EP3177567A1 - Ultra-dense shell core-shell nanoparticles - Google Patents

Abstract

The invention provides a method of producing inorganic core particles comprising a dispersant shell, comprising a dispersant molecule in a high surface covering density on the inorganic core, comprising the steps of providing an inorganic particle, ligating an organic linker onto an inorganic particle, thereby obtaining an inorganic core linker coated particle, providing a fluidized dispersant, preferably in form of a melt, suspension or solution, binding the fluidized dispersant to the organic linker, thereby obtaining the inorganic core particles comprising a dispersant shell, as well as particles obtainable by such a method.

Description

Ultra-dense shell core-shell nanoparticles

Field of the invention

The present invention relates to the field of polymer coated nanoparticles .

Background of the invention

Polymer shell-inorganic core nanoparticles are sought for their unique properties applicable for example in biomedical im^¬ aging, drug delivery and new therapeutics. They are also inves^¬ tigated as additives to improve bulk material properties. The creation of a dense functional polymer shell that shields the inorganic core from the environment is crucial to function.

Superparamagnetic nanoparticles (NPs) , such as Fe₃0₄ NPs, with core diameters of 3-15 nm, are used in a rapidly expanding number of applications in the biomedical field; the most common include magnetic cell labeling, hyperthermia, drug delivery, and as contrast agents for magnetic resonance imaging. Iron oxide cores are coated with polymer dispersants, to enable dispersion of NPs in aqueous solutions and organic matrices. NPs will rap^¬ idly aggregate without a polymer shell; attractive interactions between NPs or between NPs and biological molecules lead to ag^¬ gregates, precipitates and loss of function. With recent im^¬ provements in the synthesis of NPs, there has been a push to^¬ wards more and more well-defined, core-shell particle architec^¬ tures .

Monodisperse spherical iron oxide NPs can be synthesized with monodisperse magnetic and other physical properties; these NPs are stabilized with an irreversibly anchored shell of line^¬ ar, end-grafted polymer dispersants of sufficient thickness to ensure that the particle properties remain monodisperse during application. This defined core-shell architecture enables the prediction of all colloidal properties and serves as a platform for further defining and designing physical, chemical and bio^¬ logical interactions.

An advantage of separating the synthesis of core and disper^¬ sant with subsequent grafting of the dispersant to the core is that the hydrodynamic size of NPs can be precisely controlled in contrast to NPs with dispersant shells consisting of physisorbed high molecular weight (M_w) dispersants. The well-defined assembly of dispersants at the NP surface enables controlled surface presentation of functionalities. NP size, stability, dispersant shell thickness and control over functionalities presented at the NP surface are the factors that critically determine NP per^¬ formance in a biological fluid and also for many non-biological applications. The maximum achieved grafting densities on planar substrates of PEG(5kD) are in the range of 0.4 chains/nm² (on Ti0₂) [J. L. Dalsin et al . , Langmuir 2004, 21, 640-646.]; and only ~0.5 chains/nm² has been reported on iron oxide NPs after ligand replacement. The separate optimization of core and dispersant creates the problem of having to graft the dispersant into a dense, brush-like shell to the core surface; this requires con^¬ ditions under which an irreversible binding reaction occurs be^¬ tween a reactive end group on the dispersant polymer and the iron oxide core. It also requires that the reaction occurs when the polymer has a small coil size, R_G. The R_G determines the footprint of the polymer and thereby the maximum grafting densi^¬ ty that can be achieved.

Current ways to functionalize nanoparticles with polymer shells can be divided into two approaches: "grafting-from" and "grafting-to" . The grafting from approach requires polymeriza^¬ tion from initiators on the nanoparticle surface, a method that is difficult to control at high nanoparticle density or large volume reaction conditions, leading to polydispersity and diffi^¬ culties in controlling the functionality and density of func^¬ tional groups of the core-shell NPs.

The grafting-to approach uses pre-synthesized polymer lig- ands, which therefore can be monodisperse and added in mixtures of different functionalities to the NPs. The grafting-to ap^¬ proach is however limited in the polymer grafting density that can be achieved on the NP surface due to the steric constraints of the polymer coil approaching an already partially grafted NP, as the polymer requires solubility (which expands the polymer coil size) under reactive conditions to bind to the NP surface. The size of the polymer coil limits the achievable grafting den^¬ sity due to the effective hindrance on the particle surface.

The problem of finding appropriate reaction conditions to achieve high dispersant density on the particle surface has been shown to be increasingly problematic for state-of-the-art syn^¬ thesis protocols for monodisperse inorganic nanoparticle cores that require capping agents (strongly binding ligands) during synthesis to control size and shape (Park et al . , Nat. Mater. 2004, 3, 891-895 and US 2013/164222 Al); ligand controlled core synthesis requires ligand replacement from the synthesis ligand to the dispersant ligand after completed core synthesis. Mono- disperse particles have the advantage that they have identical physical and chemical properties, which otherwise vary strongly for nanoparticles and affect their reproducible application.

Such particles have hitherto not been demonstrated possible to functionalize with sufficiently dense covalently grafted polymer dispersant shells for biomedical (or other) applications.

WO 2013/110828 Al according to its English translation in EP 2 808 036 Al relates to polymer coated iron oxide particles in order to reduce vascular interactions, which is achieved by us^¬ ing specifically selected polymers with suitable functional groups. These polymers are attached using carbodiimide as rea^¬ gent as disclosed in Liao et al . , J. Mater. Chem. 2002, 12:

3654-3659. Liao et al . investigated the binding chemistry of this reaction and found that carbodiimide activation creates amino groups on the Fe₃0₄ particles, which in turn can form a bond with carboxylic acids of polyacrylic acid (PAA) . After binding of PAA, no amino groups were detected. Also the reaction is not very efficient. Only two PAA molecules are bound on each particle .

EP 0 516 252 A2 relates to magnetic iron oxide particles with a chemisorbed shell comprising glycosaminoglycan.

Colloidal stability was previously achieved through an irre^¬ versible bond of the polymer to bare Fe₃0₄ NP cores by direct (one step) grafting-to using nitrocatechol anchors and a report^¬ ed dispersant density of 2.6-2.8 M/nm² [E. Amstad et al . , Nano Letters 2009, 9, 4042-4048; E. Amstad et al . , J. Phys . Chem. C 2011, 115, 683-691; WO2011/026572 Al ] . This reported dispersant density has not been observed on truly monodisperse and spheri^¬ cal nanoparticles, and might therefore partly be a consequence of higher surface roughness and effective surface area, which influence the calculation of surface densities. Thus, the cor^¬ rect grafting density taking the entire surface of the Fe core into account is effectively below 1.0 M/nm² for these nanoparti^¬ cles used in the inventors' earlier work.

Synthesis of truly monodisperse (PDK1.05) Fe₃0₄ core-shell NPs makes use of surfactants as capping agents, e.g. oleic acid, to control the crystal growth and sphericity [T. Hyeon et al . , Journal of the American Chemical Society 2001, 123, 12798-12801; Park et al . , supra] . The as-synthesized NP core then has a dense shell of oleic acid that has to be replaced by the stabilizing dispersant. However, capping agent replacement by ligands or dispersants yielded NPs with ~5 times lower PEG grafting density (i.e. densities of only 0.5 M/nm² - see table 1 herein) and thereby colloidal stability than bare particles grafted with dispersants. The lower grafting efficiency results from that grafting-to by ligand replacement of oleic acid failed ful^¬ filling more simultaneous and mutually contradicting constraints than grafting to a bare surface. The grafting can also be inho- mogenous, leading to further compromised colloidal stability.

Therefore there remains the goal to produce nanoparticles with a dense dispersant shell. Preferably these particles would also be homogenous and monodisperse .

Summary of the invention

The present inventors succeeded in providing a new manufac^¬ turing method fulfilling these needs. The present invention is defined in the claims.

In particular, the invention provides a method of producing inorganic core particles comprising a dispersant shell, compris^¬ ing a dispersant molecule in a high surface covering density on the inorganic core, comprising the steps of: providing an inor^¬ ganic particle, ligating an organic linker onto an inorganic particle, thereby obtaining an inorganic core linker coated par^¬ ticle, providing a fluidized dispersant, preferably in form of a melt, suspension or solution, binding the fluidized dispersant to the organic linker, thereby obtaining the inorganic core par^¬ ticles comprising a dispersant shell.

The present invention also provides a particle or a prepara^¬ tion of a plurality of particles obtainable by the inventive methods .

In a further embodiment the invention provides a preparation of a plurality of particles, wherein said particles comprise an inorganic core surrounded by linkers that are chemically linked to dispersant molecules, wherein the dispersant molecules (a) are at an average density of at least 1.1 dispersant molecules per nm² of the inorganic core surface , and/or (b) form a shell of constant dispersant density and a further shell of gradually reduced dispersant density with increasing distance from the in^¬ organic core surface, wherein preferably the inorganic core is of an average size between 2 nm to 80 nm in diameter. Definition (b) relates to forming a shell of similar density to the molten polymer state. This high density shell is identifiable by small angle x-ray scattering in a solvent. Separate therefrom there is a solvable shell, distinct from the dense inner polymer shell.

An inventive preparation may comprise a plurality of parti^¬ cles with such dense polymer shells (a) and/or (b) , wherein the inorganic core of an average size between 2 nm to 80 nm in diam^¬ eter is of homogenic size in said plurality wherein the mean standard deviation of said average size is at most 10%, prefera^¬ bly at most 5%, even more preferred at most 2% of said average size, such as at most 0.8 nm, preferably at most 0.5 nm, prefer^¬ ably said preparation being obtainable by a method of as further defined herein.

Also provided is a particle comprising an inorganic core surrounded by linkers that are chemically linked to dispersant molecules, wherein the dispersant molecules are at an average density of at least 1.1, preferably at least 3.0, dispersant molecules per nm² of the inorganic core surface, and/or (b) form a shell of constant dispersant density and a further shell of gradually reduced dispersant density with increasing distance from the inorganic core surface, preferably said particle being obtainable by a method of as further defined herein. Similar as said above, this shell structure may also be due to the disper^¬ sant forming a shell of similar density to the molten polymer state identifiable by small angle x-ray scattering in a solvent and a solvable shell, distinct from the dense inner polymer shell, These particles can be provided in a preparation compris^¬ ing a plurality of such particles.

Surface densities on dispersant molecules per nm² of the in^¬ organic core surface as used herein refer to the surface of the inorganic core, and not on a spherical approximation thereof as e.g. used in reference Amstad et al . 2009, Amstad et al 2011, W02011 /026572 Al, all supra. In a direct comparison using the same analysis methods, the inventive particles have at least about 2 times higher surface densities than the particles re^¬ ported previously (see comparative examples herein) . The following detailed disclosure reads on all aspects and embodiments of the present invention, irrespective of relating to a method, preparation or particle. E.g. described method steps also disclose that the resulting product can result in an element of the particle or preparation, such as specific rea^¬ gents used in the method may lead to a chemical group or moiety bound to the particle or preparation. Elements described for the preparation can read on steps in the inventive manufacturing method. Furthermore described elements of the particles as such or of particles of the preparation can be elements of both groups. As such, the linker will be bound to the dispersant and in the final particle of the invention the former linker forms an anchor group of the dispersant that binds the dispersant to the inorganic core. Therefore, all descriptions of the linker also read on such an anchor group of the dispersant in the in^¬ ventive particle. Also, the invention relates to a preparation and all descriptions of particles also read on particles of said preparation .

The present invention is suitable to provide high density dispersant inorganic core nanoparticles . Furthermore it is pos^¬ sible to generate these particles with high size homogeneity (monodisperse) without the requirement of particle sorting by size such as by size exclusion chromatography. The generation of monodisperse inorganic particles without the necessity for size sorting is based on the metal-surfactant complex developed by Park et al . (Nat. Mater. 2004, 3, 891-895, incorporated herein in its entirety) , which provides inorganic particles, that form the core of the inventive inorganic particles. However, the in^¬ ventive method to obtain high density dispersant particles is not limited to metal cores - although preferred for all embodi^¬ ments - and includes any inorganic material, such as semiconduc^¬ tors, Si and silica materials for the core particle.

Detailed description of the invention

The present invention for the first time manages very high surface densities. In certain embodiments, this could also be combined with the technology for producing monodisperse inorgan^¬ ic nanoparticles with high surface density modification by dis- persants. Previously the strong bond to the capping agent re^¬ duced future ligand binding reactions. In preferred embodiments, an inorganic particle comprising a metal in complex with a surfactant on the particle surface is provided. The surfactant is a capping agent. A capping agent is a strongly adsorbed monolayer of usually organic molecules used to aid stabilization of nanoparticles . Example surfactants are fatty acids, such as oleic acid. The surfactant may comprise a hydrophilic group and an aliphatic chain of e.g. 5 to 30 C atoms in length. The aliphatic chain may be saturated or unsaturated, comprising one or more double or triple bonds. The hydrophilic group may e.g. be a negatively charged group, or a carbonyl group, such as a carboxylate, sulphate, phosphate.

