WO2021262323A2

WO2021262323A2 - Electrocatalysts and electrolyzers

Info

Publication number: WO2021262323A2
Application number: PCT/US2021/030707
Authority: WO
Inventors: Vinh Nguyen; Steven Dean DIETZ; Girish Srinivas
Original assignee: Tda Research, Inc.
Priority date: 2020-05-04
Filing date: 2021-05-04
Publication date: 2021-12-30
Also published as: WO2021262323A3

Abstract

A porous carbon-platinum electrocatalyst made from a carbon matrix-platinum composite, which has a carbon support and platinum is deposited into the pores of the carbon support. The platinum is at least 2 weight% of the carbon matrix-platinum composite. The carbon support has a nitrogen BET surface area of about 550 to 1800 m²/g; a packing density of at least 0.8 g/cc as determined at a compressive force of 500 kgf/cm2 on dry carbon powder; and an electrical conductivity of at least 10 S/cm at a compressive force of 500 kgf/cm² on dry carbon powder. Also, an electrolyzer, comprising: the porous carbon-platinum electrocatalyst, wherein the electrolyzer has a Tafel slope of at most 40 mV / decade.

Description

ELECTROCATALYSTS FOR ELECTROLYZERS

CROSS REFERENCE TO RELATED APPLICATIONS

The present application claims the benefit of the provisional application number 63/019,993 filed May 4, 2020 (titled CONDUCTIVE MATRIX, by Steven Dietz, attorney docket number 18-1D), which is incorporated by reference herein.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

This invention was made using U.S. government funding through the U.S. Navy SBIR Phase I (NAVSEA) contract N00024-17-P-4509. The government has certain rights in this invention.

BACKGROUND OF THE INVENTION

Thermodynamically, electrolysis of water requires only 1.23 V under standard conditions. However, in reality, the required cell voltage is always much higher. This is the result of several factors; first, the kinetic barriers associated with making and breaking chemical bonds during the Oxygen Evolution Reaction (OER) and the Hydrogen Evolution Reaction (HER) increase the voltage required at a given current density. Secondly, an additional voltage increase is required to overcome the ohmic resistance of the electrolyte membrane and other Membrane Electrode Assembly (MEA) components. Lastly, a voltage increment also results from depletion of the reactant concentration relative to the product concentration near each electrode during electrolysis. Therefore, the total overpotential of an electrolyzer is in fact the sum of the activation overpotential, ohmic overpotential, and concentration overpotential (Figure 1).

To minimize the total cell voltage due to these multiple overpotentials, several parameters need to be developed or optimized. For example, the cell voltage can be reduced by improving electrocatalytic activities at each electrode, minimizing electrolyte resistance of the membrane and at its interface, or effectively increasing mass transfer at the flow-fields. In addition, the electrolyzer operating conditions such as water flow rate and cell temperature need to be optimized to give the highest electrolysis performance. Although most of these parameters have been studied, developed, or optimized, current electrolyzers still use high loadings of Pt at the cathode (1.0 mg/cm²) to overcome these limitations, which significantly increases the total system cost.

Polymer electrolyte membrane (PEM) electrolysis used on-board submarines to generate breathing oxygen for the crew, for instance, requires expensive noble metals as electrocatalysts. These precious metals contribute approximately 25% of the cost of the electrolysis cell. Greatly reducing or eliminating the amount of noble metal electrocatalysts would significantly reduce the cost of the electrolysis cell stack.

The operating principle of a PEM electrolysis cell is shown in Figure 2. The half reaction on the anode side is referred to as the OER. Using an iridium oxide electrocatalyst, the liquid water reactant is oxidized to generate oxygen, protons and electrons. On the cathode side, the supplied electrons and protons that are conducted through the membrane are combined using a platinum electrocatalyst to create gaseous hydrogen through the HER. Pt loadings for the cathode side range between 0.5 and 1.0 mg/cm², and this represents a considerable portion of the total system cost.

The cathodes of PEM electrolyzers employ highly-dispersed nanoparticles of pure platinum supported on a carbon black. These carbon blacks are highly electron-conducting (which is important) and are largely adapted unmodified from products that are used as filler materials in electrically-conductive polymers, including rubber. For long-life, PEM cathodes need to retain a very high electrochemical surface area (ECSA) and to retain highly electrochemically active Pt sites. However, during PEM electrolysis, catalytic activity is decreased by a number of very well established mechanisms, including degradation and corrosion of the carbon support.

BRIEF DESCRIPTION OF THE INVENTION

The present invention solves the limitations of the prior are by providing a carbon materials that exhibits a combination of properties of activated carbons (high surface area) and carbon blacks (high conductivity). By forming activated carbons in-situ in the presence of carbon blacks, we teach the ability to make a carbon with both high porosity and electrical conductivity (Figure 3). This product is fundamentally different than mixing carbon black with activated carbon, like is typically done in battery electrodes to increase their electrical conductivity. In that case the activated carbon and carbon black particles remain separate and distinct, whereas the present invention is a homogenous carbon material with properties that are unique compared to either activated carbon or carbon black. BET characterization and bulk-conductivity measurements have shown that the activated carbons of the present invention have high surface areas (550 to 1800 m²/g), contain pore sizes in both the 1 to 10 nm and 11-100 nm ranges, and maintain good electrical conductivities (an electrical conductivity of at least 10 S/cm at a compressive force of 500 kgf/cm² on dry carbon powder). As a result, the electrocatalysts prepared from deposition of metal nanoparticles on these carbon matrixes show high activity and long cycle life. This high performance is attributed to the conductive mesoporous carbon matrix that has high surface area to support small and well-dispersed metal particles.

