Report

Transient laser heating induced hierarchical porous structures from block copolymer–directed self-assembly

See allHide authors and affiliations

Science  03 Jul 2015:
Vol. 349, Issue 6243, pp. 54-58
DOI: 10.1126/science.aab0492

Laser patterning polymer membranes

Porous materials are useful for membranes, filters, energy conversion, and catalysis. Their utility often depends on the ability to finely control both the pore sizes and their connectivity. Tan et al. prepared porous thin films of block copolymers mixed with phenol-formaldehyde resins (resols) on silicon substrates using a simple laser process. On exposure to ultraviolet light, rapid heating of the substrate causes polymerization of the resols and decomposition of the block copolymer. This method allows direct patterning of the films on a local scale, with tunable pore sizes and size distributions.

Science, this issue p. 54

Abstract

Development of rapid processes combining hierarchical self-assembly with mesoscopic shape control has remained a challenge. This is particularly true for high-surface-area porous materials essential for applications including separation and detection, catalysis, and energy conversion and storage. We introduce a simple and rapid laser writing method compatible with semiconductor processing technology to control three-dimensionally continuous hierarchically porous polymer network structures and shapes. Combining self-assembly of mixtures of block copolymers and resols with spatially localized transient laser heating enables pore size and pore size distribution control in all-organic and highly conducting inorganic carbon films with variable thickness. The method provides all-laser-controlled pathways to complex high-surface-area structures, including fabrication of microfluidic devices with high-surface-area channels and complex porous crystalline semiconductor nanostructures.

Hierarchically porous materials that bridge nano- to macroscopic length scales are expected to find use in a wide variety of applications including catalysis, energy conversion and storage, and membrane filtration (1, 2). Well-ordered nanoporous materials derived from small-molecule surfactant and block copolymer (BCP) self-assembly have been explored in the form of amorphous (37), polycrystalline (8, 9) and single-crystal (10) solids. Multiple time-consuming processing steps are typically required to generate the final structures. For example, removal of organic components by conventional thermal processing to create porosity typically takes several hours (68). Methods to fabricate hierarchical porous polymer scaffolds using BCPs have recently been devised that circumvent some of these complexities (11, 12). Further developments are still desirable, however, as hierarchical structures provide high surface area combined with high flux beneficial for a number of applications.

Laser-induced transient heating of organic materials has been explored as an alternative thermal processing approach to alter morphology, as well as materials properties, and to enable direct pattern transfer. For example, by laser heating in the millisecond to submillisecond time frames, the temperature stability of a methacrylate-based photoresist was enhanced by ~400°C relative to conventional hot-plate heating (13). CO2 laser ablation of polyimide and polyetherimide resulted in the direct formation of three-dimensional (3D) porous graphitic carbon network structures, but work did not involve a bottom-up self-assembly process, resulting in little control over mesostructure and pore size (14).

Three-dimensional porous structure fabrication by optical lithography has also been demonstrated using multiphoton absorption polymerization (MAP) and stimulated emission depletion (STED)–inspired direct laser writing techniques (1519). However, direct laser writing of such 3D continuous structures with sub-100-nm feature size and periodicity remains challenging due to the inherent serial mode of operation and sophisticated instrumentation, diffraction limitations of MAP excitation lasers, and stringent materials criteria—e.g., for STED resists (1619).

We demonstrate a simple and rapid method for direct generation of 3D continuous hierarchically porous structures with meso- to macroscopic shapes by combining BCP self-assembly‒directed resols structure formation with CO2 laser-induced transient heating. A schematic of the BCP-based writing induced by transient heating experiments (B-WRITE) is shown in Fig. 1A (20). Lab-synthesized ABC triblock terpolymers [poly(isoprene)-block-poly(styrene)-block-poly(ethylene oxide), ISO-38/69] as well as commercially available ABA di-BCPs (Pluronic F127) were used as examples (20, 21). Oligomeric resols were synthesized by the polymerization of resorcinol/phenol and formaldehyde under basic conditions (22). The structure-directing BCP was first mixed with resols to form an all-organic hybrid thin film by spin-coating on a silicon substrate. Samples are designated as BCP-M-R, where M represents BCP molar mass and R denotes the resols additives. As-deposited or thermally cured films were then heated by a continuous wave CO2 laser (λ = 10.6 μm), focused into a ~90 μm by 600 μm line beam, and scanned to generate a 0.5-ms dwell in air. Although the organic layer does not interact strongly with the far-infrared photons, the boron-doped Si substrate absorbs most of the laser energy, thereby heating the film. As the laser passes, cooling occurs by thermal conduction on submillisecond time frames. Laser heating temperatures were determined using a procedure described in (13). Well-defined porous resin structures were formed by the simultaneous thermopolymerization of resols (negative-tone) and decomposition of BCPs (positive-tone) during transient heating by sequential CO2 laser irradiations at locations predefined by the beam. Material in nonirradiated regions could be removed by rinsing with solvent, leaving resulting porous resin structures on the substrate.

