Introduction

The paramount challenge of the twenty-first century lies in safeguarding universal access to clean water and maintaining its quality for the entire ecosystem. The swift progress of contemporary industries has presented formidable challenges, impacting both energy consumption and contributing to environmental crises1. A notable concern revolves around the increasing production of synthetic dyes, prompted by their substantial demand, especially in the textile, pharmaceuticals, printing, food and clothing industries. These chemical dyes are being mass-produced, with global annual production reaching approximately 700,000 tons, reflecting the scale and impact of this industrial process2. One of the popular cationic dyes that are environmentally persistent, toxic, carcinogenic, and mutagenic is methylene blue (MB)2,3. Currently, the removal of MB from wastewater and aquatic environments has gained great attention from researchers, and led to significant increase in the number of publications during the last decades (Fig. 1). Dyes and other pollutants from the aqueous environment have become a challenging task in recent years. Hence, several strategies have been adopted to cope with this particular problem, including ozonation4, membrane filtration5,6,7, adsorption over heterostructure particles8, ion exchange removal9, adsorption10,11 and photocatalytic degradation12,13. The absorption process is usually preferred in terms of pre-release removal methodologies due to its ease of operation and economic cost14. Nonetheless, challenges such as low absorption efficiency, incomplete pollutant removal, and inadequate mechanical stability of adsorbents pose obstacles to the efficient elimination of pollutants15. Additionally, while membrane technology is applicable, a significant drawback exists—the fouling of membranes, requiring regular replacement when fouled or overloaded16.

Fig. 1: A general overview of MB global consumption and related treatment research.
figure 1

(a) The market size of MB, (b) application segments of MB utilization, (c) regional analysis for MB utilization, and (d) the number of publications per year for MB removal (from Scopus using “methylene blue” and “removal” keywords from 1980 to 2023).

Full size image

In recent years, there has been considerable focus on the photodegradation of pollutants. The energy from solar radiation on Earth within one hour surpasses the total energy consumed by humans throughout the entire year. The effective harnessing of solar energy has the potential to alleviate numerous energy and environmental challenges1. Since the identification of the Honda–Fujishima effect (photocatalytic water splitting on TiO2) in the 1970s, the field of photocatalysis has garnered escalating interest in multidisciplinary research. Initially, this catalyst exhibited activity only under ultraviolet light for water splitting on TiO2 electrodes due to its wide band gap (3.0–3.2 eV). Extensive efforts have been made to enhance the visible-light absorption (λ > 400 nm) of TiO2 photocatalysts. However, most modified and doped photocatalysts have demonstrated low activities, limited absorption of visible light, and relatively poor stability during the photocatalytic process17. Nevertheless, using prevalent photocatalysts like TiO2, ZnO2, perovskites18,19 necessitates continuous photo-assistance to facilitate redox reactions, significantly restricting their broad applications, particularly during nighttime20. When illumination ends, the photocatalysts engage in a photo-induced process, with photogenerated carriers (electrons and holes) being generated and participating in oxidation-reduction reactions. Once the light is extinguished, the semiconductor ceases the generation of carriers (electron-hole pairs), resulting in an immediate loss of catalytic activity1. Through the development of a materials system referred to as persistent photocatalysis, it is possible to enhance solar harnessing and catalytic activity even during overcast periods and after sunset. This solution aims to address the challenge effectively. The concept involves capturing and storing the photogenerated carriers in a suitable capacitor or battery-like material (electron storage) during the catalyst’s illumination phase. Subsequently, these stored carriers are thermally discharged at a slow rate after the removal of the irradiation light source. This process may exhibit a delayed photocatalysis effect, similar to the “delayed” persistent luminescence21,22. In recent years, various photocatalysts showcasing an intriguing round-the-clock photocatalytic activity, often referred to as the “memory” effect, have been developed. This unique characteristic was initially identified in a composite photocatalyst featuring PdO nanoparticles modifying nitrogen-doped TiO2 back in 200823,24. It was observed that a portion of their photoactivity could be retained in memory during illumination, allowing them to sustain activity even after the light source was turned off for an extended period. Furthermore, numerous photocatalysts exhibiting this memory effect have been documented for various environmental applications, including Mo-TiO225, Cu2O/TiO224, Cu2O/SnO226, and I/TiO226. Nonetheless, these materials are composed of two distinct components. Besides the essential element for photocatalysis, the photocatalyst (PC), there is an additional component, the energy storage substance, responsible for initiating the catalytic reaction under dark conditions. The photocatalysts reported with round-the-clock photocatalytic activity are predominantly metal-based semiconductors. Consequently, there is a current pursuit of alternative metal-free photocatalysts, which hold additional allure due to their enhanced cost-effectiveness and lower toxicity compared to various inorganic structures1. There are only a limited number of publications on metal-free, round-the-clock photocatalysts. For instance, Zang et al.27 introduced a metal-free photocatalyst with a memory photocatalytic effect, consisting of graphitic carbon nitride (g-C3N4) as the photocatalyst and carbon nanotubes (CNTs) and graphene (Gr) serving as supercapacitors for phenol removal. They proposed that the material retains a fraction of its photocatalytic activity by storing photogenerated electrons in CNTs and Gr, subsequently releasing them in dark conditions27. Recently, significant research endeavors have been directed toward developing a novel class of materials as highly efficient photoactive substitutes for conventional photocatalysts, including covalent organic frameworks (COFs) and porous organic polymers28,29,30. These materials offer the benefit of an easily controllable electron band structure, remarkable physicochemical stability, cost-effectiveness, and straightforward synthesis utilizing readily available resources abundant on Earth. For instance, conjugated microporous polymers (CMPs) based on triazine were developed for the photocatalytic conversion of CO2 to CO under visible light. The resulting polymers showed that the organic donor-acceptor junctions formed to improve charge separation are integrated with organic electron-withdrawing and electron-donating units on the triazine ring skeleton, allowing manipulation of the optical band gap of the CMPs28. One class of polymeric materials used as photocatalysts is donor-acceptor (D–A) polymers. The D–A structure of these polymers provides it with effective electron-hole separation and transfer31. Among polymeric photocatalysts, polyimide (PI), a polymer with a typical binary donor-acceptor (D–A) structure composed of electron-rich amine moiety and electron-deficient anhydride moiety, exhibited excellent photoactivity towards different reactions, including water splitting31,32,33,34, CO2 reduction35, and variety of application including water treatment36 and organic synthesis37,38. The PIs were synthesized using a polycondensation reaction of diamine and dianhydride monomers, which makes it easy to tailor their properties, including optical properties (i.e., bandgap energy) within the visible light range from 1.9 to 3 eV by changing the monomer. However, it is worth noting that PI has applications as triboelectric nanogenerators as charge storage. Feng et al.37 found that the use of polyimides as transition layers showed perfect charge-keeping ability with a decay rate of about 20% in 4 h. Furthermore, these structures can perform reversible redox processes involving two or four electron transfers thanks to the presence of four carbonyl groups at the imide terminals, producing high–capacity values of more than 150 mAh g−139.

