Microstrains and Stresses Analysis in Electroless Deposited Thin Pd

Worcester, Massachusetts 01609, and High-Temperature Materials Laboratory, Oak Ridge National. Laboratory, Oak Ridge, Tennessee 37831. This work is ...
0 downloads 0 Views 526KB Size
Download PDF
Ind. Eng. Chem. Res. 2006, 45, 8145-8153

8145

Microstrains and Stresses Analysis in Electroless Deposited Thin Pd Films Federico Guazzone,†,‡ E. Andrew Payzant,§ Scott A. Speakman,| and Yi Hua Ma*,† Center for Inorganic Membrane Studies, Department of Chemical Engineering, Worcester Polytechnic Institute, Worcester, Massachusetts 01609, and High-Temperature Materials Laboratory, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831

This work is aimed at the measurement with X-ray diffraction techniques of microstrains, intrinsic stresses, and thermal and hydrogen stresses in composite Pd porous metal (PM) structures and their release at high temperatures. In addition, the changes in the Pd microstructure upon heating were studied with SEM to determine the relation between stress release and microstructure changes. The initial microstrains, 0.29%, in the electroless deposited Pd layer were irreversibly released after annealing at 400 °C for 1 h in He atmosphere. The initial intrinsic stress, mostly due to the deposition method, was tensile in nature (104.7 MPa) and was also released at 400 °C for 1 h in He atmosphere. After the release of intrinsic stresses, the total stress was given by the sum of the thermal stresses (mismatch in coefficients of thermal expansion) and the H2 stresses due to the absorption of H2 in the Pd layer. The total stress (thermal + hydrogen stresses) was compressive and was released at temperatures higher than 400 °C with a significant change in Pd morphology. A model, valid in the 60-400 °C temperature range, was also developed to predict the total stress to which composite Pd membranes were exposed in the 250-400 °C temperature and the 1-5 bar H2 pressure range. The highest total stress, at 250 °C and 5 bar, was estimated to be equal to -260 MPa according to the model developed in this work. The lowest stress, at 400 °C and 1.5 bar, equaled -78 MPa. At temperatures higher than 400 °C, the model did not hold since stresses were released by plastic deformation; however, their value was lower than -78 MPa. The characterization of several composite Pd membranes prepared on porous metal (PM) supports showed that leaks formed at temperatures above 400-450 °C. Since leaks only formed at T > 400 °C, the magnitude of stresses only played a minor role in leak formation. 1. Introduction The production of H2 in composite Pd based membrane reactors has been widely studied, and its feasibility has been validated.1-5 Most probably, Pd or Pd alloy membranes supported on porous metal (PM) supports will prevail for industrial purposes due to the easy implementation of metallic modules into industrial plants. Currently, relatively thin (L < 10 µm) composite Pd-PM membranes with high H2 permeance could be achieved although selectivity decline over time at 500 °C still poses problems.6-8 Generally, composite Pd membranes are prepared at temperature and pressure conditions different from the ones they will be used in a membrane reactor. Fresh thin films are characterized by remaining “intrinsic” stresses and a large surface energy due to their fine crystallites that are automatically released at high temperatures. Therefore, at the reaction temperature of 500 °C, composite Pd membranes undergo a series of structural transformations that may lead to a H2 selectivity decrease over time. Indeed, the deposition (CVD, sputtering, electro- and electroless methods) of Pd onto a substrate leads to a composite structure including large amounts of extra energy. Part of the excess energy in these nanostructured films is stored as microstrains and surface energy, which is released by the growth of Pd crystallites at high temperatures. It was shown by calorimetric measurements that thermal relaxation in freestanding Pt and Pd nanocrystalline samples prepared by inert * To whom correspondence should be addressed. Tel.: (508) 831 5398. Fax: (508) 831 5853. E-mail: [email protected]. † Worcester Polytechnic Institute. ‡ E-mail: [email protected]. § Oak Ridge National Laboratory. E-mail: [email protected] | Current address: Massachusetts Institute of Technology, Cambridge, MA 02139. E-mail: [email protected].

gas condensation occurred in two stages.9 The first reduction in the total Gibbs free energy corresponded to the relaxation of microstrains and/or nonequilibrium crystallites boundaries, while the second reduction in the total Gibbs free energy corresponded to crystallite growth. In addition, part of the excess energy in composite structures (a film anchored to a support) is due to film/substrate interactions stored as “intrinsic” and “extrinsic” stresses within the film which are generally studied with X-ray techniques. Stresses are denoted as intrinsic when the causes of the formation of the stress are related to the material of the film itself and the conditions and method of deposition.10-12 Stresses caused by external factors are denoted as extrinsic. Extrinsic stresses are mostly due to thermal expansion coefficient mismatch between the film and the support.13-15 Intrinsic and extrinsic stresses lead to either the expansion or contraction of the Pd lattice cell. The fundamental difference between microstrains and stresses is that microstrains are located at the Pd crystallite boundaries or at the external shell of Pd crystallites,16 while stresses are due to contraction or expansion of the entire unit cell. More technically, microstrains lead to the broadening of X-ray peaks while stresses lead to the shifting of X-ray peaks (change in d spacing). The absorption of H2 in a composite Pd membrane also leads to compressive extrinsic stresses denoted as H2 stresses. Indeed, H2 loading leads to the expansion of the Pd film.17 However, since the film is attached to the support, its expansion will be limited and compressive stresses will arise which may lead to membrane failure if the adhesion force of the Pd layer is too weak compared to the force originated by the compressive stress. Membranes may fail by cracking, blistering, or buckling. The magnitude of compressive stresses after H2 absorption depends on the H2 loading n(H/Pd). When enough energy is given to the system, microstrains and

10.1021/ie060756s CCC: $33.50 © 2006 American Chemical Society Published on Web 10/28/2006

