1 The amorphous and glassy states 3

(1)

(2)

ii

(3)

Preface

The main purpose of this thesis work is the investigation of the role of the glassy state on the Dynamic Nuclear Polarization (DNP) process, paying particular attention to the ap- plication in the diagnostic field (Magnetic Resonance Imaging, MRI). So far, experimental evidences have shown that the crucial requests for a sample containing paramagnetic im- purities to be polarized by DNP are its amorphous state and the homogeneity of its glassy mixture and dispersion. On the other hand, a consistent theoretical interpretation for this phenomenon is missing, as well as a deep analysis on the “goodness” of the glassy state obtained after the amorphization process, intended for DNP-MRI application.

An alternative preparation procedure of contrast agents containing radical molecules, hyperpolarizable by DNP, is proposed in this thesis. The novelty of these glassy samples is that they are solid at room temperature. Under these circumstances, a methodol- ogy of characterization of the amorphous solids in all the fundamental aspects (thermal, spectroscopic, structural and magnetic properties) is suggested to investigate the correla- tion between the glassy state and the hyperpolarization features (such as the maximum achievable value of polarization, P%). In particular, in this way it is possible to study one of the main problems in this topic, that is the effect of the presence of nano- and micro-crystalline domains on the homogeneity of the radical distribution and, then, on the efficiency of the magnetic polarization transfer. Furthermore, in order to optimize the DNP efficiency, another crucial issue is the role of the radical concentration on the polarization transfer and whether high concentration could lead to either quenching effect or to radical aggregation.

For this purpose, several amorphization procedures of solids have been analyzed. This study shows that co-milling is the best procedure, that provides riproducibility, prevents degradation and allows a good control of the physical features of the glass and of the crystalline phase. A milled mixture of trehalose and TEMPO molecules has been chosen

vii

(8)

viii PREFACE as model system, because of the high stability of trehalose and high solubility of the TEMPO radical.

Chapters 1 and 2 briefly report the state of art regarding both the glassy state and

the preparation of amorphous samples, addressing to the issue relevant for the hyperpo-

larization by DNP in chapter 2. Chapter 3 presents a discussion about the choice of the

optimal combination of amorphization technique and model system. The characteriza-

tion of the model system perfomed by Differential Scanning Calorimetry (DSC), Electron

Paramagnetic Resonance (EPR), Solid State Nuclear Magnetic Resonance (SSNMR) and

Raman spectroscopies and X-Ray Diffraction (XRD) is described in chapter 4. The effects

of both the concentration and the amorphization degree on the physical properties of the

samples have been highlighted. Chapter 5 reports results of DNP measurements on the

model system. The effect of radical concentration on the polarization transfer has been

stressed for fully amorphized samples (12 h of milling), paying attention to the physi-

cal stability of these amorphous solids. In addiction, some alternative substrates used

in DNP-MRI have been tested for comparison. In the final part of this work, chapter 6

describes an ancillary study on the dehydration of solutions, carried out by means of a

novel calorimetric approach to investigate the role of water (possibly absorbed from the

environment) on the stability of the amorphous solids.

(9)

1 f

(10)

2 PREFACE

(11)

Chapter 1 The amorphous and glassy states

The main purpose of this chapter is to introduce the concepts of amorphous and glassy states, giving some general knowlegde about the physical properties of these systems (e.g.

viscosity, fragility) and the fundamental thermodynamic paramenters (e.g. configura- tional enthalpy) necessary to describe the relaxation process typical of a non-equilibrium condition of amorphous states.

1.1 Introduction

The glassy state is a solid but disordered metastable state characterized by some mechani- cal properties similar to those of crystals and configurational properties similar to those of liquids [1–3]. The combination of a crystal-like rigidity and a liquid-like structure makes the glass a special system with features that the crystalline solid, its counterpart, does not possess. The glasses are also called amorphous solids, the term amorphous having more general meaning, including liquids, rubbers, and substantially every system lacking the periodicity of the crystalline lattice. For this reason, glasses find applications in many fields such as food industry, where the amorphous state is crucial for the processing and storage steps [4]; in the commercial stabilization of labile biochemicals [5]; for the pro- duction of photovoltaic cells made of amorphous semiconductors (e.g. silicon) or optical fibers composed of pure (and occasionally doped) amorphous silica; for the manufacture of engineering metallic glasses and alloys with interesting mechanical (corrosion resistance) and magnetic (soft magnetism) properties [6]; for production of well known window glass (made of sand, lime and soda, or polymeric glasses). The glassy state is fundamental in

3

(12)

4 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES nature as well as in many bio-technological applications [7]. Indeed most of the water in the universe seems to be in the glassy state [8], many relevant biomolecules are also mixed with water in a glassy state. It has to be noticed the importance of the glassy state in bioprotection, e.g. some microorganisms (tardigrades) take advantage of amorphous trehalose to resist under extreme conditions of dehydration and low temperature [5, 9, 10].

The simplest and most common method to prepare a glass is a fast cooling of the melt well below the freezing point T

m

; if the cooling rate is sufficiently high, the molecular motions slow down avoiding crystallization [11, 12]. Eventually, the molecules will rear- range so slowly that they cannot order themselves in a crystalline configuration during the time of the cooling process and the “glass” results frozen in a liquid structure (for the timescales of the laboratory). During the cooling process, the system passes through a

Figure 1.1: Temperature dependence on the liquid volume V or enthalpy H at constant pressure (scheme from ref. [3]).

narrow transformation range (named glass transition) characterized, from a microscopic

point of view, by molecular relaxation times much higher if compared with those of the

liquid state (conventionally of the order of 100 s); moreover, from a macroscopic point of

view, physical, mechanical, electric, thermal and other properties change rapidly within

this temperature range. For example, volume and enthalpy change rates decrease abruptly

(13)

1.1. INTRODUCTION 5 but continuously to get a value similar to that of the crystal, where the derivatives of these two parameters give

α

_p

= ∂lnV

∂T

p

C

_p

= ∂H

∂T

p

, (1.1)

where α

p

and C

p

are the thermal expansion coefficient and the isobaric heat capacity, respectively. Thus, figure 1.1 provides the first definition of T

g

, as the cross point of liquid and glass volume or enthalpy curves, and its value usually occurs around

²₃

T

_m

. In particular, the trend of the volume or the enthalpy as a function of temperature (at constant pressure) is schematically represented in figure 1.1: (a) a vitreous state produced by a slow cooling rate from the liquid state, characterized by a glass transition at T

ga

; (b) a glass produced from the state initial condition by a faster cooling rate with a glass transition temperature T

gb

> T

ga

. Another definition of T

g

comes from the temperature

Figure 1.2: Time-temperature-transformation (TTT) scheme showing possible states re- sulting by cooling from a liquid state. The curves at x = 0 and x = 1 indicate the beginning and the end of the crystallin process, respectively. Note that x = 1 does not necessary mean a complete (100%) crystallization, in particular for polymers.

where the mechanical properties of the amorphous system become comparable to those of the crystalline state. Figure 1.2 shows three possible results of cooling from the melt corresponding to different cooling rates: in particular, in the cases of fast cooling (A and B) the final state is vitreous with different properties, in the case C the molecules have sufficient time to rearrange to give a crystalline state.

