A universal, template-free approach to porous oxide and polymer film processing

Abstract

We have developed a unique, template-free method for the processing of porous oxide and polymer films with a high degree of ordered porosity, taking advantage of micro-detonation. When mixing an alcoholic solution of ammonium or silver nitrate with an alcoxide precursor solution, a phenomenon of self-organization of immiscible organic solvent (Fuel)–nitrate(oxidant) micro-droplets occurs during film formation. Thermal activation leads to self-ignition reactions of the droplets resulting in the exothermic formation of gas species that upon expansion leave a porous structure behind. We demonstrate the method on thin TiO₂ films and show that hierarchical porosity may be obtained by increasing the film thickness. In order to prove the universality of the method we apply it to a polymer solution, and achieve a nicely ordered hierarchical porosity in a 20 μm thick polymer film.

---

Filtration/separation, catalysis, microreactors, drug delivery and many other established or new cutting edge technologies rely on a porous material component. Pore size, density, shape and ordering may constitute critical parameters that determine the efficiency of the porous component and thus the quality and throughput of a specific process.^1–4 The synthesis of porous materials is well documented. It makes use of established industrial processes such as foaming, extrusion of slurries containing fillers, gas injection, lithography, computer-aided techniques and so on.^3–5 However, controlling pore geometry, ordering and size remains a challenge in the majority of these processes.

Porous oxide films with ordered porosity, including the multifunctional oxide TiO₂, were processed using a variety of chemistry-based methods, but always taking advantage of specific templates.⁶ The use of polystyrene spheres of different sizes may help to fabricate porous thin films with ordered porosity, although it is mostly macroporosity.⁷ Long-chained molecules, e.g. polyethylene glycol, were also used in combination with sol–gel and “controlled” phase separation for producing porous films and bulk TiO₂, although in this case pore ordering and pore size control were rather poor. Other methods use infiltration of polymeric beads, selective dissolution (leaching) of one component and anodization.^8–12

In the case of porous polymer processing, various methods were reviewed by Hentze et al., but principally are in many respects similar to those described above for ceramics, e.g. foaming, phase separation, template-based, etc., with their advantages and limitations.¹³ Other methods take advantage of supercritical CO₂ as a solvent and porogen in aqueous media for the fabrication of porous polymers with some degree of ordering.¹⁴

As porous materials constitute the corner stone in a number of cutting edge applications, involved not only in the easing of technological processes, such as the heterogeneous catalysis of chemical reactions, but also in improving the quality of life, any versatile method for their processing that can also lead to saving time and energy, achieving higher materials efficiency, and eventually better properties, is desirable.¹

In this work we present a processing method that meets all the requirements above. It is based on mixing an oxidant, e.g. a nitrate salt, in an ethanolic solution (fuel) of an alcoxide, and letting the system self-organize. The idea was inspired from an article on Ammonium nitrate (NH₄NO₃)–Alcohol (ANA) mixtures for propellant technology.¹⁵ The authors had shown that for a given oxidant to fuel ratio self-ignition occurred exothermally with the formation and expansion of gaseous species, causing high thrust velocities. In the present case we will show that similar reactions proceed in thin films, but only after the self-organized ordering of immiscible microscopic droplets of salt–solvent–alcohol at the interface with the substrate. The self-ignition of these droplets upon thermal activation leaves a well-ordered porous structure. We will also show that this method can be extended to the processing of polymer films, a class of materials that have far more applications than oxide films.

First we present the strategies for preparing porous structures; the mechanisms will be discussed later.

In the case of TiO₂-films, two strategies proved very effective in inducing a porous structure with high ordering, while leading to different matrix microstructures.

a) With NH₄NO₃: In this approach Ti-n–butoxide (Ti–OBut₄) is dissolved in ethanol, hydrolyzed, peptized and then mixed with an ethanolic solution of NH₄NO₃. The solution is homogeneous, and a measurement of the particle size using photon correlation spectroscopy yielded approximately 4 nm which is usual for our TiO₂ precursor solutions (salt-free).

b) With silver nitrate, AgNO₃: the procedure is the same as above with the single difference of substituting NH₄NO₃ with AgNO₃.