Fig. 1 Transient laser heating of BCP-directed resols hybrid samples.

(A) Schematic of the B-WRITE method. (B) Thickness plots of 5 wt % ISO-69, 5 wt % resols, and 6 wt % ISO-69-R hybrid thin films heated by a single CO2 laser irradiation of 0.5-ms dwell (solid lines) compared with furnace heating for 1 hour (dashed line), all in air. (C) Thickness plots of 6 wt % ISO-38/69-R hybrid thin films heated by sequential CO2 laser irradiations of 0.5-ms dwell (solid lines) compared with furnace heating for 1 hour (dashed lines), all in air. (D and E) Optical image of macroscopic trenches in sample ISO-38-R after sequential CO2 laser irradiations (D) with corresponding SEM cross section and plan view (inset) of the resulting structures in a trench (E). Each B-WRITE trench was generated in less than 5 min.

Thermal stability was compared for organic films prepared from 5 weight percent (wt %) ISO-69, 5 wt % resols, and 6 wt % ISO-38/69-R solutions in tetrahydrofuran (THF) spin-coated on Si (Fig. 1B). The 6 wt % ISO-38/69-R hybrid thin films were prepared by mixing 2 to 3 parts ISO to 1 part resols in THF by mass (see table S1 for hybrid compositions). After thermal curing in vacuum at 100°C overnight (>12 hours), films were heated in air by either a single CO2 laser irradiation for 0.5-ms dwell or furnace heating for 1 hour. Curves in Fig. 1B show film thickness as a function of temperature (and corresponding laser power). Under CO2 laser heating for 0.5-ms dwell, ISO-69 (black) remained stable up to 550°C (35 W) and thermally decomposed entirely above 670°C (>40 W). Resols films (orange) extensively cross-linked, retaining more than 50% of the original film thickness at 670 to 805°C (40–45 W). For ISO-69-R hybrid films (blue), mixed behavior was observed. This suggested a laser heating process window to decouple BCP decomposition but retain the cross-linked resols (resin).

Initial experiments to generate nanoporous resin structures in ISO-69-R films by a single CO2 laser irradiation with ISO decomposition at 670°C (40 W) for 0.5-ms dwell confirmed these concepts, albeit with poor uniformity and coverage (fig. S1). Thermal stability of ISO-69-R films was increased dramatically by more than 400°C when heating duration was reduced from 1 hour in a furnace to 0.5 ms with CO2 laser-induced heating (compare dashed and solid blue lines in Fig. 1B, respectively). Submillisecond time frame CO2 laser heating suppresses oxidation reactions in ambient environments, enabling organic systems to reach temperatures inaccessible by conventional heating.

For more controlled polymerization of resols and to inhibit thermocapillary dewetting effects (23), we introduced sequential CO2 laser irradiations of increasing laser powers yielding thicker resin structures with similar enhanced thermal stability (Fig. 1C), but high film uniformity and coverage (Fig. 1, D and E). Scanning electron microscopy (SEM) (Fig. 2 and fig. S2) and Fourier transform infrared (FTIR) spectroscopy (fig. S2) were employed to investigate film morphology and chemical composition, respectively. Upon sequential irradiations, cured ISO-resols hybrid films (fig. S2, A and C) densified due to cross-linking of resols with increasing transient temperatures (22), whereas above 550°C, BCP decomposition occurred (fig. S2, B and D) and was completed at about 670°C (Fig. 2, A and C). This is corroborated by decreasing IR signal intensities of hydroxyl stretching vibrations in the 3500 to 3200 cm−1 band and a sharp reduction of IR signal intensities of alkyl stretching vibrations in the 3000 to 2800 cm−1 band, respectively (22, 24) (fig. S2, E and F). Resulting resin structures retained ~30 to 55% of the original film thickness (Fig. 1C).