The glassy state is not univocally determined, its physical properties depending on

(14)

6 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES the amorphization techniques used to prepare it (as detailed described in Section 1.4) and on the experimental conditions chosen for the amorphization: in the case of cooling of liquids (described above), the determining factor is the cooling rate. The value of the glass transition temperature T

g

can be taken as a parameter broadly indicating the stability of the amorphous solid.

1.2 The enthalpic relaxation

Although the glass transition temperature has been proven to be an indicator for the stability of the glassy matrix, some physico-chemical changes take place below T

g

, because of the out-of-equilibrium nature of the glassy state. Indeed, when a glassy material is stored slightly below its T

g

, it spontaneously approches a more stable state, then many physical properties of the glass tend to the equilibrium values at a constant temperature with a characteristic time. For example, considering H = H

c

+ H

_v

as the total enthalpy of the system, given by the sum of the configurational and vibrational contributions, H

c

and H

v

respectively, follow the Kohlrausch-Williams-Watts (KWW) equation [13]:

∆H

_c

(t) = ∆H

_c

(0) exp −(t/τ

c

)

^β^c

, (1.2)

∆H

_v

(t) = ∆H

_v

(0) exp −(t/τ

_v

)

^β^v

, (1.3) where τ is the average characteristic time of relaxation. The parameter β assumes values between 0 (∆H

c

or ∆H

v

constant) and 1 (simple exponential behaviour of ∆H

c

or ∆H

v

);

small values of β correspond to a large distribution of relaxation times and viceversa. The second contribution to the enthalpy H

v

responds very quickly to the temperature changes.

Under adiabatic conditions ∆H

v

= ∆H

_c

: d∆H

_c

(t)

dt = − dH

_v

(t) dt

dT

dt = −C

_v

dT

dt , (1.4)

where T is the vibrational temperature, which tends to equalize T

g

.

The phenomenon described above is called enthalpic relaxation, it is due to the molec-

ular motion of certain molecules or parts of some molecules (in the case of polymers),

and it is important for many materials stored just below the glass transition temperature

for the decreased stability of the physico-chemical properties of materials. The enthalpic

(15)

1.2. THE ENTHALPIC RELAXATION 7 relaxation can be considered as a spontaneous variation of temperature in adiabatic con- ditions, characterized by very variable relaxation times depending on the system which can be followed by calorimetry:

T = a + bt + c exp −(t/τ )

^β

. (1.5)

From the practical point of view, the viscosity of the liquid is a measure of the struc- tural relaxation time for the undercooled fluid. From the Maxwell theory of viscoelasticity, the relation between the viscosity η and the relaxation time τ is described by the equa- tion [14]:

τ = G

⁻¹_∞

η, (1.6)

where G

∞

is the istantaneous shear modulus.

As observed for other mechanical properties, an abrupt change of the viscosity η occurs during the glass transition: η reaches the tipical value of the crystalline state (about 10

¹²

Pa · s) [1]. Then, a new definition of T

g

can be done as the temperature where η reaches the value tipic of the solid state during the process of cooling from the liquid state.

Moreover, the viscosity η becomes a convenient kinetic parameter to describe the glass transition. Its behaviour as a function of temperature has been widely discussed in the literature: in the case of some network-forming glasses (e.g. ordinary silicate glasses) the viscosity shows an Arrhenius behaviour:

η = η

₀

exp(A/T ), (1.7)

while many molecular liquids deviate from the Arrhenius law; in the latter case η can be described by the Vogel-Tammann-Fulcher (VTF) equation:

η = η

₀

exp[A/(T − T

₀

)], (1.8)

where the constants η

0

, A and T

0

depend on the nature of the liquid.

Still referring of non-Arrenius systems, the Adam and Gibbs’s model describes the dependence of the τ parameter from the temperature:

τ = A exp z

^∗

∆µ kT

= A exp N

_A

∆µs

^∗_c

kT S

_c

(T )

, (1.9)

(16)

8 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES where z

^∗

is the number of molecules of a group or cluster that rearranges cooperatively and depends on the temperature, ∆µ is the chemical potential per molecule of that group/cluster, N

A

is the Avogadro constant, s

^∗c

is the configurational entropy of the small- est group “that can undergo the rearrangement”.

The deviation of τ from the Arrhenius behaviour is correlated phenomenologically with the so called fragility parameter m [15]:

m = d log τ d(T /T

_g

)

T =Tg

. (1.10)

For the systems that follow the Arrhenius behaviour, m assumes the value of 17 and the liquid is called strong: the structure does not change much with the temperature and the C

p

change at the glass transition is rather small. For higher values of m the liquid is called fragile because its structure is much more sensible to temperature changes and the ∆C

p

at T

g

is larger. Some examples of fragile and strong glasses are reported, in a Tg-scales

Figure 1.3: Strong and fragile behaviour in a Tg-scaled Arrhenius representation of liquid viscosities (scheme from ref. [3]).

Arrhenius representation of liquid viscosity, in Figure 1.3. Strong liquids are characterized

by linearity (Arrhenius behaviour), indicative of a temperature-independent activation

energy associated with structural relaxation: E =

_{d(1/T )}^dlnη

≈ constant. On the other hand,

fragile liquids show a super-Arrhenius behaviour and E increases as temperature decreases.

(17)

1.3. GLASS TRANSITION TEMPERATURE OF MIXTURES 9 A more detailed description of the macro- and micro-definitions of fragility is argumented by Blazhnov et al. [16]

1.3 Glass transition temperature of mixtures

At this point, a brief description of the properties of glassy mixtures is needed. [2, 17] The glass transition of a system consisting of two or more components assumes an intermediate value between the T

g

values of the individual constituents, which is dependent on the composition and nature of the components. In particular, the glass transition temperature is strongly dependent on the molecular weight; components that decrease the average molecular weight of a sugar mixture generally decrease the glass transition temperature of the mixture.

Mathematical models are able to predict the glass transition temperature of multi- component mixtures. For example, the Gordon-Taylor equation,

T

g

(x) = KT

_g2

+ x · (T

_g1

+ KT

_g2

)

k + x · (1 − K) , (1.11)

was originally developed for a binary mixture of polymer blends and was based on expan- sion coefficients (β =

^dV_dT

) and the assumption of ideal volume mixing [18, 19], and was subsequently rewritten in the following form:

T

_g

= x

₁

T

_g1

+ Kx

₂

T

_g2

x

₁

+ Kx

₂

. (1.12)

T

g

, T

g1

and T

g2

are the glass transition temperatures of the binary mixture, component 1, component 2, respectively; for aqueous systems, subscript 1 is the solid component (or a dry mixture of various solid components), subscript 2 is water. x

1

and x

2

= x are the mole fractions (or weight fractions). K is a parameter characteristic of the nature of the two components and can be obtained as aritmentic average of the K values obtained by solving the equation 1.11 or 1.12 for each binary (component 1 : component 2) system.