In a preferment for all embodiments and aspects of the in^¬ vention, the inorganic particle core comprises preferably a met^¬ al, e.g. selected from Fe, Cu, Au, Ag, Cr, Mn, Ti, Ni, Co, or any other element of the fourth row of the periodic table, or alloys thereof. In further embodiments the inorganic particle core comprises a metalloid, a semiconductor or consists of a non-metal material. Examples are Al, Si, Ge, or silica com^¬ pounds. The inorganic nanoparticle core can be a nanocrystal or a multidomain crystallised nanoparticle composed of more than one nanocrystal. Preferably the core comprises an oxide any thereof, preferably a Fe oxide, such as Fe₂0₃ and/or Fe₃0₄. In a further embodiment, the inorganic nanoparticle core comprises a hydride nitride or an iron sulfide, preferably mixed ox^¬ ide/hydroxide, nitride or sulfide of Fe (II) and/or Fe (III), e.g. in the form of a nanocrystal. Preferably, the inorganic na^¬ noparticle core is Fe₃0₄ (magnetite) or comprises Fe₃0₄ spiked with any other metal, preferably those mentioned above. "Metal" as used herein refers to the element, not to the state. The met^¬ al may be metallic (with neutral charge) or, as in most case of the present invention, non-metallic, especially in case of crys^¬ tallized cationic metals.

In further preferments of all inventive aspects and embodi^¬ ments, the inorganic core is magnetic, especially paramagnetic, preferably superparamagnetic. This property can be achieved by using metal nanoparticles of a material as described above, es^¬ pecially selected from the group consisting of iron, cobalt or nickel, alloys thereof, preferably oxides or mixed ox^¬ ides/hydroxides, nitrides, carbides or sulfides thereof. In a preferred embodiment the stabilized magnetic nanoparticles are superparamagnetic iron oxide nanoparticles (SPIONs) . Magnetic particles allow controlled mobility, such as for separation of enrichment of particles in a non-accessible compartment, e.g. in a patient's body by applying a magnetic field, or the capability to heat the particles by applying an oscillating field, in particular by radio wave irradiation, e.g. in the range of 10 kHz to 1000 kHz, e.g. 400 kHz.

In some embodiments of the invention, the inorganic particle core can be produced together with a surfactant. Preferably the provided inorganic particle is in complex with a surfactant on the particle surface, especially preferred in case of metal par^¬ ticles, and wherein ligating the organic linker onto the inor^¬ ganic particle is by replacing the surfactant by said organic linker, thereby obtaining an inorganic core linker coated parti^¬ cle. Inorganic particles in complex with a surfactant can be ob^¬ tained by contacting the particle with an inorganic particle core with a surfactant. An alternative may comprise the synthe^¬ sis of the particle core in the presence of a surfactant. Such a method may comprise thermal decomposition of dissolved metal- surfactant complexes. Metal-surfactant complexes in turn - with^¬ out necessary limitation to this option - may be generated by contacting a metal salt, such as a salt with a monovalent anion, preferably a halide such as a chloride (which is readily availa^¬ ble) , in particular iron chloride, with a negatively charged surfactant such as a fatty acid as mentioned above (which are also well available) , thereby forming a metal-surfactant com^¬ plex. The metal surfactant composition can be thermally decom^¬ posed to form the inorganic nanoparticle core. Thermal decompo^¬ sition may comprise heating in a solvent to temperatures above 200 °C, in particular temperatures above 280°C, e.g. between 280°C and 350°C, preferably about 310-320°C. Suitable solvents are e.g. aliphatic hydrocarbons, such as 1-octadecene . Prefera^¬ bly the surfactant comprises a carboxylic acid group, which de^¬ composes during thermal decomposition to form a metal oxide (Park et al . , supra; and Hyeon et al . , supra; both incorporated herein by reference) . Another preferred method comprises - as shown in example 3 herein - comprises providing a metal complex, such as a metal carbonyl (e.g. Fe(CO)₅, Cr (CO) ₆ or Ni(CO)₄), and contacting the metal complex with a surfactant in a solvent at elevated temperatures, e.g. between 150°C and 290°C, whereby the metal particle core forms in contact with the surfactant. In these methods, particle core size can be influenced by the con^¬ centration ratio of the metal complex and the surfactant. Pref^¬ erably the temperature in this method is gradually increased, e.g. by a temperature ramp, with the reaction already expected to start at a lower temperature, e.g. 170°C or lower, and for completion of the reaction is increased to a higher temperature, e.g. at least 240 °C.

The inorganic core preferably has an average diameter size of 1 nm to 400 nm, preferably 1.5 nm to 100 nm, especially pre^¬ ferred 1.8 nm to 12 nm, 2 nm to 20 nm, 3 nm to 40 nm or 4 nm to 80 nm. Such particles can be produced by the above mentioned method. Size may further improve the magnetic properties. E.g. a sufficiently small size may be selected to achieve superparamag^¬ netic properties. The size for such an effect is dependent on the material and can be selected by a skilled person with aver^¬ age skill in view of prior knowledge.

The inventive method contains the steps of

- ligating an organic linker onto an inorganic particle; if a surfactant is present according some embodiments, this will re^¬ place the surfactant by said organic linker, thereby obtaining an inorganic core linker coated particle;

- providing a fluidized dispersant, preferably in form of a melt, suspension or solution;

- binding the fluidized dispersant to the organic linker, there^¬ by obtaining the inorganic core particles comprising a disper^¬ sant shell.

Optimal reaction conditions aim at conditions to: (1) dis^¬ solve the reversibly bound surfactant (e.g. oleic acid), (2) maintain conditions for binding of the linker to the inorganic core, (3) fluidize the dispersant, e.g. PEG, (4)while keeping the dispersant in a low R_G (low solubility or low coil volume) conformation .

The inventors have investigated the consequences of limits of the grafting-to approach by the state-of-the art method of Park et al . [Park et al . , supra], which results in monodisperse spherical particles with a shell of a surfactant. The goal was to obtain a dispersant (e.g. PEG) stably anchored to the inor^¬ ganic core through the linker. Grafting-to through direct ligand replacement of surfactant with dispersant was compared to first priming the NP surface by modifying it with the linker and then reacting various end-modified dispersants to the linker having a reactive group that can form a bond with the end-modification of the dispersant (Fig. 1 bottom) . As depicted in Fig. 1, three different types of reaction environments were tested to minimize the footprint during grafting: (a) reaction in solution; (b) microwave (MW) induced reaction in poor solvent and (c) reaction in pure polymer melt. All selected conditions were suitable to increase the surface density of the dispersant above the previ^¬ ous value achieved by grafting to nanoparticles synthesized by the Park et al . and analogue methods. Some methods even achieved a particular high surface density (1 M/nm² and more) above the previously achieved density for the polydisperse bare core grafting method (Amstad et al . , supra) . According to the inven^¬ tion, colloidally stable, monodisperse, spherical core-shell na^¬ noparticles were provided.

The inventive method comprises a two-step modification of the inorganic core surface, which allows optimization of the conditions for linker binding and dispersant grafting to a linker-modified particle surface.

As used herein, "comprising" shall be understood as refer^¬ ring to an open definition, allowing further members of similar or other features. "Consisting of" shall be understood as a closed definition relating to a limited range of features.

In a first step the nanoparticle surface is grafted with an anchor group in form of the linker which covalently or near- covalently or by ionic bond binds to the surface atoms of the inorganic nanoparticle core. The linker carries one or more ad^¬ ditional chemical functionalities that can be reacted to (one or more) organic molecules to make up a dispersant shell. Of course more than one type of linker may also be used. A further possi^¬ bility is connecting linkers when ligated onto the inorganic particle core, such as by cross-linking. One linker (before op^¬ tional crosslinking) may have one or more such chemical func^¬ tionalities, e.g. two of the same type or of a different type - e.g. to control two or more different binding reactions with two or more different dispersants. In general, it is also possible to bind more than one dispersant to the linker - even if the linker just has one binding chemical functionality. Then a com^¬ petitive reaction between the different dispersants will occur if reacted at the same time - subsequent reactions are also pos^¬ sible .

In a preferred embodiment at least one chemical functionali^¬ ty of the linker for binding to the inorganic particle core is a nitrocatechol group with an amine functional group which is grafted to superparamagnetic iron oxide nanoparticles synthe^¬ sized with an oleic acid surfactant shell through ligand re^¬ placement under non-oxidative conditions. Any material or condi^¬ tion can of course be generalized and adapted within the scope of the claimed invention. Magnetic resonance contrast agents having inorganic non-metallic cores including, but not limited to FeO_x, FeOOH and FeC_x, can be modified with the dispersants of the present invention.

The linker may have a group suitable for forming one or more, preferably at least 2, bond(s) with the inorganic core. The linker may e.g. comprise an alcohol or carboxyl group in contact with the surface of the inorganic core, preferably a moiety selected from a 6 membered homocycle comprising oxygen or hydroxyl substituents such as described in W02011 /026572 Al (in^¬ corporated herein in its entirety) , such as benzoic acid, phenol or catechole, and/or preferably wherein the linker molecules comprise an aromatic group with an electronegative group bound thereto, more preferred a nitro group. An electronegative group may be selected from F, CI, Br, I, N0₂, COOH, OH, S0₄, P0₄. Fur^¬ ther 6 membered homocycles are selected from quinolines, pyri^¬ dines and/or derivatives thereof. Preferably the linkers are bound irreversibly to the inorganic particle core. Irreversible binding of linkers to the nanoparticle core surface as used herein refers to an adsorption constant k_on » than the desorp- tion constant k_off of the linker to the nanoparticle core surface. In particular, a parameter to select in the materials used is the effective k_off of the linker to the nanoparticle core surface. It should be sufficiently low to yield insignificant dispersant loss during long-term storage. The anchor groups of the present invention are effective to irreversibly immobilize dispersants via the linker on magnetic nanoparticles, thereby achieving good nanoparticle stability in dilute and high salt aqueous environ^¬ ments up to temperatures above 90°C. Especially preferred linker molecules comprise a moiety selected from nitrocatechol, nitro- DOPA or nitrodopamine, 4-nitro-substituted catechol groups and derivatives thereof, such as in particular 6-nitro-DOPA and 6- nitrodopamine .

The linker is preferably ligated to the inorganic particle core in an organic solvent, preferably an aprotic solvent such as DMF. Preferably the ligand has a negatively charged group, e.g. a carboxylic acid, sulfuric acid or phosphoric acid group.

The result is a nanoparticle with the maximum packing densi^¬ ty of reactive (e.g. amine) groups on the nanoparticle surface allowed by the steric constraints by the small linker molecule. Such a reactive surface can be prepared by other small linker molecules and the binding chemistry used for the required strong (preferably irreversible) linking to the nanoparticle core sur^¬ face has to be tuned to the nanoparticle core material.

In the second step the dispersant carrying one reactive group and optional additional functional groups are reacted to bind the linker. In a preferred embodiment, the dispersant is a polymer and the reaction conditions are at an elevated tempera^¬ ture sufficient to create a polymer melt or strongly collapsed polymer solution in a bad solvent of the dispersant. A melt is preferred as a melt has the minimum size per polymer coil, but considerable internal mobility of the chain segments remain and the reactive group remains accessible. The constraints imposed on mobility and reactivity by the non-solvated conditions are solved by performing the reaction at elevated temperatures or using microwave assisted reactions. In the best demonstrated mode, a melt of aldehyde-poly (ethylene glycol) reacts with ni- trodopamine coated particles at 110°C in excess polymer. As above, this can of course be generalized to other materials or conditions within the scope of the claims. Thus in preferred em^¬ bodiments, binding the dispersant to the linker is an aldehyde- amine binding reaction, and/or preferably the dispersant is in a molten state in the dispersant to the linker binding reaction. In preferred embodiments, the invention comprises raising the temperature of the dispersant above its melting temperature and performing binding above the melting temperature.

The examples demonstrate the inventive concepts for a range of different suitable reactions, with preferred reactive groups and conditions for different applications.

The dispersant is preferably a macromolecule providing ste- ric/osmotic colloidal stability in the preferred environment of the application, e.g. a polymer, such as poly (ethylene glycol) (PEG; especially for aqueous, e.g. biomedical, applications) or polyisobutylene (PIB; e.g. in applications as polymer filler ma^¬ terials such as to produce impact resistant polypropylenes ) .

Further polymers with preferred properties, uses and utilities are: polyoxazolines (including different thermoresponsive deriv^¬ atives, for biomedical applications), poly(N- isopropylacrylamide) (thermoresponsive polymer, for biotechno^¬ logical applications, separation, responsive membranes and drug delivery capsules) , polyisobutylene (in applications as polymer filler materials such as to produce impact resistant polypropyl^¬ enes) , caprolactone (low melting point, biodegradable, biomedi^¬ cal applications) , polyimide (very resistant, KEVLAR, filler ma^¬ terial impact resistant materials) , polythiophene (conductive polymers, smart materials applications) , polypropyl^¬ ene/polyethylene (filler materials) , polyacrylic acids and other polyelectrolytes (pH and electroresponsive, smart material ap^¬ plications, polyvinylpyrrolidone, polyvinylalcohol (biocompati^¬ ble, biomedical and biotechnological applications. A macromole- cule is a very large molecule commonly, but not necessarily, created by polymerization of smaller subunits. The subunits of the macromolecule or polymer may be homogenous or heterogenous. Preferred dispersants comprise hydrocarbon groups, which encom^¬ pass any polymers soluble in organic solvents. Typically, "hy^¬ drocarbon chains" include linear, branched or dendritic struc^¬ tures. Different forms of hydrocarbon chains may differ in mo^¬ lecular weights, structures or geometries (e.g. branched, line^¬ ar, forked hydrocarbon chains, multifunctional, and the like) . Hydrocarbon chains for use in the present invention may prefera^¬ bly comprise one of the two following structures: substituted or unsubstituted -(CH₂)_m- or - (CH₂) _n ^_Het- (CH₂) ₀ ^~ , dendrimers of gener^¬ ations 1 to 10 where m is 3 to 5000, n and o are independently from another 1 to 5000 and Het is a heteroatom, wherein the terminal groups and architecture of the overall hydrocarbon chains may vary. E.g. in the final particle there will be an anchor group which is formed by the linker molecule. This description includes any linear or branched hydrocarbon chains with ratios of unsaturated : saturated bonds varying from 0 : 100 to 100 : 0. In some embodiments the hydrophobic spacer comprises e.g. > 50% of subunits that are -CH₂-. In alternative or combined embod- iments at least 10% of the carbon atoms, e.g. 10% to 50 ~6 , more preferred 20% to 40%, of the hydrocarbon chains are substituted by a heteroatom. Heteroatoms may be selected from 0, N, S or N, preferably 0. Side chain substitutions can be at a C or at Het with the substituents being selected independently from het- erosubstituted or non-heterosubstituted, branched or unbranched, saturated or unsaturated hydrocarbons with 1 to 20 atoms, pref^¬ erably 2 to 10, especially preferred 2 to 6 atoms in length.