Fig. 2 SEM of B-WRITE resin structures.

(A to D) Plan views and cross sections (insets) of laser-heated 6 wt % ISO-38-R [(A) and (B)] and 6 wt % ISO-69-R [(C) and (D)] solution–derived structures of 2 to 3:1 [(A) and (C)] and 1.5:1 [(B) and (D)] ISO/resols mass ratios, respectively. (E and F) Lower-magnification images of same sample as in (C). (G and H) Plan views and cross sections (insets) of laser-heated 20 wt % F127-R solution–derived structures of 1:1 (G) and 1:2 (H) F127/resols mass ratios, respectively. (I and J) Plan views and cross sections (insets) of laser-heated 10 wt % (I) and 12.5 wt % (J) ISO-38-R solution–derived structures.

Optical images in Fig. 1D (and fig. S3) show macroscopic linear trenches in ISO-38/69-R hybrid thin films generated by sequential CO2 laser irradiations for 0.5 ms in air (B-WRITE method). Each ISO-38-R–based trench was ~2 cm long, requiring ~4.5 min to fabricate in air (Fig. 1D). SEM (Fig. 1E) and profilometry (fig. S3B) show that the ~150-μm-wide B-WRITE trench indeed consists of a ~200-nm-thick nanoporous resin thin film. Laser-heated macroscopic shape dimensions are tunable, as demonstrated for ISO-69-R hybrid films with ~350-nm-thick structures in ~550-μm-wide B-WRITE trenches (fig. S3F). Cross-sectional SEM at lower magnifications in Fig. 2, E and F, shows highly uniform structure formation under these conditions over large areas, as defined by the CO2 laser.

Resulting pore size and pore size distribution could be tailored by tuning hybrid film composition. Sequential CO2 laser-heated resin structures formed in the submillisecond time frames were compared with mesoporous structures formed by furnace heating under N2 for ≥1 hour (fig. S4). The average pore size of the N2-furnace-heated ISO-69-R structures at 450°C is 39 ± 9 nm with an in-plane lattice spacing of ~47 nm (fig. S5). Laser-heated structures obtained from ISO-69-R hybrid films with the same ISO-to-resols mass ratio of 2.4:1 show a wide pore size distribution with values ranging from ~30 to 200 nm (Fig. 2C). The increase in pore size relative to the N2-furnace-heated sample is ascribed to rapid evolution and release of gaseous decomposition products during transient heating, resulting in local pore expansion (14, 25). Hybrid film composition provides a convenient lever to tune pore size of B-WRITE structures. For example, Fig. 2D shows that the pore diameter of laser-heated ISO-69-R films was reduced to ~20 to 50 nm by decreasing the ISO-to-resols mass ratio to 1.5:1 (see Fig. 2B for ISO-38-R). From fast Fourier transform analysis, the in-plane lattice spacing was ~52 nm—i.e., close to the value obtained after furnace heating under N2 (fig. S5).

Terpolymers are synthetically challenging, so BCPs were varied to also include the commercially available Pluronics family (22). Smaller molar mass Pluronic F127 copolymer mixed with resols at 1:1 mass ratio in ethanol led to the formation of resin structures with pore sizes of 20 to 50 nm (Fig. 2G). Pore size and pore size distribution were further reduced to 10 to 30 nm by decreasing the F127-to-resols mass ratio to 1:2 (Fig. 2H), thereby approaching the 5- to 10-nm limit of size control in conventional BCP phase separation (26). Film thickness could be tuned via solution concentration. For example, maintaining ISO-38-to-resols mass ratio at 3:1, sample thickness was increased with the mixed solution concentration (table S1) from ~500 nm (6 wt %) to 1.0 μm (10 wt %) and 1.6 μm (12.5 wt %). Figure 2, I and J, show thicker laser-heated structures with hierarchical pore size distributions of 50 to 400 nm and 50 to 600 nm for 10 wt % and 12.5 wt % solution–derived ISO-38-R samples, respectively. B-WRITE–based porous structures may be accessible for films thicker than 1.6 μm by further optimizing laser heating protocols.