8146

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006

stresses are relaxed and the structural changes that occur during stress relaxation may play an important role in leak formation and leak growth. Understanding microstrains and stress relaxation in composite Pd-PM structures represents a step toward the comprehensive insight of the defect formation in composite Pd membranes. However, few studies dealt with macro stresses in composite Pd porous metal supports structures, and their relaxation at high temperatures. Hence, the main objective of the present work was to determine experimentally the microstrains, intrinsic stresses, and thermal and H2 stresses by X-ray diffraction techniques in electroless deposited composite Pd-PM structures and study their release at high temperatures. In addition, a model was developed to estimate total stresses in composite Pd-PM membranes in the 250-400 °C temperature and the 1-5 bar pressure range. Stress relaxation was also correlated with Pd microstructure changes with temperature using scanning electron microscope (SEM) techniques. Finally, the effects of microstrains, stresses, and their relaxation on leak formation were elucidated.

as a function of the inverse temperature for all K(T) values. The ln(K(T)) vs 1/T points were fitted with a straight line having the equation y ) 9.3-1114.2x, leading to the H2 content, n(H/ Pd), determined from eq 4

2. Stress Theory

σtotal ) σth + σH )

The thermal expansion of Pd and the substrate could be considered as linear functions of temperature within the temperature range (20-600 °C) of interest for this study; therefore, the coefficients of thermal expansion were considered as constants. Hence, the thermal stress (σth) along the Pd/substrate interface direction to which the thin film was subjected could be approximated by eq 1.10

σth )

(Lsub - Lfilm)∆TEfilm 1-ν

(1)

where Lsub and Lfilm are the thermal expansion coefficients of the substrate and the thin film, respectively, ∆T is the increment in temperature from a reference T0 (T0 ) 20 °C), Efilm is Young’s modulus, and ν is the Poisson’s ratio of the metal film. According to eq 1, thermal stresses are a linear function of temperature. However, the linear character of thermal stresses with temperature is only observed in the elastic region of the metal. When the thermal stresses, or any other stress, exceed the elastic limit of the film, they are released by plastic deformations. H2 stresses are equal to the H2 strain times the Young modulus (Efilm) of Pd and are given by eq 2

σH ) HEfilm

(2)

where H is the H2 strain and is a linear function of the H2 content n(H/Pd).17,18 Hence, σH ) kHn(H/Pd)Efilm with kH as a linearity constant. The H2 content, n(H/Pd), was determined from temperature and pressure values using Sieverts’ constant, which relates P0.5 to n(H/Pd) by the following linear equation

PH21/2 ) K(T)n(H/Pd)

(3)

where K(T) is the Sieverts’ constant in Pa0.5 or torr0.5 and is determined from the H2 absorption isotherm at the temperature T. K(T) was determined for temperatures ranging from 0 to 300 °C from experimental data (H2 pressure, n(H/Pd)) reported by earlier researchers.19-21 For all isotherms, 0, 30, 50, 60, 75, 80, 160, 180, 200, 250, 290, and 300 °C, K(T) was determined by calculating the slope of the linear portion of P0.5 vs. n(H/Pd) at n(H/Pd) ) 0. The natural logarithm of K(T) was then plotted

PH20.5

n(H/Pd) ) Exp

(

-1114.2 + 9.3 T

(4)

)

with P in torr and T in K. Equation 4 is only valid in the stability domain of the R phase of Pd and, for a given temperature, in a pressure range where P0.5 is a linear function of n(H/Pd). The accuracy on the H2 concentration n(H/Pd) given by eq 4 was estimated to be 0.002. The reader is referred to Guazzone22 for a complete analysis. The total stress in a composite Pd/substrate structure is the sum of all stresses. A composite Pd membrane under reaction conditions will be subjected to the total stress (σtotal) given by eq 5

[

]

(Lsub - Lfilm)∆T + kHn(H/Pd) Efilm (5) 1-ν

3. Experimental Section 3.1. Preparation of Samples for X-ray Diffraction Studies. As substrates, 0.5 µm grade C-22 Porous Hastelloy (PH) plates (3 cm2, 1 mm thick) and R-Al2O3 porous supports (3 cm2, 2 mm thick) were used. PM substrates were supplied by Mott Corporation, Farmington, CT. All substrates were cleaned in an alkaline solution (45 g/L NaOH, 65 g/L Na2CO3, 45 g/L Na3PO4‚12H2O), including 5 mL/L of a saturated solution of industrial detergent, for dirt and grease removal. PM supports were oxidized at 700 °C for 12 h to inhibit intermetallic diffusion at high temperatures.23-25 Substrates were activated and plated following the experimental protocol described elsewhere.26,27 The average thickness of the Pd thin film was determined by the gravimetric method, i.e., weight gain of the sample divided by the product of the plated surface area and the metal (Pd) density. Samples studied are listed in Table 1. The letter following the sample number designates the experiment. For instance, PH-1b stands for “sample PH-1, second experiment”. “Fresh” samples were the ones which were only dried at 120 °C overnight before any experiment. Only samples PH-2a and PH-3a were pretreated in He atmosphere at 400 °C for 1 h to release initial microstrains and intrinsic stresses present in every fresh Pd film. 3.2. X-ray Diffraction Procedures. Diffraction data were collected using Cu KR radiation on a PANalytical X’Pert Pro MPD. Parallel-beam optics, consisting of an incident-side graded multilayer parabolic X-ray mirror and a diffracted-side 0.09 rad parallel-plate flat collimator, were used to alleviate errors associated with sample displacement. Samples were loaded into an Anton Paar XRK900 reaction chamber which allowed the temperature to be varied between 25 and 900 °C and the gas environment to be varied between 1 and 1.5 atm of He or He4%H2 (static or flowing). Microstrain-size separation in Pd deposits was performed using Williamson-Hall’s method, which assumes the peak broadening (β), expressed in radians, to be the sum of a tan(θ) term for the microstrain and a 1/cos(θ) term for the crystallite size as given in eq 6.28