In addition, the present invention teaches an in-situ vapor-phase dissociative method that deposits very small Pt nanoparticles on the mesoporous carbon support, which provides a high ECSA of Pt for improved HER performance. In fact, the method produces smaller Pt particles than other traditional solution-based synthesis methods, such as depositing and reducing chloroplatinic acid precursor in hydrogen, sodium borohydride (NaBH4) or ethylene glycol (HOCH2CH2OH). Furthermore, the method doesn’t rely on solvents and therefore is easily scalable to high production volumes.

Using the carbon black (core) activated carbon (shell) carbon supports described in this invention and the disclosed method of Pt deposition, the present invention teaches the method to produce electrocatalysts with approximately 2.0 nm Pt nanoparticles with uniform distribution. The platinum particles preferably are 2.1-2.5 nm. The resulting Pt catalysts have very high ECSA and are stable enough to retain highly electrochemically active Pt sites inside the pores. As a result, the present invention allows for reducing the Pt loading without sacrificing HER activation overpotential, and thus solves the limitations of the prior art.

The examples below show that electrocatalysts of the present invention showed proper HER performance, equivalent to commercial Pt catalysts used in most electrolyzes, although the Pt loading of the catalyst of the present invention was a factor of 10 lower than the prior art. Reducing the Pt loading by a factor of 10 using the carbons of the prior art does not enable electrolyzers that operate properly (or that have the same proper HER performance). Based on the reduction to practice in the examples below, the electrocatalysts of the present invention can run the HER with platinum metal loadings of 0.1 mg/cm², while providing performance (both initially and over time) that provides a Tafel slope of less than 40 mV/decade. BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1. Summary of overpotential penalties.

Figure 2. Schematic of a PEM electrolysis cell.

Figure 3. Representation of activated carbon coated carbon black.

Figure 4. Pore size distribution of different carbons.

Figure 5. TGA of 26.28% Pt concentration on TDA-8 sample (VN133770A).

Figure 6. XRD peak fitting showed 2.1 nm Pt particle size on TDA-8 sample (VN133770A).

Figure 7. HER Tafel slope estimation (VN133770A).

Figure 8. Effects of temperature and pressure used to hot-press the MEA.

Figure 9. Optimizing MEAs: a / Amount of Ir0₂; b / Platinized titanium screen thickness; c/ Gas diffusion layer type; d/ Amount of ionomer in the catalyst layer.

Figure 10. Effects of serpentine vs. parallel flowfields and cell orientation.

Figure 11, Pt reduction methods tested.

Figure 12. Pol curves on Pt/TDA-7 (left) and Pt/TDA-17 (right) using unmodified Pt catalysts (solid) and the same catalysts after alloying with Co (dash).

Figure 13. Effect of Pt particle size on HER performance.

Figure 14 Effect of Pt loading on TDA’s carbon support.

Figure 15. Effect of TDA-8 supports’ surface area (SA) and 4-7 nm pore volume (PV) on the corresponding HER electrocatalytic performance (~ 30 wt.% Pt).

Figure 16. Cell voltage decreased with pore volume.

Figure 17. Comparison of TDA’s carbon supports with commercial Vulcan XC-72R carbon support.

Figure 18. Test for durability and regeneration of the catalyst.

Figure 19. A 5g scale up of Pt/TDA-8 electrocatalyst and its in-situ electrolyzer performane compare to a 0.25g scale.

Figure 20. Representation of activated carbon coated carbon black and with platinum nanoparticles.

DETAILED DESCRIPTION OF THE INVENTION

The present invention uses carbon supports described in PCT/US20/31337, which provides mesoporous carbons (pores 2-100 nm). The carbons were derived from carbohydrates and carbon black. We can incorporate elements such as nitrogen, sulfur, phosphorous and others into carbohydrate-derived carbons. This is done by adding compounds with the desired element to the mixture before carbonization. The elements are bound tightly into the activated carbon and cannot be easily removed. These carbons coat the carbon black which increases the electrical conductivity of the carbons. The hetero atoms in the activated carbon can be used to stabilize platinum and other metal nanoparticles on the carbons for use as improved electrocatalysts.

Four TDA carbons (numbered TDA-7, TDA-8, TDA-17, and TDA-23) were prepared in different formulations and compared with the baseline commercial carbon black XC-72R. We also varied the activation time of TDA-7 to generate different surface areas and pore size distributions of the same carbon formulation (subset samples labeled TDA-7A, TDA-7B, and TDA-7C). The characteristics of the carbons tested can be found in Figure 4 and Table 1.

Table 1. Surface areas and pore volumes of TDA carbons vs. Vulcan XC-72R tested as catalyst supports.

The BET data indicated that TDA carbons have high surface area (560-1620 m²/g, and preferably at least 873 m²/g) and contain both micro and meso pores. Although not wishing to be bound by theory, particularly preferred are pores 4-7 nm, where the Pt nanoparticles are possibly encapsulated or stabilized during formation. The data showed a wide pore volume in this range (0.0027-0.036 cm³/g). In comparison, the carbons of the prior are (an example is Vulcan XC-72R) have very low surface area (254 m²/g) and less overall pore volume than the present carbons. Although not wishing to be bound by theory, the carbons of the present invention contain a combination of these 4-7 nm pores as well as much larger pores. The combination of these two types of pores allow the Pt nanoparticles to remain stable in the carbon and the two types of pores allow water to diffuse out and allow for a high hydrogen diffusion rate, which enables the high performance of an electrolyzer using the Pt carbon material in the HER electrode. The present invention teaches a selection that is an optimal support, and methods are described how to deposit Pt nanoparticles on these carbons. We show characterization of the Pt nanoparticle size, and evaluated their electrocatalytic performance in both ex-situ RDE and in-situ electrolyzer studies. The present invention teaches surprisingly good HER performance in an electrolyzer for a surprisingly low Pt loading content.