Figure 1.4 shows a schematic representation of the Gordon-Taylor function; in particular, the difference in the trend due to different values of the parameter K (K < 1, K = 1 and K > 1 ) is enhanced.

The assumption of the Gordon-Taylor equation is an ideal volume mixing (no interac-

(18)

10 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES

0.00 0.25 0.50 0.75 1.00

T g

2 T

g 1

k = 1 k < 1

T g

x k > 1

Figure 1.4: Examples of Gordon-Taylor function for binary mixtures characterized by different K values.

tions between the two components), which assumes that the two components are miscible and their free volumes are additive. From the free volume theory, K is the ratio of the free volumes of the two components and can be calculated using the Simha-Boyler rule [18, 20]:

K ≈ T

_g1

ρ

₁

T

_g2

ρ

₂

, (1.13)

where ρ

1

and ρ

2

are the densities of components 1 and 2, respectively. Another method to estimate the parameter K is to evaluate the ratio between the ∆C

p

values of the two components

K = ∆C

_p2

∆C

p1

, (1.14)

where ∆C

p

is the change in heat capacity between the liquid-like and the glassy states.

This K value is developed on the basis of the classical thermodynamics, with an assump- tion that the system is purely conformational (the entropy of mixing in an amorphous mixture is purely combinatorial). Under this assumption, the Gordon-Taylor equation can be rewritten as the Couchman-Karasz equation:

T

_g

= ∆C

p1

x

1

T

g1

+ ∆C

p2

x

2

T

g2

∆C

_p1

x

₁

+ ∆C

_p2

x

₂

. (1.15)

For non-ideal mixing systems in which the interactions are “ not too strong”, the interac-

(19)

1.4. AMORPHIZATION METHODS 11 tion factor can be defined:

= K · ∆C

_p1

∆C

p2

(1.16)

as an indicator of the non-ideal mixing behaviour of binary mixtures (if = 1 we come back to the ideal system case).

Moreover, Truong et al [17] derived the Couchman-Karasz equation and applied it not only to binary systems but also to ternary, quaternary, and higher order (ideal mixing) systems:

T

_g

= x

₁

∆C

_p1

T

_g1

+ x

₂

∆C

_p2

T

_g2

+ x

₃

∆C

_p3

T

_g3

+ · · ·

x

₁

∆C

_p1

+ x

₂

∆C

_p2

+ x

₃

∆C

_p3

+ · · · . (1.17)

1.4 Amorphization methods

The transformation from an originally well crystallized material to an amorphous phase can be induced without passing through the conventional steps of melting and quenching of the liquid. There are many different methods to prepare amorphous solids, classified as “chemical” and “physical” if the composition of the substances changes or not during the amorphization process. Such a process of amorphization can result from a variety of processes: chemical, irradiative, thermal or pressure-induced disruption of the crystalline order, when the free energy of the crystal obviously exceeds that of an allied amorphous phase [21–23]. Each method is involving a lot of energy in different micro/local pro- cesses, for example powdered particles could melt in their neighbourhood due to high local temperatures achieved due to the plastic deformations.

Among the “physical” methods there are:

- liquid-cooling. It is the most familiar method of amorphization (discussed above);

it can be used on systems already liquid at room temperature or, alternatively, the liquid phase can be produced by melting at higher temperatures. Moreover, a hyper-quenched method, with a cooling rate of nearly 10

⁶

K/s, can be applied to water or metallic liquids that have strong tendency to crystallize;

- deposition of vapour onto a cold substrate: this method is one of the most powerful

methods in forming amorphous state. The kinetic energies of vapour molecules are

efficiently extracted during the condensation on to a substrate kept at a temperature

(20)

12 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES far below the hypothetical T

g

of the resulting amorphous matrix, so the deposited molecules are immobilized in a frozen disordered state;

- bombardment of crystals with high energy particles. High energy as form of radiation is supplied to the crystal to destroy the ordered arrangements of the costituents;

- compression of crystals. In this case, high energy supplied to the crystal has the form of compression depression. This kind of amorphization was firstly discovered already 25 years ago observing water ice crystals to amorphize at 77 K and 1 GPa [24];

- cold-rolling of crystals. In this case the energy has the form of a shear stress, this rout is commonly used with metals [23, 25–27]) and applicable to an inquisitive assortment of materials;

- mechanical milling of crystalline solids. By continuous grinding and milling the crystalline grains are miniaturized below perceptible nano-crystallinity;

Figure 1.5: Several kind of amorphization methods [1].

Among the “chemical” methods some examples are:

- gelation. The sample in a solution is brought into a gel state and then the cor-

responding amorphous solid is formed by removing of the extra components. For

example gelation is used to produce oxides in form of fibers or lumps;

(21)

1.4. AMORPHIZATION METHODS 13 - precipitation by chemical reactions: this method exploits the peculiarity of some chemical reactions in solution to produce precipitates that mantein the disordered conformation characteristic of the liquid state;

- dehydration of hydrate crystal. Water molecules are important ingredients in hy- drate crystals; if these molecules are extracted by rapid evacuation, the resulting anhydride cannot keep anymore the crystalline lattice in some hydrate crystals and collapse in an amorphous solid form. This method is largely used in pharmaceuti- cals [28];

- freeze-drying and spray drying. In these methods the dehydration from a solution is used by two different processes: vaporization of the solvent at high temperature and condensation of the solute component in an amorphous solid phase for the spray- drying, and sublimation of the solvent in condition of low temperature and pressure for the freeze-drying. These two methods are commonly used in pharmaceutics.

The various chemical and physical methods to produce amorphous solids are schemat- ically presented in Figure 1.5. In many cases (e.g. tri-O-methyl-β-cyclodextrin [1]) the C

p

variation at the glass transition is the same for both the sample prepared by fast quench-

ing from the melt and the sample prepared by other methods (in this case, mechanical

milling). For that reason we can say that glass and amorphous solid are quite similar

concepts, in particular, glasses are just an example of the amorphous solids formed from

the liquid states, with the disordered structure of liquids (Detailed differences between

amorphous and glass are reported in the work of Secrist and Mackenzie. [29]).

(22)

14 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES

References

[1] H. Suga. “Introduction: Some essential attributes of glassiness regarding the nature of non-crystalline solids”. In: Glassy, amorphous and nano-crystalline materials. Ed.

by J. Šesták, J. J. Mareš, and P. Hubík. Vol. 8. Hot topics in thermal analysis and calorimetry. Springer, 2011, pp. 1–19.

