Metal oxide arrays from block copolymer thin film templates

Department of Chemical & Biomolecula Newark, DE 19716, USA. E-mail: thepps@u 302 831 0215 Department of Chemistry & Biochemistry, U 29201, USA Department of Chemical Engineering, Univ 29201, USA. E-mail: lauteraj@cec.sc.edu; Fa † Electronic supplementary information Al2O3, MnO2, Co3O4, NiO, CuO, ZnO, ZrO peak positions of template oxides; tita UV-vis spectra of methylene blue photode dispersion calculations. See DOI: 10.1039 Cite this: J. Mater. Chem. A, 2015, 3, 7822


Introduction
The development of Mobil's zeolites in the 1960s sparked an avalanche of research on the design of structured materials with nanometer-scale features (>1 nm) for applications in catalysis and separations. 1Innovative uses of structure directing agents can yield nanomaterials with improved catalytic activity.Whereas small-molecule surfactants have proven to be excellent subnanometer structure directing agents, 2 block copolymers (BCPs) offer a macromolecular analog with tunable molecular weights, block choices, and compositions. 3,4][7][8][9][10] These lms can be used to direct the nanoscale structure of inorganic materials for catalysts, sensors, and optics applications. 118][19][20][21][22] The vast majority of previous investigations involving BCP templating follow two routes: (1) segregating preformed nanoparticles or (2) using in situ sol-gel methods.The former approach benets from advances in nanoparticle synthesis, which allows researchers to tune particle size, shape, crystallinity, and composition. 23Additionally, templating preformed nanoparticles is a convenient route to ensure that the inorganic material retains the desired properties of interest.Previous work by Epps and coworkers templated pre-synthesized gold nanoparticles in poly(styrene-b-isoprene-b-styrene) thin lms by tuning the particles' surface energies using ligand exchanges. 240][31][32] In all cases, preformed nanoparticle miscibility and templating generally were dominated by the particle-polymer entropic (particle size and shape) and enthalpic (ligand chemistry) interactions.
In situ sol-gel methods use a sacricial BCP to arrange metal salt precursors into nanoscale features and subsequently reduce, oxidize, or calcine the composite to simultaneously form the templated material and remove the polymer. 33,34This versatile method can be used to create many types of industrially relevant crystals, powders, and lms.Within sol-gel techniques, evaporation induced self-assembly (EISA) is a popular method for templating inorganic materials. 2In short, the polymer and metal precursor are combined in a single solution and a thin lm is coated onto a substrate.The evaporation of solvent creates a concentration gradient that acts as an ordering front that produces the nanoscale features.Features can be tuned by adjusting the polymer molecular weight, 35 sol-gel concentration, 33,36,37 and coating procedure. 38][41][42][43] Poor ordering likely was due to polymer/metal inter-and intramolecular interactions that inhibit polymer chain mobility and prevent BCP ordering. 44To maximize BCP chain mobility, Boyen and coworkers modied the EISA method by complexing Fe and Nb salts with homopolymer P2VP (h2PVP) prior to blending with the PS-b-P2VP solution. 45Thin lms of salt/h2VP/PS-b-P2VP produced well-ordered arrays with z1 mm 2 grain sizes.However, this modication to the traditional EISA method was limited to metal loading ratios of mol metal : mol 2VP-monomer # 0.2 to avoid macrophase separation.On the other hand, Russell and co-workers employed a preformed BCP template to create wellordered Au/Ag arrays for surface plasmon resonance studies. 46][49] Although the method was proven, 50 researchers have yet to use it for catalysis applications.

