral coordinating metal such as silver (I) or copper (I) and tridentate ligands

require an octahedral ion such as Fe(II), Zn(II) or Co(II) to get to the grid structure. In the case

of our grid forming compound, it is a tridentate ligand in combination with a octahedral ion.

9

The ligands mostly contain nitrogen atoms, for their good donor qualities, but also examples

of oxygen and sulfur containing compounds exist. Bi- and terpyridines are used frequently as

basis for the grid-forming ligands. The ligands shown in Figure 5 can all chelate metals and

form metallogrids. During self-assembly, the increase in preorganization could give rise to

cooperativity.26 The intermediates formed during self-assembly are kinetically labile. At the

end, the thermodynamically most stable grid structure, together with the right metal ions, is

formed. On top of that, aromatic ring systems are rigid and capable of having π-π interactions,

which are an extra stabilizing force between the ligands when being in the grid structure. It

should be noticed that formation of a grid structure is thermodynamically favored, when

compared to other structures (for example polymer structures) albeit this will be

concentration dependent. A favorable enthalpy term comes from coordination of the metal

in the coordination site of the ligand (note that a non-cyclic polymer would have non-occupied

binding sites at the end). The grid structure is the structure which can be present with the

highest amount of discrete entities which have all coordination sites occupied, making it

entropically and enthalpically favorable. An important aspect to get to the desired grid

structure, is a necessity to force it to the right geometry, which is perpendicular, and a drive

to obtaining fully occupied binding sites. Moreover, internally there also has to be an imposed

orientation. Examples are steric effects, which suppress the formation of unwanted entities.

Another example of this is stabilizing interactions, for example π-π stacking, which orientate

the ligands to form the desired grid. Lastly, external factors like the presence of counterions

or solvent molecules also have to be kept in mind.26

10

Figure 5: Ligands capable of forming grid structures by self-assembly 26,27,28,29

1.2.3 Supramolecular Grids in Solution

In solution, the supramolecular grid is not always that simple to form as some other entities

can also be present in solution. The grid formation is not unambiguous, being dependent on

the environment, the solvent, the concentration of the species, the counter anions of the

metal, the metal coordinating character and of course the identity of the ligand. In this part,

examples of ligands yielding grids and the limitations of some will be discussed.26

First of all, ligands designed for grid formation don’t necessarily form grids. The structure 2 in

Figure 5 can form a double helix, a triangle and a [2x2] grid.26 The reason for this is the poor

preorganization of the ligand. If the grid is the only structure that is wanted, a more

preorganized ligand is necessary. By changing the two inner pyridines by 4,7-phenantroline,

this preorganization can for example be introduced.

11

Another parameter capable of altering the supramolecular entities is the counter anions of

the metal ions in solution. By templating anions inside the cavity of the formed supramolecular

structure, a grid can only be formed with the right anions. An example of this is ligand 1 in

Figure 5 (DPP), which was studied for its grid formation with non-matching tetrahedral

coordinating metal ions together with other anions present in solution. With Ni(II) or Zn(II) in

combination with BF4- or ClO4-, a [2x2] was formed. Having the bigger SbF6- in solution results

in the formation of, a pentagon (5 ligands together with 5 metals, as opposed to 4 ligands and

4 metals in the [2x2] grid). The reason for this observation is that the pentagon can fit the

bigger anion SbF6- inside the cavity instead of smaller ions like BF4- and ClO4-, which fit better

into the grid structure.27 This effect is called the templating effect.26

The combination of the valency of the metal and the coordinating character of the ligand has

also to be respected. Again studies on DPP, ligand 1 in Figure 5 were done. In combination

with Zn(II)perchlorate in acetonitrile, a tetrahedral coordinating ligand and a octahedral

metal, a triple helicate instead of a grid is obtained. It should be noted that this example is

unusual. The helicate motif normally forms with ligands containing flexible linkers whereas

the 3,6-di(2-pyridyl)pyridazine is a very rigid ligand.30

Another interesting observation is made on the large ligand shown in Figure 6, made by the

group of J.-M. Lehn. This ligand has coordination cavities which should make it possible to

form a [5x5] grid.31 Interestingly, the [5x5] grid was not observed in solution together with

Ag(I) in nitromethane. What is observed instead is a [4x5] grid (Figure 6). Baxter and Lehn give

three reasons for this behavior. First, the nitrogen atoms, when not forming a complex, are in

a so called transoid conformation, which is more stable than being in the cisoid conformation.

In a transoid conformation, the nitrogen responsible for coordination are trans to each other

over the C-C sigma bond which interconnects the pyridines. This transoid conformation is

more stable because the free electron pairs of the nitrogen atoms repel each other. For

forming a complex, they have to be in the less stable cisoid formation. Secondly, the N=N

bonds are shorter compared to C=C bonds, giving the ligands in the [5x5] grid a non-straight,

domelike shape, also disfavoring its occurrence. Lastly, the Ag(I)-N bond in the middle of the

[5x5] grid would have very long bond lengths. Combining all these factors make that the [4x5]

grid is formed. They also denoted it as a [2x(2x5)] grid because the ligands are placed on two

12

sides, which can be more clearly seen in Figure 6.31 It should also be noted that the [4x5] grid

is not formed exclusively. A decanuclear helicate structure together with Ag(I) is also formed

indicating the formation of kinetically trapped structures.

Figure 6:Top: ligand designed to form a [5x5] supramolecular grid; Bottom: the formed [2x(2x5)]

supramolecular grid. Reprinted from ref. 31

The idea of using a Cu(I) catalyzed azide-alkyne cycloaddition between compounds and letting

the triazole-ring act as nitrogen-donor ligand instead of pyridine has been reported for

bipyridine32 and terpyridine33,34 analogues that could successfully be used for metal

coordination. Furthermore, there is one report of a triazole grid-forming DPP analogue. This

DPP analogue was prepared based on 3,6-(bisethenyl)pyridazine (figure 7) onto which an

oligoethyleneglycol was cycloadded yielding the grid-forming ligand. After the cycloaddition,

the obtained product successfully formed a [2x2] grid together with [Cu(CH3CN)4]PF6 in

dichloromethane (DCM). 35 In this work, we will use a related strategy to prepare terpyridine

like grid-forming ligands.

