Evanescent waves modulate energy efficiency of photocatalysis within TiO2 coated optical fibers illuminated using LEDs

Coupling photocatalyst-coated optical fibers (P-OFs) with LEDs shows potential in environmental applications. Here we report a strategy to maximize P-OF light usage and quantify interactions between two forms of light energy (refracted light and evanescent waves) and surface-coated photocatalysts. Different TiO2-coated quartz optical fibers (TiO2-QOFs) are synthesized and characterized. An energy balance model is then developed by correlating different nano-size TiO2 coating structures with light propagation modes in TiO2-QOFs. By reducing TiO2 patchiness on optical fibers to 0.034 cm2/cm2 and increasing the average interspace distance between fiber surfaces and TiO2 coating layers to 114.3 nm, refraction is largely reduced when light is launched into TiO2-QOFs, and 91% of light propagated on the fiber surface is evanescent waves. 24% of the generated evanescent waves are not absorbed by nano-TiO2 and returned to optical fibers, thus increasing the quantum yield during degradation of a refractory pollutant (carbamazepine) in water by 32%. Our model also predicts that extending the TiO2-QOF length could fully use the returned light to double the carbamazepine degradation and quantum yield. Therefore, maximizing evanescent waves to activate photocatalysts by controlling photocatalyst coating structures emerges as an effective strategy to improve light usage in photocatalysis.

The author's universal claim of a higher energy efficiency of evanescent wave energy transfer versus refraction is not proven for the basic case of 6.5 cm fiber length. In this case, the observed degradation rates are almost exactly equal. The quantum yield of the low coating strength fiber may be the highest as less light is lost to refraction, but this is compensated by higher transmission losses -which are losses all the same.
So in this case, the claimed superiority is only achieved by defining refraction losses to be inside the frame of considered effects and transmission losses outside. However, quantum yield as it is usually defined is the ratio of reaction events to absorbed photons. In this definition, photons being refracted and transmitted through the TiO2 coating (and are not absorbed) should not be considered, the same as photons transmitted through the fiber.
On the other hand, the apparent quantum yield considers the ratio of reaction events to photons supplied to the system, neglecting transmission and reflection losses. In this case, both photons transmitted through the coating and the fiber should be considered. In either case, the proposed higher quantum yield of the evanescent wave system is only achieved by an unconventional interpretation of said efficiencies.
Only in the case of the longer optical fiber the lower coating system does achieve a higher reaction rate and also a higher quantum yield. Therefore, a general claim of higher efficiency cannot be made but must be conditional in that to achieve the higher efficiency also longer fibers and larger reactors are required.
Moreover, the authors state that light that is refracted out of the fiber and not absorbed by the TiO2 layer is "lost to water". This may be true for UVC light but not for the present case of using UVA as water only has a (in this case) negligible absorption coefficient for this radiation. Moreover, in practice, not a single but multiple fibers will be used in a reactor. Light that is refracted out of the fiber into the medium will therefore in most cases not be lost but absorbed by the TiO2 coating of another fiber. This is true at least as long as the reaction medium is somewhat transparent to UVA radiation and does not contain strongly UVA-absorbing solutes. As the author's main claim is that superiority of evanescent wave energy transfer versus refraction, the model should also encompass a fair comparison with a bundle of fibers where refracted light may be absorbed by another fiber's coating.
Minor points: -The comparisons of fiber coating strength would be more appropriately given area-specifically, i.e. mg/cm² of fiber surface.
-The presence of evanescent waves was not explicitly proven, as claimed on several passages of the paper, merely inferred. The authors should be clear about this. I am not an expert on optics so I not aware if there are more explicit ways to prove this, if there are, the authors should try to generate explicit direct evidence. Alternatively, a second indirect experiment may help. For instance using uncoated fibers in a solution with a strongly UVA absorber. In this case there should also be evanescent wave energy transfer, and its intensity could be compared to the low coating fibers. If no additional evidence is provided, the authors should refrain from the claim of having proven evanescent energy transfer.
Reviewer #2 (Remarks to the Author): In this paper, the authors proposed a strategy to maximize photocatalytic optical fiber light usage and quantified interactions between two forms of light energy (refracted light and evanescent waves) and surface-coated photocatalysts. They conclude that maximizing evanescent waves to activate photocatalysts by controlling photocatalyst coating structures emerged as an effective strategy to improve light usage in photocatalysis. However, in the photocatalytic reaction, not only the light absorption process but also multiple factors such as substrate adsorption, desorption, and redox reaction affect the photocatalytic activity. Currently, the author focuses only on the light absorption process, and the comprehensive evaluation is insufficient. It can be recommended to be considered for publication in this journal after major revision as suggested by following comments.
The present manuscript describes a new approach for photocatalyst-coated optical fibers. Instead of mainly relying on refraction to transfer the light into the photocatalyst, this is achieved by evanescent waves. The authors argue that refraction has the disadvantage that either, when using thin coatings, a large fraction of the light is transmitted though the coating or when using thicker coatings, mass transport limitation may be a problem. Indeed, this "backside" illumination of thicker porous layers would create opposing gradients of absorbed light flux and pollutant diffusion, which are not ideal and can lead to masstransfer limitations. As such the paper is interesting and presents an intriguing new way of performing photocatalysis.
However, I am skeptical about the proposed advantages and claimed higher energy efficiency. My assessment on whether this paper has impact sufficiently high for publication in Nature Communications is dependent on whether the authors can prove their system really is more efficient on a fair basis (vide infra). Therefore, I recommend a major revision based on the following aspects.