In this study, a series of metal-free catalysts were synthesized from polyimide of intrinsic microporosity. Thermal annealing was employed to modulate the porosity and functionalities of the resulting catalyst at varying temperatures (i.e., 530, 600, 800, and 1000 °C) (Scheme 1). The impact of porosity and chemical structure on the continuous photodegradation of methylene blue was assessed through the preparation of benchmark polyimides. State-of-the-art characterization techniques, including scanning electron microscopy (SEM) and Brunauer–Emmett–Teller (BET) surface area analysis, were utilized to scrutinize the resulting catalysts. The findings suggest promising prospects for the application of polyimides of intrinsic microporosity as photocatalysts for persistent dye degradation.

Scheme 1: Synthetic routes of 6FDA-DMN-based catalysts.
scheme 1

(a) Synthesis of 6FDA-DMN polymer, (b) 3D structure of 6FDA-DMN and its proposed carbonized version, and (c) structures of 6FDA-TrMPD and 6FDA-mPDA for comparison purposes.

Full size image

Results and discussion

Synthesis and characterization of 6FDA-DMN-based catalysts

6FDA-DMN was prepared by a one-step high-temperature polycondensation reaction at 200 °C in m-cresol and in the presence of isoquinoline (Scheme 1). The molecular structure of the 6FDA-DMN was confirmed by NMR (Supplementary Figs. 1 and 2) and FTIR analysis (Fig. 2a). The absence of 1H NMR peaks above 10 ppm and the –COOH stretching band in FTIR confirms the total conversion of poly (amic) acid to polyimides. The 6FDA-DMN polymer demonstrated excellent solubility in organic solvents (Supplementary Table 1) and high thermal stability with an onset decomposition temperature exceeding 500 °C, making it suitable for pyrolysis at elevated temperatures (Fig. 2b).

Fig. 2: Characterization of 6FDA-DMN-based catalysts.
figure 2

(a) Fourier transform infrared (FTIR) spectra, (b) Thermogravimetric analysis (TGA) thermograms, (c) Energy-dispersive X-ray (EDX), (d) Wide angle X-ray diffraction (WXRD) patterns, (e) N2 adsorption isotherms and (f) pore size distribution.

Full size image

6FDA-DMN-derived metal-free catalysts were prepared by carbonizing 6FDA-DMN polymer at a temperature above the onset decomposition temperature (i.e., 530, 600, 800 and 1000 °C) to alter its functionalities and porosity. The catalyst designation and porosity measurements are reported in (Table 1).

Table 1 Sample designation, surface area, and total pore volume of 6FDA-DMN-based catalysts
Full size table

The FTIR spectra of the 6FDA-DMN and its carbonized counterpart are shown in Fig. 2a. The pristine sample P0, displayed C–N absorption band at 1350 cm−1 whereas the bands observed at 1724 and 1789 cm–1 present the symmetric and asymmetric vibration of C=O stretching in imide groups, respectively. Meanwhile, for the carbonized samples P1–P4, the intensities of characteristic bands become weaker with the increase in carbonization temperature, suggesting the increase in the degree of carbonization of 6FDA-DMN-based catalysts. These results are in good agreement with EDX, as shown in Fig. 2c where the intensity of carbon peak increases significantly upon increasing carbonization, leading to a reduction in the intensity of other elements (Supplementary Figs. 3–6).

