Some Unusual Properties of Activated and Reduced AgNaA Zeolites BY PETER A. JACOBS* AND JAN B. UYTTERHOEVEN Centrum voor Oppervlaktescheikunde en Colloi'dale Scheikunde, Katholieke Universiteit Leuven, De Croylaan 42, B-3030 Leuven (Heverlee), Belgium AND HERMANNK. BEYER Central Research Institute of Chemistry, Hungarian Academy of Sciences, 11 Pustaszeri-6t 57-69, Budapest, Hungary Received 2nd March, 1978 Carbon monoxide, oxygen and hydrogen were found to be chemisorbed on dehydrated AgA zeolites. This was investigated in detail using volumetric sorption and temperature programmed desorption techniques. Also i.r. and mass spectrometry were used to characterize the solid and the desorbed molecules. It was found that as a result of an auto-reductive process, colour centres are created upon de- gassing of the zeolite.These centres sorb hydrogen and oxygen dissociatively, while one molecule of carbon monoxide was chemisorbed per Ag ion available in the supercage. It is proposed that linear Ag': clusters are formed upon activation, the ends of which constitute chemisorption sites for hydrogen and oxygen. Ralek et al.' reported the use of AgNaA zeolites as moisture indicators. This application was based on the observation that the brick red colour of thoroughly dehydrated AgA zeolite was very sensitive to traces of moisture. Upon addition of increasing amounts of water, the sample colour was found to change towards orange, yellow and white. At that time no explanation for the phenomenon was advanced. have rediscovered these colour changes upon activation.The authors show that upon activation silver atoms are formed. They claim that uncharged silver clusters (Ag, molecules) are formed within cubes of eight (or six) Ag+ ions. The process of auto-reduction of transition metal ions in zeolites is not unique for Ag+, since it previously has been reported for cupric ions in Y ~ e o l i t e . ~ The silver zeolite system still shows other peculiarities. Upon thorough degassing of reduced samples, it was shown that hydrogen to a limited extent can be formed by oxidative thermal desorption from zeolites Y and m~rdenite.~ This property was combined with the ease of photochemical reduction of hydrated silver ions in zeolite. In this way, the silver+zeolite system could be used for the cleavage of water into oxygen and hydrogen using a photochemical and thermochemical step.6 It also was recently found that upon addition of ammonia to AgA the new mole- cules triazane (N3H5) and cyclotriazane (N3H3) were formed in supercages of zeolite A.3 In the present work, we report on other unusual properties of silver exchanged A zeolite. The development of bright colours upon activation could be related to a process of auto-reduction.We also attempt to explain the unexpected chemi- sorptive properties for hydrogen, oxygen and carbon monoxide in such a system. In recent single crystal X-ray diffraction studies of the same system, Seff et aL2* 56P. A . JACOBS, J. B . UYTTERHOEVEN AND H . K. BEYER 51 EXPERIMENTAL MATERIALS A commercial zeolite Na-A from Union Carbide Corporation, Linde Division, was purified as described earlier for Silver A zeolites with different silver cation content were obtained after exchanges in 0.005 mol dm-3 AgN03 solution.Afterwards the samples were washed until complete disappearance of any anions in the washing waters and dried at ambient temperature in the dark. Only freshly prepared and white samples were used in further work. Sample notation and anhydrous unit cell composition are given in table 1 . TABLE AN ANHYDROUS UNIT CELL COMPOSITION OF A ZEOLITES For particular applications other compositions were prepared in the same way. A ZK-4 zeolite was synthesized according to known procedures 9s lo with the following unit cell composition : where TMA represents tetramethylammonium cations. The sample was calcined in flowing oxygen at 773 K, and saturated with ammonia before contacting it with water. The silver zeolite A was obtained after repeated exchanges in an excess of 0.005 mol dm-3 AgN03 solution with a ratio of zeolite to exchange solution of 0.5 gdm3.Chemical analysis showed a sample with the following anhydrous unit cell