The examples given above all report the formation of grids in solvents other than water. In

this, we aim to make grids in water, this for opening the gate to biological applications. The

reason chosen for our ligand instead of DPP is because DPP doesn’t form grids in water with

Cu(I). This phenomena was discovered by studies on DPP with hydrophilic poly(ethylene

glycol) chains polymer groups on it (number 9, Figure 5). In DCM, this compound is able to

form the desired grid. In water on the other hand, the grid is not formed, even though having

hydrophilic polymer groups bound to the ligands. Reason for this behavior is a

disproportionation reaction, meaning, Cu(I) gets partially oxidized to Cu(II) and reduced to

13

Cu(0). This result was apparent both when the Cu(I) was titrated to the ligand in a watery

solution and also when the grid was self-assembled in DCM first and subsequently transferred

to water. 29

Figure 7:Cu(I)AAC reaction for obtaining grid forming ligand

In summary, formation of only [nxn] grids without other supramolecular species being found

is negatively affected by having a ligand which can form complexes in the cisoid conformation.

The ability to form a domelike shape is also undesirable. But, like mentioned previously, a

drive to obtain the fully occupied binding sites by getting the maximum amount of

coordination, the highest amount of π-π stacking between ligands and the right choice of

anions, makes that exclusively forming supramolecular grids is possible.31 Ligand 8 is included

in Figure 5 as this compound is interesting for future work as hydrazides are relatively easy to

make.

1.3 Supramolecular Hydrogels

In polymer science, supramolecular chemistry and interactions is of major importance. The

special properties of polymers like Kevlar and nylon rely on cooperative non-covalent

interactions, more precisely on hydrogen bonding. For a lot of applications, including self-

healing hydrogels, the specificity and directionality of these non-covalent supramolecular

interactions can be used. Like mentioned before, this can be done by replacing covalent

crosslinks by directional dynamic non-covalent interactions. The use of supramolecular

chemistry and self-assembly in the polymer field is expected to be a promising path for smart,

adaptive and self-healing materials. Zhang et al. (2012)36 referred to supramolecular gels as

‘[…] the most promising materials in this field [..]’ when talking about self-healing hydrogels.

This also is the application what our supramolecular grid was designed for, to be used in

supramolecular self-healing hydrogels. In this part, hydrogels and in particular supramolecular

14

hydrogels are introduced. Furthermore, the phenomenon of the shear-thinning effect and its

importance will be discussed.

1.3.1 Hydrogels

Hydrogels are three-dimensional structures that are held together by covalent or physical

crosslinks which can contain a substantial amount of water.1,2,3,4 The amount of water they

can absorb ranges from 10% to a thousand times their dry weight.37 The monomers for

building the polymer structures are hydrophilic, examples being carboxyl, amino and hydroxyl

groups. These hydrophilic groups are the main reason for the hydrogel being able to hold

water. Water holding capacity depends on the amount of hydrophilic groups as well as the

crosslink density. Higher crosslink density leads to a less stretchable polymer network and

therefore a lower maximum swelling capacity.38 Besides water, also nutrients, medicines and

other compounds of interest can be inserted in the hydrogel as aqueous solutions.1,5

Different classifications for hydrogels exist,1,38,11 the type of polymer chain can be used for

instance, namely ionic or neutral chains.11 More useful for our application is the type of

crosslinking, which can be permanent or non-permanent. In a permanent hydrogel, the

crosslinking is covalent. Because of this, these hydrogels are also called chemical hydrogels.

An example are PHEMA (poly(2-hydroxyethyl methacrylate)) hydrogels as used for contact

lenses. A crosslinked hydrogel of PHEMA is made by radical polymerization, for which HEMA

(2-hydroxyethyl methacrylate) is polymerized together with EGDMA (ethylene glycol

dimethacrylate) as bifunctional crosslinker.37 Other reactions for crosslinking include Michael

type additions, 1,3-dipolar cycloadditions of alkynes and azides and Schiff base formation.11

The equilibrium swelling in an aqueous solution of such covalent hydrogels is depended on

the crosslink density that can be quantified with MC, which is a parameter to approximate the

average molecular weight between the crosslinks of the hydrogel.37 Chemical crosslinking has

the advantage that optimization for the required application is easy. It is mostly used when

the application requires though and stable hydrogels.11 Limitations of permanent hydrogels

are the chemical techniques used for crosslinking leading for instance to traces of metal

catalysts, functional groups of unreacted crosslinker or photoinitiators. Also the use of UV light

for initiating the crosslinking has its disadvantage, for example in vivo crosslinking is something

15

which is not possible.11 Other limitations are the permanent shape making them non-reusable

and non-reshapeable.

A non-permanent hydrogel is called a physical, dynamic, reversible or a supramolecular

hydrogel.1,37,38,7 The driving force for gel formation is molecular self-assembly.11 The forces

holding together the hydrogels are secondary forces, such as hydrogen bonding, ionic

interactions, and/or molecular entanglements,37,38 which all are transient forces.37,11 Gel

formation can be induced in water, without a pronounced change of volume. Consequently,

no crosslinking reagents are necessary. Instead of permanent crosslinking during the synthesis

of the chemical hydrogel materials the crosslinks are dynamic and they are constantly binding

and dissociating. The consequence of this is that a weaker gel is obtained, which can be more

easily sheared and reversibly deformed by mechanical forces. This dynamic character also has

its advantages as it gives self-healing and shear-thinning properties to the hydrogel.11,1

Examples are PVA-glycine hydrogels, gelatin and peptide hydrogels. Stimuli-responsive

hydrogels can change their equilibrium swelling by a change in surrounding environment.

Examples of these changes are pH and temperature.39,40,41

Figure 8: Comparison of a chemical (a.) and a dynamic hydrogel (b.). Reprinted from ref. 11

1.3.2 Self-Healing Hydrogels

Self-healing polymer gels use dynamic bond formation, meaning that a bond between

different polymer chains dissociates and recombines in a dynamic equilibrium. Most self-

healing hydrogels are physical hydrogels which mostly rely on non-covalent forces as ‘cross

linkers’. These secondary forces are non-covalent binding motifs, which can be specific and

directional.11 The supramolecular interactions used include host-guest, multivalent ionic,

16

metal-ligand, hydrogen bonding, formation of stereo complexes and biomimetic interactions.

Different supramolecular interactions have different binding strengths and dynamics and one

can choose the right supramolecular interaction depending on the application of interest.