To further investigate the effect of the pyrolysis on the resulting catalysts’ morphology, WXRD analysis was carried out to identify their microstructure and inter-layer distances (Fig. 2d). All spectra exhibited broad peaks, which imply an amorphous nature. P0 demonstrated a broad peak at 2θ = 13.5° with the d-spacing value of 6.5 Å. After carbonization, a notable shift in the major peak was observed in which P1, P2, P3, and P4 exhibited major peak at 2θ = 13.8°, 20.6°,23°, 24° with d-spacing of 6.3, 4.3, 3.8 and 3.7 Å, respectively, representing (d002) in graphite40,41 Moreover, a new peak at 2θ ≈ 44° with a d-spacing of ≈2.1 Å corresponding to the (100) plane in the graphite lattice was observed after carbonization42. This represents the mixture of graphitic and turbostratic structures, contributing to the narrowing of pore size distribution41. Generally, the turbostratic structure is more favorable than graphitic because it has more spacing between the planes due to disordered arrangements of the molecules, which means high porosity and surface area. However, the lower d-spacing means lower inter-plane distances provide excellent molecular sieving properties, which is a high-demand property in the synthesis of gas separation materials43.

Porosity remains a crucial factor for efficient catalytic activities due to the importance of surface area. Therefore, the nitrogen adsorption isotherms were measured at –196 °C and presented in Fig. 2e along with their corresponding pore size distributions (Fig. 2f). All 6FDA-DMN derived catalysts displayed a typical type II isotherm with a rapid increase at a relatively low p/p0 region (<0.05), which is mainly ascribed to the existence of microporous structures44. It was noticed that as the pyrolysis temperature increases, the BET surface area increases, which aligns with previously reported behavior. For instance, P0 demonstrated a BET surface area of 521 m2 g–1, which increased by 40% upon carbonization at 1000 °C for P4 (729 m2 g–1). Moreover, the total pore volumes at a relative pressure p/p0 of 0.85 were varied between 0.31 and 0.34 for all samples. Furthermore, the pore size distribution (Fig. 2f) revealed that carbonizing 6FDA-DMN facilitated small pores formation in the carbon molecular sieve structure, indicating pore shrinkage and chain tightening. It is worth mentioning that the polymer was carbonized in order to increase its surface area and thus improve its adsorption properties and maybe catalytic activities.

The surface morphology of the 6FDA-DMN-derived catalysts was characterized by SEM analysis and presented in Fig. 3. The carbonized samples exhibited surface porosity with average pore diameter ranges between 281 and 481 nm. No clear trend was observed for the pore size changes at the surface of the catalysts. However, it was noticed that increasing the temperature from 530 to 800 °C led to a significant increase in pore diameter from 378 to 481 nm, which could be attributed to the chain rearrangement and graphite formation at elevated temperatures. Surprisingly, the average diameter was decreased upon increasing the temperature to 1000 °C. This notable reduction in pore diameter could be attributed to chain compaction and reduction in internal d-spacing as noted from WXRD, where P4 displayed a higher intensity peak at 2θ ≈ 44° with a d-spacing of ≈2.1 Å, representing further shrinkage in pores at very high temperature (Fig. 1d).

Fig. 3
figure 3

Scanning electron microscope (SEM) with pore diameter distribution for the surface of the prepared 6FDA-DMN-derived.

Full size image

To further investigate the photophysical properties of the photocatalyst, photoluminescence (PL) and fluorescence lifetime spectroscopy analyses were conducted on the obtained materials, including pristine and carbonized polymers at different annealing temperatures (P0–P4) (Supplementary Fig. 7a, b). The photoluminescence spectra exhibited a peak at ~640 nm, attributed to the charge recombination of photogenerated electron-hole pairs. Notably, the PL intensity strongly increased with higher annealing temperatures. However, the acceptor-donor moieties in the polymer backbone effectively quenched the photoluminescence peaks. This quenching is likely due to efficient charge separation in P0, resulting from donor-acceptor functionalities that generate a significant dipole moment compared to the carbonized materials45. As known, the uneven charge distribution of chemical bonds comprising component ions or atoms with varying electronegativity in a molecular structure results in the creation of dipole moment46. As the polymer is carbonized, carbon-carbon bonds form, leading to an even distribution of electron density and symmetrical charge distribution. Consequently, the dipole moment promotes molecular charge transfer upon excitation, with electron-hole pairs distributed separately over the acceptor and donor moieties (Supplementary Fig. 7c)45.

