Solid State Ionics 176 (2005) 955 – 963 www.elsevier.com/locate/ssi
Polyphosphate composite: conductivity and NMR studies S. Haufea,*, D. Prochnowb, D. Schneiderb, O. Geierb, D. Freudeb,*, U. Stimminga,* a
Department of Physics E19, Technische Universita¨t Mu¨nchen, James-Franck Str. 1, 85748 Garching, Germany b Abteilung Grenzfla¨chenphysik, Universita¨t Leipzig, Linne´str. 5, 04103 Leipzig, Germany Received 28 October 2003; received in revised form 28 October 2004; accepted 6 December 2004
Abstract A polyphosphate composite [NH4PO3]6[(NH4)2SiP4O13] with high ionic conductivity was prepared and characterized by chemical analysis, X-ray diffraction, thermal gravimetry, impedance spectroscopy, and different NMR techniques. Experiments were carried out in dry and humid atmospheres and in temperatures ranging between 20 8C and 300 8C. During initial heating (activation) of the composite to 300 8C, a weight loss of 7% due to release of ammonia occurs resulting in a material of the formal composition of [HPO3]3[NH4PO3]3[(NH4)2SiP4O13]. This material is thermally stable between room temperature and 300 8C. The conductivity was found to be remarkably high (0.1 S cm1) in a water-rich environment. Mainly ammonium species could be found in the 1H MAS NMR spectra of the non-activated composite, whereas during the activation process another signal due to bridging hydrogen increases. Temperature-dependent 2D and 1D exchange NMR spectroscopy, PFG (pulsed field gradient) NMR, and SFG (stray field gradient) NMR measurements were performed in order to compare diffusion coefficients and exchange rates with dc conductivities. The small differences between experimentally obtained conductivities and those determined from the measured self-diffusion coefficients by means of the Nernst–Einstein equation hints to an ammonium vehicle as charge carrier, but can be also explained by H+ conductivity. The slow NMR exchange rates between hydroxyl groups and ammonium ions exclude proton conductivity via hydroxyl groups. D 2004 Elsevier B.V. All rights reserved. Keywords: Proton conductivity; Inorganic composite; Impedance spectroscopy; MAS NMR; SFG NMR; PFG NMR
1. Introduction Oxygen ion conductors and proton conductors for fuel cells in the temperature range 200–600 8C are still in a developmental stage [1]. Inorganic material with water included [2–4], without water [5–7], salts of inorganic oxygen acids [8–12] and oxide ceramics [13,14] have been intensively investigated. Polyphosphate glasses [15], aluminum polyphosphate [16] and ammonium polyphosphate [17–19] were rarely considered, even though these materials exhibit high conductivities in the middle temperature range (0.1 S cm1 at 300 8C), since the nature and dynamics of the species responsible for the charge transport are not yet clear.
In this study we use a polyphosphate composite consisting of NH4PO3 and (NH4)2SiP4O13 [17–19], which exhibits better properties compared to the single components, if we consider conductivity and thermal stability as well. The material has been characterized by chemical analysis, X-ray diffraction, thermal gravimetry [20], 1H MAS NMR, and 31 P MAS NMR spectroscopy [21]. The main point of this paper is to obtain a model of the conductivity in the composite material by comparing the results of impedance spectroscopy [20] with those of NMR exchange spectroscopy, and NMR diffusometry [21].
2. Experimental * Corresponding authors. D. Freude is to be contacted at Fax: +49 341 9739349. S. Haufe, Present address: Sartorius AG, August-Spindler-Str. 11, 37079 Gfttingen, Germany. E-mail addresses:
[email protected] (S. Haufe)8
[email protected] (D. Freude)8
[email protected] (U. Stimming). 0167-2738/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2004.12.004
2.1. Preparation, X-ray, elemental and TG characterization The composite was prepared by procedures proposed by Kenyo and Inaba [17] (*) and Averbuch-Pouchot and Durif
956
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
[22] (**) in the following way: NH4 H2 PO4 þ ðNH2 Þ2 COYNH4 PO3 IYNH4 PO3 II 4NH4 PO3 þ SiO2 YðNH4 Þ2 SiP4 O13
ð4Þ ð44Þ
In the first step, Eq. (*), 11.502 g (0.1 mol) ammoniumdi-hydrogen–phosphate and 6.006 g (0.1 mol) urea were ground together and brought to reaction for 2 h at 200 8C in NH3 atmosphere in an Al2O3 crucible. The melting cake was dissolved in hot water and filtered. The NH4PO3 was precipitated as a white powder by adding methanol and is denoted as APP I. After drying in vacuum and tempering (24 h) in NH3 atmosphere at 280 8C, the phase changed into the water insoluble, long chained modification NH4PO3 II (APP II). In the second reaction step, Eq. (**), 7.807 g (0.08 mol) of the fine ground ammonium-polyphosphate II were combined with 0.500 g (0.008 mol) silicon dioxide for 30 min at 440 8C in NH3 atmosphere. After having been tempered for 24 h at 250 8C in NH3 atmosphere, the composite was finely ground. The obtained material is denoted as ASiPP and has a composition [NH4PO3]6 [(NH4)2SiP4O13] (see Section 3.1). In order to determine the qualitative composition and to examine the phase purity, the material was analyzed by Xray diffraction (diffractometer ID3000, Seifert GmbH) with a Cu-Ka source, Ge-monochromator, and a scintillation counter. The measurements were performed at room temperature with an anode voltage of 40 kV and an anode current of 30 mA. The quantitative composition of the material was determined with an elemental analyzer (Vario EL, Elementar Analysensysteme GmbH). The sample was combusted in an oxygen atmosphere. Nitrogen and hydrogen were measured by means of a thermoconductor. Phosphorous and silicon were determined photometrical with a UV–VIS spectral photometer (UV-160, Shimadzu GmbH) after a humid digestion [23]. For the thermogravimetric measurements (TGA7, Perkin-Elmer), the samples were heated in argon with a heating rate of 10 8C min1.
