TEMPO‐Oxidized Cellulose Nanofibers as Pseudocatalysts for in Situ and on‐Demand Hydrogen Generation via Aluminum Powder/Pure Water Reactions at a Temperature below 50 °C

Cellulose nanofibers (CNFs) prepared via 2,2,6,6‐tetramethylpiperidine‐1‐oxyl (TEMPO)‐mediated oxidation of the C6 primary hydroxyls of native cellulose to carboxylates are used as pseudocatalysts for enhancing the aluminum powder/pure water reactions. The Al powder/pure water reaction is a stepwise reaction. It starts from hydration of the outmost native Al2O3 thin layer and then the reaction of the inner metallic Al with water. At lower temperatures (<50 °C), OH− and Al3+ ions are the preliminary products of the native Al2O3 thin layer hydration. Once Al powders are mixed with pure water containing 0.1–0.5 wt% TEMPO‐CNFs, condensed networks consisting of TEMPO‐CNFs self‐establish over the native Al2O3 thin layer. Al3+ ions are captured by TEMPO‐CNFs via the formation of insoluble Al3+/TEMPO‐CNFs complexing nanostructures and the conjugated OH− ions are being restricted nearby the native Al2O3‐based thin layer via electrostatic repulsion. A highly alkaline condition (pH > 11) is dynamically generated, and as a result, the native Al2O3 thin layer dissolves rapidly via the reaction with OH− ions. The OH− ions function also as catalysts, accelerating the reaction of metallic Al with water. Al powders (2–200 μm) react promptly and a nearly 100% Al/H2 conversion is obtained at the reaction temperature below 50 °C.

(2) NaAlðOHÞ 4 ! AlðOHÞ 3 # þNaOH (3) Using superheated water or steam is another well-known approach to accelerating the Al powder/water reaction for hydrogen generation. Al powders react promptly in saturated water at temperatures of 230-370°C, and an early 100% conversion of Al to hydrogen was shown to take only several tens of seconds, even when using Al powders having an average particle size of 70 μm. [8] A recent study demonstrated further that supercritical water of high density (>200 g L À1 ) at a temperature of 382°C could generate hydrogen by reacting with Al powders having particle sizes up to 3 mm. [9] Notably, in the case of superheated water steam at a temperature of 600°C, Al powders (particle average size: 40 μm) were evaporated to atomic levels, and the gas phase-based reaction of Al/H 2 O produced hydrogen and nanosized Al 2 O 3 particles. [10] The chemistry of the hydrogen generation is principally achieved via Equation (4) 2Al þ 3H 2 Drawbacks to this approach, especially with mass production of hydrogen, are the extra energy needed to generate and maintain the superheated water and/or steam and the complexity of the reactor for the ultrahigh-temperature reaction.
The manufacture of Al powders down to nanosized particles, which enlarges their ultimate surface areas, allows the Al powders to react even with pure water at lower temperatures. For instance, Al powders with an average particle size of 140 nm reacted readily with deionized (pure) water at temperatures of 50-75°C and resulted in a nearly 100% Al/hydrogen conversion. [11] However, a number of drawbacks related to the ultrahigh reactivity of the Al nanoparticles, especially the safety of this reaction, have yet to be overcome.
In fact, Al powders with an average particle size of %20 μm also showed some reactivity in pure water. In this case, the Al/hydrogen conversion efficiency was less than 60%, even after the Al powder/pure water reaction persisted for 5 h at a temperature of 100°C in a specially designed reactor. [12] It is notable that bayerite, the by-product of the Al powder/pure water reaction at temperatures ≤100°C, accumulates on the surface of the reacted Al powders and forms a shell/core structure. [12] Blocking of water by the solidified bayerite-based shell creates another fundamental difficulty encountered in hydrogen generation via the Al powder/pure water reaction at low temperatures.