[2] Y. Liu, B. Bhandari, and W. Zhou. “Glass transition and enthalpy relaxation of amorphous food saccharides: a review”. In: J. Agric. Food Chem. 54 (2006), pp. 5701–

5717.

[3] P. G. Debenedetti and F. H. Stillinger. “Supercooled liquids and the glass transition”.

In: Nature 410 (2001), pp. 259–267.

[4] J. M. V. Blanshard and P. Lillford, eds. The Glassy State in Foods. Nottingham:

Nottingham Univ. Press, 1993.

[5] J. H. Crowe, J. F. Carpenter, and L. M. Crowe. “The role of vitrification in anhy- drobiosis”. In: Annu. Rev. Physiol. 60 (1998), pp. 73–103.

[6] A. L. Greer. “Metallic glasses”. In: Science 267 (1995), pp. 1947–1953.

[7] C. A. Angell. “Formation of glasses from liquids and biopolymers”. In: Science 267 (1995), pp. 1924–1935.

[8] P. Jenniskens and D. F. Blake. “Structural transitions in amorphous water ice and astrophysical implications”. In: Science 265 (1994), pp. 753–756.

[9] K. B. Storey and J. M. Storey. “Natural freeze tolerance in ectothermic vertebrates”.

In: Annu. Rev. Physiol. 54 (1992), pp. 619–637.

[10] A. Cesàro. “Carbohydrates: All dried up”. In: Nat. Mater. 5 (2006), pp. 593–594.

[11] D. Turnbull. “Under what conditions can a glass be formed?” In: Contemp. Phys.

10 (1069), pp. 473–488.

[12] C. A. Angell. “Structural instability and relaxation in liquid and glassy phases near the fragile liquid limit”. In: J. Non-Cryst. Solids 102 (1988), pp. 205–221.

[13] S. Brawser. “Relaxation in viscous liquids and glasses”. In: Review of phenomenol-

ogy, molecular dynamics simulations, and theoretical treatment . Ed. by American

Ceramic Society. Columbus, 1985.

(23)

REFERENCES 15 [14] W. Kauzmann. “The nature of the glassy states and the behavior of liquids at low

temperature”. In: Chem. Rev. 43 (1948), pp. 219–287.

[15] C. A. Angell. “Relaxation in liquids, polymers and plastic crystals”. In: J. Non-Cryst.

Solids 13 (1991), pp. 131–133.

[16] I. Blazhnov et al. “Macro- and microdefinitions of fragility of hydrogen-bonded glass-forming liquids”. In: Physical Review E 73 (2006), pp. 031201–600.

[17] V. Truong et al. “Analytical model for the prediction of glass transition temperature of food systems”. In: Amorphous food and pharmaceutical system. Ed. by H. Levine.

Cambridge: Royal society of chemistry, 2002, pp. 31–47.

[18] M. Gordon and J. S. Taylor. “Ideal co-polymers and the second-order transitions of synthetic rubbers. 1. Non-crystalline copolymers”. In: J. Appl. Chem. 2 (1952), pp. 493–500.

[19] M. Shamblin, J. S. Taylor, and G. Zografi. “Mixing behavior of colyophilized binary system”. In: J. Pharm. Sci. 87 (1998), pp. 694–701.

[20] R. Simha and R. F. Boyer. “On a general relation involving the glass temperature and coefficients of expansion of polymers”. In: J. Chem. Phys. 37 (1962), pp. 1003–

1007.

[21] G. N. Greaves and S. Sen. “Inorganic glasses, glass-forming liquids and amorphizing solids”. In: Adv. Phys. 56 (2007), pp. 1–166.

[22] G. N. Greaves et al. “The rheology of collapsing zeolites amorphized by temperature and pressure”. In: Nat. Mater. 2 (2003), pp. 622–629.

[23] H. Sieber, G. Wilde, and J. H. Perepezko. “Solid state amorphization by cold- rolling”. In: Materials development and processing - bulk amorphous materials, un- dercooling and powder metallurgy . Ed. by J.V. Wood, L. Schultz, and D. M. Herlach.

Vol. 8. Euromat. Wiley-VCH, 2000, pp. 3–9.

[24] W. Zheng, D. Jewitt, and R. I. Kaiser. “Amorphization of crystalline water ice”. In:

Astrophys. J. (2008). url: http://arxiv.org.

[25] J. H. Perepezko et al. “Amorphization and devetrification reactions in metallic glassy

alloys”. In: Mat. Sci. Eng. A 449/451 (2007), pp. 84–89.

(24)

16 CHAPTER 1. THE AMORPHOUS AND GLASSY STATES [26] C. Suryanarayana. “Solid-state amorphization”. In: Mechanical alloy and milling.

Ed. by C. Suryanarayana. New York: Marcel Dekker, 2004, pp. 269–372.

[27] T. Nagase, T. Hosokawa, and Y. Umakoshi. “Solid state amorphization and crys- tallization in Zr66.7Pd33.3 metallic glass”. In: Intermetallics 14 (2006), pp. 1027–

1032.

[28] J. F. Willart and M. Descamps. “Solid state amorphization of pharmaceuticals”. In:

Mol. Pharmaceutics 5 (2008), pp. 905–920.

[29] D. R. Secrist and J. D. Mackenzie. In: Modern aspects of the vitreous state. Ed. by

J. D. Mackenzie. Vol. 3. London: Butterworths Scientific Publication, 1964. Chap. 6.

(25)

Chapter 2 Dynamic Nuclear Polarization

2.1 Thermal equilibrium polarization

The principle of Nuclear Magnetic Resonance (NMR) is based on the interaction of a non-zero nuclear spin (quantum number I) with an external magnetic field. Many atomic nuclei, such as

¹

H,

³

He,

¹³

C,

¹⁵

N and

¹²⁹

Xe, have a non-zero I and can be studied with NMR. The most important is the case of

¹

H, largely used in the clinical application (Magnetic Resonance Imaging, MRI) for reasons of sensitivity (

¹

H has a stronger coupling with the external magnetic field than any other nucleus) and of natural abundance (about 80 M) in biological tissues [1].

^∗

The sensitivity of NMR experiments is determined by intrinsic nuclear polarization and extrinsic detection limits of probes and receiver circuits. The extraordinary poten- tial of NMR in providing manifold microscopic information about living and non-living matter [4–7] is somehow constrained by the technique’s intrinsic low sensitivity. Con- sidering nuclei with spin quantum number I =

¹₂

(such as

¹

H,

³

He and

¹³

C) in presence of an external magnetic fielf B

0

at thermal equilibrium, the net magnetization per unit volume, and thus the available NMR signal, arise from the polarization of nuclear spins proportional to the population difference between the up and down quantum states. For the two spin states split by an external magnetic field B

0

, the polarization in a thermal equilibrium is defined as

P

₀

= n

⁻₀

− n

⁺₀

n

⁻₀

+ n

⁺₀

= tanh γ

_n

~B

0

2k

_B

T

(2.1)

∗For a more detailed description of NMR see [2] and [3].