Inclusion complexes of cyclodextrins have moderate association constants (Ka) while metal-

ligand complexes have a high association constant, therefore our choice went to metal-ligand

interactions. With this in mind, it can be understood that a higher association constant

represents a stronger formation of the supramolecular entity. Cucurbit[n]urils are also

interesting for hydrogels because their binding strength can be varied over a range of 10

magnitudes depending on the cucurbit[n]urils and chosen guest.11

It is worth mentioning that not all self-healing gels are physical hydrogels. Chemical self-

healing gels have also been developed based on dynamic covalent bonds, for example

disulfide bonds or reversible Diels-Alder cycloadditions.42 Also, not all physical hydrogels rely

on dynamic secondary forces, an example being self-healing gels based on crystallization and

polymer-nanocomposite interactions resulting in kinetically trapped hydrogel structures.42

Figure 9: Different compounds with different Ka values. Reprinted from ref. 11

1.3.3 Applications of Self-Healing Hydrogels

Hydrogels have properties which are interesting for different biomedical applications,

including tissue engineering and drug delivery. In tissue engineering, the replacement of

organs by in vitro grown man-made organs, the hydrogel can be used as scaffold for cell

growth.43 For their use in drug delivery, drugs can be loaded into the pores and released at

the designed rate.44 An important requirement for implants is biocompatibility, to which the

high water content of hydrogels is beneficial. The drugs can be distributed to the specific tissue

of implantation but systematic delivery is also possible. It has to be kept in mind that also the

hydrogel forming material itself shouldn’t give physiochemical problems in the extracellular

17

matrix, where it is implanted, and should have mechanical resemblances to the place of

implant.1,5

A method for introducing the hydrogel into the body is by direct implantation of the

crosslinked material in the body, in other words, the drug containing crosslinked hydrogel is

made in advance outside the body before being implanted. A more convenient method is to

inject the drug containing hydrogel through a needle, instead of implanting it by surgery. This

is where physically crosslinked hydrogels, like ours, could be used for because of having shear

thinning properties,1,5 which is a decrease in viscosity when undergoing an increase in shear

strain. Physically crosslinked hydrogels are often mechanically less strong compared to

chemically crosslinked hydrogels, but have shear-thinning and self-healing properties, make

them interesting for biomedical applications. Furthermore, soft injectable hydrogels are less

susceptible to damage than more tough hydrogels. The ease of tunable mechanical properties

and good flow and recovery make injectable hydrogels a good candidate for drug delivery

applications.5

Shear-thinning properties can be characterized via rheology measurements. With this

method, the effect of shear rate on viscosity and shear stress can be obtained. In Figure 10

left, an example of such an experiment is given. The hydrogel used in this example is a

supramolecular hydrogel consisting of a host polymer, CD-HA (cyclodextrin modified

hyaluronic acid), and a guest polymer, Ad-HA (adamantane modified hyaluronic acid). In the

graph, the closed symbols are the data of shear stresses and the open symbols represent the

viscosity, this for different weight percentages of Ad-HA in CD-HA. The evolution of the shear

stress and viscosity as a function of shear rate clearly indicates shear-thinning behavior by a

decrease in moduli with increasing shear rate.5 The shear rate that the hydrogel will undergo

when injected with a syringe is thought to be higher than tested here and will lead to a further

decrease in viscosity.5

After being injected, the hydrogel needs to recover to form the crosslinked material by self-

assembly. The timescale in which this happens is important, mostly a faster recovery is more

desirable to avoid loss of material by diffusion. Some hydrogels take several minutes or even

hours to recover from shear stress, which limits their applicability.5,45 The cargo in the

hydrogel and the hydrogel material itself can diffuse out of the material before it is recovered

18

and therefore not using the method to its full capacity. Some complementary binding motifs

have very fast recoveries, including the Ad-HA and CD-HA host-guest system discussed above

which has ‘a near-immediate recovery following shear-thinning delivery, allowing optimal

retention of both material components and potential therapeutic cargo at the target site.’5 To

come to this conclusion, Rodell et al. exposed the material to cycles of oscillatory strain,

alternating a large amplitude and a low amplitude (Figure 10 right), in which the fast recovery

can be observed independent of the number of cycles.5

Figure 10: Left, Rheology measurement of Ad-HA and CD-HA; Right, oscillatory strain experiment, the full

black line is the applied oscillating strain, the red and blue lines are the G’ storage and G” loss modulus

respectively. Reprinted from ref. 5

1.4 Supramolecular Grids in Polymers and Gels

The supramolecular grids mentioned in 1.2 were not yet incorporated in polymer structures,

except the triazole DPP analogue. The question arises if these grid structures can also be

formed when they are part of a polymer chain. The DPP ligand A (Figure 11) immediately yields

a supramolecular grid with tetrakisacetonitrile Cu(I) hexafluorophosphate in DCM. Whether

ligands B, C and D (Figure 11) could also form a grid structure was investigated by UV-Vis

titration studies. The supramolecular complexation model, starting from metals and ligands in

solution, to the supramolecular grid structures, is rather complicated as a large number of

intermediate complexes can be formed. A simplified mechanism can be considered by

equilibria (f) and (g). In the case of B, the complex that is most present below 0,1

19

stoichiometric equivalents of Cu(I) is L2Cu, whereas going to more than 0.1, the L4Cu4 grid was

most prevalent indicating strong cooperativity in the grid formation. In the case of C and D,

the shift to the grid being the most prevalent entity was only observed above 0.5 equivalences

of Cu(I). The proposed explanation given for the different behavior of C and D compared to B,

was that more sterical hindrance in C and D suppresses the cooperativity effect. Therefore,

the grid of B is formed at lower Cu(I) concentrations and is also stronger than grids of C and

D.17

2 𝐿 + 𝐶𝑢(𝐼) 𝑘1→ 𝐿2𝐶𝑢(𝐼) (f)

2 𝐿2𝐶𝑢(𝐼) + 2 𝐶𝑢(𝐼)𝑘2→ 𝐿4𝐶𝑢(𝐼)4 (g)

Figure 11: Ligand A, B, C and D

Another example makes use of ligand 8 (Figure 5), a bis(acylhydrazone), with Zn(II) as the d-

metal for coordination.28 This compound is capable of forming an organogel in toluene. The

MGC, minimum gelation concentration, or the lowest concentration of the gelator that is

needed to form a gel, was sought. In toluene, above ambient temperature, this was 18 mg/ml.

