Thermodynamic Analysis of Catalyst Stability in ... - ACS Publications

Jun 7, 2018 - revised Helgeson−Kirkham−Flowers thermodynamic equa- ... by constructing oxygen fugacity−pH diagrams, which illustrate material ph...
1 downloads 0 Views 6MB Size
Subscriber access provided by Kaohsiung Medical University

Kinetics, Catalysis, and Reaction Engineering

Thermodynamic Analysis of Catalyst Stability in Hydrothermal Reaction Media Jennifer N. Jocz, Phillip E. Savage, and Levi T. Thompson Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.8b01334 • Publication Date (Web): 07 Jun 2018 Downloaded from http://pubs.acs.org on June 7, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Thermodynamic Analysis of Catalyst Stability in Hydrothermal Reaction Media Jennifer N. Jocz,† Phillip E. Savage,†,‡ and Levi T. Thompson∗,† †Department of Chemical Engineering, University of Michigan, Ann Arbor ‡Department of Chemical Engineering, The Pennsylvania State University, University Park E-mail: [email protected] Phone: +1 (734) 936-2015

Abstract Hydrothermal solutions are important media for the conversion of biomass-derived species to useful chemicals and the destruction of environmental pollutants. These solutions are aggressive and can degrade heterogeneous catalysts. This article describes a framework for understanding the hydrothermal stability of heterogeneous catalyst materials with respect to oxidation and dissolution. We applied the revised Helgeson-Kirkham-Flowers thermodynamic equation of state to determine the oxidation states and solubilities of metals and oxides in water at 150-550 ◦ C and 22-50 MPa. Design criteria for catalyst compositions were determined through correlations between metal solubility and electronegativity and between oxide solubility and cation electronegativity, ionic-covalent parameter, and polarizing power. Design criteria for aqueous solution compositions were determined by constructing oxygen fugacity-pH diagrams, which illustrate material phase changes in response to changes in pH and the oxidative or reductive strength of the solution. Combined, these criteria facilitate design of stable catalytic materials for hydrothermal reactions.

1

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Introduction Reactions in sub- and supercritical water (SCW) offer a potentially “greener" approach for numerous processes including biomass conversion, chemical synthesis, and oxidation of pollutants, 1–7 because water is non-toxic, inexpensive, and environmentally benign compared to organic solvent alternatives. The dielectric constant of subcritical (200 ◦ C

< T < 374 ◦ C) water is lower than that of water at ambient conditions (see Figure 1) and there are fewer and less-persistent hydrogen bonds, resulting in an increased solubility of organic compounds. In addition, the ion product (KW ) at these conditions is several orders of magnitude higher than that for ambient water. At supercritical conditions (≥374 ◦ C,

≥22.1 MPa), the dielectric constant is even lower than that of subcritical water and many permanent gases and most organic compounds are soluble. The elimination of interphase transport resistances allows for a single homogeneous fluid phase at reaction conditions. As a result of these changes in solvent properties, hot compressed water can strongly influence the rates and selectivities of ionic, polar non-ionic, and free-radical reactions. 8 While rates for homogeneous hydrothermal reactions can be high, techno-economic analysis suggest that improvements via the use of catalysts will be required for greater industrial adoption. 5,11,12 Many heterogeneous catalysts, however, are not stable in hot water. In addition to deactivation mechanisms commonly observed in gas-phase reactions such as poisoning, coking, and thermally induced solid-state transformations, physical and chemical changes similar to the corrosion of reactor walls and tubing can occur. 13–16 Activity loss can also result from changes in oxidation state, migration and leaching of active metals, and structural changes such as sintering and loss of surface area induced by the hydrothermal reaction environment. 5,17,18 However, it is unclear how catalyst composition influences oxidation and dissolution in sub- and supercritical water and which material properties are responsible for controlling hydrothermal stability. In this article, we use the revised Helgeson-Kirkham-Flowers equation of state (RHKF) 19 to elucidate the thermodynamic oxidation and dissolution of common catalytic 2

ACS Paragon Plus Environment

Page 2 of 25

Page 3 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

0

0.2

ρ (g/mL) 0.4 0.6

0.8

1.0

-15 -20 -25 -30

log10 KW (mol/kg H2 O)2

-35 -40 -45

80 60 40 ε 20

Figure 1: H2 O density (ρ), ion product (KW ), and dielectric constant (e) as a function of temperature and pressure. Density values were taken from the Steam Tables, KW was calculated using the Marshall and Franck correlation 9 and e was calculated using Johnson and Norton equations. 10 CP denotes the critical point of H2 O. transition metals and oxides (Au, Co, Cu, Ni, Pd, Pt, Ru, CeO2 , MoO3 , TiO2 , WO3 , ZrO2 ) across a large range of sub- and supercritical conditions relevant for hydrothermal organic chemistry (150-550 ◦ C and 22-50 MPa) and correlate key material properties with hydrothermal stability. The equilibrium calculations reported herein predict the most severe outcome for irreversible catalyst deactivation by indicating the maximum possible catalyst dissolution in the system. Furthermore, these methods for analyzing catalyst solubility could be used to determine the relative contributions of heterogeneous catalysis and homogeneous catalysis from dissolved metal ions. We also show the power of thermodynamic modeling for the selection of catalytic materials, solution composition, and process conditions that minimize changes in catalyst oxidation state and dissolution of metal into the solution.

