Aqueous Ammonia Wetting of Gas-Diffusion Media for Electrochemical Cells

GDM surface interactions with liquids have been investigated by numerous groups; liquid water is the non-wetting phase most commonly studied.²⁰ Measuring a sessile droplets static contact angle, θ_C, on a GDM surface can give insight as to the surfaces overall hydrophobicity; however, it has limitations due to deviation from Youngs equation caused by surface roughness, which can be better accounted for by the Wenzel angle θ_W approximation

$$\begin{eqnarray}&&\cos \,{\theta }_{{\rm{W}}}=\varepsilon \cos \,{\theta }_{{\rm{Y}}}\end{eqnarray} \tag{ 6 }$$

where ε is a surface roughness factor and θ_Y is the Young angle.²¹ Das et al. examined GDMs for proton-exchange-membrane fuel cells (PEMFC), describing the porous structure as not only rough but also chemically inhomogeneous due to non-uniform distribution of binding material and wet proofing (usually Teflon).²¹ Their work recommended using a sliding angle method to account for the total liquid/solid adhesion. A droplets sliding angle, θ_S, measures its propensity to detach from a surface. It is related to contact angle hysteresis, which is the difference between advancing and receding angles, θ_A − θ_R, used commonly as a qualitative method for determining droplet mobility. The sliding angle can be controlled using a tilting-stage goniometer which is discussed in the experimental section. Adhesion, f_a, can be defined through the force balance²⁵

$$\begin{eqnarray}&&{{\rm{f}}}_{{\rm{a}}}=\displaystyle \frac{\rho {\rm{V}}\ \sin \,{\theta }_{{\rm{s}}}}{\pi {{\rm{d}}}_{{\rm{w}}}}\end{eqnarray} \tag{ 7 }$$

where ρ is the liquid density, V is the liquid volume, and d_w is the wetted diameter. Measuring the sliding angle at the moment of droplet detachment, which is caused by gravity, allows for estimation of the maximum adhesion between the liquid and surface. In this case f_a is calculated on a force per length basis; the total adhesion force, F_a, can be calculated by F_a = f_aπd_w. For NH₃ systems, the liquid density of ammonia is a function of temperature and concentration; solution density, ρ_sol, is calculated using the methods in Appendix A.1. Figure 3 is a schematic showing how the geometry and angles are defined. This work will focus on room temperature conditions. Adhesion will be measured over a range of NH₃ sol concentrations on several GDM that are used in both DAFC as well as electrochemical ammonia synthesis cells. The effect of adding Teflon (PTFE) will also be assessed for PTFE loadings of 5, 10, 20, and 30 wt%.

Zoom In Zoom Out Reset image size

Breakthrough pressure, P_BT, also referred to as threshold pressure, is related to capillary pressure, P_C, and is determined by the pore structure and contact angle between a liquid and GDM. The Young-Laplace equation relates key variables

$$\begin{eqnarray}&&{{\rm{P}}}_{c}=-\displaystyle \frac{2\gamma \cos \,{\theta }_{{\rm{c}}}}{{\rm{r}}}\end{eqnarray} \tag{ 8 }$$

where r is the radius of a pore.²⁰ Liquid transport between CLs and reactant channels must travel through the GDM; therefore, there is a maximum capillary pressure, P_BT, that must be overcome. The breakthrough pressure method has been utilized by several groups to determine a GDMs resistance to liquid transport.^22,26 In this work a range of aqueous NH₃ concentrations are injected into commonly used GDM of varying thickness and PTFE loading while liquid pressure is recorded. The maximum pressure measured is considered to be the breakthrough pressure.

Aqueous NH₃ solutions were procured from the RICCA Chemical Company in nominal NH₃/H₂O weight ratios of 10, 20, and 30%, while pure deionized water was used for 0%. The highest wt% aqueous solution reported in literature was 35% NH₃.⁵ This range of concentrations is of interest since DAFCs produce liquid water as a by-product which dilutes NH₃, while H₂O cross over in ammonia synthesis systems also leads to NH₃ dilution. Key properties for each solution were calculated, along with pure NH₃ for reference, using methods discussed in Appendix and are presented in Table II for room temperature (∼20 °C) conditions.