Heating of the gel above 38 °C makes the gel transform to a sol. By changing the concentration,

the gel to sol transition temperature could be changed. At a concentration of 25 to 35 mg/ml,

the transition temperature was found to be 44 °C. Higher concentrations were not tested. By

using 1H-NMR and IR, it could be determined that the grid structure was present in the

organogel. A DSC measurement on the sol-gel transition was also performed with a 20 mg/ml

concentration of ligand 8, to gain insight in the mechanism. Upon heating, an endothermic

peak was found between 35 to 67 °C. Cooling resulted in a exothermic peak between 48 and

18 °C, on which the authors state that this “demonstrates the full thermoversibility of the self-

assembly process.”28

20

1.5 Supramolecular Metal-Ligand Interactions in Hydrogels

The idea of using grid like metal-ligand interactions as a supramolecular crosslinker in

hydrogels has not been explored, but related concepts exist. In early work, hydrogels were

made by bipyridyl-branched poly(2-methyl-2-oxazoline) together with Fe(II) and Ru(III) salts

by metal ligand complexation, when being at low temperatures and high concentrations.46 In

more recent work, terpyridine systems are frequently used. An example is a study of

terpyridines at the ends of an eight armed PEG star polymer yielding hydrogels with Fe(II) and

Ni(II).47

Another frequently used systems are catechols and comparable structures on star-polymers

together with Fe(III). These systems are also common when discussing self-healing. One major

advantage is their high association constant making them suitable for use in water under

ambient conditions and thus hydrogels. Tris- and bis- catechol-Fe(III) complexes have a

stability constant of Ks~1040. The amount of force needed to break this bond is only slightly

lower compared to breaking a covalent bond.15 With the information of these articles, it is

clear that making a self-healing gel with metal-ligand interactions as crosslinker, is possible.

To make these catechol based hydrogels, a four-armed star-polymer, PEG- DOPA4 with an

Mw~10000 is used together with FeCl3 with the ratio of FeCl3 DOPA being 3:1. DOPA stands for

dihydroxy-phenylalanine and is an amino acid which has a catechol as side group. An

advantage of this polymer is that the properties can be changed by altering the pH. At pH of

5, it is a green/blue fluid with mono-catechol complex while raising the pH to 8 leads to a

purple gel consisting of the di-catechol-Fe complex. At pH 12, a red gel of the tris-catechol-

complex is present. A change of color isn’t the only useful aspect, the properties of the gel

also change by pH. At pH 5, the system behaves viscous, so more like a fluid. The gels at pH 8

and pH 12 behave as elastic gels.15 The self-healing properties were also tested. To do so, a

covalent crosslinked gel was made by oxidizing PEG-DOPA4 with NaIO4. When comparing this

covalent gel to the supramolecular hydrogel, it turned out only the supramolecular hydrogel

intrinsic healed when they are damaged. The healing time was in the order of minutes.15

21

One major problem concerning catechol-Fe(III) based hydrogels is the gel instability in time.

Catechol can oxidize to quinone and this can react with catechol, forming a covalent crosslink.

On an hour time scale, this is not a problem,15 but on a month time scale, it is.48 As a result,

the supramolecular self-healing hydrogel loses its excellent properties in time. To overcome

the disadvantage of instability in time, analogues for DOPA were searched and 3-hydroxy-4-

pyridinonone (HOPO) turned out to be a good one. When used on a PEG star-polymer with

the same molecular weight as used for DOPA and using Fe(III) as metal ion, no instability in

time was found.48

This PEG-HOPO4 also has a gelation at physiological pH, instead of the need for a high pH like

DOPA polymers, making it interesting for injecting in the body. To test this hypothesis, it was

injected in a buffer solution of physiological pH. It turned out that by doing this, a good gel

was obtained, opening its way to biological applications.48

1.6 Poly(2-oxazoline)s

Lastly poly(2-oxazoline)s will be introduced as this is the polymer chosen for cycloaddition to

obtain the polymeric grid forming ligand. Poly(2-oxazoline)s can be made by the ring opening

polymerization of 2-oxazolines. In these compounds the R group (Figure 12) can be altered

and chosen to obtain the desired properties.49,50 It can be simple alkyl chains like a methyl or

an ethyl which results after polymerization in hydrophilic polymer. n-Hexyl as R group on the

other hand yields a hydrophobic low glass transition temperature polymer.49 Besides simple

alkyls, also chemically more interesting moieties can be introduced on this place, including

functional handles like alkenes and esters.50

Polymerization of 2-oxazolines is done by cationic ring opening polymerization (CROP), which

is a living polymerization.49,50 No intrinsic termination occurs in contrast to controlled radical

polymerization. Also, the repellence of the growing cationic polymer chains suppress reaction

between the two growing polymer chains and hinder termination, making high monomer

conversion possible.50 Thermodynamically all polymerizations are unfavorable in terms of

entropy, but the polymerization of 2-oxazolines has a negative enthalpic term due

isomerization of the iminoether to a more stable amine functional group. In total, the

22

enthalpic term is more pronounced and polymerization is thermodynamically favorable after

initiation.50

Figure 12: Polymerization mechanism of EtOx with IX as initiator and TY as terminator

Initiation of the polymerization is carried out by nucleophilic attack on the monomer (Figure

12). Lewis acids51, chloroformates52 and alkylating agents can be used as initiators. Nowadays,

alkylating agents like methyl tosylate are most popular.50 The identity of the counterion alters

the equilibrium of active cationic and non-active covalent entity and thereby the

polymerization rate (Figure 13).50

Figure 13: Equilibrium between cationic and covalent propagating species

After the monomer is started by initiation (Figure 12), the propagation step takes place. This

step is the heart of a polymerization, as the growth from a small molecule to a polymer takes

place. In this step, the equilibrium shown in Figure 13 is still existing. The propagation rate kp,

depends on the counterion, solvent, the type of monomer and temperature. kp follows

Arrhenius’ law given by formula 7. With respect to the parameters in these formulas, the

molecular weight of the polymer at a given reaction time can be calculated.

ln ([M]0

[M]t) = kp[I]0t (6)

kp = A e−

EaRT (7)

Termination of the living polymer chains by adding an end capping agent can occur on the two

and on the five place. Terminating agents like azides50,53,54, thiolates and sodium hydroxide

attack in the five position while water attacks on the two and the five positions (Figure 14)50.

23

As water is an effective terminating agent, the CROP of 2-oxazolines has to be carried out

under very dry conditions.55

Figure 14:Termination can occur on the two and on the five position of the oxazolinium chain end

Another interesting feature of polymerizing 2-oxazolines is the ease of introducing chain-end

functionalities. By choosing a terminating and initiator agent with the functionality of choice,

end-group functionalities can be introduced.50 Propargyl p-toluenesulfonate as initiator is an

interesting compound for introducing an acetylene end-group56. Having two azide end-group

functionalities can also be carried out. This can be carried out by an azide functionalized

initiator57 in combination with quenching the reaction with an azide. Another method is using

diiodoalkyl initiation. Subsequently, sodium azide can be added to terminate and get a

polymer with azide functionalities on both ends54 (Figure 15).

Figure 15: Synthesizing PMeOx (poly(2-methyl-2-oxazoline)) with azide functionalities on both chain-ends.