circuit for the simulation consisting of RV, Q V, R E and Q E was a serial connection of V and E boxes, consisting of a parallel connection of RVtQ V and R EtQ E, respectively. The specific conductivity r of the material was determined from the bulk resistance RV, the thickness d, and the sample area A: r¼
d : RV A
ð1Þ
2.3. NMR experiments Powdered samples of the composite material were used for the NMR spectroscopy. Continuous flow experiments could not be performed; therefore, the composite material was pretreated at 300 8C for 12 h in a dry atmosphere of hydrogen gas. A portion of the material was kept for an additional 12 h in hydrogen at 300 8C with a partial water pressure of 4.2 kPa, in order to simulate the moist flow experiment. 31 P MAS NMR measurements were performed on the Bruker spectrometers MSL 500 (external field B 0=11.74 T and Larmor frequency m L=202.4 MHz) and Avance 750 (B 0=17.25 T, m L=303.6 MHz) with MAS frequencies m rot=4–30 kHz. The spectrometer MSL 300 with a MAS frequency of 2–4 kHz was used for 1H MAS NMR experiments. The laser heated probe was calibrated by means of Pb(NO3)2 [24]. 1H PFG NMR self-diffusion measurements were carried out by means of the home-built spectrometer FEGRIS 400 [25] operating at 400 MHz. Pulsed field gradient (PFG) of 30 T m1 amplitude and 800 As duration, and the rf pulse sequence for a stimulated echo with observation times up to 30 ms were applied. 1H SFG NMR self-diffusion measurements were performed in the stray field of the 7 T magnet (gradient 50 T m1) of the spectrometer MSL 300 with a Larmor frequency m L=117 MHz.
3. Results and discussion
2.2. Conductivity measurements
3.1. Structural and chemical characterization
Conductivity measurements were carried out on powder pellets (a10 mm) with sputtered as well as with porous platinum electrodes in a temperature region between 50 and 300 8C in dry hydrogen, humid hydrogen (4.2 kPa H2O), dry oxygen, and dry argon by means of an impedance analyzer (AutoLab PGSTAT30-FRA2, Eco Chemie BV). A detailed description of the employed test cell can be found in [20]. Impedance spectra were recorded in a frequency range from 1 MHz to 0.1 Hz (10 points per decade) with an voltage amplitude of 50 mV in dry and 10 mV in humid atmosphere. Spectra were simulated by means of Zview software (Scribner Associates, Version 2.1). The equivalent
Fig. 1 shows the X-ray diffraction pattern of a powder sample. The pattern can be explained by the presence of only three phases: ammonium-silicon-polyphosphate and ammonium-polyphosphate in the modifications I and II, as found in Joint Committee on Powder Diffraction Standards JCPDS33-76, JCPDS22-61, JCPDS22-62. There is no hint to amorphous parts in the X-ray pattern. The mass fractions in wt.% were determined experimentally by elemental analysis, and the values expected on the basis of the reaction stoichiometry are shown in parentheses. We find values for N of 11.5 (11.5), H of 3.7 (3.3), P of 29.6 (31.7), Si of 2.9 (2.9) and O of 52.3 (50.7). The experimental value for
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
relative intensity
100
3.2. Conductivity
80 60 40 20 0
5
NH4PO3 I NH4PO3 II (NH4)2SiP4O13 10
15
20
25
30
2θ / °
35
40
45
50
55
Fig. 1. X-ray diffractogram of the composite (top) and references (bottom) for ammonium-silicon-polyphosphate (thick solid lines, after JCPDS33-76) und ammonium-polyphosphate, modification I (thin solid line, after JCPDS22-61) and modification II (thin dashed line, after JCPDS22-62).