Comment 1:
The author's universal claim of a higher energy efficiency of evanescent wave energy transfer versus refraction is not proven for the basic case of 6.5 cm fiber length. In this case, the observed degradation rates are almost exactly equal. The quantum yield of the low coating strength fiber may be the highest as less light is lost to refraction, but this is compensated by higher transmission losses -which are losses all the same.
So in this case, the claimed superiority is only achieved by defining refraction losses to be inside the frame of considered effects and transmission losses outside. However, quantum yield as it is usually defined is the ratio of reaction events to absorbed photons. In this definition, photons being refracted and transmitted through the TiO2 coating (and are not absorbed) should not be considered, the same as photons transmitted through the fiber.
On the other hand, the apparent quantum yield considers the ratio of reaction events to photons supplied to the system, neglecting transmission and reflection losses. In this case, both photons transmitted through the coating and the fiber should be considered. In either case, the proposed higher quantum yield of the evanescent wave system is only achieved by an unconventional interpretation of said efficiencies.
Only in the case of the longer optical fiber the lower coating system does achieve a higher reaction rate and also a higher quantum yield. Therefore, a general claim of higher efficiency cannot be made but must be conditional in that to achieve the higher efficiency also longer fibers and larger reactors are required.

Reply:
The reviewers make a good point that we need to more clearly articulate. We believe the comparison in energy efficiency between evanescent waves and refracted light is on an equivalent basis for two different fiber lengths (6.5 and 26 cm lengths). The light energy launched to TiO2-QOFs equals to the sum of i) the light energy absorbed by TiO2 (through evanescent waves and refracted light), ii) refraction losses (light energy transmitted through TiO2 coatings), and iii) transmission losses (light energy transmitted through the optical fiber). The light irradiance emitted from the side of the TiO2-QOFs accounts only for <1.35% of the irradiance launched to TiO2-QOFs (detailed see Comment 2), suggesting light refracted from quartz fibers is absorbed by TiO2 layers and refraction losses can be neglected in this study.
When assessing the quantum yield, which is defined as the ratio of reaction events to absorbed photons, the absorbed photonic energy comprises only the evanescent wave and refracted light energy absorbed by TiO2. Since light refracted from quartz fibers is absorbed by TiO2 layers, TiO2-QOF-Low with high quantity of evanescent waves absorbed less light energy compared with TiO2-QOF-High and TiO2-QOF-Med. Therefore, TiO2-QOF-Low showed high quantum yields for at a fiber length of 6.5 cm as shown in the original Manuscript Fig. 5 When assessing the apparent quantum yield, which is defined as the ratio of reaction events to photons launched to the system, both the light energy absorbed by TiO2 and the transmission losses should be considered. The apparent quantum yield of TiO2-QOF-Low at 6.5 cm shows no superiority compared with TiO2-QOF-High or TiO2-QOF-Med. However, in Supplementary Section 6.6, we already simulated the carbamazepine degradation rate constants and apparent quantum yields of TiO2-QOF-Low at a fiber length of 26 cm. Their values were both 2× higher than those of TiO2-QOF-High at a coating length of 6.5 cm in a 92 mL reactor. This simulation was further confirmed by our additional experiments in response to the reviewer's comment (Fig. 5d), in which the 26 cm TiO2-QOF-Low also showed 2× rate constants and 2× quantum yields than the 6.5 cm TiO2-QOF-High when degrading carbamazepine in a 70 ml reactor.
These clarifications have been revised in the revised Manuscript: Fig. 5d comparison between degradation rate constants and apparent quantum yields by the UV irradiated TiO2-QOF-High at a coating length of 6.5 cm and TiO2-QOF-Low at a coating length of 26 cm in a 70 mL reactor. Lines 52-53, "(i.e., moles of pollutants degraded per mole of photons absorbed by photocatalysts)", Lines 273-274, "(i.e., moles of carbamazepine degraded per mole of photons absorbed by TiO2 coating layers)", Lines 283-290, "When assessing the apparent quantum yield (moles of carbamazepine degraded per mole of photons launched to optical fibers), both the light energy absorbed by TiO2 and the transmission losses should also be considered. The apparent quantum yield of TiO2-QOF-Low at 6.5 cm shows no superiority compared with TiO2-QOF-High or TiO2-QOF-Med. However, its apparent quantum yield can be further increased by bundling multiple TiO2-QOFs with a single LED as reported before or, as we show here, extending the length of TiO2-QOF-Low to utilize the returned radiant energy and increase the photocatalytic reactive sites." Lines 295-299, "therefore, the 26-cm TiO2-QOF-Low was estimated to achieve 2× degradation rate constants and 2× apparent quantum yields compared with TiO2-QOF-High at a coating length of 6.5 cm (Supplementary Section 6.6). The higher rate constant and apparent quantum yield of 26-cm TiO2-QOF-Low was further experimentally confirmed in a 70 mL reactor (Fig. 5d). "

Comment 2:
Moreover, the authors state that light that is refracted out of the fiber and not absorbed by the TiO2 layer is "lost to water". This may be true for UVC light but not for the present case of using UVA as water only has a (in this case) negligible absorption coefficient for this radiation. Moreover, in practice, not a single but multiple fibers will be used in a reactor. Light that is refracted out of the fiber into the medium will therefore in most cases not be lost but absorbed by the TiO2 coating of another fiber. This is true at least as long as the reaction medium is somewhat transparent to UVA radiation and does not contain strongly UVA-absorbing solutes. As the author's main claim is that superiority of evanescent wave energy transfer versus refraction, the model should also encompass a fair comparison with a bundle of fibers where refracted light may be absorbed by another fiber's coating.

Reply:
The reviewer's comment is well taken regarding our comparison between evanescent waves and refracted light, and whether evanescent waves are really more energy efficient than refracted light. To quantify these energies, we measured the light irradiance 1 mm away from side surfaces of the three TiO2-QOFs, as shown in Supplementary Table 4. The light irradiance emitted from side surface of all the three TiO2-QOFs at 1 cm and 5 cm along the fiber length is less than 0.10 mW/cm 2 and 0.01 mW/cm 2 , respectively, which account only for <1.35% and <0.15% of the irradiance launched to TiO2-QOFs, respectively. Although more light energy was absorbed by TiO2-QOF-High, 80% of the light energy launched to TiO2-QOF-High was absorbed at the beginning 2 cm of the fiber, while only about 12% of the light energy was absorbed at the following 4.5 cm. Therefore, the carbamazepine degradation occurred mostly at the beginning 2 cm of the fiber, which limits the photocatalytic activity of TiO2-QOF-High. On the other hand, the light energy distribution along TiO2-QOF-Low was more uniform as shown in the original Supplementary Fig. 15. More TiO2 reactive sites was used for photocatalytic degradation with less light energy. Therefore, TiO2-QOF-Low showed a higher quantum yield. The light irradiance emitted from the side of the three TiO2-QOFs is too low to generate sufficient radicals on TiO2 coatings of another fiber even when multiple TiO2-QOFs are used. Moreover, in practice, aqueous components, such as NOM in water, absorb or scatter the light emitted from the side of TiO2-QOFs, further inhibited light reaching TiO2 coatings of other fibers. These results suggest that the light emitted from the side of TiO2-QOFs cannot be effectively utilized by the TiO2 coatings of another fibers in this study.