Regardless of the particle size, the Al powder/water reaction occurs spontaneously with hydration of the topmost native Al 2 O 3 thin layer as the initiating stage. Under low temperatures (<50°C), the breakdown of the native Al 2 O 3 thin layer via the entire hydrolysis reaction is time consuming due to the unique chemistry involved. [13][14][15][16][17] Bunker and co-workers quantitatively investigated the chemistry of hydration of the native Al 2 O 3 thin layer on aluminum using secondary ion mass spectrometry in conjunction with isotopic labeling. [18] Their experimental data indicated that hydroxide ions (OH À ) were the most prevalent product of the hydrolysis reaction of the native Al 2 O 3 thin layer. The OH À ions are mobile in the Al 2 O 3 thin layer but take a day to reach the Al 2 O 3 /Al interface. In addition, the steady-state concentration and ultimate amount of OH À ions are extremely low and insufficient due to the competitive formation of Al(OH) 3 and/or AlOOH, i.e., the formation of the hydroxide state and/or the oxyhydroxide state with the conjugated Al 3þ ions. [18] In addition to the traditional water-soluble caustic soda-based catalyst, water-insoluble, transition metal-based catalysts, such as M-B/γ-Al 2 O 3 , (M = Co, Ni), [19] Fe/AlOOH, [20] and Ni-Li-B, [21] found also capable of enhancing both the native Al 2 O 3 thin layer hydration and the Al powder/water reactions. However, 10 wt% of the catalysts were needed, due entirely to the intrinsic mechanism of the microgalvanic interactions between the solid catalysts and the Al powders.
Note here that the so-called surface-modification method was also developed for enhancing the Al powder/water reaction. [22] Fine powders of γ-Al 2 O 3 , α-Al 2 O 3 , and TiO 2 are the typical agents being used in previous studies [22] for modification of the surfaces of Al powders. The ratio of Al powder/fine powder oxide, however, was as high as 30 vol%.
Unique to both the water-soluble caustic soda-based catalyst and the water-insoluble transition metal-based catalyst, we demonstrated in this study that TEMPO-CNFs are capable of enhancing the Al powder/water reaction and this goal is achieved simply by adding only 0.1-0.5 wt% TEMPO-CNFs in pure water. TEMPO-CNFs are the 1D, cellulose-based nanostructures with a width of 3-5 nm and a length up to few micrometers, produced via 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO)-mediated oxidation of the C6 preliminary hydroxyls of native cellulose fibers to carboxylates. [23] The commercially available TEMPO-CNFs carry %2 mmol g À1 of carboxylates on their surfaces with sodium ions (Na þ ) as their counter ions. [24] TEMPO-CNFs are highly dispersible in water, forming a transparent and uniform suspension with ultra-long stabilities. [23] Al powders once are being mixed with pure water containing 0.1-0.5 wt% TEMPO-CNFs, a condensed network consisting of TEMPO-CNFs and water self-establishes over the surface of the native Al 2 O 3 thin layer. One of the two preliminary products of the native Al 2 O 3 thin-layer hydration, Al 3þ ions are tightly captured by TEMPO-CNFs via the formation of insoluble Al 3þ /TEMPO-CNFs complexing nanostructures. Then, the conjugated OH À ions are firmly protected and are being restricted within the narrow space between the TEMPO-CNF-based networks and the native Al 2 O 3 -based thin layer via electrostatic repulsion. A highly alkaline condition (pH > 11) is dynamically generated, and as a result, the native Al 2 O 3 thin layer rapidly dissolves via Al 2 O 3 þ 2 OH À þ 3 H 2 O ! 2 Al(OH) 4 À . The OH À ions function also as catalysts, accelerating the 2Al þ 6H 2 O ! 3H 2 " þ 2Al(OH) 3 reaction. In other words, TEMPO-CNFs function as pseudocatalysts. Preventing the formation of bayerite-based shells on the surface of the reacted Al powders is another advantageous functionality of TEMPO-CNFs. The TEMPO-CNF-mediated Al powder/pure water reaction is a simple, safe, clean, cost-efficient, and high-performance method and moreover this method is feasible to scale up for mass hydrogen generation at the reaction temperature below 50°C.

Reaction Behaviors of Al Powders in Pure Water Containing TEMPO-CNFs
Unique to the other type of 1D nanostructure, such as carbon nanotubes (CNTs), TEMPO-CNFs are prepared from abundant and renewable plant biomass [23] and highly biocompatible. Figure 1 shows the characteristic morphology of TEMPO-CNFs obtained using cryo-transmission electron microscopy (cryo-EM). As shown in Figure 1, TEMPO-CNFs are the truly 1D nanostructures. Water shows high affinities toward TEMPO-CNFs; water containing TEMPO-CNFs >0.6 wt% is seen as hydrogels. For preparing a water-based suspension containing the entire TEMPO-CNFs with high uniformities, the commercially available suspension containing 2-2.4 wt% TEMPO-CNFs was diluted using deionized water to a lower content of TEMPO-CNFs ≤0.5 wt%. TEMPO-CNFs are capable of capturing di-, tri-, and polyvalent metal cations via chelating interactions among the carboxylate moieties of TEMPO-CNFs and the metal cations. [25][26][27][28] Unique to the conventional chelatant, such as the well-known EDTA, the complexes of metal cations with TEMPO-CNFs are highly insoluble in water. The challenge of this study is to protect hydroxide ions (OH À ) from being consumed in situ by Al 3þ ions during the native Al 2 O 3 thin-layer hydration by using TEMPO-CNFs functioning as protectors.