100 98
Fig. 3 shows Arrhenius plots of the electrical conductivity (log(rT) vs. 1/T) which was investigated in various gas atmospheres. The measurements were made after the first heating cycle (activation), i.e. in the thermally stable phase of the material. The conductivity increases in dry hydrogen from 6.9108 S cm1 at 47 8C up to 1.7102 S cm1 at 308 8C. These results are in agreement with those of Kenyo [17] and Cappadonia et al. [19] who measured at 305 8C values of 1.0102 S cm1 and 1.4102 S cm1 , respectively. In measurements in a dry atmosphere, only small deviations in the conductivity were observed. Water, however, has an unambiguous impact on the material: in a humid atmosphere the conductivity of the material is almost three orders of magnitude higher than in a dry environment at 100 8C, and still more than half an order of magnitude higher at 308 8C. The change from the humid to the dry atmosphere and vice versa corresponds to a reversible modification of the conductivity that is caused by a small water up-take of the composite. The water up-take of the composite is about 1–2 wt.% by a partial water pressure of 4.2 kPa at 300 8C. The Arrhenius plot in Fig. 3 can be described by linear functions only if two temperature regions, one below 150 8C and one above, are considered separately. Table 1 presents the apparent activation energies for the two regions. In Section 3.5, we will present activation energies of the 1H diffusion obtained by SFG NMR in the temperature region TN150 8C. They are 0.47 eV for activation in the dry H2 atmosphere in good agreement with the value of 0.48 eV in Table 1. The SFG NMR value for the composite activated in T/ K 2
first cycle second cycle
96
+ H2O
−3
−5
DC H2 dry H2, dry O2, dry Ar, dry H2, 4.2 kPa H2O 1.6
150
200
250
300
T/ °C Fig. 2. Mass loss of the composite at a heating rate of 10 8C min1 in a dry H2 atmosphere.
− H2O
−2
92
100
400
−1
94
50
500
0
−4
90
600
1
lg (σT / S K cm-1)
oxygen results from the total mass balance. The small deviations of the experimental values from the stoichiometric values show that the composition of the composite corresponds almost exactly to the chosen reaction stoichiometry 6NH4PO3/(NH4)2SiP4O13. Fig. 2 show the loss of ammonia above 200 8C during the first heating-up. With a heating rate of 10 8C min1, a mass loss of 7% was observed at 300 8C. The second heating occurs in a thermally stable phase without any further mass change. Thermal hysteresis observed in the second cycle is due to the apparatus. The mass loss of 7% corresponds to about half the ammonium ions that react to ammonia plus H+, and half the ammonium ions remain in the composite after the activation. This gives a material of the formal composition of [HPO3]3[NH4PO3]3[(NH4)2SiP4O13] in agreement with the elemental analysis. The latter allows distinguishing between released NH3 and other products. By keeping the temperature for 6 h, it was observed that conductivity and mass are stable up to 200 8C, but a loss of about 25% in conductivity and of 2–3 wt.% were observed in the range 284–332 8C [20].
weight-%
957
2.0
2.4
2.8
3.2
1000 T-1/K-1 Fig. 3. Arrhenius plot of rT determined in various atmospheres by AC conductivity measurements. Only the three values denoted by + are determined by DC measurements on a H2–H2-celle, see Fig. 4.
958
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
Table 1 Apparent activation energies obtained from conductivity measurements Gas
E a /eV Tb150 8C
TN150 8C
Ar O2 H2 H2/4.2 kPa H2O
0.89 0.89 1.00 0.15
0.48 0.47 0.48 0.10
a humid H2 atmosphere is 0.27 eV and does not agree with the value of 0.10 eV in Table 1. DC measurements were performed on symmetrical cells [26,27], in order to distinguish between hydrogen, oxygen, and electronic conductivity. Fig. 4 shows linear current– voltage characteristics of the H2–H2-cell. The resistance decreases with the increasing temperature. Three conductivity values are obtained from Fig. 4 using the slopes of the three curves in Eq. (1). The three values are presented in Fig. 3 together with the data obtained by impedance spectroscopy. The good agreement indicates that the conductivity is determined by the ohmic resistance of the electrolyte, and that the impact of the interface processes can be neglected. DC measurements on an O2–O2-cell allow observing current densities which are five to six orders of magnitude lower than those measured in hydrogen [20]. In comparison to the H2–H2-cell, the conductivities measured in the O2–O2-cell decrease as follows: 0.8106 at 100 8C, 0.9105 at 200 8C and 1.8105 at 300 8C. The charge transport can be carried out by oxygen ions, as well as by electrons in the oxygen cell, however, the amount of electrons as charge carriers should be identical for oxygen and hydrogen cells. Consequently, the dramatically lower conductivity in the oxygen cell indicates that electron conductivity in the electrolyte can be neglected as well as in the hydrogen cell. In conclusion, proton species like H+, NH4+, or H3O+ are responsible for the charge transport in the examined composite. 3.3.