3

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 25

Methods Oxidation and dissolution equilibrium calculations The equilibrium oxidation states and aqueous metal concentrations of various materials in hydrothermal solutions from 150-550 ◦ C and 22-50 MPa were calculated from the equilibrium constants (Keq ) of the redox and dissolution reactions, respectively. The redox reactions for each material follow the form in equation (1) where the forward reaction is oxidation of a metal M(s) by O2 and the reverse reaction is reduction of the corresponding oxide MOx(s) . For these reactions, O2 comes from the water splitting reaction in equation (2). These reactions combined are thermodynamically equivalent to metal oxidation by H2 O and are not necessarily intended to represent the actual oxidation mechanism. x M(s) + O2 ↔ MOx(s) 2

(1)

1 H2O ↔ O2 + H2 2

(2)

Dissolution involves reactions of the solids with H2 O, H+ , or OH- to form aqueous inorganic species. The specific redox and dissolution reactions considered in the hydrothermal stability modeling are listed in Tables S6-S16. Keq values were calculated using equation (3) where R is the universal gas constant and ∆Grxn ( T, P)i is the molar Gibbs free energy of reaction for reaction i at temperature T and pressure P. ∆Grxn ( T, P)i values were calculated using equation (4) where νj is the stoichiometric coefficient of species j in the reaction (positive for products and negative for reactants) and ∆G f ( T, P) j is the apparent standard partial molar Gibbs free energy of formation of species j at T and P.  Keq ( T, P)i = exp

4

−∆Grxn ( T, P)i RT

ACS Paragon Plus Environment

 (3)

Page 5 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

∆Grxn ( T, P)i =

∑ νj ∆G f (T, P) j

(4)

j

The details for calculating ∆G f ( T, P) j at elevated temperatures and pressures using either the R-HKF equation of state 19 for aqueous species or the differential expression for molar Gibbs free energy for solid and gaseous species are discussed in the Supporting Information. Tables S1-S4 contain the thermodynamic properties necessary for the calculations. The methods for calculating the concentrations of dissolved catalyst species have been described previously 20 and are summarized in the Supporting Information.

Construction of oxygen fugacity-pH diagrams The O2 fugacity ( fO2 )-pH diagrams show where different species are thermodynamically favored relative to the oxidizing or reducing character of a solution ( fO2 ) and its acidic or basic character (pH). A single fO2 -pH diagram corresponds to a single temperature and pressure. The boundaries between solid species define the fO2 conditions for the equilibrium oxidation/reduction. The boundaries between solid and aqueous species define the fO2 and pH conditions for the presence of the aqueous species at some arbitrary fixed activity. These boundaries are expressed by equation (5) where νH + and νO2 are the stoichiometric coefficients of H+ and O2 in the reaction, respectively, Keq is the equilibrium constant for the oxidation or dissolution reaction, and a j is the thermodynamic activity for each species j in the reaction (not O2 or H+ ). Gas activities are expressed as fugacities ( f j , bar). From the standard state convention of the R-HKF equation of state, 21 aqueous species have unit activity in a hypothetical one molal solution referenced to infinite dilution and activities of solid phases and H2 O are taken to be unity at any pressure and temperature. If the reaction did not contain O2 or H2 , equation (5) was solved for pH, instead.  log10 fO2 =

νH + νO2



 pH −

5

1 νO2





log10 

ACS Paragon Plus Environment

∏ aj



Keq



νj

(5)

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 25

Equation (5) is derived by taking the log10 of equation (6). If the equation included aOH − , it was first converted to an a H + term by a substitution of equation (7). Then, the a H + terms were converted to pH using equation (8).

Keq ( T, P)i =

∏ aj

νj

(6)

j

aOH − =

KW a H+

pH = −log10 ( a H + )

(7) (8)

In this work, activities of 10−6 or 10−8 were used for aqueous boundary construction (corresponding to concentrations ≈ 1 or 0.01 µmol/kg H2 O). While any threshold activity can be chosen, we selected 10−6 as the threshold below which the material could be considered stable because material loss would be negligible for the timescales of lab-scale reactions. For several materials, no dissolved species were present in the diagrams at activities of 10−6 , so the activity threshold was reduced to 10−8 . When a j are constant (a j = C), equation (5) is a linear relationship between log10 fO2 and pH and corresponds to an equilibrium boundary on the diagram. On one side of the boundary is a solid species and on the other side is either another solid species or an aqueous species with a j ≥ C. The fO2 -pH boundary equations for all the materials that were modeled are listed in Tables S6-S15. Each fO2 -pH diagram also shows, for those specific conditions, the fO2 and pH of pure H2 O and where f H2 equals one bar. The pH and fO2 of pure H2 O at T and P were determined using the ion product (KW ( T, P)) and the equilibrium constant for the watersplitting reaction in equation (2) (KWS ( T, P)). From stoichiometry of the water-splitting reaction (νH2 = 2νO2 ) and the ideal gas assumption ( f i = Pi ), f H2 = 2 fO2 (in bars) and equation (9) is obtained. Under these conditions, errors from the ideal gas assumption are considered negligible and are discussed in the Supporting Information. Equation (9) can