Previous NH₃ literature was surveyed to determine what GDM materials have been utilized. Both carbon papers and cloths have been implemented in various combinations for anode and cathode electrodes.¹² In some cases, while paper vs cloth was specified, the material details were not available. Table III outlines the materials selected for testing along with key properties and prior use in NH₃ systems. GDM samples were purchased from Fuel Cell Store. Customized PTFE loadings were also done by Fuel Cell Store.

All contact angle related data as well as adhesion force measurements were completed using a Ramé-Hart tilting stage goniometer (Model 290) shown in Fig. 4a. GDM samples were fixed to a staging surface between the diffuse light source and digital camera. Automated recording of advancing & receding angles, sliding angles, wetting diameter, and volumes was accomplished with DROPImage software. Droplets were placed on the GDM surfaces using an adjustable micropipette. For the sliding angle measurements, the stage rotation rate was set to 1° s⁻¹ to reduce the effects of motion on the droplets. An image was taken at each degree until the droplet detached and fell from the surface. The angle at which this occurred was recorded as the maximum sliding angle and used to calculate adhesion force in Eq. 6. Examples of droplet behavior and advancing and receding angle data from this process are displayed in Figs. 5a and 5b. Initially, at sliding angle 0°, hysteresis is very low and differences between left and right angels can be attributed to the rough GDM surface which varies from point to point. As the sliding angle increases the droplet shifts forward causing the advancing angle to increase while the receding angle decreases. The sudden change in these angles at a sliding angle of around 10° are due to shifts in droplet pinning as the droplet wetting perimeter changes subtly to stabilize itself. Finally, at the incipient point, the droplet falls off and the maximum sliding angle is recorded, in this case it was 27°.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The breakthrough pressure, defined as the pressure at which the solution emerges from the surface of the porous layer, was measured with an experimental setup schematically shown in Fig. 4b. In this setup, the aqueous NH₃ was injected with a syringe pump to the bottom surface of the GDM sample at 300 μℓ h⁻¹ flow rate and through a 250-μm-diameter stainless-steel capillary (U-111, Upchurch). The solution passed through the GDM and emerged from its top side, forming a breakthrough on the surface of the GDM. Therefore, the breakthrough occurred as the solution emerged from the surface of the GDL, on the side opposite to the injection capillary. GDM samples, prepared in 2 cm × 2 cm cuts, were sandwiched between two 0.5-inch-thick polycarbonate plates which were tightened with four screws, one in each corner of the test section. A stainless-steel shim with a thickness of 0.05 mm was inserted around GDM samples to provide a uniform compression. The injection pressure was measured at 50 Hz sampling rate and with a pressure transducer with a pressure range up to 11.7 kPa ± 1.0% (Omega, PX163-120BD5V). On the top polycarbonate plate (the emergence side), a 6-mm-diameter through-all hole was machined to eliminate any effects the GDM compression might have on the breakthrough process. For each GDM sample, three injection trials were conducted and the mean value of the three trials were considered as the breakthrough pressure. Experiments were conducted at room temperature. Table IV lists flow conditions considered in this study. The superficial velocity in this table refers to the bulk velocity of the solution within the stainless-steel capillary and is obtained by u_f = q/πr², where u_f, q, and r are the superficial velocity, the volumetric flow rate, and the radius of the capillary, respectively.