PAOx (poly(2-alkyl/aryl-2oxazoline)) are not only interesting because the polymerization of 2-

oxazolines is living, the obtained polymers have many applications in life sciences. Interesting

is the stealth behavior, meaning the body does not recognize them as body foreign and, thus,

no or limited immune response occurs.58 This stealth behavior make PAOx ideal for drug and

gene delivery.50 PAOx are also ideal for making hydrogels. Covalent PAOx hydrogels have been

used for drug delivery and as cell culture scaffolds.59 By choosing the right monomers,

hydrophobicity and other properties of the biocompatible hydrogels can be altered. Surface

24

modifications for making non-fouling surfaces are another biomedically focused application

of PAOx.50

25

2 Results and Discussion

2.1 Synthesis Route

The synthesis route used for synthesizing compounds 3 and 4 (Figure 16) was inspired on

literature.60 First, 2,6-dibromopyridine 1 is asymmetrically substituted with triisopropylsilyl

(TIPS) protected acetylene 2 which is done via Sonogashira coupling, to give 2-bromo-6-

[(triisopropylsilyl)ethynyl]pyridine 3. Compound 3 is then reacted with n-BuLi, giving a

bromine-lithium exchange. Treating the intermediate with tributyltin chloride (Bu3SnCl) yields

the stannylated compound 4. The stannyl containing molecule 4 is coupled with compound 5

via a tetrakis catalyzed Stille coupling, yielding compound 6. This final reaction has not been

reported in literature before. Lastly, the deprotection of 6 is done with tetra-n-

butylammonium fluoride (TBAF). The synthesis of compound 3 to 7 is discussed in section 2.2

ligand synthesis.

Figure 16: Synthesis route for ligand 8

For the synthesis of the functional ligand 8, a copper(I)-catalyzed azide alkyne cycloaddition

will be used. With this reaction, different azide functionalized compounds can be attached to

protoligand 7, directly yielding the novel triazole containing grid-forming ligand 8. These azide

compounds can be oligomers, small molecules or polymers and can be used for incorporating

the grid in a polymer network, the final purpose of this project. As for the azide functionalized

compound groups, 2-(2-(2-methoxyethoxy)ethoxy)ethyl azide 13 was prepared as water-

26

soluble oligomer and azide functionalized polyethyloxazolines were prepared with 20, 50 and

100 repeating units as water-soluble polymers. The synthesis of these oligomers and polymers

is discussed in section 2.3. polymer and oligomer synthesis.

2.2 Ligand Synthesis

2.2.1 Synthesis of 2-Bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3

Figure 17: Sonogashira coupling

The first reaction step involves the asymmetric substitution of only one bromine atom of 2,6-

dibromopyridine 1, by a Sonogashira coupling reaction (Figure 17) in tetrahydrofuran (THF) by

a TIPS protected acetylene group. PdCl2(PPh3)2 and Pd(PPh3)4, which is also called tetrakis,

were both tested as catalyst for this reaction. PdCl2(PPh3)2 60 is the catalyst used in literature

while tetrakis is also commonly used in such cross-coupling reactions. The reaction worked

well with both catalysts.

These catalysts are very oxygen sensitive. Therefore, the reaction mixture needs to be oxygen

free. Different methods for getting the mixtures oxygen free were tested. A first one was by

bubbling the solvent together with 2,6-dibromopyridine 1, the catalyst, CuI and triethylamine

(NEt3) with Ar. After bubbling, the (triisopropylsilyl)acetylene 2 was added dropwise. Another

method used for getting the mixture oxygen free was by using a schlenk line with a vacuum

and Ar line. The powders in the reaction vessel were brought under argon. Subsequently, the

solvent and NEt3 were added with a syringe. Then the (triisopropylsilyl)acetylene 2 was added

dropwise. Good results were obtained using both methods.

27

In the article used for the synthesis of this compound 3,60 a three times molar excess of 2,6-

dibromopyridine 1 to(triisopropylsilyl)acetylene 2 was used to favor the mono substituted 2-

bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 instead of the disubstituted 2,6-

[di(triisopropylsilyl)ethynyl]pyridine 3A (Figure 18). After finishing and purifying the reaction

products, 2,6-dibromopyridine 2 is still intact and can be used again in another reaction,

making the excess not wasteful. Another reason number 2 is used in excess is the price of

those chemicals.

Figure 18: The three major compounds in the crude reaction mixture

The characterization of the three main compounds in the reaction mixture was not straight

forward. With ESI-LCMS, 2-bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 can be identified.

Compound 1 and 3A were not ideal for LCMS analysis. The 2,6-dibromopyridine was possibly

too polar to be found with ESI-LCMS and elutes to fast while 3A adhered to the column and

gave a very broad trail instead of a nice peak as it is too apolar. GCMS gave counts for all three

compounds (Figure 19). 2,6-dibromopyridine comes of first with a retention time of 11.7

minutes. The next compound is 3A, following at a retention time of 16.67 minutes. The

molecular ion of this compound is not visible because the m/z is only measured to 400 Da.

When measuring to 600 Da, the compounds fragment too much to be clearly identified.

Therefore, qualitative analysis of this peak could not be done, but by excluding other peaks

from being the disubstituted pyridine 3A, this peak at 16.67 could be assigned. The desired

compound 3 eluted last from the column into the mass spectrometer. All three compounds

show a characteristic mass spectrum of an aromatic compound, meaning that only the

substituents give major fragmentation while the pyridine is stable. On a silica TLC plate with

n-hexane and 15% of ethyl acetate (EtAc), the three peaks of the different compounds were

also visible. The spots were identified via larger scale preparative TLC in combination with

GCMS and 1H NMR spectroscopy. The lower spot is molecule 1, the middle spot 3 and on top

is compound 3A.

28

Table 1: tr values of GCMS, LCMS and rf values of TLC

Compound tr GCMS tr LCMS rf TLC

2,6-dibromopyridine 1 11.7 - 0.50 2-bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 17.3 0.53-6.28 0.90 2,6-[di(triisopropylsilyl)ethynyl]pyridine 3A 16.7 4.600 0.73

Figure 19: GCMS results

In the 1H NMR spectra (Figure 20), the 2,6-dibromopyridine 1 peaks overlap with those of

compound 3. The conversion of the aromatic peaks is shown when purifying the crude mixture

B to the purified compound 3. Spectrum B is a superposition of 2 molar equivalences of 2,6-

29

dibromopyridine 1 and 2-bromo-6[(triisopropylsilyl)ethynyl]pyridine 3. Upon integration, a

pure 1H NMR spectrum of 2-bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 should integrate for

3 hydrogens in the aromatic area while the TIPS protecting group, consisting of three isopropyl

groups having a proton chemical shift of 1,12 and 1,13, should integrate for 21. By combing

1H NMR spectroscopy, TLC and LCMS analysis, unambiguous identification of the desired

molecule 3 can be done.