The dispersant may have an average mass of 1 kDa to 30 kDa, preferably of 2 kDa to 20 kDa, especially preferred of 3 kDa to 15 kDa or 4 kDa to 10 kDa. Molecular weights down to about 1 kDa for hydrophilic sterically repulsive spacers such as

poly (ethylene glycol) (PEG) already provide good results. Pre^¬ ferred, however, are larger spacers resulting in dispersants having molecular weights in the low kDa range, e.g. in the range of 1.5 to 8 kDa, more preferred in the range of 1.5 to 5 kDa. These ranges are especially suitable for medical applications. For other applications higher kDa ranges may be preferred.

The linker on the other hand is usually a small molecule, e.g. with an average atomic mass of 75 g/mol to 1000 g/mol, preferably of 85 g/mol to 700 g/mol, especially preferred of 100 g/mol to 500 g/mol, even more preferred 120 g/mol to 400 g/mol, such as 140 g/mol to 300 g/mol. The small size of the linker al^¬ lows dense binding to the inorganic core during common reaction conditions, while still maintaining accessibility to the disper^¬ sant. As said above more than one linker type, e.g. 2, 3, 4, 5, 6 or more may be used, especially when more than one dispersant shall later be bound to the linkers and each dispersant is bound to a certain linker binding functionality as mentioned above. It is usually better controllable to bind different linkers at the same time than different dispersants, which may vary in optimal binding conditions.

After binding the linker to the inorganic core, the disper^¬ sant and the linker are reacted to form a bond. To this end, the linker and the dispersant each comprises one member of a pair of reactive groups, capable of forming a chemical bond in the liga^¬ tion reaction. Example pairs of reactive groups are an amine and an aldehyde, an amine and an acrylate, an amine and a carboxyl group, an amine and an ester group. Preferably the amine is a primary amine. Further pairs are constituted by a sulfonylchlo- ride and either an amine or a hydroxy-group, an acid-chloride and either an amine or a hydroxy-group, a thiol and any one se^¬ lected from a maleimide, an acrylate or an allyl-group, an iso- cyanate and either a hydroxy-group or an amine, a conjugated dien and a substituated alkene (for Diels-Alder-reaction) , an epoxide and a nucleophile (such as alkyl-halogenides , alcoxides, amines) . Especially preferred is an amine and an aldehyde that are reacted in the binding reaction, thereby forming an imine. Even more preferred, the imine is reduced to an (secondary) amine by a reducing agent, such as NaCNBH₃. The less reactive group of the pair, such as the amine, is preferably on the link^¬ er to prevent any off-reactions of the more reactive group, e.g. an aldehyde, that is then preferred on the dispersant. It may also be the other way around. Preferably binding the linker to the dispersant is a reaction without the requirement of a sol^¬ vent. Preferably a solvent is absent during the reaction or only in small amount, e.g. less than 50% (w/v) , or less than 20% (w/v) . The reaction is preferably between reactive groups that do not require a solvent, that might in other cases be required to stabilize or participate in a transition state. The inventive groups are preferably selected to only require contacting of the pair of functionalities for the reaction to occur. Leaving groups may be present though.

The linker or dispersant may comprise a leaving group that is removed in the ligation reaction. Leaving groups can be hal- ides such as Cl^~, Br^", and I^", sulfonate esters, such as tosylate (TsO^~) or mesylates, succinates or succinimid esters, perfluoro- alkylsulfonates , water or ammonia, inorganic esters such as phosphates or nitrates, thiolates, alcohols.

Further binding reaction may involve an addition, such as by an alkene or alkine group, preferably by an acrylate.

Using dispersants of the present invention it is also possi^¬ ble to reduce non-specific adsorption of the particles to other molecules, in particular biomolecules such as, but not limited to, proteins, peptides, and thus to increase e.g. good half-life time of the particles in biological fluids. This may be achieved by providing particles with stealth dispersants, e.g., but not limited to, poly (ethylene glycol) and poly (methyloxazoline) co- valently linked to linker groups as defined herein. This effect is due to the high dispersant packing densities that can be achieved with such dispersants having linear or branched, also comprising dendritic or hyperbranched, spacing groups or a com^¬ bination of such spacing groups.

In the inventive method, in preparation of the binding reac^¬ tion of the dispersant to the linker, the concentration of the dispersant can be above the solubility concentration of a sol^¬ vent whereby the dispersant is not fully dissolved, thereby ob^¬ taining a suspension and wherein the dispersant is fluidized by increasing the temperature above the melting temperature of the suspension. This will provide a fluidized dispersant at a high concentration and high temperature that is suitable to achieve the high surface densities.

In other embodiments, the dispersant may be provided in so^¬ lution at ambient temperatures, e.g. at between 10°C to 40°C. In other embodiments the dispersant may be provided as a melt with^¬ out a solvent. Although any temperature in the range of 10 °C to 200°C is suitable, high temperatures are preferred.

Preferably, binding the dispersant to the linker molecules is performed at a temperature of at least 80°C, preferably of 85 °C to 150°C. In preferred embodiments heating is assisted by mi^¬ crowave irradiation, especially microwave irradiation during the binding reaction, which will further assist the reaction in addition to the high temperature effects.

With the inventive method, high surface densities of bound dispersant molecules can be achieved, e.g. at least 1.1, prefer^¬ ably at least 1.2, even more preferred at least 1.3, at least 1.4, at least 1.5, at least 2, at least 2.5, at least 2.8, at least 2.9, at least 3, at least 3.1, at least 3.2, at least 3.3 or at least 3.4, dispersant molecules per nm² of the inorganic core surface. As said above with regard to method effects being related to product characteristics, such densities may also de^¬ fine the inventive particles or preparations thereof. Such par^¬ ticles can be selected from the particles obtained by the in^¬ ventive method. These densities are preferably determined by TGA ( thermogravimetric analysis), especially as shown in example 8.2 using air or an oxygen containing gas.

In further preferments of all embodiments of the present in^¬ vention the step of providing an inorganic particle or particle cores comprises providing a plurality of inorganic particles wherein the mean standard deviation of the particle's average size is at most 10%, preferably at most 5%, even more preferred at most 2% of the particle's average size, such as at most 0.8 nm, preferably at most 0.5 nm. Such particles may be synthesized as is with the method described above (e.g. the cores provided without size separation) or selected after size separation.

The standard deviation (SD) measures the amount of variation or dispersion from the average. The standard deviation of size distribution is the square root of its variance.

A "plurality" as used herein refers to several particles, which may differ within certain parameter thresholds in parameters such as size. The amount of the particles can be at least 100, at least 1000, at least 10000, at least 100000, at least 1 Mio., at least 10 Mio. etc.. Preferred ranges are e.g. 100 to 100 Mio.

A preparation of such a plurality of particles can e.g. be defined as a preparation, wherein

- said particles comprise an inorganic core surrounded by metal complexing linkers that are chemically linked to dispersant mol^¬ ecules,

- wherein the dispersant molecules (a) are at an average density of at least 1.1 dispersant molecules per nm² of the inorganic core surface, or any one as said above, and/or (b) form a shell of constant dispersant density and a further shell of gradually reduced dispersant density with increasing distance from the in^¬ organic core surface. The shell of constant dispersant density is an inner shell and the shell of decreasing density (with the radius or distance) is an outer shell. The inner shell can also be defined as a shell of similar density to the molten polymer state of the dispersant identifiable by small angle x-ray scat^¬ tering in a solvent, such as water for hydrophilic dispersants, and the outer shell can also be defined as a solvable shell, distinct from the dense inner polymer shell. This shell struc^¬ ture forms is solution when the outer shell may e.g. form an outreaching or brush like structure, usually heavily solvated, while the inner shell remains dense, possibly not or only resid- ually solvated without influencing its dense structure. The sol^¬ vent for the small angle x-ray scattering determination and for establishing the two-shell like structure should be a good sol^¬ vent for the dispersant, e.g. to lead to a solvation of the dis^¬ persant in the outer shell and not to a collapsing thereof. "Good solvent" relates to the excluded volume parameter of the dispersant under these conditions. E.g. a good solvent for a hy- drophilic dispersant is a hydrophilic solvent and a hydrophobic solvent for a hydrophobic dispersant.

Preferably the inorganic core is of an average size between 2 nm to 80 nm in diameter, or any one as said above. The parti^¬ cles of the preparation can be of homogenic size in said plural^¬ ity wherein the mean standard deviation of said average size is at most 10%, preferably at most 5%, even more preferred at most 2% of said average size, such as at most 0.8 nm, preferably at most 0.5 nm. This preparation can be obtainable by a method of the invention.

Further provided is a particle comprising an inorganic core surrounded by metal complexing linkers that are chemically linked to dispersant molecules, wherein the dispersant molecules

(a) are at an average density of at least 1.1, preferably at least 3.0, dispersant molecules per nm² of the inorganic core surface (or any number as said above), and/or (b) form a shell of constant dispersant density and a further shell of gradually reduced dispersant density with increasing distance from the in^¬ organic core surface or form a shell of similar density to the molten polymer state identifiable by small angle x-ray scatter^¬ ing in a solvent, e.g. water, and a solvable shell, distinct from the dense inner polymer shell - as said above for the par^¬ ticles of the preparation. This particle can be obtainable by a method of the invention.

The inventive particles having the high dispersant surface density surprisingly retain a shell having a molten state like appearance, illustrating the high density of this shell portion. Outside of this dense inner part of the shell where the density is similar to that of the molten polymer state, the dispersants have a similar structure as in previously reported particles

(Amstad et al . , supra, W02011 /026572 Al), e.g. in the case of PEG a spherical brush like structure, wherein the density de^¬ creases monotonously with distance from the particle core. In the dense molten state like shell the density of the dispersants is constant at maximum density in the entire distance range of this shell or shell portion from the core. This molten state shell is unique to the inventive particles illustrating the dif^¬ ference in density to previous particles - although and as dis- cussed above Amstad et al . , supra, W02011 /026572 Al reported high surface densities which was an artefact of calculation of surface area based on a spherical particle imprecision. This molten state shell of the inventive particles is visible in small angle x-ray scattering in water, e.g. as shown in example 10, and can be detected by this method. Preferably the molten state shell is calculated from X-ray data by the Daoud-Cotton model (as demonstrated in example 13) . The molten state like shell is preferably at least 0.3 nm thick, especially preferred at least 0.5 nm, at least 1 nm or at least 2 nm. In the molten state like shell solvents may be absent or at very low concen^¬ trations due to infiltrations. Solvent amounts are preferably below 40 % (w/v) , especially preferred below 20% (w/v) or below 10 % (w/v) . Such a solvent may be water in case of hydrophilic dispersants like PEG, polyoxazoline or PNIPAM.

The inventive particles or preparations, e.g. as produced by the inventive method, can be used for various applications, in^¬ cluding biomedical contrast agents, multifunctional nanoparti- cles for biomedical applications, nanoparticle building blocks for assembly of smart materials, therapeutic magnetic nanoparti- cles, magnetic labeling and separation. Such methods may comprise detection the inventive particles in a sample or separat^¬ ing the inventive particles within or from a sample volume, e.g. by applying a magnetic field and ordering the magnetic particles of the invention within such a field. A sample may be fluid com^¬ prising a complex mixture of various components. Further, such particles can be filler materials, stiffeners, or additives for improved conductivity, magnetic responsiveness or optical prop^¬ erties in polymer material applications.

The present invention is further illustrated by the follow^¬ ing examples without being limited to these embodiments of the present invention.

Figures :

Fig 1. Schematic representation of the direct ligand ex^¬ change method (top) and the two-step "grafting-to" synthetic procedure (down) for the synthesis of core-shell- nanoparticles .

Fig. 2. Transmission electron microscopy (TEM) images of 7.9± 0.4 nm large iron oxide nanoparticles (oleic acid ligands) . As can be seen at higher magnifications (right image) the parti- cles are monocrystalline .

Fig. 3. Exemplary depiction of the column purification (25cm Sephadex G75 superfine in Milli-Q water) results of a typical two-step synthesis (ALD-melt; Table 1, entry 15) . Fractions were collected (x-axis) every 5 mL . The yield after freeze-drying (left y-axis) and the grafting density (right y-axis) are shown for every fraction. The yields are calculated on the basis of pure Fe₃0₄ cores (inorganic fraction); Fr . 6 consists of pure PEG molecules (mass of organic weight: 99.8%), therefore the calcu^¬ lated grafting density is out of the shown range.

Fig. 4. NMR-spectra of nitro-dopamine (1) . Due to the high water content of the NMR-solvent the primary amine and the two hydroxyl-groups cannot be seen.