Liquid surface tension is related to a substances polarity. Water molecules are strongly attracted to each other due to a large dipole resulting from the differences in oxygen and hydrogen electronegativity and unsymmetrical bond shape²⁷; water therefore exhibits a high surface tension. Ammonia, which is also polar, has a smaller dipole and a lower surface tension; therefore, increasing the concentration of NH₃ reduces the solution's overall surface tension. Appendix A.3 outlines a method for calculating ammonia water solution surface tension. The effect of NH₃ wt% on contact angle is displayed in Fig. 6. For each GDM type, droplets shifted toward less hydrophobic conditions at higher levels of NH₃ due to the solutions reduced surface tension. Lower surface tension means liquid molecules are less attracted to each other and, in this case, more likely to wet the GDM surface. The surface tension is also plotted for reference in Fig. 6a for room temperature, using the methods of Appendix A.3. The effect of increasing PTFE content was probed using the Toray series GDM. Figure 6 results for water are similar to trends observed in earlier work. The addition of PTFE loadings greater than 10 wt% has diminishing effect on hydrophobicity. Figure 6c is the same data plotted with respect to NH₃ wt%; overall, no clear differences between higher PTFE loadings and contact angle were observed. Generally, after 10 wt% PTFE wet-proofing the hydrophobicity does not increase significantly, therefore, higher PTFE content may not be necessary for electrode design in ammonia systems.

Zoom In Zoom Out Reset image size

Adhesion of 18.97 ± 0.86 μℓ droplets to the GDM surface were measured on SGL 29 BA and Toray 120 5% PTFE loading GDMs. Data could be collected only up to 10 wt% NH₃ as higher concentrations lead to droplets that failed to detach even at the maximum tilting angle of 90°. This is probably due to two factors: 1.) the reduced surface tension results in less hydrophobic conditions and 2.) the density of droplets at 30 wt% NH₃ is ∼15% less than that of water so the net gravity force acting on them is reduced. Overall, these result in higher adhesion and insufficient gravity force rendering this method inadequate for these elevated concentrations. Larger volumes were not attempted as droplets greater than 20 μℓ have not been verified to give reliable results.²¹ Figure 7 displays the lower NH₃ concentration adhesion data as well as total adhesion force calculated by multiplying by the wetting perimeter. Overall, the contact angle measurements show some agreement with nominal values of perimeter normalized adhesion force as the Toray 120 with 5 wt% PTFE, which had a lower contact angle than SGL, exhibits slightly higher adhesion. However, these values fall within the error tolerances and thus a correlation cannot be stated for certain. This fact highlights the usefulness of the sliding angle methodology which may capture adhesion behavior not accounted for by the contact angle measurements alone. Increasing NH₃ had a clearer effect as adhesion was elevated by ∼16% in both cases while total force rose ∼22% for SGL and ∼6% for Toray. The results for the water adhesion are in the range of values reported in literature.^20,21 These outcomes demonstrate that the addition of NH₃ can increase adhesion force which has been shown to increase the barrier for liquid removal.