In the initial experiment, Al powders with sizes ranging from 2 to 30 μm (the average particle size was 20 μm, as recommended by the manufacturer) were used. An Al powder/TEMPO-CNFs suspension prepared with 20.02 g Al powders being dispersed in 800 mL deionized water containing 0.23 wt% TEMPO-CNFs (pH = 6.9) was kept at 36.5°C, i.e., the average temperature of human body, in a water bath under moderate mixing (150 rpm). During the first 3 h, no obvious reaction was observed in this Al powder/TEMPO-CNFs mixed suspension. However, after %3 h from the initial mixing of the Al powders with pure water containing TEMPO-CNFs, the Al powders reacted intensely with water, creating many bubbles. Hydrogen gases popped heavily for about 30 min, and the color of the Al powder/TEMPO-CNFs suspension changed from metallic to gray and finally to a purely milky-white emulsion ( Figure S1, Supporting Information).
Experiments on the abovementioned reaction of Al powder and pure water containing 0.23 wt% TEMPO-CNFs at 36.5°C were repeated in duplicate. During the course of the reactions, %20 mL of the suspension was sampled every hour. Water (pure water and free TEMPO-CNFs) was separated from the reacted Al powders via centrifugation at 13 000 rpm for 10 min at 10°C, and the pH of the supernatant of each sample was measured via a pH meter. Table 1 summarizes the average pH of three measurements (n = 3) and shows that the pH of the reaction medium (pure water containing 0.23 wt% TEMPO-CNFs) changed with time of reaction: the pH increased to a maximum value of pH 11 over 4 h, then it dropped from the maximum value of 11 to pH 10.8, pH 10.7, and pH 10.6 within 4 h. The pH value of the milky-white emulsion after standing still for 24 h exposed to open air at 36.5°C was found to be pH 9.5 due to the absorption of CO 2 from the atmosphere.
The reacted Al powders sampled during the time of reaction were separated from the water via centrifugation and then dried at 60°C in an oven overnight. The dried powders were measured via X-ray diffraction (XRD) and scanning electron microscopy (SEM). Figure 2a shows the XRD patterns of the as-received metallic Al powders, and Figure 2b-f shows the XRD patterns of the Al powders after being reacted in pure water containing 0.23 wt% TEMPO-CNFs for 1, 2, 3, 4, and 5 h, respectively. As shown in Figure  (1 3 À2), which were identified to correspond to bayerite (DB card number: 01-074-1119). These were observed for all five samples after being reacted for 1, 2, 3, 4, and 5 h, respectively. The intensities (peak height) of each bayerite peak increased with Table 1. Al powder/pure water reaction mediated via TEMPO-CNFs (0.23 wt%) at 36.5°C: changes in pH values (average value of three measurements, R.S.D: AE0.2) of the reaction media (bulk water) as the reaction proceeded from beginning to end. Water was separated from the reacted Al powders via centrifugation at 13 000 rpm for 10 min at a temperature of 10°C, and the pH was measured via a pH meter.  [29,30] However, the peaks of TEMPO-CNFs were not observed for the by-product of the Al powder/pure water reaction (Figure 2b-f ), because the free TEMPO-CNFs, after the reaction, were separated from the solidified product via centrifugation. There are some broad peaks (Figure 2d), assumable corresponding to the complexes of Al 3þ /TEMPO-CNFs.