6

ACS Paragon Plus Environment

Page 7 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

be solved for log10 fO2 of pure H2 O as shown in equation (10). 1

3

KWS = ( f H2 )( fO2 ) 2 = 2( fO2 ) 2 2 = log10 3

log10 fO2



KWS ( T, P) 2

(9)

 (10)

The value of log10 fO2 when f H2 equals one bar was determined by equation (11), which is obtained by substituting f H2 = 1 into equation (9). log10 fO2 |( f H

2

=1bar )

= 2log10 (KWS ( T, P))

(11)

Results and discussion Total solubility in pure water Figure 2 shows the total equilibrium metal concentrations (sum of all aqueous metalcontaining species) for several transition metals and oxides in pure water at 150-550 ◦ C and 22-50 MPa calculated using the R-HKF equation of state. 19 Solubilities of 0.01, 10−6 , and 10−20 mol/kg H2 O correspond to shades of red, yellow, and blue, respectively, in the various figures. In general, the total aqueous metal concentration decreases when moving down a periodic column (e.g., Ni, Pd, Pt) and as the oxidation state of a metal increases (e.g., Co, CoO, Co3 O4 ). The model predicts that Co, CoO, Ni, MoO3 , and WO2 have the highest solubilities across all hydrothermal conditions modeled in this work. Dissolution ≥ 1 µmol/kg H2 O is also predicted for Cu, CuO, NiO and WO3 in subcritical water and for Co3 O4 and MoO2 in supercritical water. The remaining materials in Figure 2 are predicted to have equilibrium aqueous metal concentrations well below 1 µmol/kg H2 O.

7

ACS Paragon Plus Environment

ACS Paragon Plus Environment

8 ����������� (°�)

-20

-15

-10

Molality log10 kgmol  water

Figure 2: Total aqueous metal concentration (log10 [mol/kg H2 O]) of different materials in pure H2 O at 150-550 ◦ C and 22-50 MPa calculated from the R-HKF equation of state. The materials are arranged according to their approximate location in the periodic table and metals and their oxides appear within the same column.

�������� (���)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 -5

Industrial & Engineering Chemistry Research Page 8 of 25

Page 9 of 25

A majority of the materials in Figure 2 undergo significant changes in solubility on going from sub- to supercritical water which correspond with changes in the solvent properties (Figure 1). The solubilities of TiO2 , WO3 , NiO, Cu and CuO decrease because their aqueous species have strong ionic character and the lower dielectric constant in SCW does not support ions as well. On the other hand, the solubilities of MoO2 , Co, Co3 O4 , and Ni increase in low density SCW because their most abundant aqueous species have no charge and negative conventional Born coefficients, which produce an inverse relationship with dielectric constant. Values of effective Born coefficients for neutral aqueous species are obtained by regression of experimental data or from correlations 22,23 and can be either positive or negative. These coefficients are positive if the species attracts H2 O dipoles (like ions) or negative if the molecule repels H2 O dipoles and disrupts the electrostatic dipole-dipole interactions within the solvent. 22 -10

Molality [ Log10(mol/kg H2O) ]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

-15 -20

22 MPa

30 MPa 40 MPa 50 MPa

PtO

50 MPa 40 MPa

PtOH+ 30 MPa 22 MPa

-25 Pt2+

50 MPa 40 MPa

-30

30 MPa

-35

22 MPa

-40 -45

200

300 400 Temperature [°C]

500

Figure 3: Pt species concentrations: Pt2+ , PtOH+ , and PtO concentrations in pure H2 O at 22, 30, 40 and 50 MPa and 150-550 ◦ C The weak dependence of solubility on temperature and pressure for CoO, MoO3 , Pt, Ru, RuO2 , WO2 , and ZrO2 is because the aqueous metal concentration is dominated by a neutral aqueous metal species. The solubilities of such species are only weakly dependent on the solvent properties. Figure 3 provides an example for the case of aqueous Pt species. 9

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 25

The neutral PtOaq species has the highest concentration at all temperatures and pressures, therefore dominating the total Pt solubility. The concentration of PtOaq is a very weak function of temperature and pressure compared to Pt2+ and PtOH+ . Individual species concentrations for the other materials are plotted with temperature in Figures S1-S15.