β ) 4e tan(θ) + λCu/(t cos(θ))

(6)

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006 8147 Table 1. List of Samples Studied Pd thickness, L (µm)

sample

pressure (bar), atmosphere

temp. range (°C)

PH-1a (fresh)

12

1, He

25-600

PH-1b

12

1, He

60-400

PH-1c

12

1.5, H2

60-500

PH-2a (pretreated in He at 400 °C for 1 h) PH-3a (pretreated in He at 400 °C for 1 h) Al2O3-3a/4a/5a/6a (fresh)

12

1, He

60-400

12

1, He

300-500

6

1, H2

300/400/500/600

experiment description Measurement of microstrains, Pd crystallite size and initial intrinsic stress. Measurement of thermal stresses in the 25-600 °C temperature range. Measurement of microstrains and stresses in the annealed sample as a function of temperature. Measurement of stresses as a function of H2 concentration in the Pd layer (n ) H/Pd). Measurement of stresses in the preannealed sample. Comparison with PH-1b Measurement of stresses in the preannealed sample at T > 400. Investigation of Pd microstructure after annealing in H2 for 48 h at 300 (Al2O3-3), 400 (Al2O3-4), 500 (Al2O3-5) and 600 °C (Al2O3-6)

where e ≈ (∆d/d)hkl is an approximate upper limit of the maximum distortion, λCu is the Cu X-ray wavelength, t is the Pd crystallite size, and θ is the half peak position. Equation 6 can also be written as eq 7

β cos(θ) )

λCu + 4e sin(θ) t

with e ) 1.25x

(7)

where x2 is the root-mean-square strain (rms), which is a measurement of the Pd lattice distortion. The peak width (β) was determined by fitting every reflection with a Lorentz peak function after subtracting the KR2 component and the background. The term β cos(θ) was then plotted as a function of sin(θ) for all available reflections: (111), (200), (220), (311), (222), (400), (331), and (420). The plotted data were then fitted with a straight line, where the slope equaled 4e and the intercept with the “y” axis equaled λCu/t. Macrostresses were determined by the d-sin2ψ method, which is the main technique for stress measurement by X-ray diffraction.29,30 The technique consists of measuring the interplanar space (d spacing) of a given reflection hkl with the sample tilted at different angles (the ψ angle). When the d spacing increases with the ψ angle, crystallites are in tension. When the d spacing decreases with the ψ angle, crystallites are under compression. Diffraction data were collected from the (422) reflection of Pd between 146 and 157° 2θ at ψ tilts of 0, (28.2, (42, (55° ψ. The peak position, and ultimately the total stress in the sample, was determined using PANalytical X’Pert Stress software, which performed a procedure which included absorption intensity correction, subtraction of linear background, Lorentz-polarization correction, stripping of the Cu KR2 peak, and then determining the peak position using a Lorentz fitting method. The variation of peak position as a function of ψ tilt was then analyzed using a uniaxial sin2ψ plot to determine the macrostrain within the plane of the coating. This macrostrain was converted to a stress using a Young’s modulus of 133.52 GPa and a Poisson’s ratio of 0.3847. 4. Results and Discussion 4.1. Microstructure of a Fresh Electroless Plated Thin Pd Film. When using only Scherrer’s equation (thkl ) 0.9λ/β cos(θ)) on the (111) and (222) reflections to determine the crystallite size (t), it was found on sample PH-1a that t222 was higher than t111 by a factor of 2. The same result (t222 > t111) was also reported on Pd black when microstrains were not accounted for.31 Hence, microstrains were also present in the electroless plated Pd-PH deposits. The initial microstrains present in the

Figure 1. Williamson-Hall plot of the PH-1a sample.

fresh Pd film and initial Pd crystallite size were estimated by performing a strain-size separation on sample PH-1a using a Williamson-Hall plot, shown in Figure 1. The initial rms equaled 2.9 × 10-3, and the initial Pd crystallite size equaled 97 nm. It has been shown in the case of nanocrystalline Pd that microstrains decreased from 0.5% to 0.05% as the crystallite size increased from 10 to 100 nm.32,33 Therefore, it was hypothesized that since microstrains decreased with increasing crystallite size, microstrains were confined to small crystallites and/or shells adjacent to the interfaces within the crystallites.16 The initial microstrains measured in PH-1a (0.29%) appeared to be higher than the values reported in the literature (0.05%) when the Pd crystallite size measured with XRD was around 100 nm. Figure 1 also shows that the (111) and the (222) reflections were considerably sharper than the (200) and (400) reflections, which was due to the anisotropic elasticity factor of Pd.16 The initial intrinsic stresses of the fresh PH-1a sample were determined using the d-sin2ψ experimental data plotted in Figure 2. An initial tensile stress of 104.7 ( 9 MPa was measured along the plane of the Pd film. The tensile nature of the initial stress could be attributed to the high melting point of Pd and the low temperature (60 °C) at which the deposition was carried out (T/Tm < 0.2).10 Indeed, during Volmer-Weber growth, the nature of the initial intrinsic stresses depends on the mobility of the metal being deposited. The mobility of the metal depends on its melting point and the temperature of the substrate at which the deposition is performed. Therefore, the initial intrinsic stress of a fresh coating depends on the Tsubstrate/ Tmelt ratio and, given a substrate temperature T, metals are categorized as low-mobility (T/Tm < 0.2) and high-mobility (T/ Tm > 0.2) metals.10 Low-mobility metals show tensile stresses

8148

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006

Figure 2. Plot of d-sin2ψ for the PH-1a sample.