Fig. 5. Typical TEM image of the particles of the invention.

Fig. 6. Pebbles results of a TEM image, (a) TEM image as measured (b) pebbles automatically identifies and marks the na- noparticles and provides (c) the statistical evaluation of the size distribution of the particles in the respective TEM-image given by Pebbles.

Fig. 7. TEM-pictures of main fractions of an ALD-melt (Table 1, entry 15) . (a) fraction I shows few aggregates with small particle-particle distances; (b) Fraction III (main fraction) show no aggregation, no free PEG whereas in (c) (fraction V) one can immediately identify huge areas of free (unbound) PEG

Fig. 8. Example for the TGA-curves of the main fraction (here: ALD-melt, table 1, entry 15)

Fig. 9. Representation of the Daoud-Cotton blob model for a spherical brush with a blob radius of an individual blob. The transition radius between melt-like and unswollen behavior is indicated by rl=r2.

Fig. 10. Radial electron density distribution (corresponding to polymer density) in the core region (blue, left horizontal) predicted by the Daoud-Cotton model, the constant density region (melt like region) (purple, middle horizontal) and the r^~4/3 decay region (brown, right) . For reasons of visibility the ASLD is in^¬ creased to 0.5.

Fig. 11. SAXS from the cores (red, squares) and the fitted function (black) , inset showing TEM images of the particles, scalebar 10 nm.

Fig. 12. Squares: SAXS from a core-shell particle with a distinct shoulder from the shell (0.1 A^-1) and the scattering from the cores in the range from 0.23 to 0.42 A^-1. SAXS from core-shell particles with different concentrations (black squares, 5 mg/ml; red circles, 1 mg/ml and blue triangles, 0.5 mg/ml) .

Fig. 13. Scattering curve of the Core-Shell particle in MQ and the respective fit function as obtained by the core-chell density model built on the Daoud-Cotton blob model.

Fig. 14. Cyclic change of temperature of core-shell parti^¬ cles from 25°C (green circles) to 50°C (red triangles) and sub^¬ sequent cooling to 25°C (cyan squares) and the respective fitted functions

Fig. 15. Normalized fit functions for the shell behavior in cloudpoint buffer in the q region of the shell for 25°C (green, top at y-axis) to 50°C (red, bottom at y-axis) and subsequent cooling to 25°C (cyan, middle at y-axis) . The inset shows the complete q range.

Fig. 16. Left: Nanoparticles from the samples of Amstad et al . 2009, supra, reported to have 2.8 M/nm² grafting density us^¬ ing a direct grafting method. Right: Nanoparticles having the same or higher grafting density using the two-step melt grafting method of the invention, trimmed down to a maximum <1 M/nm² grafting density using the direct grafting method. Although both particles are similar, have the same PEG shell, behave identi^¬ cally and have about the same dispersant density determined by e.g. SAXS, the determined density values by TGA differ due to overrepresentation in the previous work.

Fig. 17. Results of column purification of a typical two- step synthesis (ALD-melt; Table 1, entry 15) . Fractions were collected every 5 mL (x-axis) . The yield after freeze-drying (red bars) calculated on the percentage of iron oxide cores in relation to the amount at the start of the synthesis and the ap^¬ parent average grafting density (blue bars) calculated on the total organic content are shown for each fraction. Fraction 3 is the product faction and Fraction 6 and Fraction 7 consist of al^¬ most pure PEG (bars truncated) .

Fig. 18. FTIR spectrum of the main product (ALD-melt, Frac^¬ tion 3; table 1; entry 15) showing the absence of OA and dex- tran, but the presence of NDA-PEG.

Fig. 19. DLS of NP hydrodynamic diameter during T-cycling in water and PBS for NPs grafted with PEG by one-step ligand ex^¬ change (table 1, entry 2; black and grey lines with closed and open triangles respectively) and by the two-step ALD-melt method (table 1, entry 15; blue and purple curves with closed and open squares respectively) . Average and standard deviations for at least 4 measurement series per type of sample are shown. In both aqueous environments only the ALD-melt grafted particles are colloidally stable over the temperature cycle. The aggregation of the one-step ligand exchange particles leads first to in^¬ creased average hydrodynamic size of the clusters, followed by precipitation and correlating loss of recorded size. The two- step, ALD-melt grafted particles show only reversible aggrega^¬ tion in PBS and no aggregation in water.

Fig. 20. A): Pictures of different particles (ALD-melt par^¬ ticles, Table 1, entry 15 and one-step grafting-to particles, Table 1, entry 2) dissolved in fetal calf serum (FCS) and heated to 75°C for 10 hours. B) : Effective hydrodynamic diameter meas^¬ ured as function of time by DLS on the serum solutions. The hy^¬ drodynamic diameter corresponding to the main number peak is shown. It shows a removal of the proteins from the ALD-melt sam^¬ ple (blue curve) and a removal of particles from the one-step grafting-to sample (black curve) .

Examples :

Example 1 : Reagents and materials

All used chemicals were purchased at Sigma-Aldrich and used without further purification.

PEG-ALD : 0- [ 2- ( 6-Oxocaproylamino) ethyl ] -0' -methylpolyethylene glycol 5Ό00 ("PEG-aldehyde 5000");

PEG-ACRY: Poly (ethylene glycol) methyl ether acrylate;

PEG-TOS: Poly (ethylene glycol) methyl ether tosylate;

PEG-NHS: Methoxypolyethylene glycol 5,000 acetic acid N- succinimidyl ester; purchased from JenKem-USA (Mn~5000; PDI=1.1; functionality of the endgroup >95%)

Carl Roth dialyzing membranes (regenerated cellulose) with a cut-off size of 5000 Da were used to purify the nitrodopamine modified NPs. Example 2 : Synthesis of nitro-dopamine and nitro-dopamine- PEG 5000)

Scheme 1: Synthetic pathway for the synthesis of nitrodopamine and the NHS-activated coupling to the PEG5000 using a C0MU^®((1- Cyano-2 -ethoxy-2 -oxoethylidenaminooxy) dimethy1amino-morpholino- carbenium hexafluorophosphate) -activated intermediate step. Ni- tro-dopamine (1) can be used in the two-step method

2.1 Synthesis of nitro-dopamine . Dopamine hydrochloride (20 g; 105 mmol) was dissolved in 600 mL water, sodium nitrite (25 g, 362 mmol) was added and the mixture was cooled to 0°C using an ice bath. 200 mL of a 20 v/v% sulfuric acid was slowly dropped into the reaction. After removing the ice-bath the mixture was stirred at room temperature overnight. The crude product was collected by filtering the obtained dispersion. After washing the product with ice-cold water the nitrodopamine hydrogensul- fate was dried in high vacuum and stored at 4°C until further use. NMR spectra of n-dopamine is shown in Fig. 4. Also analysed was pure nitrodopamine; a mass peak at 199.10 could be correlat^¬ ed to the nitrodopamine (m=198.06+l (proton) =199.06) . Higher mass-peaks could be assigned to polymeric structures of nitro^¬ dopamine and assoziatives with the matrix materials

2.2 Synthesis of MeO-PEG-nitro-dopamine . 1 g MeO-PEG-COOH (0.2 mmol) was dissolved in 5 mL DMF (headspace grade 99.99%, anhy^¬ drous) and cooled to 4°C on an ice bath. COMU^® (95 mg; 0.22 mmol) and IV-methylmorpholine (0.24 mL; 2.2 mmol) were added and the mixture was stirred for 2 hours. Afterwards a solution of nitro^¬ dopamine hydrogensulfate (53 mg; 0.18 mmol) in DMF (0.5 mL;

99.99%) was dropped in during 5 minutes. The reaction was stirred for 2 h at 4°C and 16 h at RT . The solution was poured into 45 mL ice cold acetone and centrifuged (5000 rpm, 4°C, 10 minutes) . The precipitate was washed 2 times with acetone and afterwards dissolved in 1 mL water. The crude product was dia- lyzed against Milli-Q-water for 48 hours. Freeze drying gave 1 g (0.19 mmol) of pure product (0.95 g; yield 95%).

Example 3 : Synthesis of monodisperse Fe₃0₄ nanoparticles .

To prepare monodisperse iron oxide nanoparticles [S.-J. Park et al., J. Am. Chem. Soc. 2000, 122, 8581-8582; T. Hyeon et al . , Journal of the American Chemical Society 2001, 123, 12798-12801; E. Kang et al . , Journal of Physical Chemistry B 2004, 108,

13932-13935; J. Park, et al . , Nat. Mater. 2004, 3, 891-895; J. Park, et al . , Angew Chem Int Ed Engl 2005, 44, 2873-2877; S. G. Kwon et al., J. Am. Chem. Soc. 2007, 129, 12571-12584; J. Park et al., Angew. Chem., Int. Ed. 2007, 46, 4630-4660; C. Cavelius et al., Crystal Growth and Design 2012, 12, 5948-5955.], Fe(CO)₅ (1 mL; 1.49 g; 7.4 mmol) was added to a mixture containing 25 mL of dioctylether and 3 mL oleic acid (10 mmol) at 100°C. The re^¬ sulting solution is heated to 290°C with a heating rate of 10 K/min (reflux) and kept at that temperature for 1 hour. To gain full reproducibility and control over the temperature during the nanoparticle synthesis a thermo controller LTR 3500/S from Juch- heim-Solingen was used. During this time the color changes from the initial orange to dark black. After cooling the mixture to room temperature the crude product was poured into 175 mL nitro^¬ gen-bubbled acetone. After centrifugation (5000 rpm, 10 min, 20°C) the black product was redispersed in a small amount of toluene (~1 mL) and reprecipitated in acetone (190 mL) . This centrifugation-redispersion-precipitation step was repeated 3 times overall. The resulting oleic acid protected iron oxide na^¬ noparticles were used immediately for further reactions. The size of the particles could be controlled by the ratio of the iron pentacarbonyl and the oleic acid as well as the heating rate during the synthesis. Particles with 3.9±0.3 nm to 4.4±3 nm have been used.

Example 4: Ligand-exchange with nitrodopamine-PEG5000-OMe (one- step-method) .

The oleic acid-covered nanoparticles (25 mg) were dispersed in 20 mL DMF (headspace grade 99.99%, anhydrous) and nitrodopa- mine-PEG5000 (example 2.2) (500mg; 1.7 mmol) was added. This so^¬ lution was purged with nitrogen for 2 minutes, put in an ultra^¬ sonic bath for 60 minutes, allowed to stand at room temperature for 15 hours, ultrasonicated again for 60 minutes and precipi^¬ tated in cold acetone (180 mL) . After centrifugation and 5 washing steps (redispersion in methanol - centrifugation with 5000 rpm, 10 minutes, 20°C) the obtained core-shell iron oxide NPs (FeOx-PEG) were characterized using TEM, TGA, DLS .

Example 5: Core-shell-nanoparticles via two-step grafting-to reaction

5.1 Ligand-exchange (oleic acid to nitrodopamine) . The oleic ac^¬ id-covered nanoparticles (~3 g - whole amount of previous syn^¬ thesis) were dispersed in 20 mL DMF (headspace grade 99.99%, an^¬ hydrous) and nitrodopamine hydrogensulfate (n-dopamine) (150 mg; 0.5 mmol) was added. This solution was purged with nitrogen for 2 minutes, put in an ultrasonic bath for 60 minutes, allowed to stand at room temperature for 15 hours, ultrasonicated again for 60 minutes and precipitated in pure acetone (180 mL, cold) . Af^¬ ter centrifugation and 5 washing steps (redispersion in methanol - centrifugation with 5000 rpm, 10 minutes, 20°C) the obtained nitro-dopamine modified iron oxide NPs (NP-NH₂) were used direct^¬ ly for grafting with PEG.

Scheme 2: Schematic representation of the different grafting-to reactions. Aldehyde reacts under condensation forming an imine, PEG-tosylate with nucleophilic-substitution to form a secondary amine, the NHS-activated coupling leads to a stable amide-bond and acrylates undergo a Michael-addition to form secondary or tertiary amines.

5.2 Synthesis of PEG-core-shell nanoparticles using grafting-to reaction in solution. The as-synthesized nitrodopamine nanopar^¬ ticles (50 mg) were dissolved in dry DMF (5 mL, headspace grade 99.99%, anhydrous), the respective PEG-derivate (500 mg, 1.7 mmol, NHS, TOS, ALD) was added and the mixture was stirred at room temperature for 16 hours. After pouring the reaction mixture into 195 mL cold acetone and centrifugation (5000 rpm, 10 minutes, 4°C) the precipitate was dissolved in water (2 mL) , filtered through a 0.45 ym syringe filter to remove aggregates and purified as described below.

5.3 Synthesis of PEG-core-shell nanoparticles using microwave assisted reactions. The as-synthesized nitrodopamine nanoparti^¬ cles (50 mg) were dissolved in dry methanol (3 mL) the respec^¬ tive PEG-derivate (500 mg; -1.7 mmol; NHS; ACRY; TOS; ALD) was added and this mixture was heated in a CEM lab microwave to 140°C for 90 minutes. After cooling to room temperature the re^¬ action mixture was poured into 190 mL cold acetone and centri- fuged for 10 minutes with 5000 rpm (4°C) . The precipitate was dissolved in water (2 mL) , filtered through a 0.45ym syringe filter to remove aggregates and purified as described below.