For comparison, Figure 2g shows XRD patterns for the same lot of the as-received Al powders (average particle size: 20 μm) after reacting with pure water at a temperature of 36.5°C but without adding TEMPO-CNFs in pure water. XRD peaks corresponding to the residual metallic Al (DB card number: 00-004-0787), 2θ = 38.4533 (1 1 1), 2θ = 44.6965 (2 0 0), www.advancedsciencenews.com www.advenergysustres.com 2θ = 65.082 (2 2 0), and 2θ = 78.214 (311), were clearly observed, even after 4 days of reaction. Bayerite was also identified as the major by-product of the Al powder/pure water reaction at a temperature of 36.5°C after 4 days of reaction. Morphology data give new insights into the kinetic behaviors of the Al powder/pure water reaction under the mediation of TEMPO-CNFs. Figure 3 shows SEM images together with energy dispersive X-ray spectroscopic (EDS) mappings of oxygen and aluminum elements in the cross section of the Al powders, after reacting with pure water under the mediation of TEMPO-CNFs at a temperature of 36.5°C for 5 h. As shown in the SEM images, after the reactions, the round (parent) shapes of the as-received Al powders no longer exist. Under the mediation of TEMPO-CNFs, the parent Al powders, after the reaction, were digested 20 µm 20 µm  A small amount of the as-produced milky-white emulsion, which contains %7.2 wt% of the fine bayerite particles, was diluted with pure water containing 0.1 wt% TEMPO-CNFs and was then analyzed using a LB-550 dynamic light-scattering nanoparticle size analyzer for estimating the particle size of the as-produced bayerite. The average particle size of bayerite was found to be 102.2 nm ( Figure S2, Supporting Information). Note that the as-produced fine particles of bayerite are uniformly dispersed and remained stable for a month in pure water containing a small amount of TEMPO-CNFs, but they sank shortly when placed in pure water alone ( Figure S3, Supporting Information).
The amount of free Na þ ions that originated from the TEMPO-CNFs (the counter ions of TEMPO-CNFs) involved in the as-produced milky-white emulsion was measured using cation-exchange chromatography. Its amount was found to be 2.30 AE 0.04 mmol L À1 (n = 2), suggesting that the counter ions of TEMPO-CNFs were replaced by Al 3þ ions via the formation of the Al 3þ /TEMPO-CNF complexing nanostructures during the course of the Al powder/pure water reactions.
A water-soluble species of aluminum, mainly aluminate ions, i.e., Al(OH) À 4 , was produced when dissolving the native Al 2 O 3 thin layer via Equation (1) and was involved in the as-produced milky-white emulsion. Its amount was measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES) and was found to be 0.34 AE 0.04 mmol L À1 (n = 5).
Note that the supernatant obtained via centrifugation of the asproduced milky-white emulsion at 13 000 rpm for 10 min at 10°C was the samples used for measuring the amount of free Na þ ions and Al(OH) 4 À ions. Figure 4 shows SEM images together with the EDS mappings of elements of oxygen and aluminum in the cross section of Al powders, after reacting with pure water without TEMPO-CNFs at a temperature of 36.5°C for 4 days. As shown in the SEM images and the EDS mappings, the parent shape of the as-purchased Al powders (2-30 μm) remained unchanged. Moreover, many core/ shell structures, with the residual metallic Al as core and bayerite as shell, were identified by both XRD and EDS mapping. Five core/shell structures were selected, the thickness of the bayeritebased shell ranged from 3.5 to 8.4 μm, and the metallic Al-based core ranged from 2.6-22 μm, as identified from the SEM images (Figure 4b). Also, many small holes were observed in the bayerite-based shells.
Next, the reactivity in pure water of Al powders having average particle sizes of 75 and 150 μm, respectively, were examined. In the case of pure water containing 0.23 wt% TEMPO-CNFs, both the 75 μm Al powders and the 150 μm Al powders were completely converted into bayerite and hydrogen at the reaction temperature of 36.5°C, as identified by EDS mapping and XRD analysis ( Figure S4 and S5, Supporting Information). The completed reaction took %22 and 26 h for the 75 and 150 μm Al powders, respectively. The parent Al powders were also completely digested into fine bayerite particles.
In fact, under the mediation of TEMPO-CNFs, the Al powder/ pure water reactions occurred favorably even at room temperature. Figure S6a, Supporting Information, shows the XRD patterns of the 150 μm Al powders after 2 days of reaction in pure water containing 0.23 wt% TEMPO-CNFs at room temperature (8-22°C). XRD peaks of the metallic Al were not observed, indicating a completed reaction of the parent Al powders with water. In contrast, the 150 μm Al powders maintained their metallic XRD peaks ( Figure S6b, Supporting Information) and metallic colors in pure water without TEMPO-CNFs at room temperature, even after being immersed in pure water for 2 weeks ( Figure S7, Supporting Information).