The precursor solutions made with the two different salts were spin or dip-coated on Si, Al, or stainless steel substrates with very similar results. Drying was achieved at a temperature of 200 °C, and the films were eventually annealed at 500 °C to drive crystallization of the matrix. Scanning electron microscopy (SEM) micrographs of one-layer samples dried at 200 °C temperature are presented in Fig. 1A. The pores show different sizes with a nearly circular shape (isolated pores have ellipsoidal or triangular morphologies), but they all have smooth pore walls. Their distribution is reminiscent of close packed spheres with irregular hexagonal packing. An analysis of the size distribution of the pores revealed that 80% of the population was centered on 500 nm. The area fraction was approximately 50%. When the coating sequences or thickness are increased, a hierarchical pore structure forms, Fig. 1C. We noticed that the pores of the upper layer are slightly off-center with respect to those of the underlying layers, so that we may infer that pore self-organization is very similar in all layers. In Fig. S1, ESI,† we demonstrate that the pores are open, using backscattered electron microscopy.

Fig. 1 (A) Secondary electron (SE) micrograph of a large area of a one layer TiO₂ thin film on an oxidized silicon substrate, processed using NH₄NO₃ as the oxidant and dried at 200 °C for 30 min; ordered porosity is embedded in a TiO₂ network; scale bars 10 μm (left) and 1 μm (right). (B) Hierarchical, ordered porosity in a 20 μm thick PVDF–TrFE film that was deposited by dip-coating on glass; scale bars 100 μm (left), 20 μm (right). (C) Hierarchical porosity in a thin TiO₂ film obtained via dip-coating a Si-substrate in a sol containing AgNO₃and subsequently dried at 200 °C for 30 min; scale bar 1 μm. (D) is a high magnification of (B) where the hierarchical porosity is conspicuous. (E) Schematically summarizes the mechanism of pore formation in TiO₂ films: (I) self organization of immiscible fuel–oxidant micro-droplets upon spin-coating; (II) cross section with droplets having a thin Ti–O network skin; (III) self ignition of micro-droplets upon thermal activation and gas expansion (inset) pulverizing the skin; (IV) porous film. (F) A sketch illustrating the formation of hierarchical porosity in PVDF-based films: (I) self-organization of immiscible droplets throughout film thickness; (II) self-ignition at different thickness levels and (III) a macroporous thick film (see also Movie S1, ESI†).

Using AgNO₃ instead of NH₄NO₃ in principle leads to the same porous microstructure with an amorphous connecting matrix consisting of a mixture of TiO₂ and Ag_xO.¹⁶

As expected, the pore distribution and morphology are largely dependent on the concentration of the salt in the precursor solution. We generally observed that the pore area fraction decreased with decreasing salt concentration, and their morphology became more irregular (see Fig. S2, ESI†).

Pore morphology may also be controlled by a proper choice of substrate. For instance, spin-coating of an Au–Pd terminated Si-substrate resulted in various pore morphologies, as shown in Fig. 2A, although a close look reveals that the reaction zone was circular, as indicated by the “aureole” surrounding the pores (see inset in Fig. 2a).

Fig. 2 Microstructure manipulation. (A) TiO₂ film deposited on an Au–Pd terminated silicon substrate, showing corrugated pore walls; in the right SE micrograph the roundness of the initially self-organized micro-droplets is conspicuous from the “aureoles” surrounding the pores; the irregular pore morphology is the result of a sluggish reaction. (B) A porous TiO₂ film that was made using AgNO₃ as the oxidant and subsequently UV irradiated to drive the reduction of the Ag-particle from Ag-oxide. (C) basket-weave anatase–brookite microstructure obtained after heat treatment at 500 °C, UV-irradiation and subsequent silver dissolution in an aqueous nitric acid solution. (D) The anatase structure obtained using NH₄NO₃ as the oxidant showing bigger equiaxed grains (see also Fig. S3, ESI†).