17

(26)

18 CHAPTER 2. DYNAMIC NUCLEAR POLARIZATION where γ

n

is the gyromagnetic ratio of the nucleus, T is the temperature, k

B

is the Boltz- mann constant and ~ is the Planck constant. If the two populations are equal, the resulting macroscopic magnetization is null (no NMR signal). However, under thermal equilibrium conditions, slightly higher energy is associated with the down state, thus n

⁻0

will be slightly smaller than n

⁺0

and a weak polarization appears: at room temperature and for a standard laboratory magnetic field (1÷10 T) only one nuclear spin over 10

⁵

÷ 10

⁶

effectively contributes to the NMR signal. From the equation (2.1) it is clear that the thermal equilibrium polarization increases with increasing the magnetic field B

0

and de- creasing the temperature T . This has been the motivation for developing higher field NMR and MRI systems, although practical problems such as costs, radiofrequency pen- etration depths, and tissue contrast, increase dramatically with increasing the magnetic field.

2.2 Hyperpolarization techniques

To overcome the sensitivity limitations of the NMR, hyperpolarization techniques, such as Brute Force, Para-Hydrogen Induced Polarization (PHIP), Optical Pumping and Dy- namic Nuclear Polarization (DNP), have been developed, thus opening up unexpected experimental possibilities [1]. These methods, briefly described below, increase the popu- lation difference and then the NMR sensitivity by creating a non-equilibrium distribution of the nuclei: in the limit of polarization equal to 1, all the nuclear spins are on the same lowest energy level.

The brute force approach. As mentioned above, the thermal equilibrium polarization

increases by increasing the magnetic field and by decreasing the temperature. The brute

force method used the advanced techniques developed in the latest decades which make

accessible very high field and very low temperature ranges: for example, cooling down

the sample to liquid helium temperature (4 K) and applying a magnetic field of 20 T, the

nuclear polarization can increase up to 1000 times. Then, if the sample (commonly a noble

gase such as

³

He or

¹²⁹

Xe) is rapidly brought to 1.5 T and body temperature (in vivo MRI

application) with negligible losses of polarization, it will remain still hyperpolarized at

body temperature. However, to reach polarization levels suitable for molecular imaging,

the brute force method of hyperpolarization should be performed at impracticable low

(27)

2.3. DYNAMIC NUCLEAR POLARIZATION. 19 temperatures (few mK). Because of the great technical challenges and costs associated with these extremely low temperatures, this method has not yet been used for in vivo applications.

The PHIP method. The Para-Hydrogen Induced Polarization technique takes advan- tage from chemical reactions involving parahydrogen (the state where the hydrogen nuclei are in a singlet ground state) that increase the nuclear polarization. The hydrogenation mechanism operates by transfer of the parahydrogen molecule as a unit onto the substrate molecule containing, for example,

¹³

C nuclei. When a substrate molecule containing

¹³

C is hydrogenated with parahydrogen, the spin order of the parahydrogen molecule is sub- sequently transferred to the

¹³

C nucleous (in the substrate molecule) and converted to nuclear polarization. The non-equilibrium spin order of the parahydrogen molecule is then converted to nuclear polarization of the

¹³

C nuclei in the substrate molecule. However, this method is limited to molecules including double or triple C-C bonds; only in this case they could be parahydrogenated.

The optical pumping method. In this hyperpolarization technique, the angular mo- mentum is transferred from the electronically pumped spins of alkali metal atoms (such as Rb) to the nuclear spins of unpolarized noble gas atoms (such as

³

He). In the first step, suitable electronic transitions of the alkali metal atoms are driven by circularly polarized light to selectively pump the ground state electrons to an excited state. Loosely bound (van der Waals) molecules or binary collisions creates a spin exchange via Fermi-contact hyperfine interactions between the electron spins and the noble gas nuclear spins promot- ing the polarization transfer from the optically pumped electron spins to the noble gas nuclear spins. This technique has already been successfully exploited but its capability is strictly limited to noble gases, especially

³

He and

¹²⁹

Xe.

2.3 Dynamic Nuclear Polarization.

As seen for “Brute force”, low temperatures and high magnetic fields increase the polariza-

tion (eq.2.1), but also the gyromagnetic ratio γ plays an important role. Under reasonable

conditions, such as T ≈ 1 K and B

0

≈ 3 T the polarization reaches values still insufficient

for the application in MRI. For example, for

¹³

C nuclei, under the above mentioned con-

(28)

20 CHAPTER 2. DYNAMIC NUCLEAR POLARIZATION ditions the nuclear polarization is about 0.1% while the electronic polarization is about 95%, due to the high value of the electronic γ

e

, three orders of magnitude higher than the nuclear one γ

n

. The DNP technique takes advantage of the high polarization of the unpared electron spins, usually carried by a polarizing agent to transfer the polarization to the coupled nuclear spins of the target, usually by a MicroWave (MW) irradiation near the electron resonance frequency. Tipically the polarized agent belongs to the Trityl or to the TEMPO families while the most used targets are pyruvate or fumarate, as demon- strated by Ardenkjaer-Larsen et al. [8], for liquid state NMR and under the conditions above mentioned (1 K and 3 T) for

¹³

C nuclei, the signal-to-noise ratio can be enhanced by a factor > 10,000. In the solid state the nuclear polarization can be increased up to 70

% (pyruvic acid [9]). There are three possible DNP mechanisms leading to nuclear hyper- polarization: Overhauser Effect, Solid Effect and Thermal Mixing. A detailed description of the theory of DNP is outside the purpose of the present work; further information about this hyperpolarization method can be found in ref. [2].

2.4 Conventional preparation for DNP

As above mentioned, the DNP is largely used in MRI to prepare hyperpolarized contrast agents for in vivo metabolism investigation. In order to be suitable for MRI application, the target sample has to be brought to high (room) T and low B and dissolved into an injectable solution (DNP-dissolution). Under these conditions the hyperpolarized state is a non-equilibrium state, but by rapid heating and dissolving, and reducing as much as possible losses of polarization, the DNP technique can be fruitfully used for tracking metabolic events in vivo [10, 11]. This is the case of the molecular imaging, which differs from traditional imaging mainly because it uses special probes (biomarkers) that chemi- cally interact with their surroundings and in turn alter the image according to molecular changes occurring within the area of interest (e.g. tumoral tissue). It’s well known that although every NMR active nucleus can be hyperpolarized, the nucleus mainly used for in vivo MRI applications is the

¹³

C nucleus because of its presence in metabolic probes.