during deposition which increase linearly with the film thickness. These stresses are mostly located at the grain boundaries and arise from the “zipping” of grain boundaries. According to Hoffman, which was cited by Koch,10 “the interatomic forces at the grain boundaries tend to close any existing gap, with the result that the neighboring crystallites are strained in tension”. Weissmuller and Lemier34 also reported that grain boundary stresses in Pd-H tended to expand the crystal lattice. For the particular case of Pd deposited at a temperature of 60 °C, the T/Tm ratio (333 K/1823 K) equaled 0.18, so that Pd could be considered as a low-mobility metal, which was in agreement with the tensile nature of the initial stress in the coating. 4.2. Pd Microstructure Changes with Temperature. 4.2.1. Morphology Changes with Temperature. Pd layers (6 µm thick) were deposited on sintered R-Al2O3 porous supports and annealed at different temperatures (300, 400, 500, and 600 °C) for 48 h in H2. R-Al2O3 porous supports were used to avoid intermetallic diffusion at 500-600 °C. The morphology of the annealed Pd layers (samples Al2O3-3a/4a/5a/6a), studied by SEM, is shown in Figure 3a-d, respectively. No significant changes were noticed in the shape of Pd clusters for the samples annealed at 300 and 400 °C. Indeed, the clusters in both samples had similar size and sharp edges. Pd clusters in the 500 °C annealed sample did not show sharp edges, indicating that the Pd cluster sintering process started at a temperature between 400 and 500 °C. At 600 °C, Pd clusters were hardly discerned anymore; instead, a distinguished uniform Pd layer with Pd crystals as large as 5 µm were present. 4.2.2. Crystallite Growth and Microstrain Release with Temperature. Figure 4 shows the Williamson-Hall plot of sample PH-1a annealed for 1 h in He atmosphere at different temperatures (400, 500, and 600 °C). The experimental data obtained at 20 °C, “(20-He) A”, were also added for comparison purposes. At temperatures equal to or higher than 400 °C, plotting β cos(θ) as a function of sin(θ) led to a single straight line having a very low slope value (rms ) 3.2 × 10-4) and passing through the origin, indicating that microstrains, initially present in the Pd deposit, were released reaching a value of 0.03% at 400 °C and that crystallites grew very large (∼0.5 µm). Performing the microstrain-size separation at room temperature after heat-treatment, “(20-He) B”, led to the same results as those found at temperatures equal to or higher than 400 °C. Therefore, initial microstrains were irreversibly released after treatment in He atmosphere at 400 °C for 1 h. The study of Pd crystallite growth with X-ray diffraction was not possible since their size was already large (50-100 nm) at the fresh stage. However, crystallite growth occurred after microstrains were released.9 Therefore, Pd crystallite growth in electroless plated films started to occur at temperatures ranging between 350 and 400 °C. The hypothesis that Pd crystallites started to grow at T > 350-400 °C can be

substantiated by the fact that, as seen in Figure 3b, no significant structure modification occurred at 400 °C in H2 atmosphere. 4.3. Release of Intrinsic and Thermal Stresses with Temperature. Upon heating sample PH-1a, extrinsic thermal stresses were added to the already existing intrinsic stress. The total stress in sample PH-1a was measured at several temperatures (250, 400, 500, and 600 °C) in He atmosphere. Figure 5 shows the total stress to which the thin Pd film was subjected as a function of temperature. Initially, before any hightemperature treatment, the film was under a tensile stress of 104.7 MPa. At temperatures higher than 250 °C, the thin film was subjected to compressive stresses with a magnitude of -25 MPa at 250 and 400 °C and -13 MPa at 500 and 600 °C, as seen in Figure 5. When cooled to room temperature (25 °C), after the high-temperature treatment, the thin film was back under a tensile stress although the magnitude of the tensile stress was lower (41 MPa) than the value at the fresh state (104.7 MPa). The difference in the tensile stress before and after exposure to high temperatures indicated that the initial intrinsic stress was released. Thermal stress release started at a temperature between 250 and 400 °C since the film was under the same compressive stress at those two temperatures. Further stress release occurred at 500 °C since the film was under a less compressive stress than at 400 °C. The stress-temperature plot seen in Figure 5 was characteristic of the first high-temperature treatment of the fresh PH-1a sample. The stress-temperature plot during the temperature treatment of sample PH-1b is shown in Figure 6. During the heat treatment of sample PH-1b, the thermal stress was found to be, within the temperature range studied (60-400 °C), a linear function of the temperature as predicted by eq 1. The negative slope indicated that the previously heated thin Pd film expanded more than the PH substrate by 1.25 × 10-6 m/(m K). Since the thermal stress was a linear function of temperature, the elastic region for a Pd sample annealed at 600 °C was found to be 60-400 °C. To better understand the release of stresses that occurred during the heat treatment of sample PH-1a and PH-1b, sample PH-2a was pretreated in He atmosphere at 400 °C for 1 h to release the initial microstrains present in the film. The total stress of sample PH-2a was then measured at 60, 200, and 400 °C and also plotted in Figure 6. As seen in Figure 6, the stresstemperature function of PH-2a corresponded to the same stresstemperature function of sample PH-1b, which indicated that the heat-treatment for 1 h at 400 °C was sufficient to release the initial intrinsic tensile stress. Furthermore, the elastic region of a thin Pd film annealed at 400 °C was also found to be 60400 °C. Sample PH-3a, also pretreated at 400 °C for 1 h in He, was studied in order to elucidate the thermal stress release at temperatures higher than 400 °C. Figure 6 shows the thermal stress vs temperature behavior shown by sample PH-3a in the 350-500 °C temperature range. The stress at 350 and 400 °C was similar to the stress shown by samples PH-1b and PH-2a at the same temperatures. However, at 450 °C, the thermal stress started to relax, and at 500 °C, compressive stresses were totally released (-2.7MPa) as a result of plastic deformations. Hence, the elastic region of thin Pd films annealed at 400 °C was found to be 60-400 °C and the thermal stresses within the determined elastic region were given by eq 8 with T in °C.