Fluorescence lifetime spectroscopy revealed the decay curve of photogenerated electrons. The decay curves for P1–P4 samples were well-fitted by mono-exponential function decays, indicating a single component involved in charge carrier recombination, whereas P0 showed the best fit with a bi-exponential function. The average lifetimes (τav) obtained were 733, 610, 583, 580, and 581 ps for P0, P1, P2, P3 and P4, respectively. The charge carrier lifetime decreased from P0 to P4 as functional groups were removed from the structure. A shorter fluorescence lifetime implies rapid electron-hole recombination and vice versa. Thus, moving from P0 to P4, the lifetime decreased, indicating a slower recombination rate for P0 compared to the others.

Photocatalytic degradation of MB using 6FDA-DMN-derived catalysts

To assess the photocatalytic capabilities of the synthesized catalysts, methylene blue was employed as a model pollutant during visible light exposure. Before conducting the photocatalytic experiments, optical bandgaps were determined through Tauc plots (Supplementary Fig. 8), revealing bandgap values of 1.9, 1.8, 2.9, 3.0, and 3.1 eV for P0, P1, P2, P3, and P4, respectively. Furthermore, a control experiment was conducted to measure the degradation of methylene blue in an aqueous solution without a photocatalyst exposed to visible light for 6 h. The results indicated negligible photolysis of methylene blue in this context, as shown in Fig. 5a.

Prior to the photocatalytic degradation test, the adsorption capacity of the catalysts was measured in the dark after 2 h (Fig. 4a). P0 effectively eliminated around 60% of the dye. All prepared catalysts demonstrated commendable adsorption capabilities, with respective adsorption capacities (Qads) of 8.7 mg g−1, 9.7 mg g−1, 5.47 mg g−1, 6.12 mg g−1, and 3.87 mg g−1 for P0, P1, P2, P3, and P4, respectively. P0 and P1 exhibited the highest adsorption capacity in the absence of light, with dye removal efficiencies of 56% and 63%, respectively. Conversely, materials carbonized at higher temperatures (i.e., P2, P3, and P4) demonstrated lower adsorption efficiency, attributed to the loss of polar functional groups during pyrolysis. However, the 6FDA-DMN-based catalysts displayed outstanding photoactive properties under visible light, owing to their acceptor-donor polymer nature, as previously reported32. P0–P4 showcased photocatalytic activity under visible light, with MB photodegradation efficiencies ranging between 61% and 89% (Fig. 4a, Supplementary Fig. 15). These results underscore the crucial role of functional groups in photocatalytic efficiency, while carbonization, although increasing porosity, diminishes photocatalytic activity due to the absence of functional groups.

Fig. 4: The photocatalytic performance of the 6FDA-DMN-derived catalyst.
figure 4

(a) the adsorption and photodegradation of MB, (b) pseudo-first-order plot for MB photodegradation, (c, d) the continuous photodegradation of MB in dark using P0 and P1, respectively (memory effect representation), (e, f) the effect of pre-irradiation of P0 and P1, respectively, in solid-sate on the photodegradation of MB in dark.

Full size image

The kinetics of MB degradation using 6FDA-DMN-based catalysts were assessed and illustrated in Fig. 4b and Supplementary Table 2. Generally, the degradation of the dye increased with longer exposure to light. In the first 30 min, the degradation rate was slow but accelerated with continued light exposure. This behavior could be attributed to the availability of active sites on the catalyst’s surface. Due to the high porosity of these materials, the initial degradation rate might be affected by the substantial amount of MB adsorbed on the surface under dark conditions. During the initial period, the catalyst might degrade the adsorbed MB molecules, freeing up space for additional MB molecules to be adsorbed and then degraded. Subsequently, the degradation rate increases significantly as more active sites become available for the dye molecules. All samples adhered to the pseudo-first-order kinetics, which is typical for photocatalytic catalysts, exhibiting degradation rates ranging from 0.004 to 0.0114 min–1. Furthermore, P0 and P1 demonstrated shorter half-times (the time required to degrade half of the initial quantity of MB) compared to the other samples, indicating their rapid photocatalytic degradation behavior.

Interestingly, upon the cessation of light irradiation (after a 4-h exposure period), an intriguing shift in the behavior of MB concentration evolution in the dark was explicitly noted in the P0 and P1 samples. Despite the absence of light, the gradual disappearance of the MB dye continued to occur (Fig. 4c, d and Supplementary Table 3). This persistent degradation hints at the presence of a memory effect ingrained within the polymer structures, wherein the photocatalytic activity persists even in the absence of external light stimulation. In light of this discovery, a series of pre-irradiation experiments and analyses were meticulously conducted to delve deeper into the underlying mechanisms and elucidate this intriguing phenomenon further.