Beyond pore formation and morphology control it also appeared possible to manipulate the microstructure via the salt choice. Irradiating the films made with AgNO₃ with UV or visible light resulted in a porousTiO₂–Ag nanocomposite with a tunable Ag particle size (via the exposition time, see our work on plain TiO2–Ag films) (Fig. 2B).¹⁶ We also observed that the salt type had a strong influence on the microstructure of the connecting matrix. AgNO₃ induced a mesoporous basket-weave structure that was composed of elongated TiO₂ nanocrystals (Fig. 2C). Structural characterization of these films using Raman scattering revealed a mixture of anatase and brookite polymorphs (see Fig. S3, ESI†). This is an interesting nanocomposite structure that should be well suited for photocatalysis using daylight, e.g. for water purification.¹⁷ In contrast, NH₄NO₃ induced an equiaxed anatase grain microstructure with well developed grain boundaries (Fig. 2D and S3, ESI†).

The route for making porous polymer films is very similar to that described above, and attests to the universality of the method developed here. The only starting materials necessary are a polymer in granular form and a polar organic solvent (fuel) that is able to dissolve both the polymer and nitrate salt (oxidant). We successfully used this method to produce porous Polyvinylidene fluoride (PVDF) and PVDF–TrFE (trifuluoroethylene) copolymer films, but the method is further applicable for a variety of other polymers, including PMMA, Polylactides, etc.Fig. 1B and D show a porous film of PVDF–TrFe, approximately 20 μm thick, fabricated by dip-coating a glass substrate in a polymer solution containing ca. 15% of NH₄NO₃. We obtained a hierarchical pore structure that is principally similar to that of the TiO₂ depicted above, different only in that larger pores are obtained; the mean pore size is around 10 μm (for the upper layer).

The mechanisms underlying pore formation may be expressed in simple terms, although the reactions taking place during the film and pore formation might be complex. In the case of oxide films, the precursor solution is first homogenous with all of the species dissolved in ethanol. As mentioned above, the particle size measured in solution was approximately 4 nm, usual for this type of precursor solutions, and should correspond to hydrous titania obtained by hydrolysis, condensation and subsequent peptization reactions, described in detail in reference.¹⁸ Upon spin or dip-coating, a polymeric film forms whereupon most of excess ethanol is evaporated. From the experimental observations of the final microstructures we infer that this new situation results in a phase separation with the formation of a Ti–O network and self organized droplets containing hydrolysis products, e.g. butanol, water and solvated salt (that is not soluble in the Ti–O network), and probably some amount of ethanol that could have also diffused into the droplets during the Ti–O network formation. The driving force is purely thermodynamic: the free energy of the system is minimized for phase separation upon evaporation of the solvent (the nitrate salt is not soluble in the TiO₂ network), while the monodisperse distribution might be the result of the kinetics controlling the network formation, and thus its dimensions. Experimental observations also indicate that the droplets preferentially “condense” at the interface with the substrate, with the formation of a thin skin of the Ti–O network at the interface with air, obviously to lower the droplet–air interfacial energy (see, Fig. S4, ESI†).

A moderate heating of the films thus obtained leads to the self-ignition of the self organized alcohol–salt mixture droplets following an exothermic reaction of the type:¹⁵

C_nH_{2n + 1}OH + xNH₄NO₃ → nCO₂ + xN₂ + (1 + 3x − 2n)H₂O + (3n − x)H₂ + Q (1)

where x is the molar ratio of NH₄NO₃ to alcohol and Q is the energy released per 1 mole of alcohol. According to Sasoh et al., x controls the ignition, the pressure and the energy released. In controlled ignition experiments they demonstrated that an optimal x should be in the range of 2 to 4.¹⁵ In our case a typical solution contains 1 mole of Ti–(OBut)₄, 4 moles of H₂O (for the hydrolysis of Ti–(OBut)₄), 0.735 moles of NH₄NO₃ (AgNO₃) (constituting the empirically determined optimum concentration of salt for a well ordered porosity) and 106 moles of ethanol. Assuming that the whole butoxide is hydrolyzed and that all of the ethanol is evaporated during coating we calculate a value of approximately 0.18 for x. This figure is most probably too underestimated as not all of the butoxide is expected to be hydrolyzed (no excess water was used), and also a definite amount of the butanol should evaporate. Thus we should expect a higher fraction of salt in the droplets, if self-ignition had to occur. The molar ratio of the NH₄NO₃–butanol mixture that resulted from the dynamics of the film formation and droplet self-organization, though not known, is nonetheless sufficient to undergo self-ignition upon thermal activation, with the formation of product gases that are not necessarily those indicated in eqn. 1 (indeed an analysis of product gases during thermal gravimetric experiments of a TiO₂ precursor solution, with AgNO₃ as the oxidant, using IR-spectroscopy indicated the formation of N₂O and NH₃, in addition to CO₂ and probably H₂, see Fig. S5, ESI†). The expansion of the product gases leads to the pulverization of the skin, leaving a pore. This mechanism has been observed in situ under the effect of the electron beam in the SEM. However, it was not possible to record the events due to rapid kinetics. In the case of Au–Pd terminated substrates we have seen above that the pores had irregular morphologies. We surmise that this should be the consequence of a preferential adsorption of hydrogen, ammonia and nitrogen oxide on Pd (the Gibbs free energy of absorption of hydrogen on Au–Pd alloy is ca.−11 kJ mol⁻¹ of H₂; ammonia and N₂O also have a high adsorption energy on Pd), thus weakening the gas expansion pressure so that it no longer suffices for the removal of the whole oxide skin and results in the corrugated pore morphology shown in Fig. 2B.^19,20 A schematic summary of the mechanisms outlined above is displayed in Fig. 1E.

