Preserving nanoscale features in polymers during laser induced graphene formation using sequential infiltration synthesis

Characterization

Evidence for the successful incorporation of alumina into the PES membranes after SIS treatment can be seen in X-ray photoelectron spectroscopy (XPS) measurements (Fig. 1b), which show an absence of alumina before treatment and a strong intensity Al 2p peak after treatment consistent with Al₂O₃. Alumina incorporation was further confirmed by differences in membrane weight before (71.6 ± 0.1 g) and after (85.9 ± 0.5 g) SIS treatment, which suggests the treated films are 17% alumina by weight. The alumina content as determined by thermogravimetric analysis (TGA) was slightly higher at 28% (Supplementary Fig. 1a). This discrepancy could be explained by mass loss during the reaction of TMA with the PES membrane during SIS, making the actual loading higher than the mass change would suggest. Overall, however, the chemical composition of the original PES polymer is unchanged in the resulting organic–inorganic composite membrane, as confirmed by FTIR measurements (Supplementary Fig. 1b) and in agreement with the previous work³⁷.

The LIG pyrolysis mechanism has been studied previously for other polymers, though not specifically for PES, and is thought to involve thermal decomposition of C–O and C–N bonds due to the rapid increase in temperature, followed by growth of ring clusters forming graphitic structures³⁸. In the case of PES (and other sulfonated polymers), the transformation results in clusters with some insertion of sulfur in the graphene skeleton in the form of C-S-S and -C=S bonds²². Here, the conversion of PES to LIG is observed as a visual color change in the membranes from white to black (Supplementary Fig. 2) and is verified using Raman spectroscopy, which shows the presence of D, G and 2D bands characteristic of graphene containing materials (Fig. 1c). Optical images of the membranes lased at increasing laser powers (Supplementary Fig. 2) show that a critical laser power is required for conversion to LIG, as has been seen previously for other polymers¹³. The images also show that without the SIS treatment, the membranes transition through two regimes: at low but non-zero power, the membranes soften at their glass transition and become transparent (Supplementary Fig. 2); at a critical laser power, they exhibit the expected visual color change, though to not nearly as dark a color as the SIS-treated membranes and at higher laser powers than those required to convert SIS-treated membranes. SIS-treated membranes do not soften but instead graphitize directly, giving the first indication that the incorporation of alumina allows the membranes to resist changes to their nanoscale features during graphitization. Although the critical laser power that is required for LIG formation depends on lasing parameters such as the spacing between the laser scanning lines and laser speed, this same trend was observed at all setting tested, including the minimum spacing between scanning lines.

Cross-sectional scanning electron microscopy (SEM) images (Fig. 1d–f) demonstrate the powerful impact of SIS treatment on the membrane’s physical stability during lasing. Figure 1d shows the cross-sectional structure of a treated PES microfiltration membrane before lasing, along with a higher magnification image showing the micropores (Fig. 1g). When PES membranes without SIS treatment are lased, the membrane’s structure collapses into a dense bottom layer (Fig. 1f, k) and an exfoliated top layer (Fig. 1j). The membrane’s total cross-sectional thickness decreases to 38–58 μm compared to the initial 140 μm (Fig. 1d). In contrast, after lasing, SEM images of SIS-treated membranes have a thickness ranging from 90 to 132 μm (Fig. 1e, Supplementary Figs. 3–6) depending on the plane of the cross-section and the laser power used, showing that most of the membrane thickness is retained. Both the top 40.2 ± 1.0 μm of the membrane (high contrast in Fig. 1e, h), which is taken to be the lased region, and the remaining 85 ± 1.0 μm bottom layer (Fig. 1e, i) maintain an open and porous structure very similar to the starting membrane (see SI and Supplementary Fig. 7 for more detail about the membrane’s surface and cross-section structure). This indicates that graphitization happens in place without any macroscale deformation.

