Enhanced Organic Photocatalysis in Confined Flow through a Carbon Nitride Nanotube Membrane with Conversions in the Millisecond Regime

Bioinspired nanoconfined catalysis has developed to become an important tool for improving the performance of a wide range of chemical reactions. However, photocatalysis in a nanoconfined environment remains largely unexplored. Here, we report the application of a free-standing and flow-through carbon nitride nanotube (CNN) membrane with pore diameters of 40 nm for confined photocatalytic reactions where reactants are in contact with the catalyst for <65 ms, as calculated from the flow. Due to the well-defined tubular structure of the membrane, we are able to assess quantitatively the photocatalytic performance in each of the parallelized single carbon nitride nanotubes, which act as spatially isolated nanoreactors. In oxidation of benzylamine, the confined reaction shows an improved performance when compared to the corresponding bulk reaction, reaching a turnover frequency of (9.63 ± 1.87) × 105 s–1. Such high rates are otherwise only known for special enzymes and are clearly attributed to the confinement of the studied reactions within the one-dimensional nanochannels of the CNN membrane. Namely, a concave surface maintains the internal electric field induced by the polar surface of the carbon nitride inside the nanotube, which is essential for polarization of reagent molecules and extension of the lifetime of the photogenerated charge carriers. The enhanced flow rate upon confinement provides crucial insight on catalysis in such an environment from a physical chemistry perspective. This confinement strategy is envisioned not only to realize highly efficient reactions but also to gain a fundamental understanding of complex chemical processes.


Membrane holder assembly
A draft of the home-made membrane holder is shown in the Figure  Smembrane -the effective area of the CNN membrane, mm 2 ; Stube -the cross-sectional area of carbon nitride nanotube, mm 2 ; dmembrane -the diameter of the effective area of CNN membrane, mm; dtube -the external diameter of carbon nitride nanotube, mm.

Calculation of residence time
The residence time (τ) was calculated as where τ -the residence time, ms; V -the cumulative volume of cavity in the CNN membrane, cm 3 ; ν -the flow rate of MB solution, cm 3 ms -1 ; D -the inner diameter of carbon nitride nanotube, cm; L -the thickness of the CNN membrane, cm; ntube -the number of carbon nitride nanotubes on the effective area of the CNN membrane.
Calculation of turnover number and turnover frequency of single carbon nitride nanotube The turnover number (TON) of single carbon nitride nanotube was calculated as where TON -the turnover number; namine -the number of converted amine molecules; ntube -the number of carbon nitride nanotubes on the effective area of the CNN membrane; d -the density of amine, g mL -1 ; V -the volume of amine, mL; M -the molar mass of amine, g mol -1 ; w1 -the imine content after the reaction; w2 -the imine content before the reaction (0.75%); NA -the Avogadro constant, 6.02 × 10 23 mol -1 .
The turnover frequency (TOF) of a single carbon nitride nanotube was calculated as where TOF -the turnover frequency, s -1 ; TON -the turnover number; t -the duration of the experiment, s.

Calculation of apparent quantum yield of N-benzyl-1-phenylmethanimine
The apparent quantum yield (AQY) of imine was calculated as: where AQY -the apparent quantum yield of imine; namine -the number of converted amine molecules; nphoton -the number of incident photons; d -the density of amine, g mL -1 ; V -the volume of amine, mL; M -the molar mass of amine, g mol -1 ; w1 -the imine content after the reaction; w3 -the imine content after the dark reaction without catalyst (1.60%); NA -the Avogadro constant, 6.02 × 10 23 mol -1 ; I -the light intensity, W cm -2 ; Smembrane -the effective area of the CNN membrane, cm 2 ; t -the duration of the experiment, s; λ -the wavelength of incident light, m; hthe Planck's constant, 6.626 × 10 -34 J s; c -the speed of light, 3 × 10 8 m s -1 .
Calculation of the actual flow rate through a single carbon nitride nanotube The actual flow rate through a single carbon nitride nanotube (Q2) was calculated as: where Q2 -the actual flow rate through a single carbon nitride nanotube, m 3 s -1 ; Q0 -the imposed flow rate, m 3 s -1 ; ntube -the number of carbon nitride nanotubes on the effective area of the CNN membrane.

Calculation of enhancement factor
The enhancement factor (ε) was calculated as: where ε -the enhancement factor; Q2 -the actual flow rate through a single nanotube, m 3 s -1 ; Q1 -the theoretical flow rate through a single nanotube, m 3 s -1 .

Calculation of slip length
The slip length (ls) was calculated as: where ls -the slip length, nm; R -the inner radius of nanotube, nm; Q2 -the actual flow rate through a single nanotube, m 3 s -1 ; Q1 -the theoretical flow rate through a single nanotube, m 3 s -1 .