Figure 20:1H NMR spectra of compound 1 (A), the crude reaction mixture (B) and compound 3 (C). In the box, a

zoom of the aromatic area is shown.

In literature, the purification of 3 was described by column chromatography on silica with n-

pentane and DCM as eluent, starting from 90% n-pentane to 70% in the end. When using these

solvents, no clean product was obtained. The fraction containing most of the desired

compound 3 still had a reasonable amount of 2,6-dibromopyridine 1 in it, which is undesirable

for the next reaction step. On TLC with silica as stationary phase, these compounds came very

close to each other. To improve the separation other eluents were tested on TLC (Table 2).

These solvents were also tested on TLC’s with neutral aluminum oxide as stationary phase but

this resulted in bad retention and almost no separation. The best result was obtained with

silica using EtAc in n-hexane. n-Hexane with EtAc starting from zero percent of EtAc was

30

chosen as eluent system for the column chromatography. The silica column was first run with

0.5% of NEt3 in n-hexane before loading the compounds, to make the silica less acidic. Even

with these optimized solvents, still no completely pure compound could be obtained. The

most pure fraction of compound 3 still had 1 and 3A as minor impurities in it (analyzed by

GCMS and TLC). In the 1H NMR spectrum, the aromatic area integrated for 3,93 hydrogens

when setting the integration of the TIPS groups to 21. Because there are 3 compounds

present, the disubstituted 3A and 2,6-dibromopyridine 1 being the minor ones and the desired

2-bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 being the major one, it’s not possible to

unambiguously determine the percentage of the 2,6-dibromopyrimidine 1 that still is present

in the 2-bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3.

Table 2: TLC results of the crude reaction mixture with different eluents.

Eluents % second solvent rfA rfB rfC remark

n-pentane - - - - - one trail

n-pentane DCM 10 0,47 0,26 - only 2 spots

n-pentane DCM 20 0,83 0,58 0,46 spots

n-hexane - - 0,13 - - one spot

n-hexane CHCl3 10 0,42 0,38 - trail, only 2 spots

n-hexane DCM 10 0,21 0,14 - trail, only 2 spots

n-hexane DCM 20 0,49 0,38 0,08 trail

i-octane DCM 20 0,22 0,18 - peak overlap

n-hexane EtAc 4 0,81 0,59 0,39 spots

n-hexane EtAc 10 0,86 0,57 0,47 spots

n-hexane EtAc 15 0,90 0,73 0,50 spots

n-hexane EtAc 40 0,98 0,93 0,80 spots

As is already stated above, the disubstituted compound 3A does not give problems in the next

reaction step. Therefore, another less time consuming method, compared to column

chromatography, was searched to purify the crude mixture. The diiodopyrimidine 1 has got a

reasonable boiling point at standard pressure, being 255°C.61 This opens the way for other

purification methods, including vacuum distillations. A Kugelrohr setup, which is de facto a

short distance vacuum distillation with a rotating oven was used for removing 1 from the

crude. The pressures needed to evaporate the 2,6-diiodopyrimidine 1 out were calculated (SI

2). Pressures of 0.4 mbar could be reached with the used setup, making a temperature of 85°C

ideal to remove 2,6-dibromopyridine 1 from the crude mixture. Subsequently, it was also tried

to distil over molecules 3 and 3A. The high temperatures needed led to boiling of all content

31

in the flask, by which also drops containing catalyst came over in the trap. Another problem

when distilling 3 and 3A over was that at temperatures above 150°C, a black tar was formed,

as it indicates degradation of the products. For the final purification method, the Kugelrohr

was used in a first step to distill all of the 2,6-dibromopyridine 1 out of the crude mixture.

Subsequently, a flash column chromatography was done on silica with 30% EtAc in n-hexane

to get rid of the catalyst and all the other entities present in the crude mixture. Following this

procedure, 2-bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 was obtained, which only

contained a minor amount of 2,6-[di(triisopropylsilyl)ethynyl]pyridine 3A. In the 1H NMR

spectrum, the aromatic area integrates for 2,96 while the TIPS groups integrate for 21 (SI 5).

With this it can be calculated that the amount of 3A in 3 is around 3%. Even though this value

is not very reliable as the integrating on 1H NMR spectra has some inaccuracy, it does give an

idea of the amount of disubstituted pyridine 3A in our otherwise clean compound 3 that was

utilized for the next reaction.

2.2.2 Synthesis of 2-[(Triisopropylsilyl)ethynyl]-6-(tributylstannyl)pyridine 4

Figure 21: Substitution of Br with tributylstannyl by first a halogen-lithium exchange

followed by reaction with Bu3SnCl

2-Bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 was first treated with n-BuLi to give a lithium

intermediate. Subsequently, tributyltin chloride (Bu3SnCl) was added to prepare the

stannylated compound 4. The conditions initially used for this reaction were based on

literature.60 A mixture of 3 in THF was treated with n-BuLi at -78°C after which it was allowed

to heat till -20°C. Bu3SnCl was added when the mixture was at -78°C again after which it was

allowed to heat to room temperature. In the first reaction attempt, the right product was not

obtained. The 1H NMR peaks are published in literature and contain a double doublet at 7,40

and two doublets at 7,28 and one at 7,27. These reported values did not correspond to those

of our reaction product (Figure 22).

32

Figure 22: Crude 1H NMR spectrum of stannylation with literature conditions

The proposed compound that was formed is 2-[(triisopropylsilyl)ethynyl]pyridine 9 (Figure

22). With LCMS, the [M+H]+ was found at m/z 260 and also 1H NMR spectroscopy confirmed

the formation of 9 (Figure 22) as the proton chemical shifts of 2-

[(triisopropylsilyl)ethynyl]pyridine 9 correspond to those reported in literature.62 Initially,

impurities of compound 3 were blamed for failure of the reaction. However, further

purification of product 3 did not help. For a new reaction, a new dry bottle of Bu3SnCl was

used because there were questions about the purity of the old SnBu3Cl (SI 13) and the

solutions were prepared in the glove box. The same reaction conditions were used and the

obtained product was again compound 9.