31
tively, to adjoining phosphorous atoms via oxygen. Fig. 5 shows 31P MAS NMR signals at 9.4 ppm and 22.6 ppm due to the Q1 and Q2 units. An additional signal at about 0 ppm can be explained by mono-phosphate units (Q0), indicating that the APP II sample still contains an amount of the starting material. Grimmer [28] gives a chemical shift of about 0 ppm for NH4H2PO4. The mean number n of phosphorous atoms (or Q units) in a linear polyphosphate chain can be determined from the ratio of signal intensities I(Q1) and I(Q2) arising from the terminal and middle units, respectively: I Q1 þ I Q2 : ð2Þ n¼2 I Q1 This gives a mean chain length of 340 Q units for APP II. The 31P MAS NMR spectrum of ASiPP shows in addition to the signals of Q1 and Q2 units two more signals at d=30.0 and 35.4 ppm, which can be assigned to Q3 units. The large intensity of this signal shows a strong branching of the phosphate chains in ASiPP. Fig. 5 presents a numeric signal addition of the two basic compounds, ASiPP and APP II, and compares it with the experimentally obtained signal of the non-activated composite material. A good agreement can be observed. Another comparison of the 31P MAS NMR spectra of nonactivated with activated composite materials shows a decrease of Q1 units and an apparent equivalent increase of Q3 units upon activation of the composite, however, Q0
ASiPP 1
Q Q0
Q3
Q2
2 Q1 Q
APP II
0
-40
ASiPP + APP II
P MAS NMR studies
Polyphosphates consist of Q1 and Q2 units, which are phosphorous atoms single- and 2-fold coordinated, respec-
activated complex
I / mA cm-2
50 T = 100 °C T = 200 °C T = 300 °C
40
non-activated complex
30 20
0
-100
/ppm 10 0 0
100
200
300
400
500
U / mV Fig. 4. DC measurements on a H2–H2-celle.
600
Fig. 5. 31P MAS NMR spectra of the starting materials ASiPP and APP in the inlet. The spectrum on the top presents a numeric addition of the experimentally obtained spectra of ASiPP and APP II and can be compared with the experimentally obtained signal of the non-activated composite material on the bottom. The signal of the composite activated in a dry H2 atmosphere is given in the middle.
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
assigning the increasing signal to Q3 units is doubtful. Also Q2(H1) units, which represent a protonated Q2 unit, give rise to a signal in this range [29]. Activation means deammoniation and the exiting ammonia molecule leaves a proton that combines with a phosphate oxygen to form a hydroxyl group. Therefore, the signals at 30 and 35 ppm can be assigned to Q3 units and Q2(H1) units as well. If we assign all signals in this range to Q2(H1) units and add it to the intensity of the Q2 units giving a signal at 23 ppm, we obtain a mean chain length of 6 Q units for the non-activated and 330 Q units for the activated composite. It can be concluded beyond a doubt from this 31 P MAS NMR study that after activation there exists a significant amount of either Q3 or Q2(H1) units, which means that the activated composite cannot be considered a catena-phosphate. Q3 units hint to branching and Q2(H1) units hint to hydrogen bridging between chains, while both units should result in an increased thermal stability of the composite compared to ammonium polyphosphate. 3.4. 1H MAS NMR studies The 1H MAS NMR spectrum of APP II is shown in Fig. 6a. A signal with a chemical shift of 7.2 ppm is observed, and can be assigned to the ammonium ions in the polyphosphate, since a similar shift has been observed for ammonium ions in zeolites [30]. The spectrum of ASiPP (Fig. 6b) shows an additional signal at 9.0 ppm which we attribute to protons in bridging positions. The spectrum of the non-activated composite (Fig. 6c) consists of only one signal at 7.3 ppm, however, after activation at 580 K and cooling down to room temperature, a second signal appears at 10.5 ppm (Fig. 6d). The signals at 9.0 und 10.5 ppm can be explained by hydroxyl protons with hydrogen bridges to other oxygen atoms. Hydrogen nuclei in bridging positions give rise to chemical shifts in the range 5–19 ppm [31]. For the starting material ASiPP, the existence of OH groups is connected to
(d) (c)
(b) (a) 16
14
12
10
8
6
4
2
/ppm Fig. 6. 1H MAS NMR spectra of APP (a), ASiPP (b), the non-activated composite (c), and the in dry H2 activated composite (d) at B 0=11.7 T and room temperature.