Performance

SIS-treated LIG membranes maintain the same permeability within uncertainty before and after lasing (872 Lm⁻² h⁻¹ bar⁻¹) (Fig. 2a). Without SIS treatment, membranes show a dramatic decrease (from 1124 Lm⁻² h⁻¹ bar⁻¹ to 35 Lm⁻² h⁻¹ bar⁻¹) in permeability, owing to pore-closure in the subsurface.

In addition to maintaining their permeability, SIS-treated membranes also exhibit relatively low sheet resistance. Conductivity measurements of the LIG-coated membranes with and without SIS treatment (measured by the Van der Pauw method) as a function of laser power (Fig. 2b) show that the SIS-treated membranes achieve a sheet resistance of 37.7 ± 0.7 Ω □⁻¹ or a conductivity of 660 S/m, based on the thickness of the conductive region seen in Fig. 1e. This sheet resistance is slightly higher than LIG formed from polyimide polymer, which show sheet resistance values as low as 15 Ω □⁻¹, but is comparable to CNT membrane coatings which exhibit similar sheet resistances^3,39. In contrast, the sheet resistance of lased membranes without SIS treatment (PES-LIG), which only graphitize and become conductive at higher laser powers (Supplementary Fig. 2), is an order of magnitude higher, in excess of 1000 Ω □⁻¹. Their conductivity is also highly anisotropic: the two-point probe conductivity of these membranes is much higher along the direction of the laser path compared to perpendicular to the laser path.

Beyond high electrical conductivity, during operation, conductive membranes must be electrochemically stable enough to sustain either a capacitive voltage to electrostatically repel foulants or a faradaic current to electrochemically degrade foulants, strip scalants, and generate gas bubbles to remove accumulated contaminants. To verify their electrochemical stability, SIS-treated, lased membranes were subjected a reducing current of 10 mA/cm² sufficient to perform water electrolysis⁴⁰. Testing was performed on strips of the membrane surface that were dipped into 0.1 M NaCl electrolyte solution. To prevent contact between the electrolyte and the electrical contact wire due to wicking, only a small fraction of the lased area was dipped into the solution, far from the electrical contact point. In this configuration, membranes showed no loss in performance up to at least 14 days of continuous operation (Fig. 2c). An initial drop in the voltage required to drive 10 mA cm⁻² of current density is attributed to the wicking of water further up the test strip during the first day of testing. Thus, laser-scribed SIS-treated membranes make excellent candidates for separations that require conductive membranes, or any technology where templated conductive structures are required.

Mechanism of improved stability

Given these favorable results, we explored the mechanisms behind the improved stability of SIS-treated membranes during lasing (i.e. suppression of deformation) and the improved conductivity of the LIG formed. One possible mechanism behind the structural stability during lasing is a change in the glass transition temperature (T_g) of the membrane due to the addition of alumina. However, DSC analysis (Fig. 3a) shows only a minor increase in the glass transition temperature (T_g) from 212 °C to 230 °C after SIS treatment. These values are consistent with previously reported T_g values for PES membranes^41,42 and further indicate that the inclusion of alumina does not prevent the polymer in the membranes from undergoing a glass transition. SEM images of the membranes after the DSC measurements (i.e. after heating them above their glass transitions) (Fig. 3b, c) show that the PES completely loses its original porosity, while the alumina infiltrated sample remains mostly porous, with slight deformation. This suggests that, despite the similarity in T_g values, the presence of alumina alters the rheological properties of the SIS-PES membranes and stabilizes the membrane structure under elevated temperatures above the T_g of PES (the laser irradiation increases the PES temperature well above 230 °C)¹³. To test this hypothesis, we performed dynamic mechanical analysis (DMA) to measure the tensile storage (G′) and loss (G′′) modulus of PES and SIS-PES as a function of temperature (Fig. 3d). While the PES sample undergoes a full transition from a glassy state at 200 °C to terminal flow (i.e. liquid-like) behavior at 275 °C with only a mild entanglement plateau, SIS-PES shows a slight relaxation above 230 °C, but then exhibits a prolonged plateau in G′ (around 0.1 GPa) up to the instrumental limit of 400 °C. The solid-like properties of the SIS-PES sample are also reflected by the minimal sample elongation during testing, especially compared to PES, which yields over 100% and prematurely ends testing (Supplementary Fig. 8). The dramatic difference in mechanical properties between the two samples and the extended plateau in G′ of SIS-PES suggests that the alumina has formed a continuous network that stabilizes the original membrane structure well beyond the T_g of PES⁴³.