One advantages of our carbon technology are low cost and high purity and most importantly, our ability to tailor the pore size to hold encapsulated Pt nanoparticles; these advantages stem directly from our use of refined sugar precursors. The invention uses carbons based on refined sugars, such as sucrose or high fructose corn syrup, because the resulting carbons have the high intrinsic purity needed for use in electrochemical devices. Conventional activated carbons made from natural products, such as coconut shells, coal or wood have 2-5% ash content, compared to <1 % for our carbons. Because of their high purity, our porous carbons have electrical conductivities that are comparable to carbon blacks (10-30 S/cm), while conventional activated carbons have conductivities of < 2 S/cm. Many starting materials for carbons are possible, and many are based on natural by-products. Starting with impure carbon precursors such as coconut shells results in carbons with impurities, which must be removed in expensive subsequent processing. Because refined sugars are purified by crystallization in high volume for the food industry, they are a very pure but inexpensive starting material for carbons.

By controlling the process conditions and adding pore formers, we have developed the capability to independently control the distribution of pores with diameters between 1 and 10 nm and pores from 11 to 100 nm. The pore diameter in the mesopore range is controlled by the amount of pore former in the formulation, while we control formation of micropores during the activation step. The smaller mesopores (2-10 nm) help disperse the Pt nanoparticles and limit their size growth. The larger mesopores (>10 nm) enable rapid transport of water in and out of the carbon when used in an electrolyzer. The present invention also may use methods to control the pore size distribution of these carbons that are found in US Patent 6,297,293, 2001 , US Patent 6,737, US Patent 7,167,354, 2007, US Patent 7541312, 2009 and US Patent 9,120,079, which are incorporated by reference herein. PCT/US20/31337 also describes making the base carbon material (prior to depositing the Pt), and is also incorporated by reference herein. All of these patents and patent applications share coinventors of the present invention and all are owned by the same owner at the time of filing.

The first step in the hydrolysis of common sugar (sucrose) is the cleavage of an ether linkage to produce a 50/50 mixture of glucose and fructose (Scheme 1). Alternatively, the ratio of glucose and fructose can be changed by starting with mixture of pure fructose and glucose.

Scheme 1

Regardless of which sugar is used, the next step is the acid hydrolysis of these compounds (generically called hexoses since they all have the same molecular formula, ObH^Ob). The acid hydrolysis of hexoses generates mainly 5-hydroxymethylfurfural, levulinic acid and formic acid (Scheme 2). Then with continued heating the 5-hydroxymethylfurfural and levulinic acid polymerize with the further loss of water to leave behind char, which is mostly carbon with a small amount of residual hydrogen and oxygen. When the char is carbonized (i.e. >500°C in an inert atmosphere) the polymer decomposes, evolving mainly CO2, CO and water to give a high purity carbon.

Scheme 2.

Hexoses

(Fructose and Glucose) 5-Hydroxymethylfurfural Levulinic acid Formic acid

In addition, we can add nitrogen (and other heteroatoms) to the carbons to improve the electrochemical properties, or to stabilize other metals associated with the surface, such as platinum nanoparticles. With the use of nitrogen containing compounds and control of their quantity, we can control the type and concentration of the nitrogen functional groups in the carbon.

As shown schematically in Figure 5, the general procedure for preparing these porous carbons was to: Step 1: Mix ingredients. Step 2: Dry in air at 220°C. Step 3: Crush and screen. Step 4: Carbonize and activate in a rotary furnace under a flow of carbon dioxide (or steam) at 800-950°C for different lengths of time.

During heating the carbohydrate decomposes to make a char. The nitrogen or other heteroatom containing compounds (if present) and carbon black are homogeneously dispersed throughout the char, preventing agglomeration of the particles. During carbonization, the char decomposes to give a high purity carbon. To activate the carbons, rotary kiln is used and he final product ball- milled and screened to minus 325 mesh.

Table 2 shows formulations that were made using dextrose (glucose) and sucrose as the sugar source and two carbon blacks (purchased from Columbian). Raven® 410 is an industrial grade carbon black with a mean particles size of 101 nm and a BET surface area of 26 m²/g. N330 is a commonly used carbon black for rubber with a particle size range of 28-36 nm and a BET surface area of 78-88 m²/g.

Table 2. Formulations of TDA carbons.

The mixtures were charred at 200-220°C in polytetrafluoroethylene polymer trays. Vegetable oil was added to form a paste and prevents foaming. The samples were devolatilized at 350-950°C under nitrogen and activated at 850-1000°C under flowing carbon dioxide. Using this approach results in high surface area carbons with higher conductivities and packing densities than are possible with conventional activated carbons or other physically mixed carbon black and activated carbon particle mixtures.

The general procedure for preparing these porous carbons was to: Step 1: Thoroughly mix ingredients in mixer or paint shaker. Step 2: Dry in air at 200°C. Step 3: Carbonize under flowing nitrogen at 350-950°C. Step 4: Activate in a rotary furnace under flowing carbon dioxide or steam at 850-1000°C for different lengths of time.

During heating the carbohydrate decomposes to make a char that coats the carbon black particles. The carbon black is homogeneously dispersed throughout the char, preventing agglomeration of the particles. During carbonization, the char decomposes to give a high purity carbon.

The specific surface area and the pore size distribution of the carbons were measured by nitrogen adsorption on a Micromeritics Gemini VII instrument using Density Functional Theory. These carbons have very large pore volumes which enables us to uniformly deposit 1-10 nm Pt nanoparticles throughout the material. Table 3 shows the pore size distribution in the 2-100 nm range of representative carbon samples. As shown in Figure 10, the carbons have most of their pore volume in the 2-100 nm range.

Table 3. Surface areas and pore volumes of the carbons.

Table 4 shows the elemental analysis of the carbons. TDA-7 and TDA-8 have the lowest nitrogen contents because no nitrogen compound was added to the initial formulation. TDA-20 and TDA- 23 have higher nitrogen content because ammonium bicarbonate was added to the initial formulation. These carbon formulations also have sulfur because the carbon black contains bound sulfur which helps attract the polysulfides to the carbon surface.