The formation of pores in polymer films primarily takes place via a very similar mechanism to that depicted above. To fabricate porous polymer films the polymeric powder (in our case PVDF or PVDF-TrFE) is dissolved in dimethylformamide (C₃H₇NO (DMF), an organic polar solvent that dissolves both NH₄NO₃ and PVDF, and lots of other polymers) and the solution is then mixed with a solution of NH₄NO₃ in DMF, yielding x of approximately 0.14. Also in this case, phase separation at the microscopic scale takes place during coating and network build-up with the formation of NH₄NO₃-rich droplets dissolved in DMF, and embedded in the still uncured matrix. As the polymer films are approximately 20 μm thick the droplets are self-organized in nearly close-packed ordering throughout film thickness. The subsequent heat treatment at 130 °C (for PVDF–TrFE) or 150 °C (for PVDF) delivers enough activation energy to drive the self-ignition of fuel (DMF)–oxidant (NH₄NO₃) mixture. The reaction also takes place at room temperature, but the kinetics are very slow, with reaction times of 6 h and over. In the 20x accelerated video in the supporting information (Movie S1, ESI†) the whole process may be observed, first with the formation of self organized “dimples” of DMF–NH₄NO₃ covered with a thin skin of polymer (Fig. S4, ESI†), and subsequently, after some degree of network build-up and diffusion of excess DMF into the “dimples” (to attain an optimum fuel–oxidant ratio), rapid ignition and pore formation occur, including those beneath the film surface. We also observed that the reaction started at one edge of the film (evidenced via the opaque appearance) and advanced with a definite reaction front interface into the gel-like areas (Movie S2, †). The product gases are similar to those above. Due to droplet self-organization throughout the film, a hierarchical network of pores with high connectivity is formed. The mechanism discussed above is schematically summarized in Fig. 1F.

Further to the well known applications of porous materials for catalysis, filter membranes, etc., we describe two specific applications of the films fabricated here. Often it is desirable to make ordered, on substrate nanostructures, e.g. noble metals for nanobiosensing and molecular detection.^21,22 These ordered structures can be easily processed when the porous films are deposited on conducting substrates such as indium thin oxide (ITO)-coated glass, metals, etc. A simple electrochemical or electrophoretical deposition process would then lead to ordered arrays, e.g. of noble metals, that are strictly confined to the pores. As an example, gold nanostructures are shown in Fig. 3A. We call this application: on-pore lithography. It goes without saying that the oxide film may be removed, if desired, via a simple chemical etching process.

Fig. 3 Selected applications of porous TiO₂ and PVDF. (A) On-pore lithography: gold nanostructures are selectively deposited in the pores of a TiO₂-layer that was deposited on SnO₂–In₂O₃ (ITO) coated glass. (B) Porous PVDF that was functionalized with hydroxyl-apatite. (C) In-growth of osteoblasts into the pores of PVDF (SE-micrograph of a super-critically dried sample). The osteoblast cells were artificially colored to single them out from the gray background.