Material properties and hydrothermal solubility The solubilities of the metals and oxides were fit to models with independent variables representing chemical and electronic properties of the materials. Our objective was to construct a function of material properties that correlated the calculated solubilities with the fewest number of terms. Solubility was chosen as the dependent variable because it is a quantifiable measure of hydrothermal stability that can be used to compare materials. A strong correlation exists between pure metal solubility (Sm , mol/kg H2 O) and electronegativity (Figure 4a) and the second-order polynomial in equation (12) captures this effect (χ is electronegativity in Pauling units) for 400 ◦ C and 50 MPa. This correlation is physically meaningful because the aqueous metal-containing species used to model solubility are in higher oxidation states than the pure metals and metals with lower electronegativities are (generally) more easily oxidized. Therefore, the metals are also more easily dissolved as electronegativity decreases. Log10 (Sm )|400o C,50MPa = 15.15χ2 − 77.78χ + 87.96

(12)

For the oxide solubility data (Sx , mol/kg H2 O), we considered a second order polynomial. The base model is shown in equation (13) where αi represents first order coefficients, β ij represents second order coefficients, and xi and x j denote the independent variables (material parameters). Log10 (Sx ) = α0 + ∑ αi xi + ∑ β ij xi x j i

10

i,j

ACS Paragon Plus Environment

(13)

Page 11 of 25

0 R-HKF Solubility / log10 (mol/kg H2 O)

Co

R-HKF solubility / log10 (mol/kg H2 O)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

-4 Ni

-6

Cu Ag

-8 Pd Rh

-10 Ru

Pt

Au

-12 1.8

2.0

2.2

2.4

ZnO MoO3 *

-5 Rh2 O3 Co3 O4 WO3 * Fe2 O3 Fe2 O3

2.6

NiO

CuO CoO Al2 O3 α Al2 O3 γ

ZrO2

-10

ZrO2 CeO2

TiO2

-15

Electronegativity (P.u.)

CoO

ZnO

RuO2

-15

-10

-5

0

Correlation Solubility / log10 (mol/kg H2 O)

(a) Metal solubility correlation

(b) Oxide solubility correlation

Figure 4: Catalyst solubility analysis: Figure 4a compares the calculated solubility of different metals at 400 ◦ C, 50 MPa with a polynomial function of electronegativity (in Pauling units) in equation (12); Figure 4b compares metal oxide solubility values at 400 ◦ C and 50 MPa obtained from the R-HKF model with the best-fit correlation as a function of polarizing power, cation electronegativity, and ionic-covalent parameter in equation (14). The independent variables considered for fitting the oxide solubility data included cation radius (rcation , Å), cation electronegativity (χion , Pauling units), Ionic-Covalent Parameter (ICP, unit-less), oxide electronegativity (χoxide , eV), chemical hardness (η, eV), and polarizing power (P, Å-2 ). These parameters were chosen because they have previously been used to describe other electronic and chemical properties of oxides 24 and are listed in Table S5. ICP is related to the acid strength of the cation and the ionic or covalent nature of the metal-oxygen bond. Polarizing power is calculated from z/r2 where z is the formal charge on the cation and r is the ionic radius. Chemical hardness is defined as half the bandgap of the oxide. 24 We used LinearModelFit in Mathematica to fit equation (13) to the solubility data in Table S5 and then we used an iterative function to systematically eliminate terms. The quality of the model was measured on each iteration using the Akaike Information Criterion with correction for finite sample size (AICc). AICc measures the goodness of fit while taking into account model complexity (i.e., number of terms), with lower AICc

11

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 25

values representing better models. If the removal of a term decreased the AICc value, then the original model was replaced with the better model of fewer terms. This analysis was repeated until the best model was obtained, shown in equation (14). According to the model expression, an increase in the ionic-covalent parameter (ICP) and a decrease in polarizing power (P) will result in a decrease in oxide solubility. The calculated solubility goes through a maximum with cation electronegativity (χion ). Log10 (Sx )|400o C,50MPa = −105 − 41( ICP) + 187χion + 74χ2ion + 1.1χion P

(14)

A comparison between the solubility values obtained from the R-HKF equation of state and the expression in equation (14) is shown in Figure 4b. This is the first general correlation for metal oxide solubility in SCW of which we are aware. From the high scatter in Figure 4b, it is unclear if equation (14) will provide an accurate estimation of the solubility of other oxides or mixed metal oxides. Additional oxide solubility data or alternative material descriptors could potentially improve the correlation. Density functional theory calculations have been used to generate more sophisticated descriptors for oxide performance, however these calculations are more complex than the thermodynamic solubility calculations with the R-HKF model. To provide rapid estimation of oxide solubilities, the correlation should comprise material properties that are easily obtained or measured. Alternatively, further subcategorization of the oxides could achieve stronger correlations with the simple descriptors. The compromise with further subcategorization is that some oxides may identify with multiple categories (or none), thus increasing the difficulty of estimating solubility. Until a better global correlation for oxide solubility is found, thermodynamic properties (∆G of , CP ( T ), So ) of new oxide materials should be measured or estimated and hydrothermal solubility calculated using the R-HKF equation of state.

12

ACS Paragon Plus Environment

Page 13 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Oxygen fugacity-pH diagrams During hydrothermal reactions, concentrations of reactants and products will change the oxidizing potential and pH of the solution. As a result, the system may be operating under conditions that alter the catalyst. Oxygen fugacity ( fO2 )-pH diagrams are useful for understanding how these changes in acidity and concentrations of oxidizing or reducing species affect catalyst stability, showing regions of thermodynamically favored species in water as a function of pH and fO2 . These diagrams can also be used to design conditions that minimize catalyst dissolution and changes in the catalyst oxidation state through addition of buffers to control pH and addition of oxidizers or reducers to control fO2 . The diagrams for Co appear in Figure 5 as a representative example. Diagrams for additional materials including Ce, Mo, Ni, Pd, Pt, Ru, Ti, W, and Zr and at additional temperatures and pressures are shown in Figures S16-S25.