Comment 3:
The comparisons of fiber coating strength would be more appropriately given areaspecifically, i.e. mg/cm² of fiber surface.

Reply:
The linear TiO2 mass coating density (mg/cm) has been revised to the area-specific TiO2 coating density (mg/cm 2 ) accordingly in the revised manuscript.

Comment 4:
The presence of evanescent waves was not explicitly proven, as claimed on several passages of the paper, merely inferred. The authors should be clear about this. I am not an expert on optics so I not aware if there are more explicit ways to prove this, if there are, the authors should try to generate explicit direct evidence. Alternatively, a second indirect experiment may help. For instance, using uncoated fibers in a solution with a strongly UVA absorber. In this case there should also be evanescent wave energy transfer, and its intensity could be compared to the low coating fibers. If no additional evidence is provided, the authors should refrain from the claim of having proven evanescent energy transfer.

Reply:
We appreciate the suggestion, and performed additional experiments. The existence of evanescent waves on the interface between quartz and air/water has already been proven by previous studies 1 . The presence of evanescent waves in the current manuscript was demonstrated by tracking the irradiance loss in a UV irradiated uncoated fiber immersed in methylene blue (MB) solutions as shown in Supplementary Fig. 9. When launching UV light to the uncoated fiber, the irradiance loss increased by around 10 times with increasing MB concentrations from 1.3 to 65 mg/L. The irradiance loss was due to the absorption of evanescent waves generated on uncoated quartz fiber surfaces by MB. This test supports the presence of evanescent waves.
This information is added to the revised Manuscript: Lines 186-187: "The existence of evanescent waves was also proven by tracking the irradiance loss in a UV irradiated uncoated fiber immersed in methylene blue solutions (Supplementary Fig. 9)." The reply is also added to the revised Supplementary Information: Section 5.2: "The existence of evanescent waves on quartz optical fiber surfaces was demonstrated by tracking the irradiance loss in a UV irradiated uncoated fiber immersed in methylene blue (MB) solutions as shown in Supplementary Fig. 9. When launching UV light to the uncoated optical fiber, the irradiance loss increased by around 10 times with increasing MB concentrations from 1.3 to 64.5 mg/L. The irradiance loss was due to the absorption of evanescent waves generated on uncoated quartz fiber surfaces by MB. This test supports the presence of evanescent waves." Reference: [

Responses to comments of Reviewer 2:
In this paper, the authors proposed a strategy to maximize photocatalytic optical fiber light usage and quantified interactions between two forms of light energy (refracted light and evanescent waves) and surface-coated photocatalysts. They conclude that maximizing evanescent waves to activate photocatalysts by controlling photocatalyst coating structures emerged as an effective strategy to improve light usage in photocatalysis. However, in the photocatalytic reaction, not only the light absorption process but also multiple factors such as substrate adsorption, desorption, and redox reaction affect the photocatalytic activity. Currently, the author focuses only on the light absorption process, and the comprehensive evaluation is insufficient. It can be recommended to be considered for publication in this journal after major revision as suggested by following comments.

Comment 1:
If the photocatalyst layer is thick, TiO2 existing at the interface with the fiber absorbs light, but TiO2 far from the fiber may not receive sufficient light. In that case, TiO2-QOF-High, which is disadvantageous in terms of mass transfer, shows low activity. The author concludes that the ratio of refracted light to evanescent waves is a factor in the difference in activity, but it is necessary to exclude that the difference in mass transfer affects the activity.

Reply:
We appreciate the comment. In response to the reviewer's comment, we evaluated the impacts of external mass transfer and internal mass transfer of the TiO2 coating layers on the photocatalytic performance of different TiO2-QOFs. The external mass transfer describes the diffusion of pollutants from bulk solutions to the TiO2 coating surface. It is a function of the Reynolds number 1 . Therefore, external mass transfer of the three TiO2-QOFs are the same in the current manuscript because the reactor setup is the same. On the other hand, the internal mass transfer describes the diffusion of pollutants and radicals inside the porous TiO2 coating layers. It is an intrinsic property of the TiO2 coatings and determined by the nature of TiO2 and the coating structures 1,2 . The internal mass transfer could be evaluated using the Thiele modulus (ϕ) and the internal effectiveness factor (η) as shown in Eqs. 1 and 2, respectively 2 .
The calculated ϕ and η of the three TiO2-QOFs are shown in Supplementary Table 5. ϕ of the TiO2 coating layers decrease from 1.20×10 -4 in TiO2-QOF-High to 1.95×10 -5 in TiO2-QOF-Low, while all η of the three TiO2-QOFs equal to 1.00. According to Weisz's criteria, in which the internal mass transfer is neglectable when ϕ < 0.3 and η ≈ 1 5 , the TiO2 coating layers on the three TiO2-QOFs have no internal mass transfer limitation.
Therefore, the same external mass transfer and the neglectable internal mass transfer of the TiO2 coating layers on the three TiO2-QOFs suggest there is no difference in mass transfer and thus excluding the mass transfer effect on the photocatalytic activity of TiO2-QOFs.  Fig. 14c) of different TiO2 coating layers."

Supplementary
The discussion above is also added to the revised Supplementary Information Section 6.3. Reference: [

Comment 2:
It is necessary to observe the leakage of refracted light from the each POF. If no refracted light can be observed, it means that TiO2 layer is completely absorbed the leakage light and the light absorption process does not affect the difference in photocatalytic activity. Mass transfer and adsorption due to the difference in the thickness of the TiO2 layer may affect the photocatalytic activity.