Reaction Behaviors of Al Powders in Pure Water
Containing 3Na-EDTA The advantageous performance of TEMPO-CNFs in accelerating the Al powder/pure water reactions was demonstrated by comparison experiments using tri-sodium salt EDTA (EDTA-3Na) as accelerators. EDTA is a well-known chelating agent capable of capturing Al 3þ ions via chelating interactions. [31] Al powders with sizes ranging from 2 to 30 μm (the average particle size was 20 μm) were used. As demonstrated experimentally, EDTA-3Na is also capable of accelerating of the Al powder/pure water reactions, but at low temperatures, for example, at the reaction temperature of 36.5°C, it took 2 days (48 h) to complete the reactions. Table 2 summarizes the pH values of the water bulk during the reactions. The pH value increased from the initial value of pH 6.9 to pH 10.5 after 1 h of reaction. The pH value, however, then remained almost constant at 10.5 during the rest of the reaction. XRD patterns were also used to estimate the degree of the Al powder reactions with water. Figure S8, Supporting Information shows the related XRD measurement data, and it can be seen that Al powders remained essentially as metal during the first 7 h. Peaks corresponding to the metallic Al were clearly observed, even though the reactions lasted as long as 28 h. All the metallic Al peaks disappeared after 48 h of reaction. Bayerite was also the major solidified products of the EDTA-3Na-mediated reactions, as shown in the XRD data. Note that the Al powders showed excellent inert properties in pure water containing EDTA-2Na (pH = 3.8), with the concentration of EDTA-2Na identical to that of EDTA-3Na.

Reaction Behaviors of Al Powders in Pure Water Containing Sodium Alginate
To further demonstrate the superior properties of TEMPO-CNFs in accelerating the Al powder/pure water reactions, we also performed the reactions at 36.5°C with sodium alginate (Na-ALG) as accelerators. Na-ALG is a linear, polysaccharide-based copolymer with β-D-mannuronate and α-L-guluronate, each carrying a carboxylate moiety, as the basic blocks. Na-ALG is truly soluble in water and can capture Al 3þ ions via gel formation. [32] Al powders with sizes ranging from 2 to 30 μm (the average particle size was 20 μm) were used. As shown in the XRD measurement data ( Figure S9, Supporting Information), the Al powders remained stable in the metallic phase for as long as 30 h. Peaks corresponding to the metallic Al were recognizable after 98 h of reaction, but metallic Al peaks were not observed after 101 h of reaction. After the reaction, a purely milky-white emulsion was also obtained. Table 3 summarizes the pH value of the water bulk during the reaction. Note that bayerite is a minor component of the solidified product in the Al powder/pure water reactions mediated via Na-ALG, as is indicated by the XRD data.   Figure 4b shows the SEM images of five core/shell structures selected for measuring the thickness of the related shell and the core. Figure 4c shows the SEM image of the fine structures of the bayerite-based shell.

Chemistry of the Al Powder/Pure Water Reactions Mediated via TEMPO-CNFs
The Al powder/water reaction is a stepwise reaction. It starts from the hydration of the native Al 2 O 3 thin layer that forms naturally on the surface of Al; the reaction of the metallic Al with water starts once the native Al 2 O 3 thin layer is being broken down either partially or completely. Pitting defect is a wellestablished model and has long been used for classifying the corrosion mechanism for passive metals, such as aluminum and stainless steel. [33][34][35] In the pitting defect mechanism, foreign ions, for example the halide ions [36,37] in water, are key elements affecting the kinetics of pitting via the mechanism of anodic polarization. In the case of the aluminum/water reaction, anionic defects such as aluminum vacancies (O 2À ) are driven toward the Al 2 O 3 /Al interface, and cationic defects such as oxygen vacancies (Al 3þ ) are concurrently driven toward the Al 2 O 3 /water interface by the anodic polarization through the native Al 2 O 3 thin layer. [18] Water molecules are mobile in the native Al 2 O 3 thin layer. They react with O 2À ions at the Al 2 O 3 /Al interface and form hydroxide (OH À ) ions, as is expressed in Equation (5) The OH À ions produced via Equation (5) function as a catalyst, accelerating the Al/water reactions.