Round-the-Clock Photocatalytic Degradation of the MB dye

Among the 6FDA-DMN-derived catalysts investigated in this study, both the P0 and P1 demonstrated sustained photocatalytic activity persisting even after the cessation of light exposure for up to 14 h. This intriguing phenomenon warrants further in-depth investigation. In the existing literature, numerous methodologies have been employed to assess persistent photocatalysis, showcasing performance even under dark conditions46. One such approach involves maintaining the reaction post-light-off and monitoring dye degradation over a specified duration. Following the photodegradation process, the light source was deactivated, and the catalysts were left in contact with the remaining dye in the solution. Notably, a gradual disappearance of dye color was observed over the course of 14 hours, ultimately resulting in the degradation of approximately 90% of the residual dye (Fig. 4c, d, Supplementary Figs. 9–13). Nevertheless, to corroborate the presence of the memory effect in P0 and P1, both catalysts underwent pre-illumination by exposure to visible light for 4 hours in their solid state before being applied to the dye solution under dark conditions27. Subsequently, the dye concentration gradually decreased until it reached its maximum reduction. P0 demonstrated a 41% deviation compared to the blank samples (non-irradiated), while P1 exhibited a 37% decline in dye concentration. This observation validates the memory effect, wherein the polymers retained the effects of light exposure and subsequently released it gradually to degrade the dye molecules.

Scavenger experiments were performed using P1 to investigate the active species generated in the photocatalytic, as shown in Fig. 5. Ethylenediaminetetraacetic acid (EDTA), p-benzoquinone (pBQ), H2O2, and isopropanol (ISP), were used as scavengers to capture holes (h+), superoxide radicals (•O2–), electrons (e–), and hydroxyl radicals (•OH), respectively by adding 4 mmol of each. The photodegradation of MB can proceed via the following three routes47,48,49

$${{{\rm{\bullet }}}\, O}^{2-}+{MB}\to {H}_{2}O+{{CO}}_{2}$$
$${{{\rm{\bullet }}}\, {OH}}^{-}+{MB}\to {H}_{2}O+{{CO}}_{2}$$
$${{{\rm{\bullet }}}\, h}^{+}+{MB}\to {H}_{2}O+{{CO}}_{2}$$
Fig. 5: Scavenging tests and catalyst regeneration of P1.
figure 5

(a) The scavenger’s effect on the photocatalytic degradation under light-on and light-off, (b) the scavenger effect under light-off, (c) regeneration of the catalyst P1 under light-on, (d) regeneration of the catalyst after light-off.

Full size image

By irradiating the photocatalyst with photons holding energy that exceeds their band gap energy, electron-hole pairs were generated. By the interactions of electrons e– and holes h• with dissolved oxygen and water, the active spacious were formed according to the following reactions50:

$${O}_{2}+{e}^{-}\to {{{\rm{\bullet }}}\, O}^{2-}$$
$${h}^{+}+{H}_{2}O\to {{{\rm{\bullet }}}\, {OH}}^{-}$$

From Fig. 5a, the degradation was significantly suppressed after the addition of isopropanol and slightly affected by the addition of EDTA under light-on. Moreover, a notable decrease in photocatalytic activity (>60%) was observed when •OH is captured from the system by isopropanol, which reflects that the hydroxyl radical plays a crucial role in MB degradation. At the same time, the holes show a minor contribution with 23% inhibition of the photocatalytic activity during the light-on conditions upon adding EDTA. Furthermore, the addition of pBQ and H2O2 did not affect the degradation performance, implying that superoxide radicals •O−2 and electrons were not the main reactive oxygen species responsible for the degradation of MB as the following suggested mechanism:

$$P1+{hv}\to P1\left({h}^{+}\right)+P1({e}^{-})$$
$${h}^{+}+{H}_{2}O\to {{{\rm{\bullet }}}{OH}}^{-}$$
$${{{\rm{\bullet }}}\, {OH}}^{-}+{h}^{+}+{MB}\to {degradation\; of\; MB}$$

On the other hand, another test was performed by adding the scavengers after the light was turned off (after illumination of P1) to explore if there were any differences before and after the light off. Figure 5b shows no significant changes in the behavior observed in the mechanism after turning the light off. Moreover, based on the scavenging experiments •OH was identified as the reactive oxygen species responsible for the degradation. Thus, the presence of •OH can lead to partial oxidation of aromatic compounds and ring opening, as previously reported51,52,53, leading to the degradation of MB.

To test the catalyst’s stability and reusability, the catalyst P1 was isolated after the first run, washed with water and acetone, dried in the oven at 100 °C, and used for further cycles. As shown in Fig. 5c, d the photocatalyst exhibited good reusability with only a 23% decrease in performance under light-on conditions and 14.6% after light-off after five cycles. The photochemical efficiencies based on the total number of photons incident on the reactor for the disappearance of MB were evaluated at different wavelengths 470, 580, and 650 nm. The catalyst in the MB solution was irradiated by a 300 W lamp, which applied different band-pass filters for 4 h. The average irradiation intensity was determined to be 1.14, 1.651, and 2.808 mW cm−2 by a power meter (Newport, Model 2936-R). The apparent quantum efficiency (AQE) was calculated using Eq. (1)54

$${AQE}\left( \% \right)=\frac{2\times {nMB}\times {N}_{A}\times {hc}}{P\times S\times {\lambda }_{{in}}\times t}\times 100$$
(1)

Where nMB is the moles of degraded MB molecules, NA is the avocado’s number, 2 is the number of reacted electrons, h is Planck’s constant, c is the speed of light, P is the power density of incident light, S is the incident area, λin is the wavelength of the incident monochromator light and t is the duration of reaction54. The apparent quantum efficiency (AQE) for the P1 during photocatalytic MB degradation as a function of the monochromatic wavelength is plotted in Supplementary Fig. 14. The trend of the AQE and irradiation wavelength for MB degradation agreed well with the absorption edge for P1 photocatalyst. The AQE values were 10  ±  1.5% at 470 nm, 4 ± 1% at 580 nm, and 2 ± 1% at 650 nm.