σth ) -0.157T + 44.8

(8)

It is interesting to note that if a fresh composite Pd-PH

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006 8149

Figure 3. Pd-R-Al2O3 deposits annealed in H2 at (a) 300 (Al2O3-3a), (b) 400 (Al2O3-4a), (c) 500 (Al2O3-5a), and (d) 600 °C (Al2O3-6a) for 48 h (mag: 6000, scale marker: 5 µm).

Figure 4. Strain-size separation for PH-1a at different temperatures. The letters A and B correspond to the scan at room temperature before high temperatures (A) and after high temperatures exposure treatment (B).

Figure 5. Stress release in sample PH-1a as a function of temperature.

structure was taken and pretreated at 400 °C for 1 h in He, its initial microstrains and initial intrinsic stress were relaxed although its general microstructure was hardly affected (see

Figure 6. Stress in thin Pd films as a function of temperature: (]) PH1b, (0) PH-2a, and (4) PH-3a.

Figure 3b). The decrease of microstrains at elevated temperatures (400 °C) with minor crystallite growth was also observed earlier.16 4.4. H2 Stresses. The expansion of the Pd lattice, as a result of H2 absorption, constrained by the support leads to large compressive stresses that may, eventually, cause the membrane to fail. Therefore, it was of great interest to estimate the maximum compressive stresses to which a composite Pd membrane would be subjected during membrane characterization and/or membrane operation. Therefore, the fundamental relation between H2 stress and n(H/Pd) was investigated. The total stress of sample PH-1c was measured in a flowing 4% H2-balance N2 mixture atmosphere at a temperature of 60 °C and a total pressure of 1.5 bar (H2 partial pressure of 6 kPa). The total stress of sample PH-1c was then measured at 60, 100, 150, 200, 300, and 500 °C along with the lattice parameter of Pd. Figure 7 shows Pd lattice expansion (∆a/a0 in %) as a function of temperature in He atmosphere and in the 4%H2N2 balance atmosphere. As expected, at 60 °C, the absorption

8150

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006

Figure 9. Stresses arisen in the Pd thin film due to H2 uptake. Figure 7. Pd lattice expansion due to H2 uptake and temperature in sample PH-1c.

Figure 10. H2 stress as a function of H2 content n(H/Pd). Figure 8. Lattice increase due to interstitial H2 for sample PH-1c.

of H2 led to an expansion of the lattice in addition to the one caused by dilation. As the temperature was increased and the total pressure kept constant, H2 desorbed from the Pd film leading to the asymptotical approach of the lattice parameter to the lattice parameter values measured in He atmosphere. The Pd lattice expansion solely due to the contribution of H2 content ((aH - a0)/a0) was determined by subtracting the contribution of lattice dilation due to temperature from the measured lattice parameter. The H2 content n(H/Pd) was then determined making use of eq 4 with P equal to 45.6 torr (6 kPa) and T ranging from 60 to 500 °C. The values of (aH - a0)/a0 were then plotted as a function of n(H/Pd) as shown in Figure 8 and fitted with a straight line going through the origin. The Pd lattice parameter increased linearly with n(H/Pd) according to eq 9 with a different n(H/Pd) coefficient than the proportional constant reported by Eastman et al.,18 which equaled 0.19.

a - a0 ) 0.30n(H/Pd) a0

3

(9)

Indeed, in this case kH equaled to 0.3 and not 0.19 which was due to the fact that the film was adhered to the substrate and could only expand in the direction perpendicular to the surface.35 The total stress induced by the Pd lattice expansion due to H2 loading and thermal stresses is shown in Figure 9 for sample PH-1c. The absorption of H2 at 60 °C (n ) 0.017H/Pd) led to a 0.17% expansion in addition to the expansion due to the temperature. However, the large expansion of the Pd lattice was constrained by the support, which did not expand upon H2 introduction in the chamber and caused the switch from a 40 MPa tensile stress in He to a -60 MPa compressive stress in H2 as seen in Figure 9. As the temperature was increased, the

H2 desorbed and the compressive stress slowly decreased from -60 MPa at 60 °C to -10 MPa at temperatures higher than 200 °C, which corresponded to the stress values recorded in He atmosphere. The H2 stress component was derived by subtracting the elastic thermal stress given by eq 8 from the total measured stress (eq 5). The H2 content, n(H/Pd), was calculated using eq 4 with P equal to 45.6 torr and T ranging from 60 to 500 °C. Figure 10 shows the H2 stress component (σH) as a function of n(H/Pd). As expected, σH was a linear function of n(H/Pd) in the 60-400 °C temperature range. At 500 °C, eq 8 was no longer valid due to the stress relaxation by plastic deformations, and the 500 °C experimental data point fell outside the predicted stress line. Therefore, for any composite Pd-PH structure pretreated at 400 °C for 1 h, the H2 stress is given by eq 10

σH ) -5150n(H/Pd)

(10)

The total stress in the composite Pd structure is given by the sum of eqs 8 and 10

σtotal ) -0.157T - 5150n(H/Pd) + 44.8

(11)

where σtotal is in MPa, T is in °C, and n(H/Pd) is in mol H/mol Pd. Equation 11 is only valid within the elastic region of Pd, which was shown to be 60-400 °C. At temperatures above 400 °C, the elastic energy accumulated in stresses was released by plastic deformation. 4.5. Estimation of Total Stresses for 250 < T < 500 °C and 1 < PH2 < 5 bar. Knowing the fundamental principles and equations of thermal and H2 stresses, it was possible to estimate the total stress to which a composite Pd-PH membrane was subjected by using eq 11. The use of eq 11 implied that the composite Pd membrane was already preannealed in He at 400 °C for 1 h.