Figure 5: fO2 -pH diagrams for Co-H2 O system. fO2 is in bar, and pH is in log10 (mol/kg H2 O). Boundaries for aqueous metal species are defined as a j = 10−6 (≈ 1 µmol/kg H2 O) so within the shaded regions a j ≥ 10−6 . Striped regions indicate the presence of multiple aqueous metal species (e.g. Co2+ and CoOH+ ) with a j ≥ 10−6 . The fO2 and pH values of pure H2 O are plotted as •. In an fO2 -pH diagram, the boundaries between solid species indicate the value of

13

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

fO2 at which the two solids coexist in equilibrium. At fO2 values greater or less than these boundaries, only one solid phase is present. For example, the boundaries between Co(OH)2(s) and Co3 O4(s) in Figure 5 show the fO2 conditions at which both oxides exist at the system temperatures and pressures indicated above the diagrams. Because H+ and OHdo not participate in the redox reactions between solid species, these solid-solid boundaries are independent of pH. Boundaries between solids and neutral aqueous species are also independent of pH for the same reason. As expected, the oxidation state increases with increasing fO2 . Activities (a j ) of 10−6 and 10−8 were used to construct boundaries between solids and aqueous metal species (see Methods section). The shaded region within the boundaries indicates that a j ≥ 10−6 or 10−8 for all conditions in the region. For example, Co2+ and CoOH+ in Figure 5 are bounded by equilibrium reactions with Co(s) , CoO(s) , and Co3 O4 (s) . Inside these red regions, the ions have activities ≥ 10−6 . From the solubility calculations, the ion activity coefficients γ j were found to be ≥ 0.95 (the majority of species were ≥ 0.99) among all materials modeled. As these values are very close to unity and γ j = 1 for neutral aqueous species, the activities of aqueous metal species are approximately equal to the m

concentration (a j = γ j mΘj ≈

mj mΘ

where mΘ = 1 mol/kg H2 O).

All diagrams in this paper contain two reference conditions: the fO2 and pH of pure H2 O (black dot) and the fO2 value at which f H2 is one bar (dashed horizontal line) plotted at the system temperature and pressure. A third reference can be made from log10 fO2 = zero, where fO2 is one bar. It is important to note that the fO2 and pH of the aqueous solution can change with time during batch operation. For example, the Co-H2 O systems in Figure 5 show that at all conditions, the fO2 of pure H2 O is sufficient for Co oxidation. As the O2 is consumed during Co oxidation (or H2 is released), however, the fO2 of the system decreases. In batch, this reduction of fO2 will occur until the fO2 of the system reaches the Co/CoO equilibrium boundary (resulting in a partially oxidized catalyst) or until all the Co is oxidized (large H2 O:Co ratio). During flow operation, however, the

14

ACS Paragon Plus Environment

Page 14 of 25

Page 15 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Co-H2 O system is expected to eventually reach a steady-state fO2 value corresponding with that of the feed (H2 O) and Co will be entirely oxidized to Co3 O4 . Therefore, the pure H2 O reference identifies the dominant species at equilibrium for a catalyst-H2 O flow system. The fO2 -pH diagrams for Co (Figure 5) and Ni (Figure S20) show that both materials form cations (≥ 1 µmol/kg H2 O) at moderate to low pH values and negative ions at very high pH values. In general, Co ions form over a wider range of hydrothermal fO2 -pH conditions than Ni ions. A small decrease in pH (relative to the conditions for pure H2 O) favors formation of Co2+ and CoOH+ at concentrations ≥ 1µmol/kg H2 O. Across all temperatures and pressures, the reference points for pure H2 O on the Co and Ni diagrams are located in the regions for Co3 O4 and NiO, respectively. This means that under H2 O flow conditions, Co will form Co3 O4 and Ni will form NiO, eventually. One strategy for improving the hydrothermal stability of Co and Ni catalysts would be to operate under reducing and slightly basic conditions that correspond to the regions of Co(s) and Ni(s) . In the fO2 -pH diagrams for the Mo-H2 O system (Figure 6) and the W-H2 O system (Figure S24), H2 MoO4(aq) and H2 WO4(aq) species were omitted from the diagrams because the regions for these aqueous neutral species dominate the entirety of the diagram and obscure analysis of the other species. At all conditions, Mo and W (and their oxides) dissolve to form significant (≥ 1 µmol/kg H2 O) amounts of H2 MoO4(aq) and H2 WO4(aq) . Additionally, small increases in pH result in further formation of Mo and W anions. In pure H2 O flow, MoO3 and WO3 are predicted to be the thermodynamically favored oxidation states. While both Mo and W oxides have very poor hydrothermal stability, the amount of dissolution could be reduced by operating the system under reducing and acidic conditions. Unlike the results for Co, Ni, Mo, and W, calculations for Pt, Pd and Ru (Figures S16, S21, and S22, respectively) predict good hydrothermal stability in pure H2 O, other than the predicted formation of RuO2 in pure H2 O flow. This prediction was not realized in batch