Note that, in the case of hot water (>90°C), the breakdown of the native Al 2 O 3 thin layer was accomplished mainly because of the so-called uniform defect mechanism. Three essential stages are involved [18,38] : 1) the native Al 2 O 3 thin layer is converted to aluminum oxyhydroxide, i.e., AlOOH (boehmite), with the thickness almost identical to that of the thickness of the native Al 2 O 3 thin layer; 2) the metallic phase of aluminum reacts with water and produces pseudoboehmite at the AlOOH/Al interface; and 3) a certain amount of bayerite, Al(OH) 3 , is generated on top of the pseudoboehmite, as expressed in Equation (6)-(8), respectively Hydration of the native Al 2 O 3 thin layer via the pitting defect generates hydroxide ions, OH À , and the overall reaction can be described by Equation (9) However, without any protections, the OH À ions were consumed immediately via the formation of Al(OH) 3 , as is described in Equation (10) Al 3þ þ 3OH À ! AlðOHÞ 3 (10) In case of water containing TEMPO-CNFs, TEMPO-CNFs have captured Al 3þ ions via the formation of insoluble Al 3þ / TEMPO-CNFs complexing nanostructures. This capturing of Al 3þ ions protects the conjugated OH À ions, which consequently change the chemistry involved in the breakdown of the native Al 2 O 3 thin layer. A dissolution model, illustrated in Figure 5, is proposed to explain the chemistry involved in the Al/water reactions mediated via TEMPO-CNFs. Basic concepts of this dissolution model are summarized as follows: 1) Regardless of the temperature of the water, the native Al 2 O 3 thin layer on the surface of aluminum will spontaneously react with water. In other words, Al 2 O 3 hydration is a spontaneous type of reactions. 2) Al 2 O 3 hydration generates two preliminary species of ions, i.e., Al 3þ ions and OH À ions. The overall surface area of the native Al 2 O 3 thin layer is the key parameter determining the ultimate amount of Al 3þ ions and OH À ions produced during the hydration. 3) Both Al 3þ ions and OH À ions are immediately consumed via the formation of more stable compounds of aluminum hydroxide and/or aluminum oxyhydroxide. In other words, the lifetime of both the free Al 3þ ions and the free OH À ions is extremely short. 4) The resulted aluminum hydroxide and the aluminum oxyhydroxide accumulate on the surface of the reacted Al powders, forming shell/core structures. 5) In the case of water containing TEMPO-CNFs, a certain amount of TEMPO-CNFs attach to Al 2 O 3 , forming condensed networks carrying plenty of negative charges (carboxylates) over the surfaces of the native Al 2 O 3 thin layer. 6) Of the two species of the preliminary ions, Al 3þ ions are captured via the formation of insoluble Al 3þ /TEMPO-CNFs complexing nanostructures, and the conjugated OH À ions are pushed toward Al 2 O 3 via electrostatic repulsion. 7) A highly alkaline condition is dynamically generated, and breakdown of the native Al 2 O 3 thin layer is accomplished in a faster manner via dissolution, as described in Equation (11) The naked metallic Al reacts with water via OH À ions as the catalyst, and the overall reaction can be simplified as in Equation (12) and (13)  2) of the reaction media (bulk water) as the reactions proceeded from beginning to end. Water was separated from the reacted Al powders via centrifugation at 13 000 rpm for 10 min at a temperature of 10°C. www.advancedsciencenews.com www.advenergysustres.com In other words, TEMPO-CNFs function as a pseudotype of catalysts.
Note here that, in case of the Al powder/water reactions underwent with OH À ions as catalysts, the parent Al powders, after the reactions, shall be digested into fine particles of Al(OH) 3 , as is implied from Equation (11)- (13). Note also that 0.1-0.5 wt% is the desirable range of concentration of TEMPO-CNFs as pseudocatalysts. At a concentration <0.1 wt%, TEMPO-CNFs were at an insufficient amount to enhance Al powder/pure water reaction; at a concentration >0.5 wt%, TEMPO-CNFs form hydrogels, retarding water-diffusion.