Skipping the 100°C thermal curing step and directly laser-heating as-deposited ISO-resols samples achieved further process simplification. Figure 3A displays macroscopic porous resin lines on a clean Si substrate prepared by subjecting as-deposited ISO-69-R samples to an increased number of sequential CO2 laser irradiations and removal of nonirradiated components by rinsing with THF (see stage III in Fig. 1A and SEM in fig. S6A). Selective removal of resin material with a single CO2 laser irradiation at 30 W of 25-ms dwell enabled macroscopic line-width adjustments and sidewall smoothening (see SEM in Fig. 2F and depth profile in Fig. 3B). More complex shapes were generated by removal of resin using laser scans in other directions, as shown for orthogonal scans in Fig. 3A.

Fig. 3 Macroscopic pattern inversion and carbonization of B-WRITE structures.

(A and B) Optical image of complex shapes (A) and corresponding depth profile (B) of laser-heated ISO-69-R after THF rinsing. Depth profile was acquired in the region denoted in (A). (C) Raman spectra (785-nm excitation) of laser-heated ISO-69-R formed by B-WRITE (light gray) and after carbonization at 600°C (gray) and 800°C (black). (D) Current/voltage plot of laser-heated porous carbon sample carbonized at 800°C.

A final furnace heating step under N2 at temperatures up to 800°C resulted in carbonized and electrically conductive nanoporous carbon films without structure collapse. Raman spectroscopy shown in Fig. 3C indicated negligible graphitic carbon in laser-heated ISO-69-R samples, corroborating the organic nature of the B-WRITE porous resin structures. After furnace heating at 600°C, D and G bands centered at ~1325 and 1588 cm−1, respectively, appeared in the Raman spectrum, indicating the conversion of phenolic resins into disordered carbon (22). At 800°C, D and G bands narrowed and shifted to ~1314 and 1598 cm−1, respectively, suggesting a higher degree of graphitization (22). Cross-sectional SEM (fig. S6B) revealed that B-WRITE structures contracted by 50 to 60% in the out-of-plane direction during carbonization at 800°C, but otherwise stayed intact. The electrical conductivity of B-WRITE films carbonized at 600°C and 800°C under N2 was ~8 S/cm (fig. S6C) and 270 to 750 S/cm (Fig. 3D), respectively. This electrical conductivity is higher than that of dense carbon samples heated under identical conditions at 800°C (~52 S/cm) (fig. S6D) but lower than polycrystalline graphite (1250 S/cm) (27). The high electrical conductivity of B-WRITE carbon structures after furnace heating is attributed to the connectivity of the 3D continuous network. We speculate that porosity facilitates carbonization and release of oxygen/hydrogen species to form graphitic-like clusters (fig. S6E).

We constructed proof-of-concept microfluidic devices with highly porous organic channel floors for potential sensing or catalysis applications (Fig. 4, A to C, and fig. S7). A dense resols film was spin-coated onto Si and cured at 100°C. A trench was then formed with a single CO2 laser irradiation at 30 W of 25-ms dwell, completely decomposing the film, followed by spin-coating a 6 wt % ISO-38-R film as an overlayer. B-WRITE application formed nanoporous resin structures in the trench. The device was finally sealed with a polydimethylsiloxane film (Fig. 4A). From optical and electron microscopy, as well as profilometry, the microfluidic device channel was ~13 mm long (port-to-port), 630 μm wide, and 1.4 μm tall, containing ~200-nm-thick nanoporous resin structures at the channel bottom, increasing surface area (fig. S7). Bright field and fluorescence optical micrographs in Fig. 4, B and C, confirmed that tetramethylrhodamine (TRITC) dye dissolved in dimethyl sulfoxide (DMSO) was flowing through the microfluidic channel using a pressure-controlled pump.

Fig. 4 Hierarchical porous resin structure formation and use in microfluidic device, on lithographically defined surfaces, as well as for pattern transfer into Si.

(A) Optical image of a microfluidic device fabricated by CO2 laser writing on the probe station. (B and C) Bright-field (B) and fluorescence (C) optical micrographs of TRITC dye dissolved in DMSO passing through the microfluidic channel. (D to N) SEM micrographs. Plan view (D) and cross sections [(E) to (G)] of hierarchical structures from coupling photolithography, BCP-directed self-assembly, and CO2 laser heating. Inset in (E) shows the identified area at higher magnification. Cross section of excimer laser-induced crystalline Si nanostructures employing B-WRITE–derived organic template (H). Higher resolution cross section (I) and plan view (J) of area as indicated in (H). Plan views [(K) and (M)] and cross sections [(L) and (N)] of periodically ordered mesoporous gyroidal resin template [(K) and (L)] and resulting excimer laser-induced crystalline Si nanostructures after template removal [(M) and (N)].