Reply:
In response to the reviewer's comment, we performed additional experiments to measure the light irradiance 1 mm away from side surfaces of the three TiO2-QOFs as shown in Supplementary Table 4. The light irradiance refracted from side surfaces of all the three TiO2-QOFs at 1 cm and 5 cm along the fiber length is less than 0.10 mW/cm 2 and 0.01 mW/cm 2 , respectively, which account only for <1.35% and <0.15% of the irradiance launched to TiO2-QOFs, respectively. Therefore, the light refracted from quartz optical fiber is mostly absorbed by TiO2 coating layers. Although more light energy was absorbed by TiO2-QOF-High, 80% of the light energy launched to TiO2-QOF-High was absorbed at the beginning 2 cm of the fiber, while only about 12% of the light energy was absorbed at the following 4.5 cm. Therefore, the carbamazepine degradation occurred mostly at the beginning 2 cm of the fiber, which limits the photocatalytic activity of TiO2-QOF-High.
On the other hand, the light energy distribution along TiO2-QOF-Low was more uniform as shown in the original Supplementary Fig. 15. More TiO2 reactive sites was used for photocatalytic degradation with less light energy. Therefore, TiO2-QOF-Low showed a higher quantum yield.
This information is added to the revised Manuscript lines 246-247: "This dissipated radiant energy was mostly absorbed by the TiO2 coating layers (details see Supplementary In response to the comment, we also evaluated the impacts of external mass transfer and internal mass transfer of the TiO2 coating layers on the photocatalytic performance of different TiO2-QOFs. The external mass transfer of the three TiO2-QOFs are the same in the current manuscript because the reactor setup is the same, while the TiO2 coating layers on the three TiO2-QOFs have no internal mass transfer limitation (Supplementary Table 5).
These results exclude the difference in mass transfer of the three TiO2-QOFs affecting the photocatalytic activity.
Tests of the carbamazepine adsorption by TiO2-QOFs were conducted in the original manuscript Supplementary Fig. 14a, in which carbamazepine concentrations did not change in dark in the presence of TiO2-QOFs. We also performed additional experiments in response to reviewer's comment to test whether carbamazepine is adsorbed by TiO2, which confirmed that 5 g/L of TiO2 suspension cannot adsorb carbamazepine at an initial concentration of 2 μM and other experimental conditions commensurate with our study conditions ( Supplementary Fig. 14c).
These are added to the revised Manuscript: Lines 266-268: "Control tests confirmed that carbamazepine was not adsorbed by TiO2-QOFs in the dark or not degraded when light was launched from LEDs into an uncoated optical fiber ( Supplementary Figs. 14a and 14b)." Lines 276-278: "but not to the mass transfer limitation (Supplementary Section 6.3) or carbamazepine adsorption ( Supplementary Fig. 14c) of different TiO2 coating layers." This result is also added to the revised Supplementary Information Fig. 14c.

Comment 3:
In photocatalytic decomposition experiments, the change in concentration of CBZ until the adsorption equilibrium is reached should be shown. If it does not adsorb, what is the mechanism by which decomposition is considered to be proceeding? Tests of the carbamazepine adsorption by TiO2-QOFs were conducted in the original manuscript Supplementary Fig. 14a, in which carbamazepine concentrations did not change in dark in the presence of TiO2-QOFs. We also performed additional experiments in response to reviewer's comment to test whether carbamazepine is adsorbed by TiO2, which confirmed that 5 g/L of TiO2 suspension cannot adsorb carbamazepine at an initial concentration of 2 μM and other experimental conditions commensurate with our study conditions ( Supplementary Fig. 14c). Therefore, the decomposition of carbamazepine occurred not on surface but in bulk solution and pores in TiO2 coating layers. It was attributed to photocatalytic generated hydroxyl radicals (HO•) from TiO2 coating layers.

Reply
These are revised in the revised Manuscript: Lines 45-46: "Upon absorption of light, photocatalysts generate hole-electron (h + -e -) pairs", Lines 82-84: "the interactions between TiO2 and the evanescent wave energy for the generation of hydroxyl radicals (HO•) to degrade a refractory pollutant (carbamazepine) in bulk solution and pores in TiO2 coating layers (Fig. 1b)".

Comment 4:
What is oxidized and reduced in the photocatalytic reaction? Should be shown with a band structure model of the prepared photocatalyst.

Reply:
The redox reactions and band structure model of TiO2 P25 are added to the revised Manuscript Fig. 1b

Comment 5:
The photocatalytic decomposition data presented by the authors showed the highest activity in POF with the least photocatalytic coating. Is the thinner the coating of the photocatalyst layer, the higher the activity? Is there a peak in the relationship between photocatalytic amount and activity?

Reply:
a. Regarding the question of "Is the thinner the coating of the photocatalyst layer, the higher the activity". According to our energy balance model, the TiO2 layer structure parameters (p and za) are both important to control the ratio of refracted light to evanescent waves, while the TiO2 coating thickness is not critical. This is added to lines 258-260 of the revised Manuscript: "The above modeling results suggest the TiO2 layer structure parameters (p and za) are important to control the light energy within TiO2-QOFs, while the TiO2 coating thickness is not critical." b. Regarding the question of "Is there a peak in the relationship between photocatalytic amount and activity".
We fabricate two more TiO2-QOFs, including TiO2-QOF-Low'' at p of 0.018 and za of 139.50 nm and TiO2-QOF-Low' at p of 0.026 and za of 127.97 nm, using different coating conditions as shown in Supplementary Table 6. These two TiO2-QOFs are then compared with TiO2-QOF-High, TiO2-QOF-Med, and TiO2-QOF-Low. As shown in Supplementary Table 6, the carbamazepine degradation rate constants increased with increasing p and decreasing za from TiO2-QOF-Low'' to TiO2-QOF-Low, but remain the same with further increasing p and decreasing za from TiO2-QOF-Low to TiO2-QOF-High. A turning point exists where the degradation rate constants reach a plateau. On the other hand, quantum yields of carbamazepine degradation by the UV irradiated TiO2-QOF-Low was the highest, at 0.0248. Increasing p and decreasing za or decreasing p and increasing za resulted in a decrease in quantum yields.  Table 6)."