15

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 6: fO2 -pH diagrams for Mo-H2 O system. fO2 is in bar, and pH is in log10 (mol/kg H2 O). Boundaries for aqueous metal species are defined as a j = 10−6 (≈ 1 µmol/kg H2 O). The striped regions indicate the presence of both HMoO4 - and MoO4 2- with a j ≥ 10−6 . The fO2 and pH values of pure H2 O are plotted as •. experiments, which showed no oxidation of Ru after 60 minutes in SCW at 400 ◦ C and 24-40 MPa. 20 One possible explanation for this discrepancy is that the experiment did not reach equilibrium. Oxygen diffusion in bulk Ru is slow, 25,26 so while a thin film of RuO2 may form on the surface, subsequent oxidation of the bulk Ru would not be observed during the experiments. The fO2 -pH stability results for Ce (Figure S17), Ti (Figure S23), and Zr (Figure S25) also predict good hydrothermal stability and CeO2(s) , TiO2(s) , and ZrO2(s) are the thermodynamically favored oxidation states in pure H2 O at all temperatures and pressures examined in this study. The dissolution of TiO2(s) (shown in Figure 7) and ZrO2(s) into aqueous species is insensitive to fO2 and pH with the exception of conditions of high temperatures (≥ 500 ◦ C)

and relatively low supercritical pressures. One explanation for this sudden instability

could be errors in the ∆G f ,j values for certain ions at high temperatures (≥ 500 ◦ C) and low water densities. Specifically, the aqueous species in highest concentration at high temperatures and low pressures also had the highest heat capacities. This relationship sug-

16

ACS Paragon Plus Environment

Page 16 of 25

Page 17 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Figure 7: fO2 -pH diagrams for Ti-H2 O system. fO2 is in bar, and pH is in log10 (mol/kg H2 O). Boundaries for aqueous metal species are defined as a j = 10−6 (≈ 1 µmol/kg H2 O). The striped regions indicate the presence of multiple aqueous species with a j ≥ 10−6 . The fO2 and pH values of pure H2 O are plotted as •. gests that the thermal contribution in the calculation of ∆G f ,j becomes more influential at higher temperatures and dominates the small solvent contributions at low SCW densities. While the R-HKF model can be a useful tool for predicting catalyst solubility, it does have limitations if the material-dependent parameters contain large errors and if SCW densities are well below 0.2 g/mL.

Engineering hydrothermal solutions to enhance catalyst stability We calculated the fO2 and pH values of simple solutes (representative of reactants or products for some hydrothermal reactions) in equilibrium with H2 O at different temperatures, pressures, and concentrations to illustrate the effect of these species on the fO2 and pH of the solution (see Section S1.3.2 and Figure S26 in the Supporting Information). The values for these solutions were calculated by finding the reaction equilibrium for the components in the solution (e.g., CO2 reacts with water to form carbonic acid, bicarbonate, and carbonate) and the corresponding concentrations of O2 and H+ at that equilibrium.

17

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 25

Overall, the solutes that we modeled have the largest effect on the solution fO2 and pH at 150 ◦ C. As temperature increases, the solution points converge on the point for pure H2 O. The convergence is more significant at 22 MPa than 50 MPa, likely from the lower ion product and dielectric constant at 22 MPa. The presence of CO2 decreases the pH due to the formation of carbonic acid and the presence of NH3 increases the pH due to the formation of NH4 + . Methane (CH4 ) has very little effect on the fO2 and pH of the solution, while formic acid significantly reduces the fO2 and slightly decreases pH. This is because formic acid decomposes to H2 and CO2 .

(a)

(b)

(c)

Figure 8: fO2 -pH diagrams for Ni, CeO2 , and simple aqueous solutions. fO2 is in bar, and pH is in log10 (mol/kg H2 O). Boundaries for aqueous metal species in 8a and 8b are defined as a j = 10−6 (≈ 1 µmol/kg H2 O). In 8c, the solute concentrations are in molal (mol/kg H2 O) and the shaded region indicates the stable operating conditions for a Ni/CeO2 catalyst. Figure 8 illustrates how the fO2 -pH diagrams can be used to design a hydrothermal reaction system. Consider a Ni/CeO2 catalyst for a hydrogenation reaction in SCW at 400 ◦ C and 30 MPa.