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006 8151 Table 2. Characteristics of Composite Pd Based Membranes Studied membrane

support (grade)

thickness (µm)

permeable surface (cm2)

permeance at 500°C (m3/m2 h bar0.5)

final selectivity (H2/He) at 500°C

ref

Pd-1a Pd-1b Pd-2 Pd-Cu

PSS (0.1 µm) Pd-1a PH (0.1 µm) PH (0.1 µm)

17 19 5.6 ∼10

23 23 120 120

20.6 23.5 39 30 at 450 °C

683 478 818 733

M3-a in Guazzone et al.26 M3-b in Guazzone et al.26 Ma-42 in Guazzone22 Guazzone et al.38

The total stress calculated at 250, 300, 400, and 500 °C for H2 pressures ranging between 1.5 and 5 bar (∆P ) 0.5-4 bar) is plotted in Figure 11, which shows that a composite Pd-PH membrane preannealed at 400 °C will always be under compressive stresses in H2 atmosphere. The calculated stress values at 500 °C are obviously an indication of what would be the stress without relaxation. Indeed, at temperatures higher than 400 °C, the elasticity region of Pd is exceeded and stresses relax leading to lower values than the predicted ones. However, experiments on samples PH-1a, PH-2a, PH-3a, and, especially, PH-1c suggested that at 500 °C composite Pd-PH structures were still under small compression stresses. The maximum stress to which these composite Pd membranes were subjected at 250 °C and 5 bar was estimated to be compressive with a magnitude of -265 MPa (see Figure 11). Since membranes are subjected to tensile stresses in He (or any inert gas) atmosphere and compressive stresses in H2 atmosphere, cycling between He and H2 atmosphere may lead to creep. However, this hypothesis was not tested due to the large number of cycles (usually thousands) that are needed to study creep in these systems. Furthermore, composite Pd membranes are meant to work under steady state conditions and would only experience a change in mechanical conditions during startup and shutdown procedures. 4.6. Relation of Microstrains and Stresses with Leak Formation in Composite Pd Membranes. Stresses were in several occasions thought to play an important role in leak formation in composite Pd membranes prepared by the electroless deposition method.36 Table 2 lists the characteristics of composite Pd membranes prepared on PM supports by the electroless deposition method. Figure 12 shows the He leak measured after 50-100 h in H2 atmosphere of membranes Pd1a/1b/2 and Pd-Cu at every temperature the membranes were exposed to. The He leak of all membranes remained relatively low and grew very slowly over time at temperatures equal to or lower than 400 °C. However, after 50-100 h (the time was not the same for all membranes) in H2 at 450 °C, a significant increase in He leak was measured for all membranes. After 50100 h (the time was not the same for all membranes) in H2 at 500 °C, an even larger increase in the He leak was observed.

The release of microstrains and intrinsic tensile stresses irreversibly occurred at 400 °C. Recent results in our laboratory showed that the He leak after around 1800 h at 400 °C in H2 atmosphere was around 5.32 × 10-4 m3/(m2 h bar), which corresponded to a selectivity (H2/He) of around 30 000. Since significant leaks only started to be observed at 400-450 °C, it could be concluded that the release of microstrains and initial intrinsic tensile stresses did not, or to a minor extent, lead to leak formation. Moreover, the experimental data collected for all membranes at 250 °C in He and H2 atmosphere suggested that holding a composite Pd membrane at 250 °C in H2 atmosphere did not lead to leakages. That is, even after excursion up to H2 pressures equal to 5 bar at 250 °C, where compressive stresses were the highest and equaled -260 MPa, no leaks were measured in the composite Pd membrane at 250 °C. Therefore, composite Pd membranes prepared on PM supports by the electroless method could hold compressive stresses as large as -260 MPa. Since composite Pd membranes could hold large stresses at 250 °C without developing leaks, the formation of leak at 450 °C was not related to the cracking of the Pd layer since the total stress was lower at 450 °C than at 250 °C. Also, it is important to consider the morphology of defects when studying the origin of leaks in composite electroless Pd membranes. Pd membranes that develop leaks slowly over time at temperatures equal to or higher than 500 °C have pinholes on their surface. Figure 13 shows the morphology of a Pd layer electrolessly deposited on a porous metal support heat treated at 500, 550, 600, and 650 °C. The higher the temperature, the higher the number and size of pinholes seen on the surface. Pinholes are detrimental for membrane selectivity since they penetrate the Pd permselective layer by an intricate network of grain boundaries and connected porosity. If leaks in composite Pd membranes were due to thermal and hydrogen stresses, the morphology of defects would be cracks. That is, if the total stress was higher than the adhesion force between the Pd layer and the support, cracks would appear on the surface creating large leaks. Cracking of the Pd layer may occur if the Pd layer undergoes the R to β transition. For instance, the hydride β phase has a minimum of 0.5 n(H/Pd) dissolved in the lattice at temperatures lower than 200 °C.

Figure 11. Total stress calculations for a Pd composite membrane as a function of pressure difference at 250, 300, 400, and 500 °C.

Figure 12. He leak as a function of temperature for membranes Pd-1a/ 1b/2 and Pd-Cu.

8152

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006

Figure 13. Pd surface morphology after annealing in H2 at (a) 500, (b) 550, (c) 600, and (d) 650 °C for 48 h (mag: 1500, scale marker: 20 µm).

composite Pd membranes prepared by the electroless method. Sintering usually occurs at a temperature ranging between 0.2 and 0.4 of the melting point (in K), and for the case of Pd, 400-450 °C lies within this range (melting point of Pd ) 1552 °C). It appears then that stress relaxation, grain growth, and pinhole formation are all the consequence of a single phenomenon: sintering. At high temperatures, grains grow and small cavities coalesce to form bigger cavities, which are the pinholes. The sintering of Pd clusters will be reported by Guazzone and Ma.37 5. Conclusions

Figure 14. Cracks observed on a Pd layer supported on porous stainless steel after the Pd layer was exposed at 2 bar of H2 at 180 °C.