Table 4. Elemental analysis of the carbons.

To determine the electrical conductivity of the carbon powder themselves, the bulk conductivity of the porous carbons were measured as a function of pressure (Figure 11). The results are shown in Table 5 at a compression force of 500 kgf/cm² on the dry powder. Higher surface area carbons typically have lower density because they are more porous, but they do not pack into high density materials. The packing densities of the carbons of this invention are much higher than those of the commercial high surface area carbon blacks (BP2000). In an embodiment, the packing density is at least 0.8 g/cc and 500 kgf/cm². More preferably, the packing density is at least 1.0 g/cc and 500 kgf/cm².

Table 5. Characteristics of Exemplary Carbons of the Invention and a commercial carbon black.

Ex-Situ RDE Screening: After Pt deposition, the resulting electrocatalysts were characterized by TGA and XRD to determine the Pt concentration and nanoparticle size, respectively. Typical TGA data collected from a sample is shown in

Figure 5. The initial weight loss (~ 6 wt.%) at 100°C was due to weakly bond water absorption on the carbon surface. The next weight loss (~ 10 wt.%) at 100-300°C was probably from the acetate dissociation from residual platinum precursor. Beyond 300°C, the large decrease in weight (> 60 wt.%) was due to the major carbon oxidation. The remaining weight at the end was from thermally stable Pt metal. Therefore, this leftover residual (26.28 wt.%) was the concentration of Pt on the carbon support. The XRD pattern of this sample is shown in Figure 6. Peaks located at 40° and 46° result from Pt diffraction. By applying the Schemer equation, we estimated this particular electrocatalyst contains approximately 2.1 nm Pt particles. Several electrocatalysts were prepared on different carbon supports, at different concentrations, and with varied deposition methods. After similar TGA and XRD characterizations, all electrocatalysts were screened for HER intrinsic catalytic activity by the ex-situ RDE testing procedure.

Polarization curves (voltage vs. current) collected from the ex-situ RDE testing procedure were then plotted to determine their Tafel slopes at an early stage of the polarization. These Tafel slopes are very useful parameters to screen the HER catalytic activity of our catalysts because they are closely related to the electrochemical kinetics of the cathodic reaction. Figure 7 shows a Tafel fitting line of a polarization curve. The plot indicated a low Tafel slope of 42.1 mV/dec for this sample (VN133770A). This means the HER activation overpotential increased 42.1 mV for every 10x of current density increment (mA/cm²). Because lower cell voltage means better electrolyzer performance, we are looking for the lowest Tafel slopes. Several new catalysts were investigated Table 6 shows the estimated Tafel slopes of all tested electrocatalysts. The first row is the commercial catalyst (TKK, TEC10E50E 46.4 wt.% Pt/Ketjen). The second row is also from the commercial carbon support (Vulcan XC-72R), but it was deposited with Pt nanoparticles by TDA’s vapor phase dissociative procedure. The rest of the table includes all TDA electrocatalysts made from different supports and synthesis processes. The list is organized in the order of increasing Tafel slopes.

Table 6. Summary of electrocatalysts properties and ex-situ RDE results. All samples 0.1 mg/cm2 on the electrode.)

TKK has been known as one of the best Pt/C commercial electrocatalysts. The Tafel slope of 42.5 mV per decade was measured for the TKK catalyst, which indicates high HER kinetics for such low Pt loading (7.0 pg-Pt/cm²). The present invention teaches catalysts with even lower Tafel slopes than the TKK material. These preferred electrocatalysts are VN133769B (19.10 wt.% Pt/TDA-17, 38.0 mV/dec), VN133773B (32.56 wt.% Pt/TDA-7B, 38.6 mV/dec), VN133773A (32.7 wt.% Pt/TDA-17, 39.0 mV/dec), AR12841 (39.67 wt.% Pt/TDA-8, 39.1 mV/dec), AM1090B (32.7 wt.% PtsCo/TDA-17, 40.3 mV/dec), and VN133770A (26.28 wt.% Pt/TDA-8, 42.1 mV/dec). These electrocatalysts were used for in-situ testing described below.

We used an ultrasonic sprayer to improve the microstructure and distribution of the catalyst on the GDL, producing more reliable, consistent results. With the ultrasonic sprayer we sprayed our selected baseline catalyst (VN133770A) on MB30 GDL and used the electrode to optimize the MEA and hardware parameters.

MEA Optimization: We first optimized the temperature and pressure used to hot press an MEA. Since the glass transition temperature of the polytetrafluoroethylene membrane is approximately 130°C, we softened the polymer at around this temperature and pressed the MEA to adhere the gas diffusion electrode (GDE), the lrC>2 on platinized titanium screen, and the membrane. Too high of pressure would short-circuit the MEA, increasing its open circuit overpotential. On the other hand, too low of pressure would not form good contact between the electrodes and the membrane, reducing charge transfer at the interphase layers. Figure 8 shows that pressing with 5 Kg/cm² at 135°C for 3 min did not short-circuit the MEA and resulted in good adhesion between layers with low overall resistance.