While the infiltrated alumina is responsible for the mechanical structural resilience of the membrane during lasing, it is unlikely to be responsible for the improved conductivity and electrochemical stability of the LIG formed. Cross-sectional SEM and energy-dispersive X-ray spectroscopy (EDX) cross-sections of SIS-PES (Fig. 3e, f) show that after lasing, the top of the film is absent of aluminum introduced by SIS, likely due to sublimation under the high temperatures induced by the laser. At the LIG/PES interface, there is a small region where the alumina appears to have ripened into nanoparticles coating the polymer/LIG film but the conductive region of the resulting film is completely absent of the infiltrated alumina in its original crosslinked structure. Note that Fig. 3e is an SEM image of a membrane lased at lower laser power (14%) than the SEM image shown in Fig. 1e, which is why the thickness of the conductive layer is different. Further analysis of the PES-LIG and SIS-PES-LIG using XPS does not reveal significant chemical compositional differences. Both materials show primary C 1s peaks at a binding energy of 284.4 eV in XPS fine scans, consistent with sp2-bonded carbon¹³, with some additional higher binding energy peaks (Supplementary Fig. 9). While the PES-LIG shows a greater intensity of higher binding energy carbon peaks (286.2 eV), suggesting more ether carbon remains in the films after laser treatment, it is unlikely this difference in composition would lead to such a drastic difference in sheet resistance. Al 2p fine scans of the SIS-PES-LIG (Supplementary Fig. 10) also show that the aluminum within the film has remained in an oxide form, eliminating the possibility for the formation of Al metal. Raman spectra of PES-LIG and SIS-PES-LIG also do not exhibit significant enough differences to account for differences in conductivity. The D to G peak ratios in Raman spectra of carbonaceous materials are often used to make qualitative statements about the nature of LIG such as the crystallite size of graphitic clusters⁴⁴. The peak ratios of the Raman spectra for PES-LIG and SIS-PES-LIG (Fig. 1c) suggest similar crystallite sizes ranging from 7–10 nm. However, LIG from untreated PES have broader peaks indicating increased disorder in these films relative to the SIS-treated samples⁴⁵. In addition, some spots on the untreated PES films showed significant fluorescence (Supplementary Fig. 11a), indicating the presence of regions with minimal LIG coverage. Surface SEM images of the lased PES membrane without SIS show a heavily exfoliated structure with order-of-magnitude larger features (Supplementary Fig. 11a), and continuous regions of non-porous polymer underneath and in between, which is consistent with these Raman spectra. In contrast, spectra from SIS-treated PES after lasing showed consistent bands (Supplementary Fig. 11b) and SEM images of these membranes show a more homogeneous structure, similar to the pore-structure of the starting membrane.

The drastic differences in conductivity are therefore likely due to the less homogenous coverage of the lased surface by LIG. While membranes with and without SIS treatment show similar LIG thickness after lasing (Fig. 1e, f), lower magnification top-down SEM images of PES-LIG (Fig. 3g) reveal that after laser scribing, the laser creates linear trench structures of graphitized regions, separated by large gaps of non-porous polymer. These gaps between LIG structures also explain the high degree of anisotropy in conductivity mentioned earlier. A structural anisotropy is also visible in SIS-treated membranes (Fig. 3h). Previous work has shown that differences in lasing conditions can lead to vastly different LIG morphologies, in part due to differences in the resulting polymer temperature and anneal rate¹⁴. Here, regions outside of the direct laser path are still graphitized without any loss of porosity, but they likely do not reach the same annealing temperature and therefore exhibit different morphology.