959
the relatively high concentration of Q0 units. For the activated composite material, the deammoniation process creates OH groups. Ammonia has a mass fraction of 13.9 % in the non-activated composite material. The thermogravimetrically determined mass loss is 7%, corresponding to about half the ammonium ions that react to ammonia plus H+, and half the ammonium ions remain in the composite after the activation. The H+-cation becomes attached to an oxygen atom of a Q2 unit building a Q2(H1) unit with a hydrogen bridge to another oxygen atom located in the same chain or a neighborhood chain. From the chemical shift of 10.5 ppm the distance r(OHd d d O)c0.17 nm can be determined by means of an equation given by Brunner and Sternberg [31]: rðOH: : : OÞ=nm ¼
4:65 : d=ppm þ 17:4
ð3Þ
The crystallographic structure of the activated composite is unknown. Therefore, some distances were determined by means of the CERIUS program using the X-ray data of the starting materials. For ASiPP, the minimum (terminal) oxygen–oxygen distances are 0.37 nm (intra-chain) and 0.35 nm (inter-chain). For APP, the values are 0.426 nm (intra-chain) and 0.376 nm (inter-chain). Intra-chain and inter-chain denote minimum distances for two atoms in one chain and two neighborhood chains. It is remarkable that inter-chain distances are shorter than intra-chain distances. The subtraction of 0.1 nm (O–H distance in the hydroxyl group) from the oxygen–oxygen distance gives a minimum value r(OHd d d O)c0.25 nm (for ASiPP inter-chain). This value, however, is derived from the structure of the starting material. In the activated composite, we expect smaller distances between the chains. Therefore, the above calculated distance of 0.25 nm seems to be comparable with the value of 0.17 nm that was determined from the chemical shift by Eq. (3). Temperature-dependent 1H MAS NMR studies of the activated composite show a significant line broadening above 77 8C, a merging of the two signals at about 157 8C and a line narrowing of the merged signal to a line width (full width at half maximum) of about 0.5 ppm at about 307 8C [21]. Chemical exchange between ammonium protons and hydroxyl protons causes this effect which allows the estimation of an exchange rate of about 350 s1 at 157 8C [21]. A more precise determination of the exchange rate between species giving rise to the signals at 7.0 ppm and 10.5 ppm in the low temperature spectra was performed by exchange spectroscopy (EXSY). The NOESY pulse sequence and a one-dimensional 1H MAS NMR technique were applied [32] with two offsets, series of 20 mixing times and temperature steps of 10 K between room temperature and 175 8C. Exchange rates increase from k=50 s1 (room temperature) to 3000 s1 (175 8C). The apparent activation energy of 0.84 eV can be obtained from the Arrhenius plot in Fig. 7, if only the temperature range 108–175 8C is
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
exchange rate k/s-1
960
apparent activation energy of 0.47 eV for the in dry H2 activated composite and 0.27 eV (SFG NMR) or 0.31 eV (PFG NMR) for the in humid H2 activated composite. The energies are discussed in the next section.
1000
3.6. Combination of dynamic parameters The Nernst–Einstein law gives the DC-conductivity (r dc) as a function of the concentration (C) of the charge carrier, the charge of one particle (e), the self-diffusion coefficient (D) and the temperature (T):
100
2.2
2.6
3.0
rdc ¼
3.4
1000 T-1 / K-1 Fig. 7. Results of 1D EXSY NMR measurements of the composite activated in a dry H2 atmosphere. The temperature dependence of the exchange rate is given for the two signals at 7.3 ppm and 10.5 ppm (dotted line).
considered. The strong temperature dependence in this range is a hint that spin diffusion can be excluded as the cause for the observed exchange between the two signals. The exchange rate minimum in the temperature range between 108 8C and room temperature, however, seems to be an NMR artifact. It can be explained by a combination of spin diffusion and exchange. Thus, exchange rates cannot be determined in the temperature region below 108 8C. 3.5. 1H NMR diffusion studies PFG NMR and SFG NMR techniques have been used for the measurements of the self-diffusion coefficient. The reason for the application of the less sensitive SFG NMR was that the upper temperature limit for this probe is 600 8C, whereas the PFG NMR probe works only up to 200 8C. For the composite activated in a dry H2 atmosphere, we could not measure the self-diffusion coefficient by NMR at temperatures slightly above room temperature, since the 1 H mobility was too low. Fig. 8 presents self-diffusion coefficients of the composite which was activated in a humid or in a dry atmosphere. The Arrhenius plot gives the PFG humid SFG humid SFG dry
D / m2s-1
10-9 10-10
e2 CD ; kB T
ð4Þ
were k B denotes the Boltzmann constant [33]. The onedimensional Einstein equation (for three-dimensional carrier substitute the factor 2 by 6) gives D as a function of the jump length d and the mean time s between two jumps: D¼
d2 : 2s
ð5Þ
The concentration of the particles which carry one elementary charge can be determined in the following way. Unit cells consist of 4NH4PO3 for APP and 1(NH4)2SiP4O13 for ASiPP. Values for the volumes were taken from X-ray data and are 0.33353 nm3 and 0.55397 nm3 for APP and ASiPP, respectively. The stoichiometry of 6NH4PO3/(NH4)2SiP4O13 corresponds to 1.5 unit cells APP and 1 unit cell ASiPP. 50% deammoniation gives 5 residual ammonium ions in a total volume of 1.50.33353 nm3+10.55397 nm3=1.05426 nm3. Assumed we have ammonium as charge carrier, we obtain a charge carrier concentration C=4.741021 cm3. To estimate the jump length, we use the mean distance of two phosphorous atoms in the structure of APP and ASiPP: d P–P=0.288 nm. After determining C and d, we have all the parameters for the Eqs. (4) and (5), in order to determine the conductivity from the experimentally obtained values of the self-diffusion coefficient or the exchange rates, which are inversely proportional to the mean time between two jumps. Table 2 shows the dynamic NMR data and Table 3 compares the conductivities which were experimentally obtained by impedance spectroscopy (IS) with those which were determined by means of the Nernst–Einstein law from the PFG, SFG, and EXSY NMR data in Table 2. Fig. 9
10-11
Table 2 Experimentally obtained exchange rates and self-diffusion coefficients
10-12
Composite material
D PFG/109 cm2 s1
D SFG/107 cm2 s1
k EXSY/s1
10-13
Activated in dry H2 Activated in humid H2
– 1.6–154.0c
0.3–2.2a 2.2–44.0d
50.2–628.0b –
1.5
1.7
1.9
2.1
2.3
2.5
2.7
2.9
1000 T-1/K-1 Fig. 8. Self-diffusion coefficients determined by SFG NMR and PFG NMR for the composite activated in a dry or humid H2 atmosphere.