We next optimized the electrodes by varying the amount of Ir0₂ and the thickness of the platinized titanium screen at the anode, the gas diffusion layer type, and the amount of ionomer in the catalyst layer at the cathode. Each of these parameters contributes to the total overpotential resistance of the electrolyzer. We believe that by optimizing these parameters, we can reproducibly fabricate high quality MEAs and obtain consistent testing procedures to evaluate and develop our catalyst. Figure 9 shows the results from testing several different MEAs to vary these parameters. In summary, we found that a loading of 3.33 mg-lrC>2/cm² was sufficient to electrocatalytically split the water at the anode. The platinized titanium screen backing layer with a thickness of 4 mil could form excellent contact on the anode side. Also, a hydrophobic gas diffusion layer helped to remove water more effectively at the cathode, resulting in increased hydrogen diffusion and lower mass transport resistance at the cathode. AvCarb MB30 with Teflon treatment gave the lowest overpotential over the entire polarization curve. Lastly, 33.3 wt.% ionomer content in the catalyst layer was the optimal amount to effectively conduct protons without entirely coating the carbon’s surface area and blocking the pores, subsequently increasing the electronic resistance and reducing the gas diffusion, respectively.

Hardware Optimization: We continued optimizing our electrolyzer testing by varying the flow pattern of liquid/gas at the electrodes. One flow pattern could have better liquid/gas supply or removal than another and therefore could improve mass transport of the reactants and products. In order to choose the right flow-field, we tested our best MEAs under serpentine and parallel flow. Figure 10 shows that the serpentine flow structure had similar or slightly better electrolysis performance than the parallel flow. We then tested different cell orientations using the serpentine flow-field. One test was oriented so that the MEA was standing vertically. Another orientation was to let the MEA lay flat with the anode horizontally sitting on top of the cathode. Interestingly, the horizontal orientation showed much lower mass transfer resistance than the vertical orientation at high current densities (Figure 10). We believe that this horizontal orientation assisted water diffusion at the electrodes due to the gravitational force. This setup also gave more headspace at the flow-fields to more effectively remove the gaseous oxygen and hydrogen products.

At this point we had optimized all electrolyzer parameters so that we could be confident that any differences in performance were due to the electrocatalysts and not to random variations in the MEA preparation or electrolyzer assembly. The parameters that were optimized are shown I Table 7. For all of the following tests, we used these same test parameters except the electrocatalyst at the cathode. The Pt loading on the cathode was always kept at 0.1 mg-Pt/cm².

Table 7. Electrolyzer test parameters optimized.

Pt deposition Method: After optimizing all electrolyzer test parameters, we looked for best procedure to deposit Pt on the carbon supports. We compared the in-situ vapor-phase dissociative method with other traditional solution-based reduction methods. The in-situ vapor-phase dissociative method doesn’t rely on solvents and therefore is easily scalable to high production volumes. Using this method, XRD fitting showed very small Pt nanoparticles (2.4 nm) deposited on the carbon support even when the Pt concentration was quite high (48.48 wt.% on TDA-8). In comparison, we deposited Pt on the same carbon support but used solution-based synthesis methods. For the first method in acetone, XRD analysis of the resulting catalyst showed a larger Pt particle size (3.0 nm) on TDA-8 even when the concentration of the deposited Pt was lower than that of TDA’s procedure (40.81 wt.%). For the second method in urea, XRD analysis again indicated a larger Pt particle size (3.3 nm) for a lower Pt concentration (39.67 wt.%) than TDA’s method. As shown in Figure 11 , TDA’s in-situ vapor-phase dissociative method with smaller Pt particles gave lower overpotentials at high current density (> 1.0 A/cm²) compared to the traditional solution-based methods. The results indicated that the catalysts prepared from solution reduction methods had more mass transport resistance. Therefore, we conclude that TDA’s vapor-phase dissociative method produces better catalysts and will be used in future catalyst preparations.

Effect of Cobalt Addition: In addition to monometallic Pt catalysts, a Pt₃Co/C catalyst was also tested to investigate the potential benefit of alloying that has been observed in fuel cell oxygen reduction reaction (ORR) catalysts. Two catalyst samples (32.56 wt.% Pt/TDA-7 and 32.7 wt.% Pt/TDA-17) were synthesized and tested as baselines. These catalysts were then modified with Co to produce a targeted composition of PhCo/C. As

Figure 12 indicates, there was no enhancement in performance observed in the polarization curves after alloying with Co. In fact, the incorporation of Co resulted in higher resistance, possibly due to blocking the active Pt sites for HER.

Effect of Pt Particle Size: To test the influence of Pt particle size on electrolyzer performance, we prepared a number of electrocatalysts using in-situ vapor-phase dissociative method on a variety of carbon supports. We purposely selected carbons with different surface area, micropore volume, and mesopore distribution in order to form different Pt particle sizes. As shown in Figure 13, we found that smaller Pt particles resulted in improved catalyst performance because smaller Pt particles increased effective surface area of the catalyst. Therefore, the HER was faster, resulting in less resistance throughout the polarization curve.

Effect of Pt Concentration on the Carbon Support: Different Pt concentrations on the carbon support result in different catalyst layer thicknesses on the GDL when the catalyst concentration on the electrode is maintained at 0.1 mg-Pt/cm². Depending on its thickness, the catalyst layer can affect proton and hydrogen gas diffusion. In theory, a thinner catalyst layer formed from a highly concentrated Pt catalyst will make the protons conduct faster to the catalytic sites via shorter traveling paths. It also will be more effective at removing the generated hydrogen from the cathode side, which lowers the mass transport resistance. However, in prior art, high concentrations of Pt nanoparticles tend to agglomerate, forming larger Pt particles, and therefore, lowering the catalyst’s effective surface area. To investigate the effect of Pt concentration on a support, we prepared electrocatalysts with different Pt loadings on TDA-7 carbon support ranging from ~20 to 50 wt.% Pt. The Pt loading on the electrode was maintained at 0.1 mg-Pt/cm², resulting in different thicknesses of the catalyst layers on the GDLs. Higher Pt concentrations on the support produce thinner electrode coatings. As shown in Figure 14, the Pt particle sizes of the prepared catalysts increased with concentration. We found that the best Pt loading on the support was about 30 wt.%. This is surprising that such a high loading can be accomplished while still keep the platinum particle size from 1-10 nm.