Templating procedure
The SPICE method is depicted in Scheme 1.Each of the subprocesses are explained in detail; however, it should be noted that sub-processes can be substituted (e.g., thermal annealing instead of solvent vapor annealing, dip-coating instead of spincoating, calcination instead of ozone etching, etc.).
Spincoating lms.Films were made using a Laurell WS400-6NPP-Lite spincoater.Toluene solutions of PS-b-PEO (1 wt%) were stirred for 1 h and subsequently syringe ltered (0.2 mm, PTFE) prior to lm casting.Approximately 40 mL of solution were quickly syringed onto substrates spinning at 3000 RPM for 30 s. Films were further dried under nitrogen.Using a Filmetrics F20 instrument, thicknesses of as-cast lms were analyzed on silicon wafers.Further processing (detailed below) was carried out for lms deposited on silicon oxide wafers.
Pattern formation.Perpendicularly-oriented hexagonally array patterns of PEO domains in a PS matrix were made by solvent annealing PS-b-PEO lms with toluene/water in a high humidity chamber. 51Films were placed in a bell jar with a beaker of 4 mL of toluene.The bell jar was positioned within a high humidity chamber.Humidity was controlled by bubbling nitrogen through deionized water that was maintained at 60 C. Aer exposure to saturated toluene vapor at room temperature for 18 h, the bell jar was opened to simultaneously allow toluene evaporation and humidity exposure.Films were exposed to the high humidity environment for 15 min before being removed from the chamber.It should be noted that although arrays of perpendicular PEO domains were achieved by toluene solvent annealing alone, the addition of high humidity (>90% RH) increased grain sizes from less than 1 mm 2 to greater than 4 mm 2 .
Immersion and metal complexation.Table S1 (ESI †) details the processing conditions for each metal oxide sample.For Mg, Al, Mn, Fe, Co, Ni, Cu, Zn, Sn, Ru, or Ce, a 1.0 M precursor in ethanol solution was prepared.For Zr solutions, 0.5 M solutions in ethanol/water (equal volume) were mixed to adequately dissolve the precursor.For Ti solutions, 0.25 g HCl solution, 0.7 g titanium tetraisopropoxide, and 5 mL 2-propanol were stirred for 30 min.All annealed PS-b-PEO lms were immersed in a precursor solution for 15 min.Immersion allowed the metal ions to selectively complex with the PEO domains.Aer submerging the BCP lm in the precursor solution, excess precursor solution was removed by rinsing with 2-propanol and drying in a nitrogen stream.
EISA TiO 2 .For comparison purposes, a lm was made using the EISA method.Aer stirring, 0.2 mL of the titania solution was combined with 0.8 mL of the PS-b-PEO toluene solution (1 wt%), and a lm was spincoated as previously described.Cast lms were annealed in saturated toluene vapor (2 h) without water, which would have undesirably hydrolyzed the titanium precursor.
Etching template.For all lms, template etching and metal oxidation were simultaneously achieved using a UVO cleaner for 30 min.Residual carbon was removed by rinsing with toluene and drying under a dry nitrogen stream.
Gold nanoparticle addition.To demonstrate additional modularity of SPICE, gold nanoparticles (AuNP) were spincoated onto the SPICE-and EISA-templated TiO 2 aer polymer removal to enhance the photocatalytic degradation of MB.AuNPs were synthesized according to the Brust-Schiffrin method. 52AuNPs (3.5 AE 1.0 nm diameter) were stabilized in toluene solutions with dodecanethiol. 24

Film characterization
Atomic force microscopy (AFM) images were recorded with a Bruker Multimode Nanoscope V system.AppNano ACL tips (190 kHz resonant frequency, 58 N m À1 spring constant) were used in tapping mode.Image processing and FFTs of micrographs were performed with ImageJ.
X-ray photoelectron spectroscopy (XPS) measurements were performed on a Kratos Axis Ultra DLD instrument equipped with a monochromated Al Ka X-ray source and a hemispherical analyzer.All analyzed metal oxide arrays were deposited on silicon oxide wafers.Binding energies were calibrated using Si 2p at 103.3 eV.
Transmission electron microscopy (TEM) images were taken with a Hitachi H8000 operated at 150 kV.PS-b-PEO lms containing titania precursor were peeled and mounted onto copper mesh grids according to a method previously described in the literature. 53The composite lm was prepared for peeling by depositing a thin layer of carbon (<10 nm) using a Hitachi carbon evaporator.A droplet (<1 mL) of 25 wt% poly(acrylic acid) (PAA) was drop-cast and allowed to dry overnight at atmospheric conditions.A razor blade was used to remove the resulting solid, which adhered to the polymer composite lm.The PAA was dissolved in a large volume of water.The resulting oating lms were mounted onto copper mesh grids or silicon nitride membrane window grids and allowed to air dry for at least 15 min before loading into the TEM.