Ultimately, the temperature that is achieved during laser irradiation plays a critical role in the resulting pyrolysis process and is affected by a combination of factors, such as the total absorption of IR light by the polymer and alumina (when present), the heat capacity of the membranes with and without alumina (which can be qualitatively deduced from the DSC data (Fig. 3a)), the structural changes due to polymer softening, the loss of porosity (without alumina), the rate of heat dissipation, and the total mass of polymer irradiated. While the relative impact of each of these factors was outside the scope of this work, it can be reasonably concluded that the presence of alumina increases the resulting membrane temperature during laser irradiation, since there is a reduction in critical laser power required for graphitization (Supplementary Fig. 2): the SIS-PES membranes can reach a higher temperature with lower laser energy and thus require less laser power to achieve the temperature required for graphitization.

The saturation in sheet resistance of the SIS-treated membranes above a laser power of 20–22% is also partially explained by the anisotropy in lased area. Initially, increasing the laser power is correlated with improved conductivity (Fig. 2b), as a greater fraction of the top surface of the membrane is converted to LIG and the laser continues to penetrate deeper into the membrane bulk. At 14% power, the laser is able to convert only parts of the membrane to LIG, which is seen as dark regions with low alumina content in surface SEM images, EDS maps, and EDS line scans (Supplementary Fig. 12). Cross-sectional images at 14% power show that the dark regions are part of a hemispherical path scribed by the laser (Supplementary Fig. 12). As the laser power is increased, the radius of this path grows and neighboring paths overlap until full surface coverage is achieved at which point the conductivity also saturates. We analyzed Raman spectra of the lased surfaces at different powers to see if higher laser powers changed the graphitic nature of the formed LIG. Between 14 and 18% laser power, there is an increase in the D to G peak intensity and area ratio, a narrowing of the D band full width at half max (Fig. 3i), and an emergence of a 2D peak (Supplementary Fig. 13). These changes plateau at higher powers, mirroring the conductivity measurements. These trends seen in the Raman spectra indicate that although initially, higher laser power leads to less disorder in the type of defects found on the LIG and increased stacking of graphitic clusters, defects found on the graphitic regions persist even at high powers⁴⁶. Thus, continued improvement in the sheet resistance of the membranes will require alternative approaches, such as other chemical treatments besides alumina.

Interfaces and interfacial properties play a central role in many technologies other than membranes at the water and energy nexus². PES membranes are taken as a prototypical example of porous polymers where nanosized features need to be preserved during laser scribing. The method of combining SIS of organometallic precursors into polymers with LIG formation described in this paper can be generalized to other applications where maintaining micro and nano-sized features of polymers at temperatures well above their T_g is desired. An even broader advantage of this approach is the improved mechanical and chemical properties that are observed at these high temperatures, without the need to change the chemistry of the underlying polymer²⁸. Other modification approaches that could potentially be used, such as crosslinking of the polymer, would alter the polymer chemistry, potentially altering the material’s LIG forming properties. Since stability of LIG formed from polymers is a known concern⁴⁷, the method described herein could be beneficial for all other polymers used for LIG formation as well.

In summary, we present a simple, solvent-free process for making conductive membrane coatings without altering underlying polymer structure. We demonstrate how infiltration with alumina stabilizes the PES membrane against deformation above the glass transition temperature³⁷, allowing it to maintain its structure during laser treatment. These membranes are shown to be more conductive than LIG formed directly from the bare polymer, are electrochemically stable and maintain their permeability after lasing. These results demonstrate the immense versatility of hybrid polymer-ceramic materials as a promising class of materials to be used in conjunction with the LIG process.

Source: Resources - nature.com