To investigate what exactly went wrong with the reaction, it was tried to substitute the

bromine atom by something else instead of Bu3SnCl. A deuterium atom was used as analysis

of a deuterium substitution is easily seen by 1H NMR spectroscopy and LCMS. As such, it could

be checked if it was possible to do a simple substitution reaction on the bromine atom of 2-

bromo-6-[(triisopropylsilyl)ethynyl]pyridine 3 after lithiation. For this deuteration reaction,

the same procedure was used as before yielding 2-[(triisopropylsilyl)ethynyl]-6-

(deuterio)pyridine 10 (Figure 23). This structure was confirmed with 1H NMR spectroscopy

(Figure 23) and with LCMS that revealed the [M+H]+ at m/z 261. Based on this model reaction,

33

it could thus be concluded that the bromine/lithium exchange by n-BuLi is not the problem,

but the attack of the lithium intermediate on Bu3Sncl is.

Figure 23: Crude 1H NMR spectrum of deuterated product 10

The reaction of methanol-d4 (MeOD) with the lithium salt intermediate of compound 3 is

thought to be fast, giving rise to such a high amount of compound 10. In the stannylation

reaction, a hydrogen comes into the reaction mixture after successfully having added the n-

BuLi. A possible hydrogen source could be the degradation of THF. THF can namely be

deprotonated by strong organolithium bases.63,64,65

Other reaction conditions were searched to avoid this side reaction. In more simple

stannylation reactions on 2-bromopyridine with n-BuLi and tributyltinchloride, the reaction

has been reported to proceed at -78°C instead of ambient temperature.66,67 Furthermore, the

reaction mixture is often kept at-78°C before the tributyltinchloride is added, instead of first

allowing it to heat till -20°C. Using this optimized procedure, the desired 2-bromo-6-

[(triisopropylsilyl)ethynyl]pyridine 4 could successfully be obtained, which was confirmed by

1H NMR spectroscopy, 13C APT NMR spectroscopy and LCMS. Two major products are formed

using these new reaction conditions, namely compound 4 and 9. In Figure 21, the crude 13C

APT NMR is shown, the proton chemical shifts of both compounds 4 and 9 were also confirmed

(Figures SI 3 and SI 4). These 13C chemical shifts were reported in literature and were also

calculated and confirmed with ChemBioDraw 13 for redundancy.

34

Figure 24: 13C APT NMR spectrum of Crude reaction mixture with new conditions

Purification of the stannylated compound 4 was done with column chromatography on

aluminum oxide with n-pentane as eluent. The 2-[(triisopropylsilyl)ethynyl]-6-

(tributylstannyl)pyridine was obtained with a yield of 26 %. The reported yield is 45 %.60

2.2.3 Stille Coupling Reactions

Figure 25 : Synthesis of 4,6-bis[4’-((triisopropylsilyl)ethynyl) pyrid-2’-yl]pyrimidine 6

For the synthesis of 4,6-bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]pyrimidine 6, 2-[(tri-

isopropylsilyl)ethynyl]-6-(tributylstannyl)pyridine 4 should be coupled to 4,6-diiodopyrimidine

5. This can be done via a Stille coupling with the addition of a catalyst. This exact reaction is

not reported in literature. In literature, compound 4 is used together with 3,8-dibromo-4,7-

35

phenanthroline in a Stille coupling with toluene as a solvent.60 4-Iodopyrimidine has also been

used in a Stille coupling with 3-bromopyridine in toluene under reflux.68 Tetrakis was used as

Pd catalyst in both instances. The coupling of compound 4 and 6 was attempted in toluene

under reflux and tetrakis as catalyst as similar reactions had been carried out under those

conditions. The mixture was stirred overnight and a black precipitation was formed. Analysis

was done by 1H NMR spectroscopy and LCMS. With LCMS, a very broad band on the baseline

was visible. Because compound 6 is very apolar, it could be that this compound sticks to the

column, making LCMS hard for the identification. To be completely sure that the desired

compound was not present in the reaction mixture, a column on the crude mixture was

performed. The fractions were analyzed via 1H NMR spectroscopy confirming that compound

6 was not formed.

Stille coupling reactions sometimes need reaction times of several days 68,69,70,71. The

possibility of a too short reaction time arose, because a lot of the stannylated compound 4

was still present in the crude reaction mixture. Therefore a new reaction was done with

toluene under reflux, but now it was allowed to react for eight days instead of one. While the

reaction was followed by 1H NMR, it was clear that during the reaction, some peaks changed

in intensity relative to each other. In Figure 23 the change of the relevant peaks is shown. As

can be seen, after 48 hours, no peak was present at 9.2, while after 96 hours a peak arose at

that particular shift. In the sample after 140 hours, this peak was even bigger. After eight days,

the reaction was stopped. Again, the reaction mixture was black and had a black precipitate.

After evaporating the solvent, column chromatography was performed on silica which was

first treated with NEt3. The eluent was DCM with methanol (MeOH) going from 0 till 10% of

MeOH with NEt3 added in the end. The different fractions were analyzed by 1H NMR

spectroscopy and the fraction that looked most promising was also controlled by HSQC but

the desired product could not be identified indicating that compound 6 could not be obtained

using these conditions. Unreacted 2-[(triisopropylsilyl)ethynyl]-6-(tributylstannyl)pyridine 4

was obtained and besides compound 4, also a lot of 2-[(triisopropylsilyl)ethynyl]pyridine 9 was

formed. This compound 9 is a thermal degradation product of 2-[(triisopropylsilyl)ethynyl]-6-

(tributylstannyl)pyridine 4, which due to the long reaction times can be formed when not

being able to couple to another species present in the reaction mixture.

36

Figure 26: The 1H NMR spectra after 48, 96 and 140 hours. The peak at 9.32 changes relative to the peak

at 9,40.

Compound 5, which was not stored in a fridge, did not look completely clean in the 1H NMR

spectrum (SI 1). Also the LCMS analysis looked rather suspicious, as a lot of impurities were

found in what should have been a clean compound (SI 11 and SI 12). With HSQC, it was

determined that it nevertheless was mainly the right product. Therefore, the Stille coupling

was attempted with another compound, namely 4,6-dichloro-2-phenylpyrimidine 5A, also

called fenclorim, which has two chlorine atoms instead of the iodine atoms of 4,6-

diiodopyrimidine 5 (Figure 27). Fenclorim also has a phenyl ring instead of a hydrogen atom

on the carbon between the two nitrogen atoms. In prospect of the grid forming ability when

using compound 6A instead of 6, this phenyl group is rather interesting. Like the terpyridine

analogue of ligand 8, there also exists a terpyridine analogue with a phenyl group on that place

instead of a hydrogen. In the corresponding [2x2] grid structure, this phenyl group provides

π-π interactions that stabilize the supramolecular grid. Because this fenclorim 5A was stored

in a fridge and was more clean, together with the added value of having a phenyl group in the

middle, the next Stille coupling was carried out with fenclorim 5A instead of 4,6-

diiodopyrimidine 5.