The porous polymer (and oxide) films may be functionalized with desirable nanoparticles such as hydroxyapatite (HAP) for biomedical applications, TiO₂ for photocatalysis (to make photocatalytic membranes), noble metals for catalysis etc.Fig. 3b shows a PVDF film that was functionalized with HAP nanoparticles. The films are highly biocompatible (see Fig. S6, †), and in Fig. 3C we show the in-growth of osteoblast cells into the porous structure. These films may be used on orthopedic implants and should promote implant–osteo integration for a better and faster healing process.

In summary a versatile and universal method for preparing porous thin films of oxides and polymers has been described. The addition of an oxidant, here a nitrate salt, to a precursor solution containing an organic solvent resulted in the separation, upon spin or dip-coating, of ordered micro-droplets of nitrate-organic solvent in a matrix network. Relatively low thermal activation energy was sufficient for the self-ignition of the “propellant” mixture to drive micro-detonations, subsequently leaving an ordered porous structure. In thick films hierarchical porous structures were obtained. We have shown that the method applies well to titanium oxides and PVDF-based polymers, but it is by no means restricted to only these materials.

Acknowledgements

Financial support of this work is provided by the European council and the Land of Schleswig-Holstein Project # TraFo 08139. Thanks are due to Salah Habouti (IMST) for gold electrodeposition. We gratefully acknowledge the assistance of S. Eichholz from NETZSCH application laboratory for thermal analysis, Selb, Germany for conducting the TG experiments and FTIR-evolving gas analysis.

References

1. J. T. Hupp and K. R. Poeppelmeier, Science, 2005, 309, 2008–2009.
2. D. A. Edwards, J. Hanes, G. Caponetti, J. Hrkach, A. Ben-Jebria, M. L. Eskew, J. Mintzes, D. Deaver, N. Lotan and R. Langer, Science, 1997, 276, 1868–1871.
3. P. Colombo, Science, 2008, 322, 381–383.
4. J. N. Weber and E. W. White, Science, 1972, 176, 922–924.
5. M. L. K. Hoa, M. Lu and Y. Zhang, Adv. Colloid Interface Sci., 2006, 121, 9–23.
6. H. Zhang and A. I. Cooper, Soft Matter, 2005, 1, 107–113.
7. D. K. Yi and D. Y. Kim, Nano Lett., 2003, 3, 207–211.
8. M. C. Fuertes and G. J. A. A. Soler-Illia, Chem. Mater., 2006, 18, 2109–2117.
9. J. Konishi, K. Fujita, K. Nakanishi and K. Hirao, Chem. Mater., 2006, 18, 6069–6074.
10. D. G. Shchukin and R. A. Caruso, Chem. Mater., 2004, 16, 2287–2292.
11. E. S. Toberer, T. D. Schladt and R. Seshadri, J. Am. Chem. Soc., 2006, 128, 1462–1463.
12. D. Wang, Y. Liu, C. Wang, F. Zhou and W. Liu, ACS Nano, 2009, 3, 1249–1257.
13. H.-P. Hentze and M. Antonietti, Rev. Mol. Biotechnol., 2002, 90, 27–53.
14. R. Butler and I. Hopkinson, J. Am. Chem. Soc., 2003, 125, 14473–14481.
15. A. Sasoh, T. Ogawa and K. Takayama, Shock Waves, 1999, 9, 291–294.
16. M. Es-Souni, M. Es-Souni, S. Habouti, N. Pfeiffer, A. Lahmar, M. Dietze and C.-H. Solterbeck, Adv. Funct. Mater., 2010, 377–385.
17. H. Pan, B. Gu and Z. Zhang, J. Chem. Theory Comput., 2009, 5, 3074–3078.
18. B. L. Bischoff and M. A. Anderson, Chem. Mater., 1995, 7, 1772–1778.
19. A. Maeland and T. B. Flanagan, J. Phys. Chem., 1965, 69(10), 3575–3581.
20. A. C. A. de Vooys, M. T. M. Koper, R. A. van Santen and J. A. R. van Veen, J. Electroanal. Chem., 2001, 506, 127–137.
21. L. Qin, S. Park, L. Huang and C. A. Mirkin, Science, 2005, 309, 113–115.
22. J. N. Anker, W. P. Hall, O. Lyandres, N. C. Shah, J. Zhao and R. P. van Duyne, Nat. Mater., 2008, 7, 442–453.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c1ra00200g

---