Supplementary
The discussion above is also added to the revised Supplementary Information: Section 6.4: "We fabricate two more TiO2-QOFs, including TiO2-QOF-Low'' at p of 0.018 and za of 139.50 nm and TiO2-QOF-Low' at p of 0.026 and za of 127.97 nm, using different coating conditions as shown in Supplementary Table 6. These two TiO2-QOFs are then compared with TiO2-QOF-High, TiO2-QOF-Med, and TiO2-QOF-Low. As shown in Supplementary Table 6, the carbamazepine degradation rate constants increased with increasing p and decreasing za from TiO2-QOF-Low'' to TiO2-QOF-Low, but remained the same with further increasing p and decreasing za from TiO2-QOF-Low to TiO2-QOF-High. A turning point exists where the degradation rate constants reach a plateau. On the other hand, quantum yields of carbamazepine degradation by the UV irradiated TiO2-QOF-Low was the highest, at 0.0248. Increasing p and decreasing za or decreasing p and increasing za resulted in a decrease in quantum yields. "

Comment 6:
What solvent was used for dip coating of TiO2? During the photocatalytic reaction, is the coating come off without calcination?

Reply:
TiO2 suspension was prepared by dispersing TiO2 in double deionized water (18.2 MΩcm), and no organic solvents were utilized. This information is added to lines 324-325 of the revised Manuscript: "TiO2 suspension was prepared by dispersing TiO2 (P25) in double deionized water (18.2 MΩ-cm), and no organic solvents were utilized." In response to the reviewer's comment, we performed an indirect test to demonstrate that TiO2 coating is stable during the reaction. Tests are conducted to evaluate the radiant energy dissipation and carbamazepine degradation rate constants of TiO2-QOF-Low for 3 cycles. For each cycle, TiO2-QOF-Low was immersed in the carbamazepine containing solution under stirring for 4 h. After each cycle, the used TiO2-QOF-Low was dried before running the next cycle. Supplementary Fig. 17 shows that both radiant energy dissipation and carbamazepine degradation rate constants remain unchanged for the 3 cycles of testing, suggesting TiO2 coating is stable during the reaction.
These are added to the revised Manuscript: Lines 301-303: "The TiO2 coating in TiO2-QOF-Low is also stable, because the radiant energy dissipation and degradation rate constants remained unchanged for 3 cycles of carbamazepine degradation ( Supplementary Fig. 17)." This information is also added to the revised Supplementary Information: Section 6.7: "We performed an indirect test to demonstrate that TiO2 coating is stable during the reaction. Tests are conducted to evaluate the radiant energy dissipation and carbamazepine degradation rate constants of TiO2-QOF-Low for 3 cycles. For each cycle, TiO2-QOF-Low was immersed in the carbamazepine containing solution under stirring for 4 h. After each cycle, the used TiO2-QOF-Low was dried before running the next cycle. Supplementary Fig. 17 shows that both radiant energy dissipation and carbamazepine degradation rate constants remain unchanged for the 3 cycles of testing, suggesting TiO2 coating is stable during the reaction."

Comment 7:
TiO2-QOF-High, which has a large amount of deposition, should have more TiO2 far from the optical fiber surface than other POFs. Why is za of TiO2-QOF-Low the largest? We appreciate the comment. We revised the definition of zn on lines 165-168 of the revised manuscript as "zn is the normalized distance between the fiber surface and TiO2 nanoparticles at a TIR spot, which is equal to the average distance from the center of an evanescent field to the closest surrounding TiO2 nanoparticles within the evanescent field". While za equals to the average value of zn at all TIR spots along the fiber length. The definition of za indicates that evanescent wave energy dissipates when they reach the closest TiO2 nanoparticles, while they return to the fiber if not reach TiO2. Our model shows that za is largest in TiO2-QOF-Low and smallest in TiO2-QOF-High among the three TiO2-QOFs. Although TiO2-QOF-High has a large amount of TiO2 deposition, these TiO2 nanoparticles are far from the optical fiber surface and they do not absorb evanescent waves. The above information is revised in the revised manuscript lines 251-254: "The value of za in TiO2-QOF-High (7.7 nm) is much smaller than the value of Λ from quartz to water (50-120 nm) ( Supplementary Fig. 11), suggesting evanescent wave energy dissipates when they reach the closest TiO2 nanoparticles and a negligible amount of them returned to the optical fibers to give a highest ratio of EE,dis' to EE,g." <b>REVIEWER COMMENTS</B> Reviewer #1 (Remarks to the Author):

Reply
Even though the authors have clarified some issues in their rebuttal and improved the manuscript in many aspects, I am still very sceptical about the proposed superiority of their system.
The way I understand it, the authors merely shift the light losses from refraction to transmission. Since the former is not accurately determined (see below) and deducted from the absorbed photon flux but the latter is, this results in their supposedly higher quantum yield. The authors claim that the transmission losses can be counteracted by just using a longer fibre length but also the refraction losses are stongly attenuated by using fibre bundles. So in either case, there is no real advantage.
The only potential advantage would be that the light is indeed more evenly dissipated and this may prevent local photon oversaturation and its associated efficiency losses. However, the authors do not provide any evidence for oversaturation in the case of the highly coated fibers so this is pure speculation.
Overall, the paper is quite intruiging and well written. However, as there seems to be no real benefit of the new system the impact is not sufficiently high to warrant publication in Nature Communications.
On the determination of the refracted light: The methodology of determining the refraction losses is highly questionable. If the irradiance is measured 1 mm away from the fiber, it is already quite dispersed. In fact, the total area illuminated at 1mm distance to the fiber is 780 times larger than the cross section of the fiber at 6.5 cm length and even 3120 times larger for 26 cm length. This needs to be taken into account when comparing irradiances. As such I suspect the actual reflection losses are many times the value stated by the authors and may therefore not be so easily discounted. An accurate determination would entail an integration of the total power (or photon flux) emitted by the fibre over its entire length (either experimentally or computationally based on few measured data points).
Their data also suggests that the highly coated fibre refracts significantly more light (almost three times) than the low coated one. The higher refraction losses of the former may readily explain the differences in the observed (supposed) quantum yield.