Hydrogen Generation via Al Powder/Pure Water Reaction Mediated by TEMPO-CNFs
Hydrogen generated during the Al powder/pure water reaction mediated via TEMPO-CNFs was identified using gas chromatography (GC). Table S1, Supporting Information, summarizes the GC conditions. Three standard gas samples, containing 0.1% H 2 , 5.0% H 2 , and 15.0% H 2 , respectively, were used to obtain the calibration curve, and a calibration curve (Y = 3198, 494.61X þ 332, 463.14) with a linearity R 2 = 0.9997 was obtained. All reactions and sample collections were performed under atmospheric pressure and open air ( Figure S10, Supporting Information). The reaction temperature was kept constant at 45°C via a water bath. Sampling of the gases during the reaction was performed with a 100 mL syringe, and a total of 300 mL was collected for each sample. Each sample was analyzed in triplicate using GC, and areas of the H 2 peak for each analysis are summarized in Table S2, Supporting Information. The as-recorded chromatograms of the nine samples are given in supporting Figure S11, Supporting Information. Peaks corresponding to oxygen (O 2 ) and nitrogen (N 2 ) were also detected, because both the reaction and the sample collection were performed under open air. Figure 6a shows the concentration of hydrogen in each sample. The sample collected after 60 min of reaction had 1.55% H 2 , indicating that the native Al 2 O 3 thin layer was partially broken within 1 h after the Al powders were mixed with pure water containing 0.23 wt% TEMPO-CNFs. A higher concentration (11.81% H 2 ) was found in the sample collected at 90 min, and the highest concentration (17.92% H 2 ) was found in the sample collected at 105 min of reaction. The concentration of H 2 dropped to 11.3% in the sample collected at 135 min, 7.11% in the sample collected at 165 min, 5.01% in the sample collected at 195 min, and 0.53% in the sample collected at 300 min.
The color of the emulsion of Al-power/pure water-TEMPO-CNFs changed from the initial metallic to gray at %90 min and finally to milky white after %195 min of reaction.
For quantifying the hydrogen yield, the ultimate volume of hydrogen generated via the Al powder/pure water reaction was measured by monitoring the amount of water displaced by hydrogen during the time of reaction. A photograph showing the equipment setup for measuring the hydrogen volume is given in Figure S12, Supporting Information. Figure 6b shows the volumes of hydrogen generated via the Al powder/pure water reaction with and without the addition of TEMPO-CNFs in pure water at 45°C. The advantageous performance of TEMPO-CNFs as pseudocatalysts have been experimentally demonstrated: 1) the induction time is being reduced from 90 to 60 min, 2) the hydrogen generation ratio is being enlarged from 0.70 to 5.77 mL min À1 (the linear range of 100-220 min was used for calculating the hydrogen generation rate), and 3) the yield is being elevated from 12.4% (111/897 mL mL À1 ) to 84.5% (758/897 mL mL À1 ) after the reaction persisted for 220 min. The yield of hydrogen is calculated based on Equation (14) [21] Figure 5. Schematic representation of TEMPO-CNFs prepared via regioselective oxidation of C6 primary hydroxyls of the native plant cellulose fibers to carboxylates under the mediation of TEMPO and the key chemistries involved in the Al powder/pure water reaction mediated by using TEMPO-CNFs. Condensed networks consisting of TEMPO-CNFs and water self-established over the native Al 2 O 3 thin layer of the Al powders. The native Al 2 O 3 thin layer starts to hydrate, producing Al 3þ ions and OH À ions, and TEMPO-CNFs capture Al 3þ ions via the formation of insoluble Al 3þ /TEMPO-CNFs complexing nanostructures. The conjugated OH À ions are restricted within the narrow space between the TEMPO-CNF-based networks and the native Al 2 O 3 thin layer via electrostatic repulsion. The native Al 2 O 3 thin layers dissolve rapidly by reacting with OH À ions, and the as-received Al powders, therefore, lose their native Al 2 O 3 thin layers. The naked metallic Al powders react with water via OH À ions as catalysts, producing hydrogen (H 2 ) and Al(OH) 3 (bayerite).
where Y (%) is the yield of hydrogen, V is the volume of water displaced by hydrogen, W is the weight of Al powders (0.66 g), and 24.45 (L) is the volume of 1 mol hydrogen (298 K and 1 atm).