Zoom In Zoom Out Reset image size

Figure 8 shows time series data from the liquid breakthrough process in Toray and Sigracet carbon papers. In all cases, pressure increases until a clear breakthrough point is observed. Pressure then falls quickly as fluid passes through the GDM forming a droplet on the surface. All runs conducted with carbon papers resulted in breakthrough while no breakthrough was obtained in runs with the ELAT sample. In Fig. 8a, higher NH₃ concentration resulted in a longer injection duration before breakthrough which may be attributed to enhanced pore wetting due to lower surface tension. The pressure increased more slowly at higher concentrations as the lower surface tension enabled the solution to percolate more readily into the GDM pores thereby relieving the buildup of reservoir pressure. Therefore, it is expected that more pores are filled during the breakthrough process for the lower surface tension fluid. Considering the surface tension and the capillary number of the 10 wt% solution as datum, the surface tension and the capillary number of the 20 and 30 wt% solutions can be compared to qualitatively characterize transport phenomena. The surface tension of 20 and 30 wt% solutions are 10.2% and 17.7% less than the surface tension of the 10 wt% solution. Likewise, the capillary number of the 20 and 30 wt% solutions are 18.1% and 27.9% greater than the capillary number of the 10 wt% solution. This can be translated as the transport phenomena for the higher percentages of 20 and 30 wt% is more toward the stable displacement flow regime on the drainage phase diagram.^23,28 Therefore, the percolation of the solution at higher concentrations (20 and 30 wt%) occurs relatively more uniform within the plane of the GDM, resulting in a relatively lower slope in Fig. 8a. While this explanation cannot be verified with the given setup, it may make an interesting candidate for X-ray based imaging studies.²⁹ Increasing GDM thickness also increased the duration of time before breakthrough, however, for the same concentration it did not affect the pressure increase rate significantly. Thicker samples (of the same series) lead to higher breakthrough pressures as shown in Fig. 9. PTFE content had minimal effect on breakthrough pressure, meaning that a 10% PTFE loading can be implemented without restricting flow through the GDM. Finally, in Fig. 9c, increasing NH₃ concentration in all cases lead to lower breakthrough pressure. Lower pressure can be attributed to both the lower surface tension, predicted in A3, as well as the lower measured contact angle. Overall, and based on the previous section's results, the Toray 30 series carbon paper with 10% PTFE may provide the highest surface hydrophobicity with minimal through plane transport resistance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In DAFC and NH₃ electrochemical synthesis systems, bulk transport of reactants and removal of liquid byproducts is accomplished using pressure driven convective flow in channels. Decreasing contact angle and increasing adhesion force can have significant impact on liquid management and should be considered in future work. For example, several studies have analyzed droplet and gas bubble detachment behavior and highlighted that higher adhesion requires higher flow velocity for removal which also means more parasitic pumping losses.²⁰ These results mean that regions where lower surface tension are present may require specific flow management strategies such as elevated flow rates. Increased adhesion and lower density also reduce the usefulness of gravity driven flow. All these items should also be considered for Air/NH₃ Sol. scenarios where gas bubbles rather than droplets are present. Another tool for electrode design currently being developed is coupled continuum and pore network models which link CL with GDM and channel physics.³⁰ The results uncovered in this work are useful to these pore network model inputs and demonstrate that liquid transport along an electrode may vary due to surface tension gradients. Finally, the effect of temperature on these properties may be of interest in future work as it impacts surface tension and viscosity.

The density of the aqueous NH₃ solutions, ρ_sol, plotted in Fig. A·1, were estimated using a model developed by Conde³¹

$$\begin{eqnarray}&&{\rho }_{\mathrm{sol}}=x{\rho }_{{\mathrm{NH}}_{3},{{\rm{T}}}_{{\mathrm{NH}}_{3}}^{* }}+(1-{\rm{x}}){\rho }_{{{\rm{H}}}_{2}{\rm{O}},{{\rm{T}}}_{{{\rm{H}}}_{2}{\rm{O}}}^{* }}+{\rm{\Delta }}{\rho }_{{{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}}\end{eqnarray} \tag{ A·1·1 }$$

where x is the wt% of NH₃, ${\rho }_{{\mathrm{NH}}_{3},{{\rm{T}}}_{{\mathrm{NH}}_{3}}^{* }}$ and ${\rho }_{{{\rm{H}}}_{2}{\rm{O}},{{\rm{T}}}_{{{\rm{H}}}_{2}{\rm{O}}}^{* }}$ are the pure substance densities as a function of temperature, and ${\rm{\Delta }}\ {\rho }_{{{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}}$ is the excess density approximated by

$$\begin{eqnarray}&&{\rm{\Delta }}\ {\rho }_{{{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}}=[{\rm{x}}(1-{\rm{x}})-{\mathrm{Ax}}^{2}(1-{\rm{x}})]{\rho }_{{\mathrm{NH}}_{3},{{\rm{T}}}_{{\mathrm{NH}}_{3}}^{* }}^{0.5}{\rho }_{{{\rm{H}}}_{2{\rm{O}},{{\rm{T}}}_{{{\rm{H}}}_{2}{\rm{O}}}^{* }}^{0.5}}\end{eqnarray} \tag{ A·1·2 }$$

where A is a parameter and a function of normalized solution temperature, ${{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}$, made dimensionless by the critical temperature of water, ${{\rm{T}}}_{{\rm{c}},{{\rm{H}}}_{2}{\rm{O}}}$ = 647.14 K

$$\begin{eqnarray}&&{{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}=\displaystyle \frac{{T}_{\mathrm{sol}}}{{T}_{{\rm{c}},{{\rm{H}}}_{2}{\rm{O}}}}\end{eqnarray} \tag{ A·1·3 }$$