Effects of Pore Volume at 4-7 nm: The present invention teaches the surprising relationship between the electrolyzer performance and the carbon support surface properties. To investigate this relationship, we produced a series of TDA-8 carbons at different thermal activation time. The TDA-8 carbon batches formed a variety of pore volumes at 4-7 nm, ranging from low 0.0027 cm³/g to high 0.0376 cm³/g. After Pt deposition (~ 30 wt%) using TDA’s in-situ vapor-phase dissociative process, the resulting electrocatalysts were then sprayed and tested in a 5.28 cm² active area electrolyzer. All tests were done using the standard testing procedure. The polarization curves can be found in Figure 15. Without wishing to be bound by theory, the overall cell voltage is lower with higher pore volume in the 4-7 nm range. From XRD fitting, the Pt nanoparticles deposited on these carbon supports are about 2.1-2.5 nm. These particles that are entrapped in 4-7 nm pores (double or triple their particle size) are better stabilized and therefore suffered less from agglomeration. In addition, 4-7 nm pores are large enough to remove the migrated water during operation of an electrolyzer, preventing flooding and blocking of the Pt sites. At the same time, the hydrogen produced by the catalyst can also diffuse out, lowering the concentration overpotential. As shown in Figure 16, the cell voltage (at 1 A/cm²) is inversely proportional to the pore volume at 4-7 nm. In summary, the cell overpotential can be reduced by almost 200 mV with high pore volume at 4-7 nm (> 0.0199 cm³/g).

Preferred electrocatalyst candidates: We have described (above) several catalysts that have high intrinsic catalytic activity with low Tafel slopes compared to the commercial TKK catalyst using ex-situ RDE testing procedures. After in-situ electrolyzer screening tests, we found the four best catalysts that showed excellent electrolysis performance compared to the others and these are 26.28 wt.% Pt/TDA-8 (1508 m²/g, 2.1 nm Pt), 28.65 wt.% Pt/TDA-23 (1287 m²/g, 2.1 nm Pt), 29.46 wt.% Pt/TDA-7 (800 m²/g, 3.0 nm Pt), and 32.7 wt.% Pt/TDA-17 (1379 m²/g, 2.2 nm Pt). We then compared the performance of these catalysts with another catalyst prepared using the same in-situ vapor-phase dissociative method but using conventional Vulcan XC-72R (254 m²/g, 3.1 nm Pt), which is currently the most widely used catalyst support. Due to its much lower surface area, the Pt particle size formed on Vulcan XC-72R was larger (3.1 nm) than for catalysts synthesized on the carbons of the present invention. Figure 17 shows that the catalysts on the carbon of the present invention outperformed 29.24 wt.% Pt/XC-72R. This result indicates that the outstanding performance of the electrocatalysts was due to the unique carbon support and not the platinum deposition procedure. At this low platinum loading (0.1 mg-Pt/cm²) on the electrode surface, the Vulcan material cannot compete.

Electrocatalyst Durability and Regeneration: To test the durability of our catalyst, we prepared an MEA (membrane electrolyte assembly) and tested the catalyst in-situ using a lab scale electrolyzer with an active area of 5.28 cm². After break-in conditioning, the electrolyzer was held at 1.0 A/cm² for 10 hours. Figure 18 shows the recorded voltage slightly increased over time. It started at about 1.85 V and finished at about 1.95 V after 10 hours of testing. Without wishing to be bound by theory, this increasing of overpotential was likely due to flooding at the cathode. Water migrated from the anode to the cathode and built up over time. If the water was not effectively removed, it would flood the cathode and prevent the hydrogen product from diffusing off of the catalyst sites, increasing the transport resistance. To investigate this, we removed the MEA from the electrolyzer cell and regenerated the catalyst by heating at 110°C for 1 hour. After this thermal treatment, the MEA was tested again. As indicated, the electrolyzer performance was recovered at about 1.85 V, the same voltage it started at.

Example 1 , Carbon Preparation and Characterization: We prepared different carbon supports by varying the formulation of our carbon sources (sucrose and dextrose). Carbon black was also included in these formulations to increase the electrical conductivity. The mixture of ingredients often contained carbohydrate, carbon black, ammonium bicarbonate and vegetable oil (Table 8). After heating the mixture in nitrogen at 220°C and crushing the forming solid into granules, it was carbonized into chars in a furnace under a flow of nitrogen (600-1000°C). Following carbonization, the chars were activated under carbon dioxide at 950°C to increase their surface area. After cooling to room temperature, the carbon was ground, screened (-325 mesh size), and stored for further analyses. To obtain certain pore size distribution and surface area, the carbonization and activation time, as well as the nitrogen and carbon dioxide flow rates, were adjusted appropriately. After each preparation, the specific surface area and the pore size distribution of the carbons were measured by nitrogen adsorption on our Micromeritics ASAP 2020 instrument using Density Functional Theory. Carbon supports that showed interesting surface properties were down selected for Pt deposition in the next step. Table 8. Formulations of TDA carbons.

Example 2, Pt Deposition and Characterizations: Activated carbons were loaded with Pt using vapor deposition method, which is by the thermal decomposition of platinum(ll) acetylacetonate. In this method, Pt nanoparticles were deposited on carbons by an in-situ vapor-phase dissociative process. First, Pt(acac)2 precursor was mixed with our activated carbon in an appropriate proportion for at least 15 min. Next, the dry mixture was heat-treated at 250°C for 4 hours in air. During the heat treatment the carbons reduced Pt²⁺ to Pt° and captured the metal nanoparticles inside their pores. The catalyst powders were then stored in a sealed container for analyses and usage. The structures of the electrocatalysts were characterized by X-ray diffraction (XRD) analysis using a Scintag Theta/Theta XDS2000 X-ray diffractometer with a Cu target X-ray tube. The Pt diffraction peaks were then fit and the crystallite size was calculated using the Scherrer equation to determine the particle size of the deposited Pt nanoparticles. In addition, the concentration of Pt on carbon supports was accurately measured by thermogravimetric analysis (TGA Q5000, TA Instruments Inc.). In this process, the samples were heated to 800°C under air at a ramp rate of 20°C/min to burn out all carbon supports. The residual mass (which is pure Pt) was then used to calculate the percentage of Pt on carbon.