The fO2 -pH diagram for Ni (Figure 8a) shows that NiO(s) is favored in pure

H2 O and Ni(s) is favored at fO2 < 10−30 bar, so sufficient H2 or another reducing agent should be added to the reactor to prevent NiO(s) formation. The fO2 -pH diagram for Ce (Figure 8b) shows that CeO2(s) is stable at fO2 > 10−48 bar and pH > 8. Ce ions (> 1 ppm), however, are thermodynamically favored at fO2 < 10−48 and pH < 8 so if the solution becomes too reducing or acidic, Ce dissolution may occur. By using various solutes, one can engineer a solution wherein both Ni(s) and CeO2(s) are stable. The equilibrium fO2 and 18

ACS Paragon Plus Environment

Page 19 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

pH results for various aqueous solutions at 400 ◦ C and 30 MPa are plotted in Figure 8c. From the options plotted in Figure 8c, a feed with > 0.1 mol/kg H2 O formic acid or ammonia will provide thermodynamically stable flow operation of Ni/CeO2 . Specifically for the design of a hydrogenation reaction, formic acid in concentrations > 0.1 and < 5 mol/kg H2 O would generate H2 in situ for the reaction while preventing oxidation and dissolution of the Ni/CeO2 catalyst. Additional considerations when evaluating catalyst oxidation and dissolution in hydrothermal reaction media include metal complexation with organic species and oxidation and dissolution rates. While the fO2 and pH of the solution may be acceptable for stable catalyst operation, it is possible for metal atoms to leach from the catalyst through the formation of metal coordination complexes (e.g., carbonate, sulfate, nitrate, chloride, and acetate complexes) with aqueous species present during hydrothermal reactions. Published methods 27 for calculating free energies of aqueous metal complexes at high temperatures and pressures provide a route for predicting metal loss from the formation of complexes with reagents and reaction intermediates. It is also possible that a catalyst can be used successfully at conditions where oxidation or dissolution are thermodynamically favored, provided that oxidation and dissolution rates are slow enough such the catalyst remains active for an acceptable lifetime. At present, little is known about the rates of catalyst oxidation and dissolution and future work is required to determine the influence of temperature, pressure, and solution composition on the oxidation and dissolution rates of catalytic materials.

Conclusion Results for the calculated solubilities of transition metals and oxides in water at 150-550 ◦C

and 22-50 MPa identified a variety of hydrothermally stable materials (Au, Pt, Pd,

Ru, RuO2 , TiO2 , ZrO2 , and CeO2 ) and illustrated the “worst case scenario” for catalyst

19

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

deactivation by dissolution. That is, if oxidation and dissolution rates are fast, the catalyst would quickly reach its new equilibrium state, but total oxidation and dissolution would not exceed the equilibrium limit. Of course, the dissolution kinetics may be slow and more forgiving in practice. New correlations between the hydrothermal solubilities and readily available material properties allowed for rapid estimation of metal or metal oxide solubility and hence identification of materials with improved hydrothermal stability. For example, metal alloy catalysts should have large average electronegativities and oxides should have relatively large ionic-covalent parameters and small polarizing power values to minimize dissolution. The fO2 -pH diagrams predicted phase changes of the catalytic materials in response to changes in the composition of hydrothermal reaction solutions. Together with fO2 and pH calculations for different solutes in aqueous solution, these diagrams provide the information needed to design hydrothermal solutions (e.g., temperature, pressure, and solute concentration) that improve the catalyst stability.

Acknowledgement The authors gratefully acknowledge contributions from David Hietala and Dr. Lucas Griffith and financial support from the University of Michigan. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE 1256260.

Supporting Information Available • Supporting_Information.pdf: Section S1, expansion of methods for calculating solubility and constructing fO2 -pH diagrams; Section S2, additional figures of calculated equilibrium concentrations of aqueous metal species and fO2 -pH diagrams of ma-

20

ACS Paragon Plus Environment

Page 20 of 25

Page 21 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

terials and solutions at additional temperatures and pressures; Section S3, tables of thermodynamic properties, parameters, and reaction equations used to model catalyst oxidation and dissolution and solution fO2 and pH. This material is available free of charge via the Internet at http://pubs.acs.org/.

References (1) Savage, P. E. Organic Chemical Reactions in Supercritical Water. Chem. Rev. 1999, 99, 603–622. (2) Elliott, D. C. Catalytic hydrothermal gasification of biomass. Biofuels, Bioprod. Biorefin. 2008, 2, 254–265. (3) Savage, P. E. A perspective on catalysis in sub- and supercritical water. J. Supercrit. Fluids 2009, 47, 407–414. (4) Azadi, P.; Farnood, R. Review of heterogeneous catalysts for sub- and supercritical water gasification of biomass and wastes. Int. J. Hydrogen Energy 2011, 36, 9529–9541. (5) Yeh, T. M.; Dickinson, J. G.; Franck, A.; Linic, S.; Thompson, L. T.; Savage, P. E. Hydrothermal catalytic production of fuels and chemicals from aquatic biomass. J. Chem. Technol. Biotechnol. 2013, 88, 13–24. (6) Furimsky, E. Hydroprocessing in Aqueous Phase. Ind. Eng. Chem. Res. 2013, 52, 17695– 17713. (7) Choudhary, T.; Phillips, C. Renewable fuels via catalytic hydrodeoxygenation. Appl. Catal., A 2011, 397, 1–12. (8) Akiya, N.; Savage, P. E. Roles of Water for Chemical Reactions in High-Temperature Water. Chem. Rev. 2002, 102, 2725–2750. 21