Equation 11 predicts a total compressive stress of -2561 MPa, which is a very large stress and is well above the adhesion force of Pd to the support. Cracking was experimentally observed when the Pd layer undergoes the R to β transition. Figure 14 shows a Pd layer that unfortunately, due to a power loss, was exposed to H2 at a pressure of 2 bar. The temperature of the furnace started to decrease from 250 °C, and when it reached around 180 °C, the H2 flux doubled due to the cracking of the layer. In fact, as already stated, when the hydride β phase is not formed, composite Pd membranes on PM supports can hold compressive stresses up to a total of at least -260 MPa without cracking or peeling off. Nevertheless, leaks started to appear and grew at 450 °C, which corresponded to the same temperature at which significant changes in the microstructure of Pd and Pd crystallite growth were observed. Also, extrinsic (thermal + H2 stresses) stresses started to relax at 450 °C. Therefore, it appears that the stress release is not the main mechanism for leak development in

Freshly prepared composite Pd-PH structures were characterized by small Pd crystallites (50-100 nm), an initial tensile intrinsic stress of 104.7 MPa, and an initial microstrains value of 0.29%. Initial microstrains and initial intrinsic stresses were released by heat treatment in He at 400 °C for 1 h with no visible change in microstructure. After annealing at 400 °C, thermal stresses were found to be a linear function of temperature in the 60-400 °C elasticity region of Pd in agreement with the literature. At temperatures higher than 400 °C, thermal stress relaxation occurred with visible changes in microstructure. The Pd lattice parameter and H2 stresses were found to be linear functions of H2 loading n(H/Pd). Stress estimations for the pretreated composite Pd membranes at 400 °C showed that the Pd thin films were always under compressive stresses in H2 atmosphere. During membrane characterization, the maximum compressive stress was reached at 250 °C and a pressure of 5 bar and equaled -265 MPa. Leaks only formed at temperatures higher than 450 °C and under experimental conditions at which the stresses were minimum in the Pd-PH samples. From the apparent minor role of stresses in the leak formation revealed by the detailed stress analysis, it could be concluded that the leak formation was mainly caused by the formation of pinholes.

Ind. Eng. Chem. Res., Vol. 45, No. 24, 2006 8153

Acknowledgment The authors would like to thank the financial support provided by Shell International Exploration and Production, Inc. and Shell Hydrogen. This research was also sponsored in part by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Freedom Car and Vehicle Technologies, as part of the High Temperature Materials Laboratory User Program, Oak Ridge National Laboratory, managed by UTBattelle, LLC, for the U.S. Department of Energy under contract number DE-AC05-00OR22725. Nomenclature kH ) proportionality constant, no units K(T) ) Sievert’s constant, Pa0.5 n(H/Pd) ) hydrogen loading in the Pd sample, mol H/mol Pd PH2 hydrogen pressure, torr or kPa or bar L ) coefficient of thermal expansion, m/(m K) e ) approximate upper limit for maximum distortion, no units E ) Young modulus of the film (Pd), Pa T ) absolute temperature, K t ) Pd crystallite size, Å (in calculations) Greek Symbols β ) peak breath, radians  ) root-mean-square strain (rms), no unit Η ) hydrogen strain λCu ) Cu X-ray wavelength, 1.45046 Å σth ) thermal stresses, Pa σH ) hydrogen stresses, Pa ν ) Poisson ratio of the film (Pd) Literature Cited (1) Lin, Y.-M.; Rei, M.-H. Study on the hydrogen production from methanol steam reforming in supported palladium membrane reactor. Catal. Today 2001, 67 (1-3), 77. (2) Wieland, I. S.; Melin, I. T.; Lamm, I. A. Membrane reactors for hydrogen production. Chem. Eng. Sci. 2002, 57 (9), 1571. (3) Barbieri, G.; Violante, V.; Di Maio, F. P.; Criscuoli, A.; Drioli, E. Methane Steam Reforming Analysis in a Palladium-Based Catalytic Membrane Reactor. Ind. Eng. Chem. Res. 1997, 36 (8), 3369-3374. (4) Matzakos, A. N.; Wellington, S. L.; Mikus, T.; Ward, J. M. Integrated flameless distributed combustion/steam reforming membrane reactor for hydrogen production and use thereof in zero emissions hybrid power system. U.S. Patent 6,821,501, 2004. (5) Wellington, S. L.; Matzakos, A. N.; Mikus, T.; Ward, J. M. Integrated flameless distributed combustion/membrane steam reforming reactor and zero emissions hybrid power system; US 2003/0068260 A1, April 10, 2003. (6) Jun, C.-S.; Lee, K.-H. Palladium and palladium alloy composite membranes prepared by metal-organic chemical vapor deposition method (cold-wall). J. Membr. Sci. 2000, 176 (1), 121. (7) Su, C.; Jin, T.; Kuraoka, K.; Matsumura, Y.; Yazawa, T. Thin Palladium Film Supported on SiO2-Modified Porous Stainless Steel for a High-Hydrogen-Flux Membrane. Ind. Eng. Chem. Res. 2005, 44 (9), 30533058. (8) Tong, J.; Suda, H.; Haraya, K.; Matsumura, Y. A novel method for the preparation of thin dense Pd membrane on macroporous stainless steel tube filter. J. Membr. Sci. 2005, 260 (1-2), 10. (9) Tschope, A.; Birringer, R. Thermodynamics of nanocrystalline platinum. Acta Metall. Mater. 1993, 41 (9), 2791. (10) Koch, R. The intrinsic stress of polycrystalline and epitaxial thin metal films. J. Phys.: Condens. Matter 1994, 6 (45), 9519-9550. (11) Rajamani, A.; Sheldon, B. W.; Chason, E.; Bower, A. F. Intrinsic tensile stress and grain boundary formation during Volmer-Weber film growth. Appl. Phys. Lett. 2002, 81 (7), 1204-1206. (12) Floro, J. A.; Hearne, S. J.; Hunter, J. A.; Kotula, P.; Chason, E.; Seel, S. C.; Thompson, C. V. The dynamic competition between stress generation and relaxation mechanisms during coalescence of VolmerWeber thin films. J. Appl. Phys. 2001, 89 (9), 4886-4897. (13) Murakami, M. Deformation in thin films by thermal strain. J. Vac. Sci. Technol. A 1991, 9 (4), 2469-2476.