a b c d
Temperature range 200–300 8C. Temperature range 77–150 8C. Temperature range 100–200 8C. Temperature range 200–330 8C.
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
961
Table 3 Comparison of the conductivities which were experimentally obtained by impedance spectroscopy (IS) with those which were determined by means of the Nernst–Einstein law from the NMR data in Table 2 Composite, activated in humid H2 r PFG/S cm1 100 150 200 300
4
0.410 2.7104 2.9103 –
r SFG/S cm1 – – 4.1103 3.2102
Composite, activated in dry H2 r IS/S cm1
r NOESY/S cm1
2
10
9.110 9.2102 9.4102 9.7102
illustrates this comparison with a simple representation of the temperature dependences which consist of the extreme values of each curve. A conclusion from the conductivity studies is that proton species like H+, NH4+, or H3O+ are responsible for the charge transport. Self-diffusion NMR studies give information about the mean squared (diffusion) path length of the 1H nuclei in a given time what means about the mass transport. It is expected that either NH4+ or H+ are charge carrier in the dry sample. That means that either an ammonium ion jumps from one site to the next site, or ammonium dissociates to ammonia and H+ and only the latter is the charge carrier. The mass transport should exceed the charge transport in this case, since it is likely that the molecular ammonia is more mobile than the charged H+. However, the latter is not stable and should combine with an oxygen atom giving a hydroxyl group. This process is well-known for the ammonium forms of zeolites. They deammoniate at temperatures above 150 8C building an uncharged zeolitic framework with structural hydroxyl groups and fast diffusing ammonia gas [34]. It should be noted that these hydroxyl protons are not fixed. An ammonia molecule can combine with the hydroxyl proton and rebuild an ammonium ion, which contributes to the charge transport. This is the so-called vehicle mechanism of the proton transport, which should not be fused with pure H+ transport. Ammonium ions as proton vehicle and charge carrier are well-known [35]. Since the vehicle transport is very sensitive with respect to the ammonia concentration, a zeolite-based sensor can even been used as selective ammonia exhaust gas sensor for automotive applications [36]. In principle, the charge transport could be explained also by H+ as charge carrier, if the mobility of H+ would be similar to the mobility of ammonia in the case of the dissociated ammonium ion. But those bridging hydrogen atoms, which are also known as charge carrier in phosphates [37], seems not to be important for the conductivity of the composite: Conductivities determined from the EXSY NMR data are some order of magnitudes lower than the IS values. The 1H MAS NMR signals at about 10 ppm are not relevant for the conductivity of the composite. For the explanation of the significant difference between SFG and IS values at 570 K, Fig. 9, we should recall the fact that the composite is under flow condition in the IS cell, whereas the NMR probe behaves like a batch reactor.
4.410 2.7109 – –
r SFG/S cm1
r IS/S cm1
– – 0.6103 3.4103
1.5105 3.8104 4.0103 1.4102
A similar explanation can be given for the conductivity in the humid atmosphere at high temperature, since the difference between IS and SFG values at 300 8C is not significant. But a major difference exists at lower temperature, where the conductivity values determined from SFG NMR data fall below the impedance values by some orders of magnitude. A possible reason for this can be the insufficient conformance of the flow experiment in the conductivity cell and the batch experiment in the fused NMR glass tube containing the composite which was previously activated in a humid atmosphere. The water vapor pressure which was constantly 4.2 kPa in the IS experiment could not be controlled in the NMR tube. The chosen water concentration could give a conductivity contribution by hydroxonium vehicles (H3O+) at lower temperature. The water concentration can be lower in the NMR tube and, therefore, not reflected in the NMR experiment. Several activation energies are presented in this study. Conductivity measurements give 1.00 eV and 0.48 eV for Tb150 8C and TN150 8C, respectively, for the composite activated in a dry H2 atmosphere. For the composite activated in a humid H2 atmosphere, we obtained 0.15 eV and 0.10 eV for Tb150 8C and TN150 8C, respectively. EXSY (NOESY) experiments describe the chemical exchange of hydrogen atoms between ammonium ions and bridging hydrogen atoms. The corresponding activation energy is 0.84 eV for the temperature range 108–175 8C. 100 10-1 10-2 10-3
/S cm-1
T/8C
10-4
NOESY, SFG, dry IS, dry PFG, humid SFG, humid IS, humid
10-5 10-6 10-7 10-8 10-9 1.6
1.8
2.0
2.2
2.4
2.6
2.8
3.0
3.2
1000 T-1/ K-1 Fig. 9. Simplified representation (only extreme values are plotted) of the conductivities for the dry and humid samples obtained by three spectroscopic techniques, IS, SFG NMR, EXSY NMR.