5.4 Synthesis of PEG-core-shell nanoparticles using polymer melts. The freshly synthesized nitrodopamine modified nanoparti^¬ cles (-50 mg) were roughly mixed with the dry PEG-derivate (TOS, NHS, ALD) (500 mg; 1.7 mmol), purged with nitrogen and slowly heated to 110°C. After the material is completely molten a ni^¬ trogen-flow (needle) is used for mixing the melt continuously. The reaction was kept at this temperature for 90 minutes. After cooling to room temperature the black solid was dissolved in wa^¬ ter (2 mL) (ultrasonic bath for 1 hour), filtered through a 0.45 ym syringe filter (RC) to remove aggregates and purified as de^¬ scribed below.

Example 6 : Oleylamine as surfactant

In previous examples, the as-synthesized NP core has a dense shell of oleic acid that was replaced by the stabilizing disper- sant. Synthesis of truly monodisperse (SD<5%) Fe304 core-shell NPs makes use of e.g. oleic acid ligands to control crystal growth. Similarly monodisperse, but not as spherical, NPs have been synthesized using oleylamine instead of oleic acid. Oleyla^¬ mine has the advantage that it can more easily be removed from the NP surface compared to oleic acid, due to its weaker bind^¬ ing. Ligand replacement of oleylamine by nitroDOPA-PEG therefore has been demonstrated previously (Hyeon et al . , J. Am. Chem.

Soc, 2001, 123, 12798-12801). However, these NPs possessed -5 times lower PEG grafting density (-0.5 chains/nm²) and thereby significantly lower colloidal stability than NPs that were syn^¬ thesized without capping agents and grafted with nitroDOPA-PEG . The low grafting efficiency is the result of simultaneously try^¬ ing to fulfill several mutually contradicting conditions during the ligand replacement reaction: (1) to dissolve the capping agent (oleic acid), (2) to solubilize the dispersant, (3) to keep the dispersant at low coil size, quantitatively described by e.g. R_Gr which determines the grafting footprint, and (4) to provide the right conditions (protonation) of the anchor group to irreversibly bind to the core.

Chemicals were used as in example 1. PEGYlated core-shell nanoparticles were produced as in example 5.4.

Example 7: Purification of PEG-iron oxide core-shell nanoparticles .

The dark brown/black solution is applied onto a 25 cm long (3 cm diameter) column filled with Sephadex G75 (milli-Q water) . Pure water was used for the purification. Fractions of 5 mL were collected, after filtering the single fractions through 0.45 ym syringe-filters and freeze drying the products were character^¬ ized using TEM, TGA and DLS .

Example 8: Characterization of core-shell nanoparticles

8.1 Characterization of the core-shell nanoparticles using TEM.

A small amount of product was dissolved in water (nitrodopamine and PEG-modified particles) or toluene (oleic acid modified na^¬ noparticles) and dropped onto the TEM-grid (3.05 mm HR-TEM-grid, copper 300 mesh, carbon film, Gropl Austria) . All pictures were taken using a FEI TECNAI at 200kV. A typical TEM image is shown in Fig. 5.

The statistical size distribution was obtained by the use of the freeware Pebbles (http://pebbles.istm.cnr.it). Pebbles results of a TEM image: (a) TEM image as measured (b) pebbles automati^¬ cally identifies and marks the nanoparticles and provides (c) the statistical evaluation of the size distribution of the par^¬ ticles in the respective TEM-image given by Pebbles. TEM- pictures of main fractions of an ALD-melt are shown in Fig. 7.

8.2 Characterization of the core-shell nanoparticles using TGA. 1-3 mg of the respective sample were weighed into 70μ1 A10_x-cups and measured on a Mettler-Toledo TGA/DSC 1. The samples were measured in a constant flow of synthetic air (80 mL/min plus 20 mL nitrogen stream as protection gas for the balance) with a heating rate of 10 K/min. Analysis was performed using the Met^¬ tler-Toledo software (simple step-function from 150-400 °C) . A TGA-curve is shown in Fig. 8 for an ALD-melt. The grafting density and parameters for the ALD melt are:

ALD-melt :

organic content: 88.6%

r= 2.0 nm

M= 5200 g/mol

Density (PEG) = 1.12g/cm³

Density (ironoxide)= 5.24 g/cm³

V(NP)=33.5 nm³

Mass (1 NP)= 1.8xl0^"19 g/NP (core)

0.885 g PEG per gram product

1.7*10^~7 mol PEG

l.OxlO¹⁷ molecules PEG

0.115 mg ironoxide per gram product

6.5*10¹⁴ NP/g

3.1 molecules PEG /nm²

8.3 Characterization of the core-shell nanoparticles using DLS.

The count-rate and hydrodynamic size of core-shell nanoparticles redispersed in Milli-Q water were measured before and after tem^¬ perature cycling (20-90 °C) to assess the colloidal stability and the stability of the surface coating [E. Amstad et al . , Nano Letters 2009, 9, 4042-4048.].

8.4 Characterization of the core-shell nanoparticles using ATR- FTIR

Mid-IR powder spectra of the lyophilized samples were col^¬ lected using a Bruker Tensor 37 FTIR spectrometer with a Bruker Platinum Diamond single reflection ATR equipment at a resolution of 4 cm^-1 by averaging 32 scans.

8.5 Characterization of core-shell nanoparticle hydrodynamic size and tests of colloidal stability of nanoparticles in etha- nol and PBS

The different iron oxide NPs (one-step ligand exchange par^¬ ticles: Table 1, entry 2 and ALD-melt grafting-to particles: Ta^¬ ble 1, entry 15) were dissolved in water (0.5 mg/mL) and placed into a Malvern Zetasizer Dynamic Light Scattering (DLS) instru^¬ ment. After equilibrating at 25°C the hydrodynamic size distri^¬ bution was measured every hour using the built in fitting based on the CONTIN algorithm.

After 2 hours of recorded size stability in water, 10 weight% of ethanol was added to the cuvette, which was shaken intensely for 3 seconds, and placed back into the instrument. Measurements of the hydrodynamic size were performed for 8 more hours .

The stability of the NPs in phosphate buffered saline (PBS) buffer was tested by dissolving 0.5 mg of the ALD-melt grafted NPs (table 1, entry 15) and the one-step ligand exchange NPs (Table 1, entry 2) in water and PBS. The cuvettes were placed in the DLS instrument, heated to 85°C and data was collected every hour for 15 hours.

8.6 Test of colloidal stability of NPs upon heat treatment in serum

1 mL of fetal calf serum (FCS) was placed into DLS cuvettes. 0.5 mg of either the ALD-melt NPs (Table 1, entry 15) or the one-step ligand exchange NPs (Table 1, entry 2) was added to the FCS and the cuvette was gently shaken until the solution was ho^¬ mogeneous. The hydrodynamic size was thereafter measured every hour for 10 hours by DLS.

Example 9: Purification and characterization of core-shell iron- oxide-nanopartides

All PEGylated samples were dissolved in a small amount of water (2 mL of water per g of raw product) and added onto a 20- 25cm long column of Sephadex G75 superfine (3cm diameter) . During the separation process a high percentage of the nanoparti- cles stick to the Sephadex material - this lost fraction is con^¬ sidered as the unmodified (nitrodopamine-modified) and aggregat^¬ ed parts of the raw product (See TEM-analytics of single frac^¬ tions) . The fractions were collected manually, filtered through a 0.45ym syringe filter and freeze dryed to obtain the products.

Example 10: Purification and Yield

Figure 17 shows the yield and the estimated grafting density of each sample fraction collected after the Sephadex column pu^¬ rification for NPs grafted by ALD-melt under bubbling with nitrogen (Table 1, entry 15) . The yield is calculated as the per^¬ cent of iron oxide cores found in each collected fraction in re- lation to the total amount of iron oxide cores at the start of the synthesis. The grafting density is calculated from the total organic content (TOC) fraction measured by TGA for each collect^¬ ed 5mL fraction. This apparent grafting density only corresponds to a real average grafting density for sample fractions contain^¬ ing iron oxide cores and no free PEG.

A high grafting density fraction (Fraction 3) with relatively high yield compared to other fractions was selected as the product. We emphasize that yields significantly lower than 100% after such demanding purification are acceptable given the high- end applications, and that the yield can be "increased" by en^¬ larging the selection of the fraction in line with the demands of the application. The selected Fraction 3 with an average grafting density of 3.1 chains/nm² and average yield of up to 35% shows individual, well separated iron oxide cores inspected by TEM (Fig. 7b) . Aggregated cores with <0.2 chains/nm² were col^¬ lected before the main peak (Fr. 1-2), as evidenced by the sam^¬ ple TEM in Fig. 7a. Free PEG was collected in the final frac^¬ tion (s), well separated from the product fraction. Fraction 5 already shows free PEG in the background when investigated by TEM, with only few visible cores (Fig. 7c); for Fr . 6 and above essentially no particles can be imaged and only PEG is found in the sample. The much smaller and flexible PEG(5kDa) is trapped meandering for much longer time in the small-porous Sephadex G75 column, which makes size-exclusion chromatography a very efficient method for separation of densely grafted NPs from free polymer. Repeated column passes does not change the observed TOC measured by TGA, while large amounts of free polymer are removed during the first pass.

Grafting in a melt was first performed by slowly heating the NPs dispersed in pure PEG to 110°C. A flow of nitrogen was used for continuous mixing at constant temperature for 90 min. The best fractions using this method (Table 1, entries 12-15) far surpass the other methods in terms of grafting density. Grafting densities of 2-3 chains/nm² were consistently achieved. This is ~25% below the ~4 NDA/nm² possible grafting sites and at least 2 times higher than for the other grafting methods. We emphasize again (Fig. 17) that any free PEG is well separated from the NPs by column purification and that passes on the column did not change the measured grafting density. FTIR spectroscopy demon- strated that the OA had been fully replaced and the true NDA-PEG grafting density had been measured for the ALD-PEG product frac^¬ tion, i.e. Fr. 3 (Figure 18) .

The two-step melt graft-to method was also investigated on 6.4nm-core particles. The grafting densities were within the standard deviation equal and in fact near identical to those on 3.8nm cores for PEG-TOS and PEG-ALD, while no sample of PEG-ACRY was produced that could be purified by column. Also the obtained yields were similar.

Example 11: Optimization of the melt grafting method to improve the yield)

Optimally, both grafting density and yield should be high for a successful NP surface modification protocol. Although the MW-assisted TOS-PEG (Table 1, entry 9) and ACRY-PEG melt reac^¬ tions (table 1, entry 14) lead to high grafting densities (1.6 and 2.2 chains/nm²) the yields are low compared to the melt- grafted ALD-PEG (3.1 chains/nm²) . In both cases the yield is <l/3 of that for ALD-PEG. The yields were very similar, but on aver^¬ age slightly lower for 6.4-nm-core particles. This could reflect that a good dispersion of NP cores in the polymer melt is more difficult to achieve for larger cores. More efficient bubbling using inert gas could potentially improve the mixing and thereby the yield.

Bubbling with inert gas also serves a second purpose, which is to remove water that is produced during the reaction. We therefore also attempted the synthesis under vacuum suction to more efficiently achieve continuous removal of water. Thus, the NDA-coated particles were dispersed in a small amount of dry THF (lmL/50mg particles) and added to the ALD-PEG, followed by slow drying in vacuum. Subsequently, the mixture was heated under continuous vacuum suction to conduct grafting under melt condi^¬ tion. Thereby, the yield of the ALD-melt could be improved from ~16% to ~35% (Table 1, entry 16) . Although the yield was consist^¬ ently higher with the improved procedure, the high variability in the yield suggests that further optimization of the reactor vessel environment to improve NP dispersion and water removal can be made and thereby the yield could further be increased significantly. Example 12: Results and best mode

The NPs were synthesized by an optimized heat-up process leading to Fe₃0₄ NPs (a) with well-defined size (adjustable be^¬ tween 3.5-9 nm in diameter; PDI≤1.05); (b) highly spherical and (c) monocrystalline (see figure 2) . All grafting experiments were done with particles in the size range between 3.810.3 to 4.210.3 nm in diameter NPs to exclude size or curvature effects.

All grafting methods were performed with an excess of free polymer to ensure that the maximum possible grafting density could be achieved. A key step to characterize polymer-grafted core-shell NPs is the removal of residual or weakly bound poly^¬ mer from the sample. Previous works indicate that magnetic de- cantation, dialysis and centrifugation all lead to substantial amounts of residual PEG even after multiple applications, while size exclusion chromatography can separate free PEG from grafted particles. Dextran column chromatography can also be used to evaluate the colloidal stability since too sparsely grafted NPs have high affinity for dextran chromatography columns; insuffi^¬ ciently densely grafted particles therefore aggregate to the column material. All samples were therefore purified using a 20- 25 cm long and 3 cm broad Sephadex G75 filled column. This length was chosen as it generated several distinguishable frac^¬ tions for further analysis. Fractions were separated every 5mL . Samples obtained by the direct ligand exchange of oleic acid for nitrodopamine-PEG did not pass columns longer than a few cm.

These samples were therefore instead purified using multiple centrifugation in methanol/acetone mixtures and using magnetic extraction to get rid of the excess free PEG; residual PEG could be present in the samples even after purification leading to overestimation of the grafting density and was therefore thor^¬ oughly removed.

The success of the grafting methods can be compared on a set of dependent criteria such as grafting density, resulting col^¬ loidal stability, cost and ease/time of synthesis. The colloidal stability, and therefore the suitability for biomedical applica^¬ tions, is believed to strongly relate to stable and high density of PEG grafting; thus, the grafting density was our primary concern and was determined by TGA for each grafting method and fraction .