Supplementary discussion
Supplementary discussion 1.
In polar solvents, such as water, carbon nitride surface gains negative charge due to dissociation of surface hydroxyl groups possessing acidic character or adsorption of negatively charged ions from the solution. 2 Zeta-potential defines electric potential at the distance corresponding to Debye length (λD), which is calculated using the equation (1): where lB -is Bjerrum length, nm; I -ionic strength of the solution, mol L -1 .
Bjerrum length is defined by the equation (2): Ionic strength is defined by the equation (3): where ci -molar concentration of ions i, mol L -1 ; zi -charge of ion i.
Using equations (1), (2) and (3) Debye length for pure water, aqueous methylene blue solution used in the photocatalytic experiments have been calculated and results are given in the Table S5.
Polar solvents facilitate ionization of surface functional groups and give shorter λD. Contrary nonpolar solvents do not facilitate ionization of surface functional groups and give longer λD. However, it has been shown that even in non-polar solvents, nanoparticles, such as SiO2, still possess high surface charge and characterized by relatively high zeta-potential. 3 Comparison study of the CNN membrane and reference CN powder, shows that zeta-potential of reference carbon nitride is independent on presence or absence of methylene blue (Table S6).
However, zeta-potential of the CNN membrane becomes less negative upon addition of methylene blue. Such results indicate that absorption of phenothiazinium cation in the CN nanotube is significantly enhanced. Taking into account that both reference carbon nitride and the CNN membrane are made of the same material such enhancement is related to the confined environment.
Electric potential of the particle immersed into a polar solvent decays exponentially and approaches zero at infinite distance, while Debye length defines the distance at which electric potential decreases by e times (eq. 4): where ϕ0 -electric potential at carbon nitride surface, mV; r -distance from the surface, nm.
Taking into account measured zeta-potential, electric potential at zero distance (at carbon nitride surface) is calculated from the equation (5): where ζ -zeta-potential, mV; e -Euler number, e = 2.72.
The results of 0 calculation are given in the Table S6.
Given that Debye length for all electrolytes used in this work, exceeds the internal diameter of the CN nanotube (d = 40 nm), electric field gradient (E) caused by drop of surface potential with the distance may be estimated using the linear equation (Table S6): where c -potential at the axis of the CN nanotube, i.e. at the distance 20 nm from the inner wall of the nanotube, V; c -potential at the surface of the CN nanotube. Calculated using equation (4) According to the mechanism of dye degradation ( Figure S12), we propose that methylene blue is reduced to leuco-form at the expense of water as sacrificial donor of electrons and protons.
Oxidation of water is accompanied by the formation of O2, which is converted to singlet oxygen ( Figure S17) as well as other potential reactive oxygen species.
Such reactive oxygen species facilitate oxidation of leuco-methylene blue to methylene blue (MB + Cl -) according to the reaction: Redox potential of this reaction is E = +0.011 V versus NHE at pH = 7. 4 Therefore, once the residence time becomes long enough, recovery of methylene blue is facilitated. Therefore, concentration of methylene blue raises, which is registered as increase of the intensity of the absorption band at 664 nm and decrease of the degradation rate (k). Supplementary discussion 4. CNN membrane modified with Au single atoms (Au-CNN membrane).
In our previous report on single-site Au loaded N-doped porous noble carbon for electrochemical reduction of nitrogen 6 , we found that Au was nicely present as single species at a mass loading of 0.2 wt. %. Excessive amount of Au (>0.7 wt. %) would result in large nanoparticles and lead to decreased catalytic activity. Carbon nitride has N species similar with N-doped porous carbon which can stabilize Au atoms. Therefore, we employed the same method to deposit 0.2 wt. % Au on CNN membrane and successfully obtained well-dispersed single-site Au as shown in Figure 4b.
Our results show that the Au-CNN membrane is much more active in benzylamine oxidation than the CNN membrane (AQY~ 0), indicating that Au single sites play a critical role in the oxidation reaction. It is accepted that the local electric field between single metal sites and coordinated N on the surface of carbon nitride drives the transfer of photogenerated electrons from carbon nitride to the single sites, which provide active sites for reduction reaction [7][8][9] . In our case, the photoexcited electrons in CNN could be captured by the surface Au atoms and activate oxygen to yield singlet oxygen ( 1 O2, detected by EPR spectroscopy, Figure S17) and superoxide radical anion (O2 •-) as potentially important reactive species to enable benzylamine oxidation 10 . Therefore, we conclude that Au single sites could provide active sites for oxygen activation reactions.

Schemes
Scheme S1. Oxidative aza Diels-Alder reaction. Figure S1. SEM images of CNN incorporated AAO membrane.                   Tables   Table S1.   Calculated as a product of specific pore volume (cm 3 g -1 ) and mass of the photocatalyst (g) taken for degradation of methylene blue using data from the corresponding reference [b] Calculated as a ratio between pore volume (cm 3 ) and volume of the methylene blue solution (cm 3 )

Figures
[c] Calculated from first order rate constant 0.0033 min -1 [d] Calculated from first order rate constant 0.003 min -1 [e] Irradiation area instead of the SBET area of catalyst [b] Due to low acidity of benzylamine, λD has been arbitrary assigned to >10 5 nm, to emphasize that it is significantly longer than λD in aqueous environment