Example 3: For comparison, we also deposited Pt on the same carbon support using solution- based synthesis methods. The first method used chloroplatinic acid hexahydrate as the Pt precursor in acetone. According to this procedure, the Pt solution was slowly impregnated on the carbon support, followed by drying at 60°C overnight and reducing under hydrogen at 300°C for 1 hour. For the second solution-based synthesis method, we also used chloroplatinic acid hexahydrate as the Pt precursor but dissolved it in an aqueous solution of urea instead. The aqueous urea/Pt solution was then added to the carbon support, followed by heating to 90°C to hydrolyze urea and subsequently adsorb Pt(IV) on the carbon. The Pt was then reduced in ethylene glycol at 120°C.

Example 4: In addition to monometallic Pt catalysts, a Pt₃Co/C catalyst was also tested to investigate the potential benefit of alloying that has been observed in fuel cell oxygen reduction reaction (ORR) catalysts. The catalyst was prepared in two steps. The first step was deposition of Pt on the activated carbon using TDA’s in-situ vapor-phase dissociative procedure to obtain Pt/C (~ 30 wt.% Pt). The resulting catalyst was then modified with Co to produce a targeted composition of PtsCo/C. During this second step, Pt/C powder was impregnated with an aqueous solution of cobalt chloride (0.4 M) with an atomic ratio of Co:Pt of 3 and was dried at 80°C for 24 h. Next, the dried catalyst powder was annealed at 900°C in 10 vol% H2/N2 for 5 h to form Pt-Co alloy nanoparticles. Finally, the annealed catalyst powder was leached in 1 M nitric acid at 80°C for 24 h to remove the excess Co, washed with copious deionized water, dried at 80°C for 24 h, and then ground manually in a mortar, forming the final dealloyed PtCo3/C.

Example 5, electrocatalysts Ex-Situ Rotary Disk Electrode (RDE) Measurement: The RDE measurements were performed in a three-electrode electrochemical cell using a Gamry Potentiostat Reference 600 Model at room temperature. An RDE (5.52 mm ID and 7.16 mm OD) with glassy carbon disk (5.0 mm diameter) was employed as the working electrode. A catalyst ink was prepared by blending the catalyst powder using aqueous suspensions containing a Nafion® ionomer solution in a 2:1 ratio, respectively. The ink was sonicated and vortexed for at least 1 hour. Before ink deposition, the glassy carbon electrode was polished with AI₂O₃ powder and dried. The catalyst ink was then deposited onto the glassy carbon disk at exactly 7.0 pg-Pt/cm². For this study, 0.5 M sulfuric acid (H₂SO₄) was used as a proton conducting electrolyte. A platinum wire and a saturated calomel (Hg2Cl2) electrode (SCE) were used as the counter and reference electrodes, respectively. Before each electrochemical measurement, the electrolyte was saturated with hydrogen by bubbling for at least 30 min. The applied potential was then stepped down from 0.08 V to -0.42 V vs. a reference hydrogen electrode (RHE) at a rotation rate of 1600 rpm. Polarization curves (voltage vs. current) were then plotted to determine the Tafel slopes. We used this method to screen for electrocatalysts with the lowest Tafel slopes for further testing in an electrolyzer. Example 6, in-situ electrolyzer testing: The selected electrocatalyst powders were first fabricated into MEAs. TDA’s electrocatalyst (for the cathode) and Nafion® ionomer (using D2020 solution) with a weight ratio of 2:1, respectively, were suspended in water and isopropanol (weight ratio of 5:3, respectively) to bring the catalyst concentration to approximately 1.0 wt.%. The ink was sonicated by probe sonicator for 10 s and bath sonicator for 30 min, and then vortexed for 2 min to obtain a homogenous solution. The ink was next sprayed onto MB30 GDL made by AvCarb at 80°C, using an ExactaCoat OP3 ultrasonic sprayer (Sono-Tek Inc.) equipped with a heating table. The GDL was weighed before and after the catalyst deposition to make sure that the loading of Pt was 0.1 mg/cm². Ir0₂ (Fuel Cell Store, for the anode) was combined with Nafion® ionomer (also using D2020 solution) to make a solution that was 5 wt.% Nafion® ionomer in Ir0₂, and suspended in water and isopropanol (weight ratio of 5:3, respectively) to bring the catalyst concentration to approximately 2.7 wt.%. The solution was then dispersed by sonicating for 30 minutes and vortexing for 2 min before use. The resulting suspension of Ir0₂ was sprayed directly onto the Nafion® 115 membrane at 80°C (loading about 3.33 mg-lr0₂/cm²), using a heating and vacuuming table. A platinized titanium screen (4 mil thick) was used as a backing layer for the Ir0₂. Nafion® 115 (8 cm²) was used as the solid electrolyte membrane, which was sandwiched between 5.28 cm² electrodes, placed in between 1/16” Gylon sheets, followed by hot pressing at 135°C at 5 kg/cm² for 3 min. Temperature, pressure, and time were optimized to produce a catalyst layer with sufficient integrity for further testing. PTFE-coated fiberglass fabric gaskets were used to seal the MEA on both anode and cathode sides (3 mil and 5 mil thick gaskets, respectively). The resulting MEA was finally placed inside electrolyzer hardware equipped with a serpentine flow field and was ready for testing.