Previously, direct ligand exchange of the oleic acid with the anchor-group modified PEG has been used; such ligand re^¬ placement requires solvents, e.g. DMF, that strike a compromise between the solubility and R_G (solvated coil size) of the PEG and oleic acid. The R_G of PEG in DMF is above its theoretical mini^¬ mum, which limits the highest obtainable PEG grafting density (Table 1, entry 2) . The resulting expansion of the PEG coil size above its minimum limits the highest obtainable PEG-molecule grafting density (table 1, entry 2) . The achieved density

0.510.2 chains/nm² after excess PEG removal is lower than required for column purification and lower than previously reported to be necessary to achieve non-aggregating particles under high salt, protein or temperature conditions, although due to very different purification and testing conditions the litera^¬ ture is not conclusive on what minimal grafting density is re^¬ quired .

Table 1. Overview of the calculated grafting densities of

PEG(5kD) obtained by different grafting-to reactions evaluated by TGA after Sephadex G75 column purification

Grafting

Yield

Entry reactant Method density

[M/nm2] [%]

1 FeO_x-oleic acid 17.415 [f]

one step grafting- ^■to

2 n-Dopamine- -PEG5000 Solution (DMF) 0.510.2 [d]

two step grafting- ^■to procedure

3 Fe₃0₄-n-Dopamine (step 1) 3.811.2

4 PEG-NHS Solution [a] 0.9+0.2

5 PEG-TOS Solution [a] 0.8+1.1

6 PEG-ACRY Solution [a] [d]

0.8+0.3;

7 PEG-ALD Solution [a]

1.2+0.6 [e]

8 PEG-NHS Microwave [b] 1.0+0.3

9 PEG-TOS Microwave [b] 1.6+0.8 <5

10 PEG-ACRY Microwave [b] 1.3+ 1 7 +3

11 PEG-ALD Microwave [b] [d]

12 PEG-NHS me1t [ c ] [d]

13 PEG-TOS me1t [ c ] 1.1+0.5 [g] <5

14 PEG-ACRY me1t [ c ] 2.2+0.4 [g] 8 +5

15 PEG-ALD me1t [ c ] 3.1+0.9 [g] 16+5

16 PEG-ALD melt [h] 3.1+0.6 35+ 15

[a] 50 mg Fe₃0₄-n-dopamine particles , 500 mg PEG-X, 3 mL DMF, 16 h at RT; [b] 50 mg Fe₃0₄-n-dopamine particles, 500 mg PEG-X, 3 mL MeOH, 90 min at 120°C; [c] 50 mg Fe₃0₄-n-dopamine particles, 500 mg PEG-X, N₂-stream, 90 min at 110 °C; [d] sample did not pass the sephadex column, sample purified by 3 precipitation steps in acetone; [e] yield <10%, higher grafting densities and yields could be obtained by adding NaCNBH₃ as reduction agent; [f] large amount of free (unbound) oleic acid present; [g] measured iden^¬ tical also for 3.8 nm and 6.4 nm particles; [h] melt was con^¬ ducted under vacuum (4-8 mbar) and rotated with 100 rpm on a ro^¬ tary evaporator

To increase the maximal density of surface grafted polymer it is necessary to change to a two-step grafting approach.

First, the oleic acid is replaced completely by nitrodopamine . This reaction can be carried out in an ideal solvent for oleic acid and nitro-dopamine, e.g. DMF, and results in ~4 M/nm² (table 1 , entry 3 ) .

Several different methods for binding end-group modified PEG to the free amine-group of the densely nitrodopamine modified iron oxide NPs can be envisioned, of which we choose to test: a) the tosylate-group reacting with primary amines under a S_N2 reac^¬ tion; b) the acrylate-group enabling a Michael-addition; c) the aldehyde-group enabling condensation (Mannich-reaction) ; and d) NHS activated acid enables the stable formation of an amide. Our goal was to obtain the highest PEG grafting density; given the applications, yields as low as <10% of the Fe₃0₄ cores after col^¬ umn purification are therefore acceptable. For polydisperse and amorphously shaped NPs we have previously shown that a PEG(5kD) grafting density of >2 chains/nm² nonmonodisperse, nonspherical nanoparticles is required for short Sephadex column purification [Amstad et al . , 2009, supra] . 25cm long Sephadex G75 column made it possible to separate and analyze several fractions to identi^¬ fy the highest grafting densities and their relative yields.

Fig. 3 shows an example of such fractionations of an ALD-melt modification (Table 1, entry 15) . A high grafting density fraction (Fr. 3) with 5-10% yield was selected as the product frac^¬ tion. Before the main peak aggregated cores with <0.2 chains/nm² were collected. After the product fraction free PEG was collect^¬ ed. The fraction with the highest average ligand density meas^¬ ured by TGA after column purification for three repetitions of each synthetic protocol is given in Table 1.

Table 1 shows that the coupling of NHS-PEG to the nitrodopa^¬ mine modified particles in DMF slightly improves the grafting density over direct ligand displacement to ~1 chain/nm², which is sufficient to enable column purification. The other reactive end-group PEGs were also attempted in DMF and resulted in simi^¬ lar grafting densities.

The same reactions were conducted in a MW reactor (Table 1, entry 8-11) . The MW-induced high heating of the NP-surface could speed up the grafting-to reactions to enable performing them at much lower solvation using MeOH; by this, the foot print of the PEG is reduced and the grafting density potentially increased. However, the MW-assisted reactions only insignificantly improved the PEG density.

The main limitation of the above strategies is that the footprint of the PEG-coil during grafting is low but not mini^¬ mized. The reaction efficiency can be increased at the same time as the polymer footprint is minimized by choosing a reaction that performs at an elevated temperature at which PEG(5kD) is in a melt and therefore retains chains mobility. The best fractions using this melt grafting method at 110°C (Table 1, entries 12-15) far surpass the other methods in terms of grafting density.

Grafting densities of 2-3 chains/nm² were consistently achieved, which is only slightly below, 82% of, the initial ~4 nitrodopa- mine/nm². This is at least a factor of 2 higher than for similar yields of the other standard grafting-to methods. Given differ^¬ ent purification and measurement methods employed in previous studies it is difficult to compare NP grafting densities in ab^¬ solute terms; however, these densities are higher than the high^¬ est reported for grafting-to onto shape and size polydisperse bare Fe₃0₄ NPs of ~2.5 chains/nm² [Amstad et al . , 2009, supra] . The amorphous shape in this work leads to underestimation of the true surface area and increases surface accessibility; this can explain the high obtained grafting density compared to our con^¬ trol under the same conditions. As we directly and consistently compare the solvent, MW and melt grafting-to approaches, we can clearly state that the latter is superior in a two-step graft^¬ ing-to approach by at least a factor of 2 and to ligand replace^¬ ment by a factor >4, which we could not quantitate due to that the ligand replacement particles did not pass the column used for purification.

The highly grafted monodisperse NPs demonstrated remarkable redispersion compared to previously observed samples with lower grafting density. The powder hydrated and redispersed directly upon addition of smallest amounts of water, indicating that the cores even in the dried state do not come in close proximity to each other. The supreme colloidal stability was further demon^¬ strated by high stability in water over prolonged period of time (1 year) and by the possibility to repeatedly filter through a 0.45ym syringe filter without loss of material. The absence of aggregation under temperature cycling, which produces aggregation of reversible and/or sparsely grafted NPs, was investigated by filtering through a 0.45ym syringe filter at a successive row of temperatures (25—»75—»25 °C; 10°C steps) . As expected the ALD- melt grafted particles were in all tests superior to the lower grafted particles by showing no aggregation or loss of material, whereas all the other grafting methods led to residues in the syringe filter.

For the overall rating of the performances of different grafting methods in addition to the grafting densities, the yields have to be taken into account. Although the microwave as^¬ sisted PEG-TOS grafting to reaction (Table 1, entry 9) and PEG- ACRY melt reaction (table 1, entry 14) also lead to very high grafting densities (1.6 and 2.2 M/nm² respectively) the yields were very low in both cases (<3% calculated on the basis of ironoxide cores) .

In summary, we have introduced a melt-based approach to densely graft sterically stabilizing shells to monodisperse, spherical superparamagnetic Fe₃0₄ NPs making them suitable for e.g. biomedical applications. Achieving colloidal stability of such NPs which require ligand replacement has hitherto been a bottleneck. A two-step grafting-to method from pure melt of ALD- PEG offered a superior and straightforward alternative to achieve this at sufficient yield (up to 10%) . The obtained grafting density and corresponding colloidal properties signifi^¬ cantly surpass those of previously presented methods. By ap^¬ proaching grafting of the maximum density of possible grafting sites, a brush density profile of the shell of the core-shell nanoparticles with unique structure (referred herein as "molten like state") and unique properties is obtained.

Example 13: Analysis of dispersant shell by small angle x-ray scattering (SAXS) 13.1 Sample preparation and SAXS measurements

The purified (example 9) Core-ALD-Shell particles were di^¬ luted in MQ water with a known concentration of 5 mg/ml. For the dilution series, concentrations of 5, 1 and 0.5 mg/ml were pre^¬ pared from the stock solution. The cloud point buffer measure^¬ ments were carried out in a solution containing 0.5 M NaCl, 0.5 M K2S04 and 0.08 M MOPS with a core-shell particle concentration of 5.43 mg/ml with temperature steps of 25, 40, 50 and 60 °C. For the SAXS experiments the solutions were prepared in type 0500 glass capillaries supplied by Hilgenberg with a nominal di^¬ ameter of 1 mm and a wall thickness of 10 ym. They were flame- sealed to rule out contamination and finished with a droplet of epoxy resin to avoid any evaporation during the course of meas^¬ urement .

The Fe₃0₄ cores were prepared as a dry powder between two layers of Scotch tape and measured for 60 s.

Measurements were carried out using a Rigaku S-Max 3000 SAXS system equipped with a copper-target micro focus X-ray tube Mi- croMax-002+ (45kV, 0.88mA), collimated through three pinholes (400, 200 and 700 ym) to achieve a beam diameter at the sample position of 280 ym (FWHM) and a Triton 200 2D multi wire gas- filled x-ray detector (200 mm diameter of active area, spatial resolution 200 ym) . Data was acquired in the q-range from 0.01 to 0.95 A^-1 with a measurement time of 28800 s for each scatter^¬ ing pattern at vacuum conditions better than 10^~2 mbar. In-situ heating experiments were carried out with the aid of a Linkam HFSX 350 heating stage. Subsequent data manipulation included background correction based on the measured transmission and ra^¬ dial integration with the SaxsGui 2.8.03 software package. In cases of large deviation in the capillary dimensions (+/- 5%) the transmission was normalized with the absorption of water to an equivalent nominal thickness. Therefore the outer diameter of the capillaries was measured using a light microscope (Leica DM 4000 M) . The thickness of the walls was equal and therefore needed no correction.

13.2 SAXS data fitting model

A theoretical model of the shell morphology was developed based on a mean field statistical approach established by Daoud & Cotton for polymer brushes. According to the Daoud - Cotton model, the brush can be represented by a sequence of concentric close packed blobs. Within every blob, chains behave as if it would be free, hence following the free chain scaling laws.

Based on this theoretical approach, the local monomer concentra^¬ tion can be computed and estimated as a function of the distance r from the grafting point. The shell of each coated magnetite core can be described as a star polymer with f branches, each one consisting of N statistical units of lengths 1, where 1 is taken to be half of the Kuhn length. Due to the spherical sym^¬ metry of the system, the polymer brush covering the colloid is represented as a sequence of concentric blobs, of radius ξ( τ ) . Each blob contains a certain number of statistical unit lengths, which within every blob behaves as an insulated polymer. Given the spherical symmetry, blobs are contained in cones; the closer to the surface the smaller the radius of the blob. Since blobs are close packed, by purely geometrical consideration the radius ξ( τ ) of the blob scales with the number of branches f and the distance from the grafting point r as:

The closer the blobs are to the grafting surface, the smaller their radius. It is possible to distinguish three distinct sta^¬ tistical regions. The first region is characterized by a constant density of monomers. In this region, the grafted chains are so closely packed that the monomers behave as if they were in a melt. Originally called core region, the term melt re^¬ gion or melt like region is used to avoid confusion with the iron oxide cores used as the grafting surface.

The regimen ri<r<r₂ is termed the "unswollen" region. Here the radius ξ( τ ) of a blob is smaller than that of a thermal blob and the polymer cannot behave as an ideal, self-avoiding chain but are affected by the solvent quality that find its expression in the excluded volume parameter via the Flory interaction parameter .

The second demarcation distance r₂ is given by r₂=l*v * f^1/2 with ^y'=v/l³ being the dimensionless excluded volume parameter. For distances r > r₂, the radius of the blob ξ( τ ) is bigger than that of a thermal blob. The r>r₂ region is the so called "swol^¬ len" region; here the blobs follow a self-avoiding polymer statistics and their radius scales with the number of monomers in it to the power 3/5: For the whole polymer shell with the distinct regimens the con^¬ centration of monomers within the blobs is given by:

The excluded volume parameter ^v for PEG in good solvent is one, therefore in good solvent and hence the intermediate regimen vanishes. For r>r₂ the number n(r) of statistical seg^¬ ments within a blob of radius ('^') is smaller than those in the thermal blob and within each blob the chains behave as if they were ideal chains. In this region, the concentration follows the usual star polymer r^~4/3 decay. A graphical representation of those distinct regimens is given in Fig. 9.

This behaviour can easily be transferred into an electron density profile as shown in Fig. 10. Note that the scattering length density (electron density) of the shell is not repre^¬ sentative of that in the real core-shell particles used here but greatly exaggerated for reasons of better visibility.