Photocatalysis experiments
We demonstrate the SPICE method through the improved photocatalytic degradation of MB, which is an industrial pollutant commonly found in textile waste streams. 20In this work, an aqueous solution of 1 Â 10 À5 M MB was stirred for 2 days and then stored in a dark environment for later use.Wafer fragments ($2 cm 2 ) and 3.3 mL of MB solution were placed in Brandtech cuvettes (220-900 nm transparency).The surface areas of the samples were hard to measure when the wafer fragments were irregularly shaped.Therefore, degradation rates were normalized by the weights of the wafer fragments.Irradiation was accomplished with a ScienceTech SF300A solar simulator coupled with a ScienceTech 500-300 power supply (350 W average output).Wafer fragments were removed from Scheme 1 The spincoat-pattern-immerse-complex-etch (SPICE) templating method decouples polymer annealing from the metal precursor gel incorporation.Traditional methods, represented by EISA, spincoat an all-in-one solution that includes both polymer and metal precursor.Although EISA has fewer steps, it can produce irregular arrays of inorganic material.Pattern formation was achieved by annealing with toluene and water vapor using standard solvent annealing techniques enclosed in a humidified glove box.
the cuvettes prior to UV-vis measurements, which were recorded with a Shimadzu UV2450 instrument.The 665 nm peak was monitored during time-lapse studies; peak area integrations from 550-740 nm were calculated.Photocatalytic experiments were reproduced three times for each sample.