Reviewer #2 (Remarks to the Author):
This manuscript has been well revised. Thus I recommend its acceptance without any change Title: Evanescent waves modulate energy efficiency of photocatalysis within nano-TiO2 coated quartz optical fibers illuminated using LEDs

Authors: Yinghao Song, Li Ling, Paul Westerhoff, Chii Shang
Manuscript ID: NCOMMS-20-44004B Before responding to the comments of the reviewers, the authors would like to express our deepest appreciation for their time and effort in reviewing this manuscript.

Responses to comments of Reviewer 1: Comment 1:
Even though the authors have clarified some issues in their rebuttal and improved the manuscript in many aspects, I am still very skeptical about the proposed superiority of their system.
Overall, the paper is quite intriguing and well written. However, as there seems to be no real benefit of the new system the impact is not sufficiently high to warrant publication in Nature Communications.

Reply:
Thank you for recognizing the intriguing parts of our paper. Below is a discussion that lays out the significance of the work, and addresses the specific point about superiority of our system.
In last 5 years, over 60,000 journal papers were published about photocatalytic processes as recorded in Web of Science database. These publications covered a wide range of fields including energy production, pollutant remediation, organic synthesis, etc. Most of these papers focused on developing new catalytic materials to enhance the quantum yields of photocatalytic processes. An important but often overlooked aspect to limiting use photocatalytic processes into engineering practice is the influence of photocatalytic reactor design, rather than material properties alone. We highlighted this in our recent review article 1 , published in late 2018 which has over 15,000 downloads and already over a hundred citations. To address the barrier of reactor design rather than material properties, we show how quantum yields are effectively increased more significantly by managing the way light reaches the photocatalyst, than recent improvements in catalyst materials themselves.
Here, we developed a novel coating strategy to create a new type of fixed-film reactor that left 3.4% of the optical fiber surface in direct contact with TiO2 and, on-average, 114.3 nm interspace distance between the fiber surface and the coated TiO2 layers. This strategy successfully reduced refraction, generated evanescent waves in the TiO2-QOF, and thus allowed even dissipation of light along TiO2-QOF to prevent oversaturation of the light delivered to the fiber and its associated efficiency losses (details see reply to Comment 2). As a result, this strategy saves 23% of the radiant energy delivered to the TiO2-QOF without compromising carbamazepine degradation. The saved energy could be utilized by extending the TiO2 coating length to further enhance pollutant degradation and quantum yields. The novel TiO2-QOF enables the design of a photocatalytic reactor that uses 77% less mass of photocatalysts but achieves up to 96% improvement in quantum yields compared with reactors built of fibers with densely coated TiO2 layers, mainly because optimizing evanescent waves to fully use transmitted light in longer TiO2-QOF-Low is more efficient than harvesting refracted light by using TiO2-QOF-High bundles (details see reply to Comments 3). Moreover, such a strategy of managing evanescent wave energy is applicable to all photocatalysts and thus emerges as an important cost consideration especially when contemplating expensive photocatalysts. By replacing TiO2 to photocatalysts with different functions, for example, hydrogen production or organic synthesis, an improvement of up to 96% in quantum yields will significantly reduce the cost of the processes. In addition to considering only refracted light in our optical fiber reactor, another key insight is the separation of photocatalyst activation by refracted light versus evanescent wave energy to activate the photocatalyst. We report for the first time here the relative importance of both mechanisms, as well as how to modulate their relative importance.
A new platform can also be established using novel photocatalyst-coated optical fibers to allow evanescent wave energy to activate photocatalysts. This new platform can quantify performance of photocatalysts more precisely by minimizing the radiation scattering by photocatalysts and water parameters, preventing aggregation of photocatalysts to maximize the interaction between reactive sites and target compounds, and allowing easier and more accurate quantification of photons absorbed by photocatalysts.
We believe these benefits grant the significance and high impacts of this manuscript to warrant its publication in Nature Communications.
These are revised from the latest revised Manuscript: Former text in Lines 305-317 "We developed a novel coating strategy that left 3.4% of the optical fiber surface in direct contact with TiO2 and, on-average, 114.3 nm interspace distance between the fiber surface and the coated TiO2 layers. This strategy successfully reduced refraction, generated evanescent waves in the TiO2-QOF, and saved 23% of the radiant energy delivered to the TiO2-QOF without compromising carbamazepine degradation. The saved energy could be utilized by extending the TiO2 coating length to further enhance pollutant degradation and quantum yields. This coating strategy also enabled the use of fewer photocatalysts, which emerges as an important cost consideration when contemplating other novel visible light active materials. Because similar optical fiber coating modulation is likely to increase quantum yields using higher wattage and lowercost visible light LEDs or even the mass abundant sunlight with visible light photocatalysts, our findings on the ability to manage evanescent wave energy by controlling patchiness and interspace distance between optical fiber surfaces and photocatalysts will have broad impacts on the next generation photocatalytic reactor design. " was revised as Lines 312-341 in the newly revised Manuscript: "Photocatalytic processes have broad application in water and air treatment, energy production, organic synthesis, and other fields. Less than 1 in 100 papers address the critical barrier to making photocatalytic reactors more effective, namely light and energy management; the other papers focus largely on discovery of new or incremental improvement in existing photocatalyst materials 10 . An important but often overlooked aspect to limiting use photocatalytic processes into engineering practice is the influence of photocatalytic reactor design, rather than material properties alone. To address the barrier of reactor design rather than material properties we show how quantum yields are effectively increased more significantly by managing the way light reaches the catalyst, than recent improvements in catalyst materials themselves. In addition to considering only refracted light in our optical fiber reactor, another key insight was the separation of photocatalyst activation by refracted light versus evanescent wave energy to activate the photocatalyst. We report for the first time here the relative importance of both mechanisms, as well as how to modulate their relative importance. Here, we developed a novel coating strategy that left 3.4% of the optical fiber surface in direct contact with TiO2 and, on-average, 114.3 nm interspace distance between the fiber surface and the coated TiO2 layers. This strategy successfully reduced refraction, generated evanescent waves in the TiO2-QOF, and thus allowed even dissipation of light along TiO2-QOF to prevent oversaturation of the light delivered to the fiber and its associated efficiency losses. The novel TiO2-QOF enables the design of a photocatalytic reactor that uses 77% less mass of photocatalysts but achieves up to 96% improvement in quantum yields compared with reactors built of fibers with densely coated TiO2 layers, mainly because optimizing evanescent waves to fully use transmitted light in longer TiO2-QOF-Low is more efficient than harvesting refracted light by using TiO2-QOF-High bundles. Moreover, such a strategy of managing evanescent wave energy is applicable to all photocatalysts and thus emerges as an important cost consideration especially when contemplating expensive photocatalysts.
A new platform can also be established using novel photocatalyst-coated optical fibers to allow evanescent wave energy to activate photocatalysts. This new platform can quantify performance of photocatalysts more precisely by minimizing the radiation scattering by photocatalysts and water parameters, preventing aggregation of photocatalysts to maximize the interaction between reactive sites and target compounds, and allowing easier and more accurate quantification of photons absorbed by photocatalysts. "