37

Instead of toluene, DMF was used, which is a more standard solvent for carrying out Stille

couplings compared to toluene72,73,74. Tetrakis and PdCl2(PPh3)2 were both used as Pd catalyst

and the new reaction scheme is displayed in Figure 27.

Figure 27: Synthesis of 4,6-bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 6A

ESI-MS analysis of the crude pointed out that 4,6-bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]-

2-phenylpyrimidine 6A was present in both the crudes at m/z 671,4. Because the spectra

looked more or less the same and both reactions were done on small scale, workup was done

on the combined product. First, a column on silica with n-pentane/DCM (DCM in a gradient

from 0 to 20 percent with a couple drops of NEt3 was carried out. The fraction containing

compound 6A was further purified by recrystallization with ethanol (EtOH) as it is a non-

solvent at room temperature and a solvent under reflux. 4,6-bis[4’-

((triisopropylsilyl)ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 6A was obtained with a yield of 57%

(Figure 28).

The Stille coupling with 4,6-diiodopyrimidine 5 was not carried out in DMF due to time

constraint. Further research is needed to find if the conditions for synthesizing compound 6A

also work with chemical 4,6-diiodopyrimidine 5.

38

Figure 28: 1H NMR spectrum of 4,6-bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 6A

2.2.4 Synthesis of 4,6-Bis[4’-(ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 7A

Figure 29: Deprotection of 6A to get to the protoligand 4,6-bis[4’-(ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 7A

The deprotection of 4,6-bis[4’-((triisopropylsilyl)ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 6A

was done with TBAF. In literature, 1.3 to 2.5 equivalents of TBAF to one TIPS acetylene group

are used.60,75 Here, 1.8 equivalents of TBAF for one TIPS group was used and the reaction was

followed by TLC. After one hour, the mixture didn’t change anymore and after one and a half

hour, the reaction was stopped. Recrystallization was tried in EtOH since compound 7A was

suspected to be a crystalline compound. No crystallization occurred so another method was

searched. Washing the crude with water, diethylether (Et2O) and n-pentane was another

39

literature inspired 60 workup which also did not prove to be successful. Likewise a simple

extraction with water and chloroform was non satisfying as well. Lastly, a column on silica with

n-pentane/DCM/MeOH in a gradient starting from 85/15/0 to 0/90/10 with 0.5 % of NEt3 was

performed. After evaporation, protonated NEt3 was in it, which was removed by resolving the

product in CHCl3 and washing with water. The right structure was proved by 1H NMR

spectroscopy (Figure 30) and confirmed by ESI-MS that revealed the [M+H]+ at m/z 359.2. The

protoligand 7A was obtained with a yield of 97%. It still contained impurities in the lower ppm

area.

Figure 30: 1H NMR spectrum of 4,6-bis[4’-(ethynyl)pyrid-2’-yl]-2-phenylpyrimidine 7A

40

2.3 Polymer and Oligomer Synthesis

The aim of synthesizing these oligomers and polymers is to attach them onto alkyne

functionalized ligand 7 via a cycloaddition. The polymers and oligomers should have an azide

functionality as end-group. Synthesis of 2-(2-(2-methoxyethoxy)ethoxy)ethyl azide 13 is

discussed in section 2.3.1 and the poly-α-methyl-ω-azido-(2-ethyl-2-oxazoline) (PEtOx-N3)

synthesis in section 2.3.2.

2.3.1 Synthesis of 2-(2-(2-Methoxyethoxy)ethoxy)ethyl Azide 13

Figure 31: Synthesis of azide functionalized monomethyl triethylene glycol

2-(2-(2-Methoxyethoxy)ethoxy)ethyl azide 13 was synthesized starting from triethylene glycol

monomethyl ether (TEG) 11 (Figure 31). Firstly, the TEG 11 was tosylated by TsCl. The used

conditions came from literature76 and the reaction was carried out in DCM. NEt3 was added in

excess and 4-dimethylaminopyridine (DMAP) was added substoichiometric as catalyst. After

workup, which involved extraction with water, purification was performed by column

chromatography. The tosylated monomethyl triethylene glycol 12 was obtained with a yield

of 70%. Analysis was done by 1H NMR spectroscopy and LCMS. The reported 1H NMR peaks

were visible and no impurities were found (SI 9). With LCMS the [M+NH4]+ peak was observed

instead of the [M+H]+ peak. Small ethylene glycol chains can act as supramolecular podands,

making them able to coordinate cations,9 such as ammonium cations. In an environment with

such a high concentration of ammonium compared to the concentration of compound 12 as

ammonium is present in the eluent, observing only the [M+NH4]+ peak is accountable.

Upon tosylation of the triethylene glycol monomethyl ether 12, the primary alcohol group of

compound 11 was transferred in a better leaving group. In the next step, a nucleophilic

substitution reaction was carried out. This reaction was also reported in literature before and

was performed with 3.5 equivalents of NaN3 in DMF at 60°C.77 After extraction with Et2O,

triethylene glycol monomethyl ether azide 13 was obtained with a yield of 84%. The expected

1H NMR shifts were found (SI 9) and the compound was confirmed with ESI-MS with the

[M+H]+ at m/z 190.2 Da and the [M+NH4]+ at 207.2 Da.

41

2.3.2 Synthesis of Poly-α-methyl-ω-azido-(2-ethyl-2-oxazoline)

Figure 32: Synthesis of PEtOx-N3 by CROP and termination by sodium azide

PEtOx-N3 was made by polymerization of 2-ethyl-2-oxazoline with MeOTs as initiator. As the

CROP of 2-oxazolines is a living polymerization that is very sensitive to nucleophiles it requires

working under dry conditions for which the reaction was carried out in a glovebox. After

polymerization, sodium azide is added as terminating agent to introduce the azide end-

group.50,53,54 Three different lengths of PEtOx-N3 were made, namely with 20, 50 and 100

repeating units. The standard optimized conditions of our group were used, being 0.05

equivalents of MeOTs and 0.15 equivalents of NaN3 with regard to 2-ethyl-2-oxazoline in ACN

at 100°C for PEtOx20-N3. For PEtOx50-N3 these equivalences where 0.02 MeOTs and 0.06 NaN3,

for PEtOx100-N3 0.005 MeOTs and 0.03 NaN3 for. After polymerization, the two times molar

excess of NaN3 had to be removed from the crude. First, an extraction with EtAc and water

was used as workup. The resulting solid was dissolved in EtAc and extracted with saturated

NaHCO3 and saturated NaCl solutions for two times. The amount of the PEtOx-