To further validate the memory effect resulting from 6FDA-DMN-derived catalysts, impedance spectroscopy was used to compare the duration of charge release. The Nyquist diagram (spectra of the real and imaginary parts of the complex impedance, \({{{\rm{Z}}}}^{* }\) (\({{{\rm{Z}}}}^{* }={{\rm{Z}}}^{\prime} +{{\rm{jZ}}}{\hbox{''}};{{{\rm{j}}}}^{2}=-1\)), as a function of the exciting frequency, was demonstrated in Fig. 6a–c to depict the electrical response of the samples. P0–P2 exhibited similar curve patterns, with experimental points dispersed and forming a roughly circular arc. This behavior suggests the presence of an electric dipole, potentially created by a resistance connected in parallel to a capacitor. However, a simple model comprising only a resistance and a capacitor element is inadequate to fully explain the results as the scattered points do not perfectly align into half circles. Consequently, it is commonly understood that incorporating a Constant Phase Element (CPE) in parallel with resistance (R) provides a more accurate representation of the circuit-fitting parameters55. Figure 6d presents the equivalent electrical circuit of the samples, with the best fit depicted by the red line in Fig. 6a–c.

Fig. 6: Impedance spectroscopy analysis.
figure 6

Nyquist plots of (a) P0, (b) P1, (c) P2 and (d) the equivalent circuit.

Full size image

The maximum imaginary impedance spectrum Z′′ decreases by around three orders of magnitude (from giga Ohm to mega Ohm) as the temperature rises from ambient to 600 °C. Additionally, it demonstrates that while the capacitance roughly stays constant in nanoFarad, the temperature causes the resistance R to decrease from tera Ohm to mega Ohm. The extracted parameters from the fit for the circuit elements are collected in Table 2. The time constant, τ, (lifetime at the semiconductor’s depletion layer) for each sample is determined by multiplying the resistance by the capacitance of the prescribed loop. The results are summarized in Table 2.

Table 2 Resistance, capacitance, and lifetime of 6FDA-DMN based catalysts
Full size table

Since a capacitor never experiences an instantaneous energy charging (storage) or discharging (release), the τ parameter could be thought of as the amount of time needed to charge or discharge the capacitor within a specific percentage of its full supply. Accordingly, the τ parameter measures the amount of time that photogenerated charges are still present at the semiconductor interface and can produce reactive oxygen species that can break down the dyes. P0 demonstrated the highest τ value of 60.66 s, indicating its potential to store and release energy over a period of time. Interestingly, P2 showed speedy storage/release of energy within 1.25 ms, indicating its bad capability of round-the-clock degradation of the dyes. Although the impedance showed a significant difference in lifetime storage/release capacity between P0, P1 and P2, further investigation is needed to better understand this behavior.

To gain a more comprehensive understanding of the observed memory effect, a study correlating structure with phenomena was undertaken to elucidate the observation. This investigation sought to determine whether the memory effect is primarily influenced by the porous structure or the chemical composition of the polymer. Two polyimides sharing a common backbone were examined, with variations in the diamine component (Scheme 1). The selected structures differ solely in terms of porosity: 6FDA-mPDA is nonporous, while 6FDA-TrMPD features a porous structure. The obtained results exhibited that 6FDA-TrMPD demonstrated a photocatalytic degradation efficiency of approximately 54% (in the presence of light), with the reaction continuing after light off to degrade 28% of the remaining dye. In contrast, 6FDA-mPDA demonstrated a degradation efficiency of 43% under light, with minimal photodegradation occurring after light off. It is worth mentioning that both 6FDA-mPDA and 6FDA-TrMPD have the same chemical composition and functional groups, but they differ in porosity. Therefore, porosity could potentially increase the capability of the catalyst to store energy and allow photodegradation of the dyes in the dark.

According to Lan et al.56, the electrical, optical, and redox characteristics of the polymer were tuned by increasing the number of benzyl units in the polymer’s backbone that donate electrons, which enhanced the photocatalytic oxygen evolution capability. Therefore, to evaluate the effect of benzyl rings on the photodegradation in the dark, 6FDA-TrMPD was used for comparison. The two polyimides (i.e., 6FDA-DMN and 6FDA-TrMPD) are made from the same dianhydride (6FDA) and different diamines (i.e., DMN and TrMPD). Both polymers demonstrated similar BET surface areas (approx. 550 m2 g–1); thus, any changes in the memory effect would be a result of monomer chemical structures (mainly the presence of benzyl rings and conjugation).