Parameter A is defined as

$$\begin{eqnarray}&&{\rm{A}}=\sum _{{\rm{i}}=0}^{2}{{\rm{A}}}_{1,{\rm{i}}}{{{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}}^{{\rm{i}}}+\displaystyle \frac{\sum _{{\rm{i}}=0}^{2}{{\rm{A}}}_{2,{\rm{i}}}{{{\rm{T}}}_{\mathrm{sol}}^{\unicode{x02021}}}^{{\rm{i}}}}{{\rm{x}}}\end{eqnarray} \tag{ A·1·4 }$$

where A₁ and A₂ are found in Table A·I. The density of each pure substance, L, ρ_L = ${\rho }_{{\mathrm{NH}}_{3},{{\rm{T}}}_{{\mathrm{NH}}_{3}}^{* }}$ or ${\rho }_{{{\rm{H}}}_{2}{\rm{O}},{{\rm{T}}}_{{{\rm{H}}}_{2}{\rm{O}}}^{* }}$ was calculated using

$$\begin{eqnarray}&&\displaystyle \frac{{\rho }_{{\rm{L}}}}{{\rho }_{{\rm{c}},{\rm{L}}}}=\sum _{{\rm{i}}=0}^{6}{A}_{{\rm{i}}}{\tau }^{{{\rm{b}}}_{{\rm{i}}}}\end{eqnarray} \tag{ A·1·5 }$$

where A and b are parameters defined for each component in Table A·II, ρ_c,L is the critical density for each component (${\rho }_{{\rm{c}},{\mathrm{NH}}_{3}}=225\,\mathrm{kg}\,{{\rm{m}}}^{-3}$, ${\rho }_{{\rm{c}},{{\rm{H}}}_{2}{\rm{O}}}=322\,\mathrm{kg}\,{{\rm{m}}}^{-3}$) , and τ is calculated by

$$\begin{eqnarray}&&\tau =1-\displaystyle \frac{{{\rm{T}}}_{L}^{* }}{{{\rm{T}}}_{{\rm{c}},{\rm{L}}}}\end{eqnarray} \tag{ A·1·6 }$$

where the critical temperature for ammonia is ${{\rm{T}}}_{{\rm{c}},{\mathrm{NH}}_{3}}$ = 405.4 K.

Zoom In Zoom Out Reset image size

Viscosity for the NH₃ solutions, η_sol, was calculated using the Grunberg-Nissan constituent model

$$\begin{eqnarray}&&\mathrm{ln}{\eta }_{\mathrm{sol}}=\sum _{{\rm{i}}=1}^{{\rm{i}}={\rm{C}}}{x}_{{\rm{i}}}\mathrm{ln}{\eta }_{{\rm{i}}}+\displaystyle \frac{1}{2}\sum _{{\rm{i}}=1}^{{\rm{i}}={\rm{C}}}\sum _{{\rm{i}}=1}^{{\rm{i}}={\rm{C}}}{{\rm{x}}}_{1}{{\rm{x}}}_{{\rm{i}}}{{\rm{G}}}_{\mathrm{ij}}\end{eqnarray} \tag{ A·2·1 }$$

where, G_ij, is a binary interaction parameter.³² This model is dependent upon temperature and concentration, values over a wide range are plotted in Fig. A·2. For binary mixtures such as in the case of this work