We further demonstrated compatibility of the B-WRITE method with complex substrates obtained from established semiconductor processing technologies. Figure 4D displays a SEM plan view micrograph of an ISO-69-R hybrid sample deposited on a photolithographically patterned Si substrate and transformed into porous resin structures by the B-WRITE method. Cross-sectional SEM in Fig. 4, E to G, shows excellent conformal adhesion of the laser-derived structures despite the complex Si surface topography.

Finally, we demonstrated an all-laser-induced pattern transfer to create several-hundred-nanometer-thick 3D porous crystalline Si nanostructures using an excimer laser-induced Si melt-crystallization process described previously (10). To that end, B-WRITE–derived ISO-69-R structures were further heated for 1 hour at 400°C under N2 to enhance adhesion to the substrate. The resin pore network was then backfilled with amorphous Si (a-Si) by chemical vapor deposition (CVD) (28) and irradiated with a single 40-ns pulsed XeCl excimer laser (λ = 308 nm) at 550 mJ/cm2 to induce a transient 40- to 50-ns Si melt, which subsequently solidified into polycrystalline Si (fig. S8). SEM in Fig. 4, H to J, shows nanoporous polycrystalline Si nanostructures after template removal, using CF4 reactive ion and wet acid etching treatments. What is notable about this experiment is the use of an organic BCP-directed resin template [as opposed to BCP-directed inorganic templates employed in (10)] that is not only stable for a-Si CVD at 350°C but also survives the transient melt of a-Si at temperatures ~1250°C, enabling high pattern-transfer fidelity (10, 29). All-organic BCP-directed structures open pathways to templates with high degrees of periodic order obtained—e.g., via thermal or solvent annealing before porous structure formation (30, 31). In a proof-of-concept experiment, SEM micrographs in Fig. 4, M and N, demonstrate that periodically ordered and networked mesoscopic crystalline Si nanostructures can be obtained after a-Si CVD into mesoporous gyroidal resin structures prepared by evaporation-induced self-assembly and pyrolysis (Fig. 4, K and L, and fig. S9) (21), excimer laser annealing, and template removal. Plan view (Fig. 4, K and M) and cross-sectional (Fig. 4, L and N) SEM images suggest that the interconnected and periodic pores of the mesoporous gyroidal resin template were conformally backfilled by the network struts of the resulting laser-induced crystalline Si nanostructures. Furthermore, the mesoporous all-organic resin template is inert to hydrofluoric (HF) acid, facilitating removal of native SiO2 layers on the Si substrate using diluted HF (or HF vapor) before a-Si deposition and enabling formation of periodically ordered porous single-crystal epitaxial nanostructures (10).

Supplementary Materials

www.sciencemag.org/content/349/6243/54/suppl/DC1

Materials and Methods

Table S1

Figs. S1 to S9

References (32, 33)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: This work was supported by the National Science Foundation (NSF) Single Investigator Award (DMR-1409105). K.W.T. gratefully acknowledges the Singapore Energy Innovation Programme Office for a National Research Foundation graduate fellowship. This work made use of research facilities at the Cornell Center for Materials Research (CCMR), with support from the NSF Materials Research Science and Engineering Centers (MRSEC) program (DMR-1120296); Cornell NanoScale Science and Technology Facility (CNF), a member of the National Nanotechnology Infrastructure Network, which is supported by the NSF (grant ECCS-0335765); Nanobiotechnology Center shared research facilities at Cornell University; and Center for Energy and Sustainability at Cornell University. We acknowledge experimental assistance from P. Carubia and T. Yuasa for attenuated total reflectance FTIR spectroscopy measurements; D. Lynch for electrical conductivity measurements; Y. Sun and Q. Zhang for microfluidic device fabrication; and T. Corso (CorSolutions LLC) for microfluidic probe station setup. We also gratefully acknowledge the expertise and experimental assistance from M. Goodman and P. Braun (University of Illinois at Urbana Champaign) for amorphous silicon chemical vapor deposition. We thank L. Estroff, Y. Gu, T. Porri, H. Sai, and M. Weathers for helpful discussions. A patent application related to this research has been filed by Cornell University.
View Abstract

Navigate This Article