Comment 2:
The only potential advantage would be that the light is indeed more evenly dissipated and this may prevent local photon oversaturation and its associated efficiency losses. However, the authors do not provide any evidence for oversaturation in the case of the highly coated fibers so this is pure speculation.

Reply:
We appreciate that the reviewer raised that we need to better articulate advantages of TiO2-QOF-low regarding to how we prepared fibers to enable light to be more evenly dissipated along the fiber and prevents local photon oversaturation and its associated efficiency losses. In response to the reviewer's comment, additional experiments and data analysis were conducted. These are show in SI, as follows: In Supplementary Fig. 15a, we show the simulated radiant flux dissipated at each coating section at a length of 1 cm along TiO2-QOFs as calculated from the energy balance model. Approximately 80% of the light launched to TiO2-QOF-High was dissipated at the first 2cm section, while only 12% of which was dissipated at the following sections. In contrast, only 43% of the light launched to TiO2-QOF-Low was dissipated at the first 2-cm section and light was more evenly dissipated along the fiber.
The large amount of light dissipated at the beginning section of TiO2-QOF-High resulted in local photon oversaturation and caused efficiency losses. This was proved by an additional experiment, which shows the degradation rate constants of carbamazepine and quantum yields by the UV-irradiated TiO2-QOF-High at a coating length of 1 cm as a function of radiant flux dissipation ( Supplementary Fig. 15b). With increasing radiant flux dissipation from 3 to 36 μW, the degradation rate constants of carbamazepine by 1-cm TiO2-QOF-High increased by 1.8 times, while the quantum yields showed a significant drop by 77%. Therefore, the radiant flux dissipated at the first 2-cm section of TiO2-QOF-High at 44 μW contributed to over 75% of carbamazepine degradation but with a low quantum yields of 0.014, while only 25% of carbamazepine degradation was attributed to the last 4-cm of TiO2-QOF-High. The local photon oversaturation and its high efficiency losses at the first 2-cm section cause the low overall quantum yield of 6-cm TiO2-QOF-High. In contrast, TiO2-QOF-Low has a more evenly dissipated radiant flux of 2 to 15 μW, which grant TiO2-QOF-Low higher quantum yields in degrading carbamazepine.
The novel fiber coating strategy we discovered and reported in this manuscript modulates the coating layer structures to generate higher portions of evanescent waves. The generation of evanescent waves results in even dissipation of light in TiO2-QOFs and thus prevents local light oversaturation and its associated efficiency losses. This is added to the newly revised Manuscript Lines 275-278: "The evanescent waves generated in TiO2-QOF-Low allows light to be evenly dissipated along the fiber, which prevents local photon oversaturation at the beginning sections of TiO2-QOFs and thus reduces its associated efficiency losses (Supplementary Section 6.2).
These are revised from the latest revised Manuscript: Former text in Lines 275-277: "but not to the mass transfer limitation (Supplementary Section 6.3) or carbamazepine adsorption ( Supplementary Fig. 14c) of different TiO2 coating layers." was revised as Lines 278-280 in the newly revised Manuscript: "We also proved that the improved quantum yield was not attributed to the mass transfer limitation (Supplementary Section 6.3) or carbamazepine adsorption ( Supplementary Fig. 14c) of different TiO2 coating layers. " We delete the text in the latest revised Manuscript Lines 270-271: "The hypothesis about why k was not compromised is shown in Supplementary Section 6.2." The information about the even dissipation of light in TiO2-QOF-Low prevents local photon oversaturation is revise in the revised Supplementary Information Section 6.2.

Comment 3:
The way I understand it, the authors merely shift the light losses from refraction to transmission. Since the former is not accurately determined (see below) and deducted from the absorbed photon flux but the latter is, this results in their supposedly higher quantum yield. The authors claim that the transmission losses can be counteracted by just using a longer fiber length but also the refraction losses are strongly attenuated by using fiber bundles. So in either case, there is no real advantage.