To test the catalysts, we used a lab scale electrolyzer with an active area of 5.28 cm² (Fuel Cell Technology Inc. The electrolyzer was connected to our electrolyzer test station that can be run unattended to quickly collect data. The electrolyzer was operated at 80°C and ambient pressure. Water was supplied at the anode at 5.0 mL/min. The current was initially held at 1.0 A/cm² for 30 min for break-in conditioning of the MEA before being polarized from 0.2-1.8 A/cm². During polarization, each current was held for 5 min to record the voltage and rested for 5 s before stepping up to the next higher current. Four polarization curves were recorded and averaged to obtain the given polarization curve. The MEA was then tested for durability by holding the current at 1.0 A/cm² for 10 h. Finally, four polarization curves were collected again to study how the electrolyzer performance might be degraded after a 10 h hold. The best performing electrocatalyst with the lowest cell voltage at 1.0 A/cm² was selected for 50 g scale up. Example 7: electrocatalyst synthesis. We are using a small rotary kiln with a 4” quartz tube to scale- up the catalyst production and to ultimately produce the 50 g batch of electrocatalyst. To prepare 50 g batches, Pt(acac)₂ precursor will be mixed with TDA activated carbon in an appropriate proportion. The mixing procedure will be done in TDA’s Speedmixer (DAC 600FVZ) at 800 rpm for 30 s and 2000 rpm for 2 min, followed by hand mixing for 1 min. The mixing procedure will be then repeated another two times. After that, the sample will be carefully placed inside the glass tube and assembled in the rotary kiln. The air flowrate will be set very low at 0.034 seem. The temperature will be then increased from room temperature to 235°C over 2 hours. The heat soak will be maintained at 235°C for 2 hours, and then the sample will be cooled down naturally to room temperature. The catalyst powders will be then stored in a sealed container for analyses and usage. Pt nanoparticle size will be determined by XRD, and Pt concentration will be measured by TGA. The electrocatalyst performance will be finally confirmed with an in-situ electrolyzer testing.

Example 8: general method of producing larger amounts of electrocatalyst. The general procedure for scaling up the electrocatalysts was to: Step 1 : Thoroughly mix metal acac (20-50 wt. % metal in final catalyst) with carbon support in Speedmixer (DAC 600FVZ) at 800 rpm for 30 s and 2000 rpm for 2 min, followed by hand mixing for 1 min. Repeat mixing two more times. Step 4: Activate in a rotary furnace under flowing air (0.034 seem) at 250-350°C for 2-4 hours until the desired metal nanoparticle size was reached.

Example 9: In-Situ Electrolyzer Testing. Error! Reference source not found, shows test parameters that were used to evaluate the MEAs.

Claims

What is claimed is:

1. A porous carbon-platinum electrocatalyst, comprising: a carbon matrix-platinum composite, wherein the carbon matrix comprises a carbon support, wherein platinum is deposited into the pores of the carbon support, and the platinum is at least 2 weight% of the carbon matrix-platinum composite; wherein, the carbon support comprises: a. a nitrogen BET surface area of about 550 to 1800 m²/g; b. a packing density of at least 0.8 g/cc as determined at a compressive force of 500 kgf/cm² on dry carbon powder; and c. an electrical conductivity of at least 10 S/cm at a compressive force of 500 kgf/cm² on dry carbon powder.

2. The porous carbon-platinum electrocatalyst of claim 1 , wherein the platinum is at least 2 weight % and at most 50 weight% of the carbon matrix-platinum composite.

3. The porous carbon-platinum electrocatalyst of claim 1, wherein the carbon support has a carbon content of at least 94 weight %.

4. The porous carbon-platinum electrocatalyst of claim 1, wherein platinum is deposited as 1 to 10 nm particles on the carbon support.

5. The porous carbon-platinum electrocatalyst of claim 1 , wherein the carbon support further comprises: a nitrogen BET surface area of about 600 to 1000 m²/g.

6. The porous carbon-platinum electrocatalyst of claim 5, wherein the carbon support further comprises: a nitrogen BET surface area of about 700 to 900 m²/g.

7. The porous carbon-platinum electrocatalyst of claim 1 , wherein the carbon support further comprises: a nitrogen BET surface area of about 873 to 1620 m²/g.

8. The porous carbon-platinum electrocatalyst of claim 1, wherein, the carbon support has at least 0.01 cm³/g of 4-7 nm diameter pores.

9. The porous carbon-platinum electrocatalyst of claim 1 , wherein the packing density is at least 1.0 g/cc and 500 kgf/cm².

10. The porous carbon-platinum electrocatalyst of claim 1, wherein the carbon support has an electrical conductivity of a least 16 S/cm at a compressive force of 500 kgf/cm² on dry carbon powder.

The porous carbon-platinum electrocatalyst of claim 1 , wherein the platinum is at most 25 weight % and at most 35 weight% of the carbon matrix-platinum composite.

11. The porous carbon-platinum electrocatalyst of claim 1, wherein the platinum comprises a platinum alloy.

12. The porous carbon-platinum electrocatalyst of claim 11 , wherein the platinum comprises a platinum-cobalt alloy.

13. An electrolyzer, comprising: the porous carbon-platinum electrocatalyst of claim 1, wherein the electrolyzer has a Tafel slope of at most 40 mV / decade.

14. A porous carbon-platinum electrocatalyst made by the steps of: a. providing a plurality of carbon black particles, a carbohydrate in the form of a refined sugar, and an oil; b. coating the plurality of carbon black particles with the carbohydrate and the oil; c. carbonizing the carbohydrate, which is coated on the plurality of carbon black particles, under a nitrogen atmosphere; d. activating the carbohydrate, which is carbonized on the plurality of carbon black particles, under a steam or carbon dioxide atmosphere to form a carbon support; e. providing a platinum precursor; and f. using an in-situ vapor-phase dissociative method to deposit platinum nanoparticles in the pores of the carbon support