Due to the poor natural abundance of

¹³

C nuclei (about 1%), the molecule of interest

(e.g. pyruvic acid) needs to be isotopically enriched in order to reach high enough con-

centration of active nuclei (i.e. 0.1 M). Pyruvic acid is widely used in metabolic imaging

(29)

2.5. OPEN QUESTIONS IN DNP: THE GLASSY STATE 21 because of its high DNP enhancement. Moreover, its long longitudinal relaxation time T

1

, high solubility in water and important role in the cellular metabolism also contribute to make the pyruvic acid the DNP substrate for excellence.

A DNP sample usually contains the molecule of interest, typically in concentrations of the order of molar (M), and unpaired electron spins, typically in concentrations of the order of tens of mM. When the target sample is a solid crystalline form at room tempera- ture, either a melting procedure or a dissolution in a glass-forming agent (such as glycerol or DMSO) must be applied before the freezing step [12, 13] to obtain a homogeneous mixing of the paramagnetic centers, fundamental requirement to make the DNP process highly efficient. Indeed (as better justified in section 2.5) when the sample is frozen at very low temperatures, the segregation of the radicals outside crystalline domains has to be absolutely avoided: solvents typically used are DMSO, glycerol, ethanol and their aqueous mixtures. Both procedures, melting and quenching, dissolution in a solvent, have several drawbacks. Melting is an additional step that must be performed on site, inhibiting any industrial and pharmaceutical production of a “ready to use” formulation. Moreover, both melting and the subsequent cooling need to be carefully monitored to ensure the absence of sample degradation during the complete melting and of recrystallization during the rapid cooling step. This means that both the target and the radical must have a common window of melting and stability, which is often not the case. The dissolution procedure implies the addition of a glass-forming agent that may introduce toxicological issues and metabolic interferences in the event of the product being intended for medical use. More- over, the presence of a sort of “dead volume” in the sample to be polarized gives rise to a suboptimal employment of the whole DNP set up: the larger the sample volume, the less efficient the cooling, the microwave irradiation and the final DNP-dissolution.

2.5 Open questions in DNP: the glassy state

A detailed theoretical interpretation of the correlation between polarization level and

avoidance of crystalline structure seems limited in the literature concerning the role of a

glassy structure in the DNP process. Based on experimental evidences, in order to have

an effective DNP process, the nuclei of the target sample and the unpaired electrons of the

stable radical species need to be suitably dispersed, so that an efficient hyperfine coupling

(30)

22 CHAPTER 2. DYNAMIC NUCLEAR POLARIZATION occurs in the mixed solid solution.

The available information on the influence of configurational properties of the substrate molecules (in solid state) on the DNP process is related to polarized targets for neutron scattering experiment, used so far to study magnetic ordering at µK temperature [14–

17]. Two papers are more important than others in this topic. The first one reports a study about the characterization of glassy properties of hydrocarbon mixtures that was performed by Takala e Niinikosky in 1990 [16] subsequently to the observation that different mixtures of the same compound (i.e. butanol in ethanol or in water) gave better and more reproducibile results in terms of DNP. The authors observed by microcalorimetry anomalies in the thermal behaviour (i.e. glass transition, re-crystallization and melting) of various butanol solutions, highlighting those patterns which are preferred. Finally, they claimed that “any cristallization [..] may result in the separation and accumulation of paramagnetic molecules, either on microcrystal or in crystalline boundaries [..] and is always associated with a dramatic reduction of DNP as well as in the increase of the nuclear spin-lattice relaxation time”. The same conclusion is obtained in ref. [18] where a DNP study at 3.35 T and 1.4 K of high- and low- crystallinity polyethylene films doped with 2,2,6,6-tetramethylpiperydin-l-oxyl (TEMPO) was carried out. According to Kumada et al. their “results indicate that TEMPO is localized in the amorphous parts, but not incorporated into the crystalline parts of the films. The results thus revealed that the polarization transfer from electron spins of TEMPO to proton spins takes place only in the amorphous parts”. The non-homogeneity of the radical dispersion in presence of a crystalline phase is considered the cause of the DNP process quenching.

Moreover, Colombo Serra underlines in her PhD thesis [19], from both computational and experimental investigations, the importance of the glassy state in DNP intended for MRI applications. In particular, several points have been noted in this study of butyric acid:

- vitreous and crystalline structure behave differently in terms of T1, due to the pres- ence of molecular motions even at very low temperature, which promote relaxation mechanism and probably also favour DNP process;

- DNP experiments on crystalline structures and glassy matrixes showed very low

polarization level for glasses compared to those of the crystals: electrons are not

evenly dispersed, but accumulated at the edges between crystalline domains avoiding

(31)

2.5. OPEN QUESTIONS IN DNP: THE GLASSY STATE 23 a magnetic ordering establishment;

- polarization level and polarization time as a function of the glass former addition show non monotonic trend, meaning that the “goodness” of the glass is not improved by the increasing glass former concentration.

- from numerical simulation carried out in the Solid Effect regime in the presence of a finite intrinsic nuclear relaxation time, the polarization is strongly reduced when the spin diffusion is inefficient and the electrons are clusterized (bad co-mixing of the target and radical, possibly due to the presence of a crystalline phase).

Therefore, it seems evident that a glassy structure is required and that different quality of the glass leads to different polarization levels and build up times, and different magnetic properties also in absence of radical species.

Final comment The aim of this thesis is to develop a methodology of preparation and structural characterization of the DNP sample that can be used in a future work to correlate the hyperpolarization efficiency with physical properties of the matrix, such as, for example, the crystallinity degree. The characterization can be performed a “priori”

test to check the goodness of the glassy matrix before it is used in DNP, thus reducing the costs of the test of a new substrate for example.

The standard procedure of preparation of a DNP sample, i.e. by quenching from a

solution, by putting it in liquid nitrogen just before the NMR measurement, produces

the glassy state at very low temperature. Thus, to overcome these limitations in the

characterization of the sample at low temperature, our attention has been addressed to

other procedure of amorphization for the preparation of suitable solid state samples to be

eventually polarized by DNP meachanisms, as describe in chapter 3.

(32)

24 CHAPTER 2. DYNAMIC NUCLEAR POLARIZATION

References

[1] K. Golman et al. “Molecular imaging using hyperpolarized 13C”. In: Br. J. Radiol.

76 (2003), S118–S127.

[2] A. Abragam and M. Golman. “Principles of dynamic nuclear polarisation”. In: Rep.

Prog. Phys. 41 (1978), pp. 395–467.

[3] C. P. Slichter. Principles of Magnetic Resonance. Springer-Verlag, 1989.

[4] R. G. Griffin and T. F. Prisner. “High field dynamic nuclear polarization-the renais- sance”. In: Phys. Chem. Chem. Phys. 12 (2010), pp. 5737–5740.

[5] C. P. Slichter. “The discovery and demonstration of dynamic nuclear polarization-a personal and historical account”. In: Phys. Chem. Chem. Phys. 12 (2010), pp. 5741–

5751.

[6] A. J. Rossini et al. “DNP enhanced NMR spectroscopy for pharmaceutical formu- lations”. In: J. Am. Chem. Soc. 136 (2014), pp. 2324–2334.