(14) Vook, R. W.; Witt, F. Thermally induced strains in evaporated films. J. Appl. Phys. 1965, 36 (7), 2169-2171. (15) Witt, F.; Vook, R. W. Thermally induced strains in cubic metal films. J. Appl. Phys. 1968, 39 (6), 2773-2776. (16) Reimann, K.; Wurschum, R. Distribution of internal strains in nanocrystalline Pd studied by X-ray diffraction. J. Appl. Phys. 1997, 81 (11), 7186-7192. (17) Baranowski, B.; Majchrzak, S.; Flanagan, T. B. The volume increase of fcc metals and alloys due to interstitial hydrogen over a wide range of hydrogen contents. J. Phys. F: Metal Phys. 1971, 1 (3), 258-261. (18) Eastman, J. A.; Thompson, L. J.; Kestel, B. J. Narrowing of the palladium-hydrogen miscibility gap in nanocrystalline palladium. Phys. ReV. B 1993, 48 (1), 84-92. (19) Gillespie, L. J.; Galstaun, L. S. The palladium-hydrogen equilibrium and new palladium hydrides. J. Am. Chem. Soc. 1936, 58, 25652573. (20) Gillespie, L. J.; Hall, F. P. The palladium-hydrogen equilibrium and palladium hydride. J. Am. Chem. Soc. 1926, 48, 1207-1219. (21) Wicke, E.; Nernst, G. H. Phase diagram and thermodynamic behavior of the system Pd/H2 and Pd/D2 at normal temperatures; H/D separations effects. Ber. Bunsen-Ges. Phys. Chem. 1964, 68, 224-235. (22) Guazzone, F. Engineering of substrate surface for the synthesis of ultrathin composite Pd and Pd-Cu membranes for H2 separation. PhD dissertation, Worcester Polytechnic Institute, Worcester, 2006. (23) Guazzone, F.; Ayturk, M. E.; Ma, Y. H. Effect of intermetallic diffusion barrier on the stability of composite Pd/PSS membranes at high temperatures. Proceedings of the 21st International Pittsburgh Coal Conference, Osaka, Japan, September 13-17, 2004. (24) Ma, Y. H.; Akis, B. C.; Ayturk, M. E.; Guazzone, F.; Engwall, E. E.; Mardilovich, I. P. Characterization of Intermetallic Diffusion Barrier and Alloy Formation for Pd/Cu and Pd/Ag Porous Stainless Steel Composite Membranes. Ind. Eng. Chem. Res. 2004, 43 (12), 2936-2945. (25) Ma, Y. H.; Mardilovich, P. P.; She, Y. Hydrogen gas-extraction module and method of fabrication. U.S. Patent 6,152,987, 2000. (26) Guazzone, F.; Engwall, E. E.; Ma, Y. H. Effects of surface activity, defects and mass trabsfer on hydrogen permeance and n-value in composite palladium-porous stainless steel membranes. Catal. Today, 2006, 118 (12), 24-31. (27) Mardilovich, P. P.; She, Y.; Ma, Y. H.; Rei, M.-H. Defect-Free Palladium Membranes on Porous Stainless Steel Support. AIChE J. 1998, 44 (2), 310. (28) Williamson, G. K.; Hall, W. H. X-ray line broadening from filed aluminum and wolfram. Acta Metall. 1953, 1 (1), 22-31. (29) Cullity, B. D.; Stock, S. R. Elements of X-ray diffraction, 3rd ed.; Prentice Hall: Upper Saddle River, NJ, 2001; p 664. (30) Klug, H. P.; Alexander, L. E. (joint author) X-ray diffraction procedures for polycrystalline and amorphous materials; Wiley: New York, 1954; p 716. (31) Bogdanov, S. V.; Moroz, E. M.; Tsylbulya, S. V. Investigation of substructure of catalysts by X-ray diffractions methods. Kinet. Catal. 1984, 25 (5), 1028-1032. (32) Sanders, P. G.; Whitney, A. B.; Weertman, J. R.; Valiev, R. Z.; Siegel, R. W. Residual stress, strain and faults in nanocrystalline palladium and copper. Mater. Sci. Eng., A 1995, 204 (1-2), 7-11. (33) Weissmuller, J.; Loffler, J.; Kleber, M. Atomic structure of nanocrystalline metals studied by diffraction techniques and EXAFS. Nanostruct. Mater. 1995, 6 (1-4), 105. (34) Weissmuller, J.; Lemier, C. Lattice constants of solid solution microstructures: the case of nanocrystalline Pd-H. Phys. ReV. Lett. 1999, 82 (1), 213-216. (35) Zabel, H.; Hjorvarsson, B. Hydrogen in thin films and multilayers. In Progress in hydrogen treatment of materials; Goltsov, V. A., Ed.; Kassiopeya Ltd., ITA NRNFM, DonSTU: Donetsk, 2001; pp 119-146. (36) Tosti, S.; Bettinali, L.; Castelli, S.; Sarto, F.; Scaglione, S.; Violante, V. Sputtered, electroless, and rolled palladium-ceramic membranes. J. Membr. Sci. 2002, 196 (2), 241. (37) Guazzone, F.; Ma, Y. H. Leak formation in composite Pd membranes prepared by the electroless deposition method. 2006, under preparation. (38) Guazzone, F.; Engwall, E. E.; Mardilovich, I. P.; Ma, Y. H. Hydrogen permeation and stability of a composite PdCu membrane. Abstract of papers of the 230th ACS national meeting, Washington, DC, August 28-September 1, 2005.

ReceiVed for reView June 13, 2006 ReVised manuscript receiVed August 29, 2006 Accepted September 18, 2006 IE060756S