962
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963
SFG NMR experiments yield for the temperature range TN150 8C values of 0.47 eV and 0.27 eV for the composite activated in the dry and humid H2 atmosphere, respectively. PFG NMR experiments yield for the temperature range Tb150 8C a value of 0.31 eV for the composite activated in the humid H2 atmosphere. Only two correlations should be discussed here. The EXSY activation energy is comparable with a value, which could be obtained from conductivity measurements, if we consider only the temperature range 108–175 8C. The comparison of the corresponding conductivities, however, excludes a proton transfer between ammonium ions and bridging hydrogen atoms as explanation for the electric conductivity, see Fig. 9. A better connection seems to exist between the diffusion of hydrogen species and the conductivity for the composite activated in the dry atmosphere, see Fig. 9. The good agreement between the SFG NMR activation energy above 150 8C (0.47 eV) and the corresponding IS value (0.48 eV) is a further hint that either an ammonium ion jumps from one site to the next, or ammonium dissociation to ammonia and H+ plays the most important role for the charge transport in the composite.
dissociation products NH3 and H+. A contribution to the conductivity arising from the isolated OH groups, which are the result of activation, can be excluded. The activation of the composite causes the substitution of half the ammonium ions by hydroxyl protons and gives free space for the diffusion of the residual ammonium ions or the dissociation products NH3 and H+. The mobility of ammonia molecules alone cannot explain the charge transport. NMR studies of the composite activated in a humid H2 atmosphere show that the ammonium ions, or H+, are the charge carrier at high temperature (300 8C). Another NMR sample technique should be used, in order to obtain a constant water vapor pressure in the whole temperature range.
Acknowledgements We are grateful to Prof. Dr. Jfrg K7rger, Johanna Kannelopulos and Allen Ehrlicher for advice. This work was supported by the Deutsche Forschungsgemeinschaft under the projects Fr 902/12-1 and by the Max-BuchnerStiftung.
4. Conclusions A composite with a stoichiometry of 6NH4PO3/(NH4)2 SiP4O13 was prepared. After an initial mass loss of 7% (mainly by deammoniation) in the first cycle, the material is thermally stable upon cycling between 50 8C and 300 8C. The electrical conductivity of the composite in a dry gas phase increases by about five orders of magnitude with increasing temperature up to 3008C. The conductivity increases continuously from 6.9108 S cm1 at 47 8C up to 1.7102 S cm1 at 308 8C. The activation energy in dry atmosphere is 1.00 eV in the lower (b150 8C), and 0.48 eV in the higher temperature regime (N150 8C). Changing from a dry to a humid atmosphere causes a reversible increase in conductivity which is associated with water uptake of the material. Only weak temperature dependence can be observed in a humid atmosphere, where the activation energy varies between 0.10 and 0.15 eV. At 100 8C the conductivity is about thousand times increased by the humid atmosphere, while at 308 8C it is increased only by a factor of about 5. DC measurements on a H2–H2- and a O2–O2-cell in addition to the impedance spectroscopy show that the composite material exhibits proton conductor properties. Contributions of other charge carriers like electrons or oxygen-ions can be neglected. NMR studies of the dry H2 activated composite can be explained by a proton conductivity that is caused by ammonium ions which act as proton vehicle and by H+ as charge carrier as well. The latter explanation is possible under the assumption of a similar mobility for the