Results and discussion
In heterogeneous catalysis, ordered and narrowly disperse features made by bottom-up processes are critical for applications requiring maximum active surface area.Thus, grain size and narrow structural dispersity inherent in BCP lms is an important aspect of the SPICE method.In our case, perpendicularly-oriented and hexagonally-packed domains of PEO cylinders in PS-b-PEO lms (30 nm thick) were obtained by toluene vapor annealing in a high-humidity chamber, as depicted in Scheme 1. 51 The average radius of the PEO domains in a toluene/water annealed neat polymer lm was 8.7 AE 0.8 nm (>2500 measurements) (Fig. 1b).The inset fast Fourier transform (FFT) exhibits 4 th order peaks, which corroborate the excellent order and large grain sizes achieved by high-humidity annealing.Grain sizes of toluene/water vapor annealed PS-b-PEO lms were greater than 4 mm 2 .The PS-b-PEO lms were used to template SPICE TiO 2 (Fig. 1d) and Au on SPICE TiO 2 (Fig. 1e) samples.On the other hand, lms that were annealed by only toluene or tetrahydrofuran vapor contained grains that were less than 1 mm 2 (Fig. 1a).Additionally, EISA-templated TiO 2 exhibited large size distributions and poor order (Fig. 1c).These results highlighted the importance of using high-humidity during the annealing step.Some metal precursors, such as the titanium alkoxide and tin chloride used in this study, are susceptible to hydrolysis and cannot be processed in humid or aqueous environments. 54By separating template formation and metal precursor inclusion into two decoupled steps, the SPICE method can take advantage of annealing techniques (using high humidity in this case) that otherwise would be precluded by the presence of the metal precursor.To further extend this idea, the authors envision bolstering the SPICE method with supplementary annealing techniques that employ external elds, shear elds, or patterned substrates.Achieving wafer-sized areas of templated material is industrially relevant for magnetic storage media and energy harvesting. 55,56etal oxides were templated into the annealed PS-b-PEO lms by a simple immersion process.BCP lms were submerged in a metal precursor solution to allow the metal ions to selectively complex with the ethylene oxide monomer units.A typical loading ratio of 0.38 was achieved, which is almost twice that achieved by Boyen and coworkers (calculations are available in ESI †). 45Metal oxide formation and polymer removal were simultaneously achieved using an ultraviolet ozone (UVO) oven.The versatility of the SPICE procedure was exemplied by templating commonly used metal oxides: MgO, Al 2 O 3 , TiO 2 , Empirical formulas for the metal oxides were determined using high resolution X-ray photoelectron spectroscopy (XPS) scans to identify the oxidation states of the metals. 57ig. 1a-e show AFM images of selected metal oxide arrays as well as corresponding FFTs in the insets.Height images of the remaining metal oxides (Fig. S1 †) and XPS peak positions of all metal oxides (Table S2 †) can be found in the ESI.† Hexagonallypacked metal oxide dots are noted in Fig. 1b, d, e, and S1; † all corresponding FFTs reveal reections that are characteristic of hexagonal packing. 2 nd order reections are noted for Co 3 O 4 , Fe 2 O 3 , MgO, MnO 2 , ZnO, SnO 2 , and TiO 2 samples.Aer polymer removal, the height of the templated materials ranged from 1-7 nm.For TiO 2 , a 30 nm thick BCP lm template produced an average oxide dot height of 5.4 AE 1.4 nm.The height reduction was attributed to the signicant mass loss during polymer template removal and was used to calculate the average oxide loading.The average radius of a templated dot was 8.5 AE 1.9 nm, which closely matches the initial PEO domain size aer solvent annealing (8.7 AE 0.8 nm). 46The similarity in PEO domain size and the SPICE dot size suggests that templated metal oxide size can be controlled by tuning the polymer molecular weight and volume fraction. 58Highly ordered PS-b-PEO lms led to highly ordered metal oxide dots, supporting the efficacy and simplicity of the SPICE templating method.
The EISA-templating method was applied to the same polymer and titania precursor solution for the purpose of comparison.Because the titania precursor was highly sensitive to water, toluene/water solvent vapor undesirably hydrolyzed the titania precursor in the EISA lm (Fig. 2a).Thus, only toluene vapor was used to anneal EISA lms.Despite, our best efforts to anneal the EISA composite with toluene vapor, the resulting titania were poorly ordered and disperse in diameter (Fig. 2b).On the other hand, the SPICE method allowed the use of highhumidity annealing conditions and produced highly-ordered PEO domains, which preferentially absorbed the titania precursor (Fig. 2c).TEM images illustrate the improvement in order and dispersity of the SPICE method.
The EISA method produced an average TiO 2 dot radius of 9.8 AE 4.4 nm.Decreased morphology control was evidenced by the line scan in Fig. 1f, AFM image (Fig. 1c), and TEM image (Fig. 2b).Dispersion calculations are available in ESI.† The FFT of EISA micrographs showed no spots or rings, indicating the absence of any signicant ordering.Poor ordering of annealed EISA lms was attributed to hydrolytic oligomerization of titania and reduced polymer chain mobility caused by coordination bonds between the titania precursor and PEO domains. 37,59Huh and coworkers studied the effect of Cd coordination with poly(4-vinylpyridine) (P4VP) and found that increasing the salt loading increased gelation and decreased ordering of their Cd/PS-b-P4VP system. 44Similarly, it is expected that PEO-metal complexes hinder polymer chain mobility during solvent vapor annealing.By decoupling the polymerordering from the metal-coordinating, the SPICE method produced narrow size distributions and highly ordered arrays.
1][62] As a further validation of our synthesis approach, a toluene solution of gold nanoparticles (3.5 AE 1.0 nm diameter) was spincoated onto SPICE-and EISA-templated TiO 2 .The resulting Au/TiO 2 catalysts contained approximately 5 mol% Au as determined by XPS survey scans (not shown).SPICE-templated TiO 2 arrays maintained their high degree of ordering and dispersity aer Au addition (Fig. 1e).
Photocatalytic studies were used to demonstrate the increased activity of SPICE TiO 2 compared to EISA TiO 2 .Aqueous solutions of MB were photocatalyzed, under irradiation, with EISA TiO 2 , SPICE TiO 2 , and SPICE-templated and EISA-templated TiO 2 coated with gold nanoparticles (Fig. 3).Cuvettes with no catalyst (Blank) and MB stored in the dark (Dark) were used as controls.More specically, the Blank cuvette indicates how much MB degradation occurs due to irradiation alone.UV-vis spectra were collected every two hours, and peak areas were integrated between 550-740 nm (Fig. S2 †).MB photodegradation was presumed to follow rst order kinetics as indicated by the following expression: 20 The rst order rate constants (k) were determined by tting linear regressions of the MB concentrations as a function of irradiation time as shown in Fig. 3.The reaction rate constants followed k Au/SPICE > k SPICE > k Au/EISA ¼ k EISA > k Blank > k Dark and are listed in Table 1.
The value of k Dark was attributed to instrumental dri and was used as the error associated with the rate constants.SPICE TiO 2 showed a 13% increase in photocatalytic activity over EISA TiO 2 during MB degradation experiments.The addition of AuNPs to the SPICE TiO 2 samples (Au/TiO 2 ) further improved the photocatalytic activity by 43%.AuNPs did not improve the catalytic activity of the EISA TiO 2 .
Photocatalytic activity has been shown to depend on catalyst crystallinity, surface hydroxyl concentration, and surface area. 63ecause the EISA and SPICE methods share the same materials and polymer removal process, it is expected that crystallinity and surface hydroxyl groups remain similar.Both SPICE and EISA processes were carried out at room temperature except for the UVO treatment, which remained below 60 C. Therefore, the anatase phase is expected for both synthetic processes. 64The improvement in photocatalytic efficiency was attributed to increased surface area, as the SPICE-templated TiO 2 surface area was increased by minimizing the TiO 2 size distribution (Fig. S3 †).Exposed surface area was estimated to be 25% larger for SPICE TiO 2 over EISA TiO 2 ; uncertainty in the measurement stems from the variations in EISA dot sizes.In addition to an improvement over EISA TiO 2 , our experiments demonstrated a 27% activity gain between SPICE Au/TiO 2 over SPICE TiO 2 .On the other hand, our EISA Au/TiO 2 exhibited no improvement over EISA TiO 2 .Numerous previous studies have demonstrated improvements to EISA TiO 2 photocatalysts by adding metal nanoparticles.Catalysts were optimized by changes in synthetic techniques and metal type and loading.In some cases, AuNP addition caused losses in photocatalytic activity. 60,65Amongst our experiments, the simple addition of AuNPs to EISA TiO 2 yielded no change, which indicated that optimization is required.On the other hand, AuNPs that were simply spincoated onto the SPICE TiO 2 yielded an improvement of 27%.The improvement in efficiency indicates that the SPICE TiO 2 can be more easily optimized and augmented than EISA TiO 2 . 61,62