$$\begin{eqnarray}&&\mathrm{ln}{\eta }_{\mathrm{sol}}=x\mathrm{ln}{\eta }_{{\mathrm{NH}}_{3}}+(1-x)\mathrm{ln}{\eta }_{{{\rm{H}}}_{2}{\rm{O}}}+x(1-x){G}_{\mathrm{ij}}\end{eqnarray} \tag{ A·2·2 }$$

where ${\eta }_{{\mathrm{NH}}_{3}}$ and ${\eta }_{{{\rm{H}}}_{2}{\rm{O}}}$ are the pure substance viscosities of ammonia and water as a function of temperature. Figueiredo et al. estimated G_ij using a quadratic function

$$\begin{eqnarray}&&{{\rm{G}}}_{\mathrm{ij}}=112.35\times {10}^{-6}{{\rm{T}}}^{2}-933.24\times {10}^{-4}{\rm{T}}+21.459\end{eqnarray} \tag{ A·2·3 }$$

which is valid for a temperature range of 283 K to 422 K.³³ The pure liquid viscosities were calculated using correlations from Fernandes et al.³²

$$\begin{eqnarray}&&{\eta }_{{\mathrm{NH}}_{3}}={10}^{-6}{{\rm{e}}}^{(6.278-1731/{\rm{T}}+5.616\times {10}^{5}/{{\rm{T}}}^{2})}\end{eqnarray} \tag{ A·2·4 }$$

$$\begin{eqnarray}&&{\eta }_{{{\rm{H}}}_{2}{\rm{O}}}={{\rm{e}}}^{(-0.318+2214/{\rm{T}}-1.9839\times {10}^{5}/{{\rm{T}}}^{2})}\end{eqnarray} \tag{ A·2·5 }$$

Zoom In Zoom Out Reset image size

Surface tension, γ, plotted in Fig. A·3, was calculated using a model developed by Conde³¹

$$\begin{eqnarray}&&{\gamma }_{\mathrm{sol}}=x{\gamma }_{{\mathrm{NH}}_{3},{{\rm{T}}}_{{\mathrm{NH}}_{3}}^{* }}+(1-x){\gamma }_{{{\rm{H}}}_{2}{\rm{O}},{{\rm{T}}}_{{{\rm{H}}}_{2}{\rm{O}}}^{* }}+{\rm{\Delta }}{\gamma }_{{T}_{\mathrm{sol}},x}\end{eqnarray} \tag{ A·3·1 }$$

where ${\rm{\Delta }}\ {\gamma }_{{{\rm{T}}}_{\mathrm{sol}},{\rm{x}}}$, is defined as

$$\begin{eqnarray}&&{\rm{\Delta }}\ {\gamma }_{{T}_{\mathrm{sol}}},{\rm{x}}=-({\gamma }_{{{\rm{H}}}_{2}{\rm{O}},{{\rm{T}}}_{{{\rm{H}}}_{2}{\rm{O}}}^{* }}-{\gamma }_{{{NH}}_{3},{T}_{{\mathrm{NH}}_{3}}^{* }})F(x)\end{eqnarray} \tag{ A·3·2 }$$

where F(x) is calculated using

$$\begin{eqnarray}&&{\rm{F}}({\rm{x}})=1.442(1-{\rm{x}})[1-{{\rm{e}}}^{-2.5{{\rm{x}}}^{4}}]+1.106[1-{{\rm{e}}}^{-2.5{\left(1-{\rm{x}}\right)}^{6}}]\end{eqnarray} \tag{ A·3·3 }$$

The surface tension for each pure substance as a function of temperature is calculated by

$$\begin{eqnarray}&&{\gamma }_{{\rm{L}}}={\gamma }_{{\rm{o}},{\rm{L}}}(1+{\rm{b}}\tau ){\tau }^{\sigma }\end{eqnarray} \tag{ A·3·4 }$$

where the parameters γ_o,L, σ, and b are contained in Table A·III for each component.

Zoom In Zoom Out Reset image size