[7] A. Rossini et al. “Dynamic nuclear polarization NMR spectroscopy of microcrys- talline solids”. In: J. Am. Chem. Soc. 134 (2012), pp. 16899–16908.

[8] J. H. Ardenkjaer-Larsen et al. “Increase in signal-to-noise ration of > 10,000 times in liquid-state NMR”. In: Proc. Natl. Acad. Sci. U. S. A. 100 (2003), pp. 10158–

10163.

[9] H. Johannesson and J. H. Macholl S. ad Ardenkjaer-Larsen. “Dynamic nuclear po- larization of [1-

¹³

C]pyruvic acid at 4.6 T”. In: J. Magn. Reson. 197 (2009), pp. 167–

175. [10] K. Golman et al. “Metabolic imaging by hyperpolarized 13C magnetic resonance imaging for in vivo tumor diagnosis.” In: Cancer Res. 66 (2006), pp. 10855–10860.

[11] J. Kurhanewicz et al. “Analysis of cancer metabolism by imaging hyperpolarized nuclei: prospects for translation to clinical research.” In: Neoplasia 13 (2011), pp. 81–

97. [12] M. Karlsson et al. “Development of dissolution DNP-MR substrates for metabolic research”. In: Appl. Magn. Reson. 43 (2012), pp. 223–236.

[13] T.-C. Ong et al. “Solvent-free dynamic nuclear polarization of amorphous and crys-

talline ortho-terphenyl”. In: J. Phys. Chem. B 117 (2013), p. 304.

(33)

REFERENCES 25 [14] D. A. Hill and J. J. Hill. “Investigation of polarized-proton target materials by differential calorimetry: preliminary results”. In: ANL-HEP-PR-81-05 12 (1980), pp. 1–66.

[15] D. Hill and M. Krumpolc. “Dynamic Polarization In Some High Hydrogen Glasses”.

In: AIP Conf. Proc. Vol. 95. Proc. Conf. on High Energy Spin Physics-1982. Brookhaven, 1983, pp. 479–484.

[16] S. Takala and T.O. Niinikoski. “Measurements of glass properties and density of hydrocarbon mixtures of interest in polarized targets”. In: Proc. 9th Int. Symp. on High Energy Spin Physics . Vol. 2. Bonn, 1990, pp. 347–352.

[17] E. I. Bunyatova. “New investigation of organic compounds for targets with polarized hydrogen nuclei”. In: Nucl. Instrum. Meth. A 356 (1995), pp. 29–33.

[18] T. Kumada et al. “Dynamic nuclearpolarization of high- and low- crystallinity polyethylene”. In: Nucl. Instrum. Meth. A 606 (2009), pp. 669–674.

[19] S. Colombo Serra. “Dynamic nuclear polarization of heterogeneous systems for en-

hanced MRI”. PhD thesis. University of Torino, 2012.

(34)

26 CHAPTER 2. DYNAMIC NUCLEAR POLARIZATION

(35)

Chapter 3 Choice of amorphization technique on a model system

The purpose of this chapter is to present the first step of the experimental activity, the in- vestigation of several amorphization techniques in order to identify the most reproducible procedure to control the structural properties of the resulting amorphous and nano- or micro-crystalline matrix. Therefore, many techniques of amorphization have been inves- tigated to compare the physical and structural features of the several amorphous solids obtained. Among the methods shortly discussed in chapter 1, fast quenching from the melt, evaporation from the solution, spray drying, freeze drying, fast quenching from the solution have been explored. However, it has been found that each method has specific advantages and disadvantages, therefore it is necessary to identify the best procedure for the aim of this project.

In particular, as described in chapter 2 the system investigated consists of a background containing a metabolically active molecule, and of a paramagnetic molecule, magnetic transfer agent in the DNP process. The background can be formed by a single component or a mixture of two or more components, one of these acting as metabolic agent, while the other(s) as stabilizing species, which allow to control the configurational properties of the mixture and to preserve them during an eventual storage (i.e industrial purview). For example, where glucose is used as metabolic molecule while lactose as stabilizer owing to its high glass transition temperature (Tg about 120 °C).

It has to be noted that in this context the target is not only a good amorphization but the co-amorphization of the background and the radical species without allowing

27

(36)

28CHAPTER 3. CHOICE OF AMORPHIZATION TECHNIQUE ON A MODEL SYSTEM confinement or aggregation in specific sites, e.g., in the edges between small crystalline domains. For this reason some amorphous and nano- or micro-crystalline mixtures will be taken into account as model system.

Moreover, a very important issue to be taken into account is the presence of water into the amorphous matrices. The latter, depending on the preparation procedure, can be residual water of the dehydration process from solutions (such as evaporation, deeply described in chapter 6, freeze drying, spray drying...) or adsorbed water from the envi- ronment if the sample is stored under uncontrolled conditions (relative humidity).

3.1 Model system

In this work, the choice for the stabilizing molecule has been addressed to some sugars and sugar mixtures due to their high stability, expecially because they are solid at room temperature. Moreover, information about the physical and mechanical properties of amorphous solids prepared with sugars is available in the literature, due to their large use in the food field. In particular, the investigated samples are based on trehalose, a well-known sugar deeply investigated both in crystalline and glassy states [1–5].

As to the choice of the polarization agent, several radicals exist but only two families were proved to be efficient for the DNP process: trityl and nitroxide radicals. The most important features required for a DNP agent are the following: solubility in the substrate of interest, stability and a narrow EPR line. The nitroxide radical TEMPO is probably the most widely used stable radical, highly soluble in water and in typical DNP solvents.

Moreover, it shows an EPR line width of few hundreds of MHz at low temperature. On the

other hand, trityl radicals are also well known for their stability, their small line width at

low temperature and their long electronic relaxation time (few seconds). Their solubility

in polar liquids (lower than that of TEMPO family) can be adapted by replacing some

external groups in their structural formula. In conclusion, the choice for the radical has

been addressed to TEMPO, because of low costs and higher solubility if compared to its

derivatives and to the trytil molecules.

(37)

3.2. COMPARING AMORPHIZATION TECHNIQUES ON THE MODEL SYSTEM 29

3.2 Comparing amorphization techniques on the model system

3.2.1 Fast quenching from the melt

As mentioned in chapter 1, this is the conventional methodology used to quench the sugar molecules from the melt in a disordered metastable state. It can be used only for the individual sugars, not always suitable for mixtures of sugars, when the components have very different melting points, because of degradation problems. This method is performed directly inside the chamber of a Differential Scanning Calorimeter (DSC), which is proven to be an important tool for the preparation of the amorphous solids, allowing a real time monitoring of crucial transitions, such as melting.

3.2.2 Fast quenching from the solution

As mentioned in chapter 2, this is the standard preparation for DNP-MRI application.