Conclusions
In conclusion, the SPICE templating method (spincoat-patternimmersion-complex-etch) was used to produce arrays of a variety of metal oxides.The method decoupled metal-PEO complexation from BCP ordering and thus enabled both exceptionally well-ordered arrays and simple sol-gel templating.A variety of metal oxides were templated; all exhibited the same high degree of ordering as their sacricial BCP thin lm template.In particular, TiO 2 arrays exhibited a high degree of ordering and narrow size dispersion, which could not be achieved using standard EISA methods.Finally, MB photodegradation studies showed that the SPICE TiO 2 (1) showed a 13% increase in photocatalytic activity over EISA TiO 2 and (2) could be augmented easily with gold nanoparticles to further improve photocatalytic activity by 27%, whereas the EISA Au/ TiO 2 showed no such improvements.The SPICE method opens exciting pathways for thin lms of ordered metal oxides in sensors, magnetic bit storage media, and energy harvesting.1.The value of k Dark was taken to be the error in the rate constants.

Fig. 1
Fig. 1 Atomic force microscopy height images of PS-b-PEO films after solvent vapor annealing with (a) toluene and (b) toluene/water.TiO 2 arrays were templated by (c) EISA and (d) SPICE; and (e) AuNPs were spincoated onto templated SPICE TiO 2 .Inset images are FFTs of the AFM images.Scale bars represent 200 nm.(f) Line scans show height profiles of the TiO 2 made by the SPICE (green, bottom) and EISA (blue, top) methods.

Fig. 2
Fig. 2 Transmission electron micrograph of (a) EISA (Ti precursor) films after annealing with toluene/water vapor and (b) toluene vapor as compared to a (c) SPICE Ti precursor film.Images are unstained; contrast is afforded by the electron density difference between PEO-titanium complex (dark) and PS (light).The scale bars represent 100 nm.

Fig. 3
Fig.3Plot of the natural log of MB concentrations against irradiation time.Error bars on data points represent standard deviations of the mean across three reproducibility studies.The slopes represent rate constants (k), which are reported in Table1.The value of k Dark was taken to be the error in the rate constants.

Further- more, we demonstrate an improvement in photocatalytic activity of SPICE TiO 2 and Au/TiO 2 over EISA TiO 2 . In addition to photocatalysis, we anticipate this method will be critical to developing advanced materials for sensors, environmental catalysis, and lithography applications.
by creating well-ordered hexagonally packed arrays of MgO, Al 2 O 3 , TiO 2 , MnO 2 , Fe 2 O 3 , Co 3 O 4 , NiO, CuO, ZnO, ZrO 2 , RuO 2 , SnO 2 , and Ce 2 O 3 .

Table 1
Photocatalytic conditions and rate constants a Error values for the rate constants are given by the magnitude of the Dark rate constant.