Reply:
To demonstrate the advantage of our system, we conducted additional experiments to further confirm that the use of the transmitted light by using longer TiO2-QOF-Low is more efficient than the use of refracted light by using TiO2-QOF-High bundles. We conducted additional experiments in which seven TiO2-QOFs were bundled together and used for carbamazepine degradation, as shown in Supplementary Figs. 17 and 18. The conclusion is that optimizing evanescent waves to fully use transmitted light in longer TiO2-QOF-Low is 44-96% more efficient than harvesting refracted light by using TiO2-QOF-High bundles depending on fiber spacings (1 mm to 7 mm). Meanwhile, TiO2-QOF-Low uses 77% fewer photocatalysts than TiO2-QOF-High. Below are details on these additional experimental results which are integrated into the manuscript: As shown in Supplementary Fig. 17a, a hexagonal arrangement for a TiO2-QOF bundle which consists of one TiO2-QOF in the center and six TiO2-QOFs at the edge was proposed. The hexagonal arrangement is the closest packing that allows TiO2-QOFs to utilize the most refracted light out of fibers. The minimum distance between two TiO2-QOF surfaces, either from the TiO2-QOF in the center to the ones at the edge or between those at the edge, are defined as the fiber spacing (S) in the bundle. The S was set as 1, 3, 5 and 7 mm in the supplementary tests. The TiO2-QOF bundle was installed in a tubular reactor of an inner length of 65 mm and an inner diameter of 24 mm. All seven TiO2-QOFs have a coating length of 6.5 cm. Light was only allowed to be launched from a UV-LED to the TiO2-QOF in the center of the bundle, i.e., the one centered in the axial of the reactor.
Supplementary Fig. 17b shows the comparisons in carbamazepine degradation rates (r) by a single fiber of TiO2-QOF-High, a single fiber of TiO2-QOF-Low, a bundle of TiO2-QOF-High, and a bundle of TiO2-QOF-Low irradiated by one UV-LED at a UV intensity of 7.02 mW/cm 2 . The S in the two bundles were 1 mm. The carbamazepine degradation rate by a 6.5-cm TiO2-QOF-High (rc-High) was 0.0065 μmole h -1 . Under the same experimental condition, the carbamazepine degradation rate by a 6.5 cm TiO2-QOF-Low was also 0.0065 μmole h -1 . But to fully utilize the incident light launched into the single TiO2-QOF-Low, it shall be extended to 26 cm. The 26 cm TiO2-QOF-Low thus showed a carbamazepine degradation rate (rc-Low) of 0.0127 μmole h -1 , which was 97% higher than rc-High (Manuscript Fig. 5d). By bundling seven 6.5-cm TiO2-QOF-High together using our proposed hexagonal arrangement, the carbamazepine degradation rate by a bundle of TiO2-QOF-High (rHigh) was 0.011 μmole h -1 , which was 67% higher than rc-High. Moreover, by bundling seven 26-cm TiO2-QOF-Low, the carbamazepine degradation rate was further improved to 0.015 μmole h -1 , as calculated by adding up the carbamazepine degradation rates by a 6.5-cm TiO2-QOF-Low bundle and that by the extended 19.5 cm portion of the 26-cm TiO2-QOF-Low in the center. The degradation rate by a 26-cm TiO2-QOF-Low bundle (rLow) of 0.015 μmole h -1 is thus about 1.4 times as high as rHigh. In TiO2-QOF bundles, r was attributed to one TiO2-QOF in the center irradiated directly by a UV-LED and six TiO2-QOFs at the edge irradiated by refracted light from the one in the center. Therefore, rHigh and rLow can be expressed as Eqs. 1 and 2, respectively. = + 6 (1) = + 6 (2) where re-High and re-Low are the carbamazepine degradation rate attributed to each of the six TiO2-QOF-High at the edge and each of the six TiO2-QOF-Low at the edge, respectively. Besides, since all the four TiO2-QOF systems received the same UV intensity, their apparent quantum yields followed the same trend as their degradation rates.
As the S affects the amount of refracted light received by the TiO2-QOFs at the edge and thus the carbamazepine degradation rate by a TiO2-QOF bundle. r as a function of S were examined and shown in Supplementary Fig. 17c using the above-mentioned experimental setup. With an increase in S from 1 to 7 mm, both rHigh and rLow decreased proportionally. By subtracting rc-High from rHigh and rc-Low from rLow, re-High and re-Low as a function of S were obtained as shown in Eqs. 3 and 4, respectively. Eqs. 3 and 4 suggest both re-High and re-Low decrease with increasing S. However, in practice, light will be delivered to multiple TiO2-QOFs in the bundle, rather than a single TiO2-QOF. Therefore, the carbamazepine degradation rates by (i) a bundle of TiO2-QOF-Low consisting of seven 26-cm TiO2-QOF-Low each irradiated by one UV-LED (r'Low), and (ii) a bundle of TiO2-QOF-High consisting of seven 6.5-cm TiO2-QOF-High each irradiated by one UV-LED (r'High), were simulated using the data we obtained from Supplementary Fig 17c. The TiO2-QOF in the center receives light launched from a UV-LED and refracted light from the surrounding six TiO2-QOFs at S = x ( Supplementary  Fig. 18a). Each of the six TiO2-QOFs at the edge receives light launched from a UV-LED, refracted light from the surrounding three TiO2-QOFs at S = x, and refracted light from two TiO2-QOFs at a further distance S' = [(x+1)3 0.5 -1] (Supplementary Fig. 18b). Therefore, the overall carbamazepine degradation rate of seven TiO2-QOFs each irradiated by one UV-LED (r') were calculated by summing the carbamazepine degradation rate by each of the seven TiO2-QOFs as shown in Eq. 5.
The r'Low and r'High as a function of S were then calculated and shown in Supplementary  Fig. 18c. Both r'Low and r'High were the highest at an S of 1 mm. At such a small S, r'Low is 44% higher than r'High. This result shows TiO2-QOF-Low which generates high quantity of evanescent waves to activate TiO2 is more efficient to degrade carbamazepine compared with TiO2-QOF-High even when they are bundled together. Nonetheless, such compact arrangement of TiO2-QOFs with an S of 1 mm cannot guarantee uniform mixing as the reactor scales up when installing more TiO2-QOFs and may compromise the degradation rates and apparent quantum yields. A larger S is thus required. With increasing S from 1 to 7 mm, the differences between r'Low and r'High increase from 44% to 96% ( Supplementary  Fig. 18c). The advantage of TiO2-QOF-Low becomes more significant compared with TiO2-QOF-High at larger S. Besides, a 26-cm TiO2-QOF-Low bundle uses 77% less mass of photocatalysts compared with that used in a 6.5-cm TiO2-QOF-High bundle.