As the x-ray scattering is given by the Fourier transformation of the scattering length density variations encountered in a system, the form factor F(q) of a core-shell particle can be obtained f °° sitif or

F q 1 = in I SLD(r) -^- dtr.q

J„ ^' qr

With ^SLD(^r) being the scattering length density profile of the particle and q being the scattering vector can be calculated for a wavelength λ of the incident x-rays and an angle Θ of the scattered waves as:

The scattering length density in SAXS is intimately related to the monomer density c by the coherent scattering length b_COh

SLD = c *

The coherent scattering length b_COh is proportional to the cumula- tive atomic number Z of the monomer and the classical electron radius .

If core size and the polymer grafting density are known, one can directly calculate the monomer concentration, the scattering length density as well as the scattering length density contrast and compare the numbers to the SAXS data.

The monomer density in the shell is calculated as follows:

As a first step, we obtain f, the number of PEG molecules at^¬ tached to the core surface calculated grafting density p_graft and the surface of the core to obtain

A monomer concentration c(r) varies with the radial distance to the center of the particle. The total number of monomers in the shell, n_m0no is given by the integral of c(r) over the shell thickness and equals the product of the number of polymer chains f and the chain length N:

n„„_KC, = 4ΤΓ J c(r) < r² * dr— ,V * /

In the case of a density decay with r^~4/3 as predicted for star polymers, it follows that:

By integration and solution of the equation the monomer concen tration near the core c (Rc) can be expressed as follows

3 5 » Λ' * ^*

In case of a high grafting density (large number of chains f ) , the Daoud-Cotton model predicts a constant density region up to a radius r<rl = l*f 1/2 'W - < (<t) as shown in Fig.10 which is added to the star polymer decay. To be of practical relevance, the melt-density regimen has to exceed the core radius. The neces^¬ sary number of chains f attached to the core can be calculated as :

v» rl ,

/ = (^-)²

The monomer concentration between the core and rl=r2 is then given by

3 5 * N * Thereafter it decays with r^~4/3 as for star polymers.

As mentioned above, the monomer density can be directly used to calculate the scattering length density. In practice, it has to be considered that the SAXS signal will depend on the scattering length density contrast of core and shell with respect to the solvent. Therefore a relative scattering length density of the shell compared to that of the core (always with respect to the solvent water) is defined. It is a measure of how much the shell contributes to the SAXS signal and therefore allows an estimate on the visibility of the shell

^SLDSWI~^SLDHZn

ASLD =^■

SLD — SLD

The model was implemented in a Mathematica 9 script and fit^¬ ted to the data in a q range from 0.045 to 0.49 A^-1, using the free variables of particle diameter, ASLD and a scaling parame^¬ ter. The value for the core radius was fixed to the values de^¬ termined separately by the TEM measurements. The value for the constant density radius was calculated based on the morphology of the shell. Fitting was carried out using a Levenberg- Marquardt algorithm minimizing fit residuals and evaluated by means of the coefficient of determination R2 for the whole fit and the estimated standard errors for each fit parameter.

In order to carry out the quantitative modelling of the core-shell size, the SLD parameters haven been determined based on the set of equations a ASLD of 0.025 is calculated as the starting value of the fit parameters together with an estimate of 8 nm for the particle radius R_P.

It has to be pointed out that the model does not take ac^¬ count any effects of instrumental resolution and polydispersity. Albeit being low, a certain "smearing" of the data is inevita^¬ ble. As no dedicated variable could be defined for this, the ASLD, which is also affected by polydispersity and instrumental resolution, was effectively used as a fit parameter in the cur^¬ rent model.

Fig. 11 shows the scattering from the dry cores (red

squares) with the fit-function (black line) for a sphere, with gaussian size distribution and the structure factor for hard spheres (black line) . The diameter was determined to be 4.6 nm with a polydispersity of 17 %. The same function was also ap^¬ plied to q> 0.14 A^-1 region of the core-shell particles in order to fit the size and polydispersity of the core signal alone.

Table 2 Core diameter (D) and polydispersity (SD) as established by SAXS and TEM

Table 2 gives a comparison of the different methods of size de^¬ termination and their respective results. It can be seen that all methods are in good agreement with each other to within ex^¬ perimental uncertainty. The Pebbles evaluated TEM diameters are in good agreement with the diameter determined by SAXS. The man^¬ ual analysis of the cores in the TEM images yields systematical^¬ ly smaller diameters which may owe to the fuzzy interface area between cores and the grey scale fitting algorithm of Pebbles that utilizes the sphericity of the nanoparticles to obtain a more precise size fit to the image data.

Nevertheless, a good agreement between TEM and SAXS methods, in particular with Pebbles fitted data, can be seen which justifies the direct application of TEM values for the modelling of the core-shell particle scattering.

TGA, grafting density, calculation of SLD.

From eq. 1 the monomer density in the dense melt like region at the core can be found to be 1.18*10²² / cm³, which seems rea^¬ sonable in comparison to 1.55 *10²² / cm³ as calculated for a PEG melt .

The scattering density of the PEG shell at the core follows from this as 7.85*10¹⁰ cm^-2. With a scattering length density of 4.11*10^1:L cm^"2 for Magnetite and 9.43*10¹⁰ cm^"2 for H₂0, ASLD for the model follows as 0.025 in comparison to 0.033 for the pure melt .

13.3 Shell

The thickness of the PEG polymer shells grafted to nanopar^¬ ticles is highly dependent on the solvent, which is crucial to their functionality in biological environments. Measurements were carried out in MQ water and buffer solutions. Figure 12 shows a typical scattering curve from a core-shell particle in water and exhibits two distinct features. One in the larger q range from 0.23 to 0.42 A^-1, which can be attributed to the core signal as in Figure 11 and a shoulder with a weak peak in the region of 0.1 A^"1, where 0.1 A^"1 in the reciprocal space corre^¬ sponds to 6.3 nm in real space.

A dilution series was carried out to check on possible par^¬ ticle-particle correlations.

The scattering curves in Figure 12 depict the scattering from particles with decreasing concentration. The shoulder around 0.12 A^-1 keeps its position and the whole signal decreases with decreasing concentration. It is therefore concluded that the peak at 0.12 A^-1 cannot be attributed to a concentration de^¬ pendent structure factor but that it is due to permanent struc^¬ tural features such as the shell. It should be noted that a structural peak at 0.12 A^-1 would correspond to an interparticle distance of 5.2 nm and could therefore only be caused by core- core interaction. Therefore the absence of a concentration de^¬ pendent structure peak in the q-range corresponding to the core diameter is a good indicator for the integrity of the PEG shell effectively preventing core agglomeration.

13.4. Size and scattering contrast of the PEG shell

Having ruled out a concentration dependent structure peak, we attribute the shoulder around 0.12 A^-1 to the PEG shell. The scattering curve can thus be used to fit the size of the shell. Figure 13 displays the scattering data of the core-shell parti^¬ cles in MQ and the respective fit with obtained R_p of 8.2 nm. The fit is consistent with a constant density of the core, a high- density region of PEG close to the core and a decrease of PEG density according to r^~4/3 in the outer part of the shell as pre^¬ dicted for star polymers in a good solvent. The monomer density at the core is found to be 1.18 * 10²²cm^~3 which compares well to the 1.55 *10²² cm^"2 in the PEG melt.

The actual curve fitting gave a value for the SLD of 0.015 (Table 3) . This is lower than the theoretical estimate Of 0.025 calculated for the particles and therefore reasonable but still reasonable as it is not only influenced by the actual confor^¬ mation of the brush but as well by experimental factors or as- sumptions of the underlying models such as homogenous hydration of the shell or the monodispersity of the shell.

Furthermore, this consistent curve fitting yielded a value for r₂ of ~4.6nm. The use of a traditional spherical brush Gauss^¬ ian or parabolic density profile did not produce a good fit to the data. The fitted value of r₂=4.6nm means a constant density, melt like region, with thickness of more than 2, almost 3 nm for the small cores used in the experiment. For larger nanoparticles with lower curvature, the thickness of this region would in^¬ crease at the same grafting density.

Generally speaking, the core-shell particles exhibit a re^¬ markable monodispersity both for core and shell features which justifies the approach used for modelling here. Substantial pol- ydispersity would smear out the features to an extent not ob^¬ served in the investigated nanoparticle system; indeed, it would prohibit the performed fit analysis.

13.5. Cloud point buffer experiments

In order to test the hypothesis of dumbbell formation as an origin for the shoulder/peak at 0.12 A^-1, the swelling behavior of the PEG shell in cloud-point buffers was monitored by SAXS measurement of PEG coated particles in cloud-point buffers. When polymer coated nanoparticles are dispersed in buffer solutions, temperatures close to the cloud point of the buffer should re^¬ sult in a collapse of the shell visible in the SAXS curve, whereas dumbbells would not be affected by temperature changes. A cyclic temperature program was chosen with an initial measure^¬ ment at 25°C, one point at elevated temperature in the vicinity of the cloud point (50°C) and a final cooling step back to the original conditions (25°C) . Free PEG polymer is known to demon^¬ strate clouding under these buffer conditions in the temperature range 70-80°C due to specific ion interactions, which can de^¬ crease for a high volume fraction of PEG. As the temperature is increased towards this "cloud point" the polymer excluded volume is expected to decrease; in our case the shell size should de^¬ crease. This transition is broader than a LCST transition in pure solvent.

The results show that the peak/shoulder is indeed responsive to changing temperature, as shown in Figure 14. It is therefore concluded that the shoulder can be attributed to the PEG shell. The shell in its initial stage at 25°C (first step, 25°C, green) has size of 8.8 nm, comparable to the size observed in MQ. Heat^¬ ing to 50°C (second step, red) brings the shell to its cloud- point which shrinks the shell radius to 7.5 nm. The final cool^¬ ing step (third step, cyan) expands the shell to 8.6 nm, which is in close proximity to the initial shell radius. The observed changes in shell size upon temperature are significant. A tem^¬ perature induced reduction in hydration is therefore observed and a structural origin of the peak/shoulder can be further ruled out.

In Figure 15, the simulation of intensity-normalized scat^¬ tering curves for the different shell sizes is shown in the q- region of the shell (0.01 to 0.1 A^-1 ) as well as an inset with the scattering curve of the whole core-shell particle. Observa^¬ ble is the shift of the shoulder in the 50°C step towards larger q-

Table 3. Parameters for shell diameter, ASLD and R2 as obtained by the curve fitting with the respective error estimates.

Table 3 shows the obtained fit parameters for the shell diameter and the SLD. R² for the fit was in all cases better than 0.99. As discussed in the "Shell" section the results are consistent as they indicate a monomer density lower than the upper theoretical limit given by the Daoud-Cotton model. In the case of the cloud point temperature cycle the values show a large variability for the different temperatures. That indicates that the quite straightforward system that is observable in MQ gets more com^¬ plicated due to necessary interactions between the buffer, its ions and the PEG polymers. The SLD parameter only governs the definition and shape of the shoulder and not its position; thus, the observed change in the shoulder position is attributed to shrinkage and subsequent swelling of the shell during a change of temperature.

This experiment shows not only the reversibility of the cloud- point behavior of the particles but also shows that the shell always maintains a clearly defined size with low polydispersity as otherwise the signal would vanish.

This observation basically indicates three points:

• Shoulder stays in the same region meaning a comparatively low response to change in environment

• Increasing temperature causes a decrease in the shoulder in^¬ tensity and shift of the shoulder to larger q (smaller ra^¬ dius) meaning no dumbbells (structure contribution)

• The change of the q position is reversible

• The shoulder can hence be attributed to the PEG shell which shows reversible shell shrinkage in the vicinity of the cloud point.

•A constant density region relating to a molten like state of this region of the shell was observed close to the iron particle, with a thickness of about 2.75 nm.

These observations have not been shown for particles grafted by the direct method used by Amstad et al . 2009, supra. In fact, for grafting densities below 1 M/nm², a shell could not be ob^¬ served by SAXS using the same experimental setup since the ASLD was too low.

Example 14 : Comparison of surface density estimation methods

The high dispersant density of the inventive particles has not been observed on truly monodisperse and spherical nanoparti- cles, and higher numbers might partly be a consequence of higher surface roughness and effective surface area. Thus, the effec^¬ tive grafting density is likely below 1.0 M/nm² also for these nanoparticles used in our early work (Amstad et al . 2009, su^¬ pra . ) .

Comparison of the absolute dispersant density per nm² re^¬ quires that the same measurement and calculation methods are ap^¬ plied to all samples. The samples should be fully comparable. It further requires that care has been taken to remove all free dispersants from solution before quantification. Therefore, stringent purification by removal of free dispersant, e.g.

through a Sephadex (dextran) column that shows high affinity to iron oxide nanoparticles, is a good way of demonstrating high colloidal stability and high dispersant density. However, the differences in purification method, characterization method and calculation method, makes numbers from the literature difficult to compare, which is well established in the field.