Before conducting the photodegradation experiment, the adsorption efficiency of MB was assessed, revealing values of 56% and 23% for 6FDA-DMN and 6FDA-TrMPD, respectively (Fig. 7). Subsequently, a notable disparity in photodegradation efficiency emerged after illuminating the samples for 4 h. Specifically, 6FDA-DMN achieved an 82% degradation of the remaining dye under light, while 6FDA-TrMPD managed only 54%. Moreover, post-light exposure, 6FDA-DMN continued to degrade the dye for up to 12 h, reaching an 88% degradation of the remaining dye, whereas 6FDA-TrMPD exhibited a lesser capability, degrading only 28% of the remaining dye. These findings suggest that both polymers possess the capacity to store energy and release it in the dark; however, the memory effect was significantly more pronounced in 6FDA-DMN compared to 6FDA-TrMPD. Thus, the presence of more benzyl rings in DMN monomer was helpful for increasing the memory effect capability. These results contribute to a deeper understanding of the influence of porosity and functional groups on the memory effect and dye degradation. This study sets the stage for the further development of metal-free catalysts for round-the-clock photodegradation and the exploration of methods to enhance their performance.

Fig. 7
figure 7

The effect of porosity and monomer type on the persistent photodegradation of MB.

Full size image

Conclusion

In conclusion, novel metal-free catalysts derived from polyimide of intrinsic microporosity were synthesized and evaluated for the photocatalytic degradation of methylene blue. These catalysts exhibited substantial surface areas ranging from 521 to 729 m2 g–1 and demonstrated remarkable photocatalytic efficacy under visible light irradiation. Notably, the pristine 6FDA-DMN polyimide and the catalyst prepared via thermal annealing at 530 °C demonstrated the highest efficiency in methylene blue degradation. Furthermore, these catalysts showcased a distinctive photocatalytic memory effect, enabling continuous photodegradation round-the-clock, attributed to a synergistic interplay of porosity, functional groups, and the presence of benzene rings within the polymer matrix. Contrastingly, the nonporous polyimide (6FDA-mPDA) served as a benchmark, lacking any memory effect, highlighting the pivotal role of porosity in this phenomenon. Additionally, a comparison with porous polyimide (6FDA-TrMPD), possessing similar porosity but differing in the number of benzyl rings on the diamine molecule, revealed a less efficient yet present memory effect compared to 6FDA-DMN.

Methods

4,4′-(Hexafluoroisopropylidene) diphthalic anhydride (6FDA, 99%), 3,3′-dimethyl-naphthidine (DMN, 98%), m-phenylenediamine (mPDA, 99%), 2,4,6-trimethylbenzene-1,3-diamine (TrMPD, 98%) isoquinoline (97%), m-cresol (99%), chloroform (≥99.5%), N-methyl-2-pyrrolidone (NMP, 99.9%), methanol (≥99.9%) and methylene blue (MB, 99%) were purchased from Merck (KGaA, Darmstadt, Germany) and used as is.

Synthesis of 6FDA-DMN

6FDA-DMN was prepared by polycondensation reaction of dianhydride (6FDA) and diamine (DMN) at 200 °C, as previously reported57. The reaction commenced by combining equimolar amounts of 6FDA and DMN in m-cresol, along with a catalytic quantity of isoquinoline (0.1 mL), in a Schlenk tube under nitrogen. The mixture was heated to 100 °C, followed by a gradual increase in temperature to 200 °C to ensure complete imidization and full conversion of poly (amic) acid to polyimide. The resulting solution was precipitated in methanol, washed thrice, and dried in a vacuum oven at 150 °C for 24 h to obtain polymer fibers. FTIR analysis of the polymer powder revealed characteristic peaks at 1789, 1724 and 1350 cm−1 for stretching of C=O asym, C=O sym, and C−N, respectively. The polymer displayed a number-average molecular weight (Mn) of 68,000 g mol−1, molecular weight (Mw) of 84,000 g mol−1, and a polydispersity index of 1.23. The 5% decomposition temperature (Td,5%) was found to be 525 °C.

Synthesis of 6FDA-DMN-based catalysts

The polymer fibers were placed in an alumina crucible and transferred to a Carbolite furnace. Carbonization of the polymer fibers occurred at temperatures ranging from 530 °C to 1000 °C, with a heating rate of 2 °C min−1, followed by an hour of isothermal treatment. Subsequently, the furnace was allowed to cool to room temperature (22 °C). The carbonized samples were collected and stored in a desiccator.