Conclusion
The sluggish kinetics of hydration reaction of the native Al 2 O 3 thin layer on Al powders is the major bottleneck encountered in hydrogen generation via Al powder/pure water reaction. This intrinsic difficulty, as is demonstrated experimentally, in this study, is overcome by simply adding 0. reaction. The conjugated OH À ions functioned also as catalysts, enhancing the metallic-Al/pure water reaction via 2Al þ 6H 2 O þ 2OH À ! 2Al(OH) À 4 þ 3H 2 " and Al(OH) 4 À ! Al(OH) 3 # þ OH À . The TEMPO-CNF-mediated Al powder/pure water reaction is safe, clean, and cost-efficient and is feasible to scale up for in situ and on-demand hydrogen generation. Moreover, the by-product of the Al powder/pure water reaction, i.e., bayerite., can be perfectly reconverted to Al powders with zero emissions via electrolysis if powered by renewable energy. Developing of methods suitable for evaluating behaviors, such as stability of TEMPO-CNFs in Al powder/pure water reaction, is under investigation by this research group.

Experimental Section
Al Powders: Three types of Al powders with purity >99.9% and average particle sizes of 20, 75, and 150 μm, respectively, were purchased from Kojundo Chemical Laboratory Co. Ltd. (Tokyo, Japan) and used as received. SEM images showing the morphologies and the actual size of the Al powders are shown in Figure S13, Supporting Information.
TEMPO-CNFs: 2-2.4 wt% TEMPO-CNFs being dispersed in deionized water were purchased from Dai-ichi Kogyo Seiyaku Co. Ltd. (Kyoto, Japan); they were produced via TEMPO-mediated oxidation of native plant cellulose fibers [23] and were used by diluting the as-received 2-2.4 wt% TEMPO-CNFs with deionized water.
Deionized Water: The deionized water was produced using distilled water followed by ion exchange via STILL ACE SA-2100E1 (EYELA).
Al Powder/Pure Water Reactions: Reactions were performed in a 1000 mL flat-neck flask. %800 mL of pure water containing 0.1-0.5 wt% TEMPO-CNFs was poured into a 1000 mL flat-neck flask and then 20-30 g of Al powders was added. The flask was placed in a water bath while mixing the pure water-TEMPO-CNFs/Al powders at 150-300 rpm with a motor-driven mixer.
Gas Sampling: The hydrogen gas was sampled via a 100 mL syringe. The total volume of each collected sample was 300 mL. All sampling was performed under open air; thus, all samples contained the produced hydrogen together with oxygen and nitrogen of open air. We also sampled the mixture of pure water-TEMPO-CNFs/Al powders during the course of the reactions to measure the change in pH and to analyze the reacted Al powders using SEM, EDS, and XRD. The sampled volume was %20 mL.
Hydrogen Volumes: The volume of hydrogen generated via Al powder/ pure water reaction was measured by measuring the amount of water replaced by hydrogen. The reactor (500 mL glass bottle) containing Al powders and pure water with and without TEMPO-CNFs was placed in an ultrasonic bath filled with water to ensure a proper mixing during the reaction. A photograph showing the equipment setup for the hydrogen volume measurement is given in Figure S12, Supporting Information.
Morphology Observation of TEMPO-CNFs via Cryo-EM: The iceembedding method with a Leica EM GP plunge freezer was used. Briefly, 3.0 μL of the solution containing 0.23 wt% TEMPO-CNFs was diluted 1/100th with deionized water and was then dropped onto a hydrophilized Quantifoil grid (R2/1, Cu 400 mesh grid) in a chamber with 90% relative humidity and a temperature of 24°C. The excess sample solution was removed from the back of the grid by blotting with filter paper for 5 s, and then the grid was plunged frozen into liquid ethane (À175°C). Frozen samples were transferred to a Gatan 914 cryo transfer holder at the temperature of liquid nitrogen and observed with a JEM 2100F cryo-transmission electron microscope operated at an accelerating voltage of 200 kV. The images were taken using the minimum dose system. SEM Images of the Cross Section of the Reacted Al Powders: After embedding the reacted Al powders in epoxy resin, the cross section was made by Ar ion beams via SM-09010 and SM-09020 (JEOL). The cross section was coated with Os to impart the conductivity. SEM images were obtained via (a) (b) Figure 6. a) Hydrogen generation via TEMPO-CNFs mediated Al powder/ pure water reaction, 20.02 g Al powders (average particle size: 20 μm) reacted with pure water containing 0.23 wt% TEMPO-CNFs at 45°C. b) Hydrogen generation curves via Al powder/pure water reaction at a reaction temperature of 45°C. Al powders (purity >99.9, particle average size 20 μm, 0.66 g) reacted with 400 mL deionized water containing 0.23 wt% TEMPO-CNFs (blue curve) and with 400 mL deionized water alone (yellow curve).