Nanoparticles have a very high surface-to-volume ratio. For spherical nanoparticles, a difference in a few A (corresponding to single atomic layers) in terms of determining the particle radius can have dramatic impact on the calculated dispersant density determination, since the area used for the calculation scales with the square of the radius. Similarly, calculating the area from the radius measured by TEM, SAXS or other common meth^¬ ods assuming a sphere, can strongly underestimate the actual ar^¬ ea of a non-spherical nanoparticle, since the minimum area for a 3D object is obtained for a sphere. The dispersant grafting den^¬ sity is calculated from the number of dispersants measured by a chosen characterization method, usually TGA which requires no assumptions on the physical properties of the organic molecules, normalized to the surface area; thus, underestimating the sur^¬ face area will artificially increase the estimated grafting den^¬ sity. TGA, however depends on the choice of thermal interval in^¬ vestigated, the atmosphere under which the thermal decomposition of organics take place, and for high organic fraction materials (densely polymer grafted nanoparticles) can have a relatively large error. Other methods making use of spectroscopy, however, have to rely on the correct assignment and integration of peaks, correct calibration to a known peak, the correct attribution of relative cross-sections or extinction coefficients. All these uncertainties are well-known to someone experienced in the field. Therefore, relative comparisons for different grafting methods on the same particles under the same conditions is the preferred way to establish superior methods rather than comparing absolute values from different types of methods and prepara^¬ tions throughout the nanoparticle literature. For the inventive two-step grafting-to method using a fluidized dispersant, this comparison was performed in relation to the previous standard direct grafting-to method.

Previously, a grafting density as high as 2.8 M/nm² has been reported using the same type of binding chemistry but using the direct grafting-to method and polydisperse iron oxide cores formed by thermal decomposition (Amstad et al . , 2009, supra) . Calculation of the grafting density for these samples relied on the above assumptions, which as illustrated in Fig. 16 will re- suit in a significant overestimation of the grafting density, when polydispersity and non-sphericity is taken into account, i.e. the effective surface area for grafting will be strongly underestimated .

Comparing the direct grafting-to method to the two-step method using a fluidized dispersant on monodisperse spherical nanoparticles elucidates the much higher grafting density achieved by the latter method compared to the first. This is a direct relative comparison of the same nanoparticle cores after grafting and removal of excess dispersants using the same TGA settings and calculations for characterization. The results re^¬ ported in Table 1 demonstrate an at least 2-3 times higher grafting density of the latter compared to the first, which are shown to translate also into a unique structure of the shell with near maximum density of the polymer close to the inorganic core as theoretically predicted.

This difference could further be studied by SAXS, as de^¬ scribed in Example 13, which could detect the shell around densely grafted particles and match the density profile to the theoretical predictions; while the direct grafting-to method did not lead to polymer shell densities observable by SAXS (the scattering length density contrast is too small for strongly hy- drated PEG) . The constant high density, melt like, region of the dispersant shell was shown to extend several nm from the inor^¬ ganic core surface. This provides the core-shell nanoparticle with the surface and interaction properties of that of a star polymer (cf. Daoud Cotton) comprising the dispersant polymer, completely masking inorganic core interactions. The physical properties of such core-shell nanoparticles are thus qualita^¬ tively different to those of more sparsely grafted core-shell nanoparticles .

Example 15: Effect of dispersant grafting density on colloidal stability in aqueous suspension

That higher brush grafting density leads to higher colloidal stability under application of physiological conditions is a well-accepted hypothesis; in biological fluids the colloidal stability of bare NPs is severely compromised due to high con^¬ centrations of ions and biomolecules such as proteins. The high^¬ er temperature than room temperature experienced in many appli- cations of magnetic NPs has also been shown to lead to reduced colloidal stability of poorly grafted NPs or of NPs with revers- ibly anchored dispersants. Temperature cycling can thus be used as an additional relevant stress test to compare the colloidal stability of differently synthesized NPs.

Redispersion of strongly aggregated, dried NPs can be chal^¬ lenging, but the densely melt-grafted monodisperse NPs demon^¬ strate remarkably easy redispersion. Dried powder dispersed in^¬ stantly upon addition of the smallest amount of water; this in^¬ dicates that the cores even in the dried state do not come in close proximity to each other. The supreme colloidal stability was further demonstrated by high stability in water over pro^¬ longed period of time (>1 year) and repeated filtering through 0.45-ym syringe filter without loss of material.

DLS showed that melt-grafted ALD-PEG NPs remained stable in size under temperature cycling in water (Fig. 19), which produced aggregation of NPs more sparsely grafted using one-step ligand exchange. ALD-PEG melt-grafted NPs showed reversible ag^¬ gregation without precipitation upon T-cycling in PBS buffer (Fig. 19) . PBS frequently leads to stronger aggregation of bare or weakly grafted iron oxide cores due to the strong interaction of phosphate ions and iron at the NP core surface. The one-step grafted NPs again demonstrated severe aggregation followed by precipitation during the same T-cycle in PBS. Experiments test^¬ ing the resistance to precipitation in ethanol likewise showed that melt-grafted ALD-PEG NPs remained stably dispersed and could be filtered through 0.45 ym syringe filter.

Example 16: Colloidal stability of NPs upon heat treatment in serum

Finally, the stability of NPs in the presence of serum was investigated. Stability at room temperature was observed over experimental time scales also for NPs with grafting densities <1 chain/nm². This can be the effect of NPs showing sufficient sta^¬ bility after adsorption of the albumin predominant in serum to not precipitate. BSA is commonly used as an easy and low-cost surface modification for biotechnological applications in ex vi^¬ vo body fluids; however, it is not sufficient to stabilize NPs or other interfaces in vivo. It was not possible to efficiently separate NPs and proteins e.g. on a magnetic column, due to the high stability; this in itself demonstrates that the interaction with highly concentrated protein solutions does not lead to strong aggregation of the cores, but it could not be used to differentiate between strongly and weakly stabilized NPs. The low effective density of the NPs (>80% of the mass is highly hy- drated PEG shell) also foiled our attempts to use centrifugation to separate protein and NPs. Several different centrifugation and filtration separation methods were tested to analyse any po^¬ tential protein content adhering to the NPs, but none were suc^¬ cessful, i.e. a single dispersion always remained. A test was instead devised that used the denaturation and precipitation of serum at high temperature (75°C) to demonstrate the difference in protein interaction between NPs grafted by the two-step melt method and the one-step ligand replacement method. Denatured proteins adsorbing to an insufficiently protected NP surface at high T are likely to aggregate with other proteins and to pre^¬ cipitate the aggregates together with the NP out of solution. If the denatured proteins cannot directly adsorb to the NP surface a much smaller fraction of NPs would be trapped by such aggre^¬ gates and precipitate.

Figure 20A shows the precipitation in the presence of serum protein. A majority fraction of NPs grafted by one-step ligand replacement precipitates with the protein when the protein dena^¬ tures; this indicates a strong and frequent interaction with the serum proteins, at least at elevated temperature. In contrast, NPs grafted by the ALD-melt method remain in solution when the protein precipitates; this indicates a negligible interaction of the great majority of NPs.

Further evidence of the difference in stability in serum was the monitoring of the effective hydrodynamic size corresponding to the main number peak measured by DLS as function of time at 75°C (Fig. 20B) . For ALD-melt-grafted NPs the predominant inten^¬ sity peak shifted from 5nm hydrodynamic diameter (interpreted as being close to the average size of serum proteins) to ~15nm which corresponds well to the size measured for NPs alone. Thus, the result indicates a removal of protein from solution while the NPs stay suspended. In contrast, the one-step ligand exchange grafted NPs showed an early onset of aggregation and a reduction in hydrodynamic size of the main population. This could corre^¬ spond to that the main residual in the supernatant after temper- ature-induced denaturation and precipitation consists of small and stable proteins in the serum, i.e. the NPs were efficiently precipitated while a fraction of protein remained. These obser^¬ vations support the interpretation of the visual inspection: a majority of NPs precipitate with the denatured serum protein for particles that do not reach high grafting densities, while densely grafted ALD-melt particles remain suspended. Thus, serum proteins are able to interact with and bind directly to the core of iron oxide NPs that are not densely grafted, while densely grafted NPs are efficiently shielded. The difference between relevant density regimes was found to be >0.5 chains/nm² of PEG (5kDa) .

Claims

Claims :

1. A method of producing inorganic core particles comprising a dispersant shell, comprising a dispersant molecule in a high surface covering density on the inorganic core, comprising the steps of:

- providing one or more inorganic particles,

- ligating at least one organic linker onto the inorganic parti^¬ cle, thereby obtaining an inorganic core linker coated particle,

- providing at least one fluidized dispersant, preferably in form of a melt, suspension or solution,

- binding the at least one fluidized dispersant to the at least one organic linker, thereby obtaining the inorganic core parti^¬ cles comprising a dispersant shell.

2. The method of claim 1, comprising raising the temperature of the dispersant above its melting temperature and performing binding above the melting temperature.

3. The method of claim 1 or 2, wherein the dispersant is a mac^¬ romolecule, preferably a macromolecule comprising a polymer, preferably PEG, polyoxazoline, poly (N-isopropylacrylamide) , pol- yisobutylene, caprolactone, polyimide, polythiophene, polypro^¬ pylene, polyethylene, polyacrylic acids and other polyelectro- lytes, polyvinylpyrrolidone, polyvinylalcohol .

4. The method of any one of claims 1 to 3, wherein the disper^¬ sant has an average weight of 1 kDa to 30 kDa, preferably of 2 kDa to 15 kDa, especially preferred of 3 kDa to 10 kDa.

5. The method of any one of claims 1 to 4, wherein the inorgan^¬ ic core is paramagnetic, preferably superparamagnetic.

6. The method of any one of claims 1 to 5, wherein the inorgan^¬ ic core comprises a metal, preferably selected from Fe, Ti, Ni, Co, or an oxide any thereof, preferably a Fe oxide, such as Fe₂0₃ and/or Fe₃0₄, or comprises a metalloid or a semiconductor or consists of a non-metal material, preferably the inorganic core comprises Al, Si, Ge, or silica compounds.

7. The method of any one of claims 1 to 6, wherein the inorgan^¬ ic core has an average diameter size of 1 nm to 400 nm, prefera^¬ bly 1.5 nm to 100 nm, especially preferred 1.8 nm to 12 nm, 2 nm to 20 nm, 3 nm to 40 nm, 4 nm to 80 nm.

8. The method of any one of claims 1 to 7, wherein the linker and the dispersant each comprises one member of a pair of reac^¬ tive groups, capable of forming a chemical bond, or further mem^¬ bers of further such pairs,

preferably wherein the pair of reactive groups is selected from an amine and an aldehyde, an amine and an acrylate, an amine and a carboxyl group, an amine and an ester group, a sulfonylchlo- ride and either an amine or a hydroxy-group, an acid-chloride and either an amine or a hydroxy-group, a thiol and any one se^¬ lected from a maleimide, an acrylate or an allyl-group, an iso- cyanate and either a hydroxy-group or an amine, a conjugated dien and a substituated alkene, an epoxide and a nucleophile, preferably wherein an amine and an aldehyde is reacted in the binding reaction, thereby forming an imine, which is even more preferred reduced to an amine by a reducing agent, such as

NaCNBH₃.

9. The method of any one of claims 1 to 8, wherein said linker or dispersant comprises a leaving group that is removed in the ligation reaction and/or wherein binding the linker to the dispersant is a reaction without the requirement of a solvent.

10. The method of any one of claims 1 to 9, wherein the linker comprises an alcohol or carboxyl group in contact with the sur^¬ face of the inorganic core, preferably a moiety selected from a 6 membered homocycle comprising oxygen or hydroxyl substituents, such as benzoic acid, phenol or catechol, and/or preferably wherein the linker molecules comprise an aromatic group with a electronegative group bound thereto, more preferred a nitro group, especially preferred wherein the linker molecules com^¬ prise a moiety selected from nitrocatechol , nitroDOPA or nitro- dopamine .

11. The method of any one of claims 1 to 10, wherein the concen^¬ tration of the dispersant is above the solubility concentration of a solvent thereby obtaining a suspension and wherein the dis- persant is fluidized by increasing the temperature above the melting temperature of the suspension.

12. The method of any one of claims 1 to 11, wherein the provid^¬ ed inorganic particle is in complex with a surfactant on the particle surface, especially preferred in case of metal parti^¬ cles, and wherein ligating the organic linker onto the inorganic particle is by replacing the surfactant by said organic linker, thereby obtaining an inorganic core linker coated particle.

13. The method of any one of claims 1 to 12, wherein the step of providing an inorganic particle comprises providing a plurality of inorganic particles wherein the mean standard deviation of the particle's average size is at most 10%, preferably at most 5%, even more preferred at most 2% of the particle's average size, such as at most 0.8 nm, preferably at most 0.5 nm.

14. A particle comprising an inorganic core surrounded by link^¬ ers that are chemically linked to dispersant molecules, wherein the dispersant molecules

(a) are at an average density of at least 1.1, preferably at least 3.0, dispersant molecules per nm² of the inorganic core surface, and/or

(b) form a shell of constant dispersant density and a further shell of gradually reduced dispersant density with increasing distance from the inorganic core surface, preferably said parti^¬ cle being obtainable by a method of any one of claims 1 to 13.

15. A preparation of a plurality of particles, wherein

- said particles comprise an inorganic core surrounded by link^¬ ers that are chemically linked to dispersant molecules,

- wherein the dispersant molecules

(a) are at an average density of at least 1.1 dispersant mole^¬ cules per nm² of the inorganic core surface, and/or

(b) form a shell of constant dispersant density and a further shell of gradually reduced dispersant density with increasing distance from the inorganic core surface, and

preferably wherein the inorganic core is of an average size be^¬ tween 2 nm to 80 nm in diameter and further of homogenic size in said plurality wherein the mean standard deviation of said aver^¬ age size is at most 10%, preferably at most 5%, even more pre^¬ ferred at most 2% of said average size, such as at most 0.8 nm, preferably at most 0.5 nm,

preferably said preparation being obtainable by a method of any one of claims 1 to 13.