Characterization

The Fourier transform infrared (FTIR) spectra of all samples were recorded on Bruker INVENIO Series FTIR spectrometer with 63 scans at a resolution of 4 cm−1 in the range of 400–4000 cm−1. Wide-angle X-ray diffraction (WXRD) patterns were collected from PanAnalytical diffractometer model Empyrean Alpha1 (Cu Kα radiation with a wavelength of 0.154 nm at 40 kV and 20 mA) with a scan speed of 0.014° s−1 in 2θ range of 4–70°. The d-spacing was estimated by Bragg’s Eq. (2):

$${{\rm{n}}}{{\rm{\lambda }}}=2{{\rm{dsin}}}{{\rm{\theta }}}$$
(2)

Thermogravimetric analysis (TGA) was performed on a PerkinElmer TGA 8000 instrument from 30 °C to 800 °C at a heating rate of 10 °C min−1 under a nitrogen atmosphere with a flow rate of 20 mL min−1. The Brunauer, Emmett and Teller (BET) surface area of the catalysts, pore size distribution, and pore volume were analyzed using the nitrogen adsorption isotherms measured at –196 °C, which were obtained using a Micromeritics ASAP 2020 adsorption analyzer. The number-average molecular weight (Mn), molecular weight (Mw), and polydispersity index (PDI) of 6FDA-DMN was obtained from gel permeation chromatography using an Agilent 1200 system with DMF and polystyrene. The microscopic morphologies of the prepared catalysts were measured using the ZEISS Gemini SEM 460 series instrument coupled with EDS; the samples were coated with a thin layer of gold/palladium (80/20%) before being loaded to the system. The images were taken at 50 pA probe current and 5 kV acceleration voltage using the SE2 detector. The optical properties (bandgap) of the obtained polymers were done using a Cary 7000 Universal Measurement Spectrophotometer (UMS) from Agilent. The photoluminescence (PL) spectra of all catalysts were obtained from powder samples that were excited with 400 nm and emission was measured at 640 nm using Duetta spectrofluorometer from HORIBA. The fluorescence lifetime measurement of the catalysts was obtained using a Quantaurus-Tau Fluorescence lifetime spectrometer. Frequency-dependent impedance measurements were made with the use of an impedance analyzer (PalmSens). The experiments were performed at room temperature by scanning the frequency between 1 Hz and 1 MHz while applying a voltage of 5 mVRMS. The samples are employed as thin films with dimensions of 13 × 10 × 2 mm. Then, they were irradiated under the light at a distance of 15 cm for 6 h. After that, they were sandwiched between two glasses that had been treated with conductive FTO electrodes.

Photocatalytic activity

The photocatalytic efficiency of the 6FDA-DMN-based catalysts was assessed using methylene blue (MB) as a representative pollutant. A photocatalytic setup included a reaction chamber, a white LED light source (emitting light in the 400 to 800 nm range, 300 W), and a fan to maintain a consistent temperature (22 °C) within the chamber. Initially, a 100 mL solution containing 15 mg L−1 MB was introduced into the reactor, followed by the addition of 100 mg of each catalyst. Stirring in darkness was employed to achieve adsorption-desorption equilibrium. The photocatalytic process commenced upon activation of the light source, with 3 mL samples withdrawn at 30-minute intervals for analysis using UV-visible absorption spectroscopy via a Duetta spectrofluorometer (HORIBA). To explore the photocatalytic memory effect, the reaction continued after the light source was switched off, with periodic sampling for analysis. Additionally, for samples subjected to pre-illumination, a 6-h light exposure was applied prior to the reaction under dark conditions.

Impedance spectroscopy

The total impedance of the circuit is calculated as follows:

$${Z}^{* }={Z}^{{\prime} }+j{Z}^{{\prime} {\prime} }=\left(\frac{1}{R}+\frac{1}{{Z}_{{CPE}}^{* }}\right)$$
(3)

In which the impedance of the CPE is defined via55:

$${Z}_{{CPE}}^{* }=\frac{1}{{A}_{0}{\left(j\omega \right)}^{n}}$$
(4)

where Z* represents the total impedance of the circuit, Z’ is the real component of impedance, Z” is the imaginary component of impedance, Z*CPE is the impedance of the constant-phase element. \({A}_{0}\) is a frequency-independent constant \(\omega\) is the angular frequency (\(\omega =2\pi f\)), and \(0 < n < 1\) is a dimensionless parameter that determines the degree of deviation from a perfect semicircle58. Equation (4) where \({A}_{0}=C\), yields the impedance of a capacitor when n = 1. The impedance curve’s intercept with the Z’ axis is the resistance R. The following relationships were used to fit the experimental scattered points.

$${Z}^{{\prime} }=\frac{R\left(1+R{A}_{0}{\omega }^{n}cos \left(\frac{n\pi }{2}\right)\right)}{1+2R{A}_{0}{\omega }^{n}cos \left(\frac{n\pi }{2}\right)+{\left(R{A}_{0}{\omega }^{n}\right)}^{2}}$$
(5)
$${Z}^{{\prime} {\prime} }=\frac{{R}^{2}{A}_{0}{\omega }^{n}sin \left(\frac{n\pi }{2}\right)}{1+2R{A}_{0}{\omega }^{n}cos \left(\frac{n\pi }{2}\right)+{\left(R{A}_{0}{\omega }^{n}\right)}^{2}}$$
(6)