References [1] T. Norby, Solid State Ionics 125 (1999) 1. [2] T. Surumi, H. Ikawa, K. Urabe, T. Nishimura, T. Oohashi, Solid State Ionics 25 (1987) 143. [3] C.K. Kuo, A. Tan, P. Sarkar, S. Nicholson, Solid State Ionics 38 (1990) 251. [4] T.R.N. Kutty, V. Jayaraman, G. Periaswami, Solid State Ionics 128 (2000) 161. [5] A.I. Baranov, V.M. Duda, D.J. Jones, J. Roziere, V.V. Sinitsyn, R.C.T. Slade, Solid State Ionics 145 (2001) 241. [6] V.G. Ponomareva, L.G. Lavrova, L.G. Simonova, N.F. Uvarov, G.V. Lavrova, E.F. Hairetdinov, Solid State Ionics 90 (1996) 161. [7] V.G. Ponomareva, G.V. Lavrova, L.G. Simonova, Solid State Ionics 119 (1999) 295. [8] B. Zhu, Solid State Ionics 125 (1999) 397. [9] B. Zhu, I. Albinsson, B.-E. Mellander, G. Meng, Solid State Ionics 125 (1999) 439. [10] B. Zhu, B.-E. Mellander, Solid State Ionics 97 (1997) 535. [11] B. Zhu, B.-E. Mellander, J. Power Sources 52 (1994) 289. [12] S. Tao, Z. Zhan, P. Wang, G. Meng, Solid State Ionics 116 (1999) 29. [13] R. Doshi, V.L. Richards, J.D. Carter, X. Wang, M. Krumpelt, J. Electrochem. Soc. 146 (1999) 1273. [14] B.C.H. Steele, Solid State Ionics 129 (2000) 95. [15] T. Kenyo, K. Inaba, Denki Kagaku 58 (1990) 649. [16] E. Montoneri, F.J. Salzano, E. Findl, F. Kulsea, Solid State Ionics 18–19 (1986) 944. [17] T. Kenyo, Y. Owaga, Solid State Ionics 76 (1995) 29. [18] O. Niemzig, Protonenleiter im Mitteltemperaturbereich (100–500 8C), Diplomarbeit, Rheinische Friedrich-Wilhelms-Universit7t zu Bonn, 1996. [19] M. Cappadonia, O. Niemzig, U. Stimming, Solid State Ionics 125 (1999) 333. [20] S. Haufe, Kompositelektrolyte: Herstellung, Charakterisierung und Leitf7higkeitsuntersuchungen, PhD thesis, Technische Universit7t Mqnchen, 2001.
S. Haufe et al. / Solid State Ionics 176 (2005) 955–963 [21] D. Prochnow, Festkfrper-NMR-Untersuchungen an Phosphatmaterialien, PhD thesis, Universit7t Leipzig, 2003. [22] M.T. Averbuch-Pouchot, A. Durif, J. Solid State Chem. 18 (1976) 391. [23] F. Ehrenberger, S., Goldbach, Methoden der organischen Elementar und Spurenanalyse (VCH Verlagsgesellschaft, Methoden der organischen Elementar und Spurenanalyse, 1973). [24] T. Mildner, H. Ernst, D. Freude, Solid-State NMR 5 (1995) 169. [25] P. Galvosas, F. Stallmach, G. Seiffert, J. Karger, U. Kaess, G. Majer, J. Magn. Reson. 151 (2001) 260. [26] S.P.S. Badwal, Ceramic superionic conductors, in: R.W. Cahn, P. Haasen, E.J. Kramer (Eds.), Materials Science and Technology, VCH, Weinheim, 1994, p. 567. [27] E. Skou, I.G.K. Andersen, E.K. Andersen, DC techniques and AC/DC combination techniques, in: P. Colomban (Ed.), Proton Conductors: Solids, Membranes and Gels—Materials and Devices, Cambridge University Press, Cambridge, 1992, p. 418. [28] A.-R. Grimmer, Beitr7ge zur Entwicklung der hochauflfsenden Festkfrper-NMR (1H, 19F, 27Al, 29Si, 31P) und ihrem Einsatz in der anorganischen Chemie, Dissertation B, 1989.
963
[29] P. Hartmann, J. Vogel, B. Schnabel, J. Magn. Reson. 111 (1994) 110. [30] D. Freude, H. Ernst, I. Wolf, Solid State NMR 3 (1994) 271. [31] E. Brunner, U. Sternberg, Progr. Nucl. Magn. Reson. Spectr. 32 (1998) 21. [32] R.R. Ernst, G. Bodenhausen, A. Wokaun, Principles of Nuclear Magnetic Resonance in One and Two Dimensions, Oxford Univ. Press, London, 1987. [33] P. Colomban, A. Novak, Proton conductors: classification and conductivity, in: P. Colomban (Ed.), Proton Conductors. Solids, Membranes and Gels—Materials and Devices, Cambridge University Press, Cambridge, 1992, p. 38. [34] M. Hunger, Catal. Rev., Sci. Eng. 39 (1997) 345. [35] K.D. Kreuer, A. Rabenau, W. Weppner, Angew. Chem., Int. Ed. Engl. 21 (1982) 208. [36] R. Moos, R. Mqller, C. Plog, A. Knezevic, H. Leye, R. Iron, T. Braun, K.J. Marquardt, K. Binder, Sens. Actuators, B, Chem. 83 (2002) 181. [37] D. Michel, J. Totz, Y.N. Ivanov, A.A. Sukhovsky, I.P. Aleksandrova, J. Petersson, Ferroelectrics 267 (2002) 303.