As in the case of fast quenching of the melt, the molecules are quenched from a liquid state in a metastable state. Unfortunately these samples are not physically stable for the huge presence of water, which considerably decreses the glass transition temperature (T

^Hg²^O

= −137 °C [6, 7]). For this reason, it is necessary to store the samples prepared by this procedure at very low (sometimes unapproachable) temperatures.

3.2.3 Dehydration from the solution

This preparation technique of glassy samples, which can be used to obtain binary or more

complex mixtures, is further investigated in chapter 6 and it . The sample must be kept at

high temperature for a sufficient time to allow the complete evaporation of water. Depend-

ing on evaporation rate, the sample may retain disorder in the solid state. A compromise

between the residence time at high temperature and the evaporation temperature has to

be found to avoid degradation and to control the eventual crystallization. Also in this

case the DSC can be fruitfully used to control the environment conditions, particularly

the relative humidity by setting the nitrogen flux inside the chamber. In chapter 6 a deep

study on the solution dehydration on controlled geometry systems, such as films or drops,

has been reported. In this study great attention has been paid to the role of water: the

(38)

30CHAPTER 3. CHOICE OF AMORPHIZATION TECHNIQUE ON A MODEL SYSTEM relative humidity is a disturbing factor as promoting agent of crystallization events and a wasteful constituent of the amorphous matrix, even if it is a fundamental element during the amorphization process determining the structural features of the final product.

3.2.4 Nano spray drying and freeze drying

Freeze drying and spray drying are techniques widely used in pharmaceuticalsm; for this reason the co-amorphization of sugars (and sugar mixtures) and radical species has been tested also by these methods. The nano spray drying produces samples consisting of nano amorphous grains. In particular, the solution is sprayed by an atomizer (control- lable size pores) in nano-drops at high temperature (typically 120 °C), which causes the water evaporation from the drops, the disordered state being retained. A temperature grandient (from 120 °C to a suitable temperature, tipically lower than 20°C) in the spray- ing chamber allows the sample to quickly reach the suitable temperature for obtaining the glassy state. However, the sprayed samples are not physically stable, because they easily absorb humidity. Indeed they have a high exposed surface and they may give further crystallization over time, in particular in case of low T

g

solutes (such as glucose, T

g

= 38

°C). Preliminary tests for the preparation of these amorphous solids have been performed by a Buchi Nano Spray Dryer B-90 in the laboratory of Prof. P. Blasi at the Department of Pharmaceutical Sciences, University of Perugia.

The freeze drying allows to produce amorphous solids from a solution, previously freezed in a glassy state, by sublimating at low temperature (typically -50°C) and under vacuum. This method includes several steps of lyophilization using also drying materials to avoid the presence of water in the final product. Moreover, some items have to be pointed out. First of all, the glass transition temperature of the solution has to be not too low so that the sample may retain an amorphous conformation during the quenching step. Secondly, the solubilization temperature has to be not too high in order to avoid the radical sublimation (in the case of TEMPO radical).

3.2.5 Milling

Mechanical alloying is a non-equilibrium amorphization process and has been widely used

in the development of new metallic systems. Milling of molecular crystals produces amor-

phous solids consisting of one or more compounds at room temperature, or lower tem-

(39)

3.2. COMPARING AMORPHIZATION TECHNIQUES ON THE MODEL SYSTEM 31 perature in case of cryo-milling. Supply of mechanical energy to a crystalline substance beyond a critical level induces lattice instability of the crystal and freeze it in an energized state that has lost the original periodicity. Obvioulsy the process must be carried out at temperatures below the range of cold-crystallization and, preferably, below the glass tran- sition temperature of the resulting amorphous solids. The properties of the amorphous solids depend on the milling conditions. Both the T

g

value and the enthalpy of crys- tallization of the solid are found to increase asymptotically with the milling time. For some binary molecular crystals, milling of a mixture results in the molecular alloys with a single T

g

varying with the composition. This indicates that the component molecules mix uniformly at molecular level to exhibit a single relaxation process over the whole composition range. Formation of solid solutions in the crystalline state is highly limited by many factors such as the shared crystal symmetry, the similar sizes of the unit cells, the similar molecular shapes, and so on. Alloying of otherwise immiscible substances in the solid state is possible only under non-equilibrium conditions. It has to be noticed that amorphization by any low temperature routs is particularly useful for substances that are unstable at high temperatures.

The high reproducibility of the procedure and good control of the presence of nano- and micro- crystals makes milling the optimal technique for the purpose of this work.

Moreover, trehalose is able to form glass solutions with several other molecular mate- rials by co-milling (driven molecular alloys) [8] then it can be actually used as mainly component of the model system.

Scanning Electron Microscopy (SEM) measurements have been carried out on trehalose

to prove the effectiveness of the amorphization by milling. SEM measurements have been

performed by a SEM Zeiss EVO 50 microscope at the Medical Veterinary Department of

the University of Teramo under the supervision of Dr. Luca Valbonetti. The crystalline

domains of the β-polymorph trehalose are known to be of the order of 100 µm, while in

figures 3.1,3.2, 3.3 and 3.4 it is evident that the crystalline domains eventually present in

the 12h milled trehalose have dimensions below the perceptible range (< 50 nm).

(40)

32CHAPTER 3. CHOICE OF AMORPHIZATION TECHNIQUE ON A MODEL SYSTEM

Figure 3.1: SEM image of 12 h milled trehalose with magnification of 202x.

Figure 3.2: SEM image of 12 h milled trehalose with magnification of 2040x

(41)

3.2. COMPARING AMORPHIZATION TECHNIQUES ON THE MODEL SYSTEM 33

Figure 3.3: SEM image of 12 h milled trehalose with magnification of 10,440x

Figure 3.4: SEM image of 12 h milled trehalose with magnification of 21,190x

(42)

34CHAPTER 3. CHOICE OF AMORPHIZATION TECHNIQUE ON A MODEL SYSTEM

3.3 Materials and methods

Anhydrous crystalline trehalose (β polymorph) from Acros Organics was used as the primary component of the amorphous samples, and the radical 2,2,6,6,-tetramethyl-1- piperidinyloxy (TEMPO) from Sigma Aldrich, was used as the polarizing agent. α- lactose monohydrate, fumaric acid, L-glutamine and anhydrous D-(+)-glucose from Sigma Aldrich have been employed for further tests to investigate the DNP properties of sev- eral matrices. α-Lactose monohydrate has been dehydrated by evaporation in the (α) anhydrous crystalline form before the amorphization.

The sample amorphization has been performed through the use of a high energy plan- etary mill (Pulverisette 7 from Fritsch) at room temperature, in the UMET laboratory at the University of Lille under the supervision of Prof. M. Descamps and Dr. JF Willart.

Each sample has been inserted in a ZrO

2

milling bowl (volume of about 45 cm

³

) contain-

ing seven ZrO

2

1 The amorphous and glassy states 3

ii

Contents

Preface vii