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(9) Marshall, W. L.; Franck, E. U. Ion product of water substance, 0-1000 ◦ C, 1-10,000 bars. New International Formulation and its background. J. Phys. Chem. Ref. Data 1981, 10, 295–304. (10) Johnson, J. W.; Norton, D. Critical phenomena in hydrothermal systems; state, thermodynamic, electrostatic, and transport properties of H2O in the critical region. Am. J. Sci. 1991, 291, 541–648. (11) Elliott, D. C.; Biller, P.; Ross, A. B.; Schmidt, A. J.; Jones, S. B. Hydrothermal liquefaction of biomass: Developments from batch to continuous process. Bioresour. Technol. 2015, 178, 147–156. (12) Yang, C.; Li, R.; Cui, C.; Liu, S.; Qiu, Q.; Ding, Y.; Wu, Y.; Zhang, B. Catalytic hydroprocessing of microalgae-derived biofuels: a review. Green Chem. 2016, 18, 3684–3699. (13) Kritzer, P.; Boukis, N.; Dinjus, E. Factors controlling corrosion in high-temperature aqueous solutions: A contribution to the dissociation and solubility data influencing corrosion processes. J. Supercrit. Fluids 1999, 15, 205–227. (14) Kritzer, P. Corrosion in high-temperature and supercritical water and aqueous solutions: a review. J. Supercrit. Fluids 2004, 29, 1–29. (15) Marrone, P. A.; Hong, G. T. Corrosion control methods in supercritical water oxidation and gasification processes. J. Supercrit. Fluids 2009, 51, 83–103. (16) Sun, C.; Hui, R.; Qu, W.; Yick, S. Progress in corrosion resistant materials for supercritical water reactors. Corros. Sci. 2009, 51, 2508–2523. (17) Ding, Z. Y.; Frisch, M. a.; Li, L. X.; Gloyna, E. F. Catalytic oxidation in supercritical water. Ind. Eng. Chem. Res. 1996, 35, 3257–3279. (18) Xiong, H.; Pham, H. N.; Datye, A. K. Hydrothermally stable heterogeneous catalysts for conversion of biorenewables. Green Chem. 2014, 16, 4627–4643. 22

ACS Paragon Plus Environment

Page 22 of 25

Page 23 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

(19) Tanger, J. C.; Helgeson, H. C. Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures; revised equations of state for the standard partial molal properties of ions and electrolytes. Am. J. Sci. 1988, 288, 19–98. (20) Jocz, J. N.; Thompson, L. T.; Savage, P. E. Catalyst Oxidation and Dissolution in Supercritical Water. Chem. Mater. 2018, 30, 1218–1229. (21) Shock, E. L.; Helgeson, H. C. Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: Correlation algorithms for ionic species and equation of state predictions to 5 kb and 1000◦ C. Geochim. Cosmochim. Acta 1988, 52, 2009–2036. (22) Shock, E. L.; Helgeson, H. C.; Sverjensky, D. A. Calculation of the Thermodynamic and Transport Properties of Aqueous Species at High Pressures and Temperatures: Standard Partial Molal Properties of Inorganic Neutral Species. Geochim. Cosmochim. Acta 1989, 53, 2157–2183. (23) Shock, E. L.; Sassani, D. C.; Willis, M.; Sverjensky, D. A. Inorganic Species in Geologic Fluids: Correlations Among Standard Molal Thermodynamic Properties of Aqueous Ions and Hydroxide Complexes. Geochim. Cosmochim. Acta 1997, 61, 907–950. (24) Matar, S. F.; Campet, G.; Subramanian, M. A. Progress in Solid State Chemistry Electronic properties of oxides : Chemical and theoretical approaches. Prog. Solid State Chem. 2011, 39, 70–95. (25) Ahn, J.-H.; Lee, W.-J.; Kim, H.-G. Oxygen diffusion through RuO2 bottom electrode of integrated ferroelectric capacitors. Mater. Lett. 1999, 38, 250–253. (26) Oh, S. H.; Park, C. G.; Park, C. Thermal stability of RuO2 /Ru bilayer thin film in oxygen atmosphere. Thin Solid Films 2000, 359, 118–123.

23

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(27) Sverjensky, D. A.; Shock, E. L.; Helgeson, H. C. Prediction of the Thermodynamic Properties of Aqueous Metal Complexes to 1000◦ C and 5 kb. Geochim. Cosmochim. Acta 1997, 61, 1359–1412.

24

ACS Paragon Plus Environment

Page 24 of 25

Page 25 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Graphical TOC Entry catalyst bed

H2O + reagents

150-550 oC, 22-50 MPa Catalyst A + Solution A

Catalyst B + Solution A

H2

+

+

OH-

O2

H2

oxidation + dissolution -

OH

H2

H

H

Catalyst

A

unstable

M

Catalyst A + Solution B

H2 +

O2

OH-

NH 4

H2 improved catalyst

+

composition

H+

-

OH

-

Catalyst

B

H+

OH

H2

stable

improved solution pH & fO2

H2 Catalyst

A

OH-

H2 NH

4

+

stable

25

ACS Paragon Plus Environment