composition : The X-ray diffraction pattern showed that the samples in all cases remained highly crystalline The number following the sample name corresponds to the degree of silver exchange. In a few cases, the number in brackets following the degree of cation exchange, denotes the degassing temperature in U ~ C U O in K. Labelled oxygen (I8O2) with 99.9 % isotopic purity was from L.C.B. Na8 TMA (A102)9 (Si02)15 Ag9 (A102)!3 (Si02)15.PROCEDURES AND METHODS Gas uptake and desorption measurements were performed in a low volume circulation system. Using a sample of kg, the accuracy is better than 3-0.005 mmol. Tempera- ture programmed desorption (t.p.d.) measurements were carried out in a Hewlett Packard 5992A gas chromatographjmass spectrometer combination, using single ion monitoring techniques. The mass scale of the mass spectrometer unit was calibrated automatically. 1.r. measurements were taken in situ with a Beckman IR12 grating spectrometer in the double beam/absorption mode, using a sample and reference cell attached to a vacuum sys tem. Sorption of hydrogen was carried out at 195 K. The " chemisorbed hydrogen " cor- responds to that amount of hydrogen adsorbed at 195 K that cannot be desorbed after a 10 min degassing at 293 K.Carbon monoxide is considered to be chemisorbed, when after room temperature adsorption (293 K) it cannot be desorbed after degassing in vacuo (1.33 mN m-") for 1 h. RESULTS AND DISCUSSION ACTIVATION OF AgNaA ZEOLITE A typical t.p.d. experiment for hydrated AgA-100 is shown in fig. 1. At low temperatures, water is the main desorption product, but above 400 K non-negligible amounts of oxygen can also be desorbed from the sample. Using a volumetric58 PROPERTIES OF AgNaA ZEOLITES system with cold trap (195 K) the oxygen desorption curve can be perfectly reproduced. It should also be noted, that when the desorption of water has ceased completely (525 K), oxygen desorption restarts again. Fig. 2 shows clearly that the rate of oxygen desorption declines in the 500-520 K region and shows two definite regions characterized by high rates of oxygen loss.In the low temperature region (c 500 K) desorption temperature/K FIG. 1.-T.p.d. experiment on hydrated AgA-100 with the desorption curve for water (a) and oxygen (b), respectively. the sample turns from white to a deep yellow, golden colour. At this stage almost complete dehydration is reached and a first quantity of oxygen is released. Con- sequently, the sample turns to a bright red colour, releasing a slightly higher quantity of oxygen. The same solid state reactions established for the auto-reduction of CuY4 zeolites can be used here: As+ As+ 0 0 0 0 0 \-/\ A-N \/\-/\ / /\ /\ /\ /\ /\ /\ A /\ Al Si A1 Si -+ Ag"+Si A1 Si+ Al+O- (1) Ag+ 0 0 v \--/ v \-/ /\ A /\ /\ Si Al+O--+Ag"+O-+ Si A1 (2) 2 0 .4 0 2 . (3) (4) The overall stoichiometry of these reactions is where ZO- represents the zeolite lattice and Z+ a Lewis site, written in the conven- tional form. Nothing changes in the reaction stoichiometry when alternative re- action schemes for Lewis acid site formation are used. The use of this stoichio- metry allows to estimate the degree of auto-reduction. Fig. 2 shows that at 600 K, the degree of Ag+ reduction amounts to almost 8 %. After contacting dry and degassed AgA-100 with I6O2 or 1 8 0 2 at 673 K and cooling to ambient temperatures, t.p.d. curves are given in fig. 3. In every case, the sample remains bright red. Only half of the amount of oxygen initially desorbed can be recovered during a second desorption experiment from a dry reoxidized sample.Upon chemisorption, these molecules are dissociatively adsorbed as revealed by the experiment with labelled oxygen. 2(Ag+ZO-) 3 *O, + Agi + 20- + Z+P . A . JACOBS, J . B . UYTTERHOEVEN AND H . K . BEYER 59 Upon oxygen desorption from a AgA-100, the initial state cannot be restored. Oxygen deficient sites and reduced silver ions are irreversibly formed. Therefore, it is not probable that chemisorbed oxygen fills up again oxygen deficient sites, but rather is dissociatively chemisorbed upon silver species containing both reduced and non-reduced silver ions, as proposed by Seff et aL2* X desorprtion temperature/K FIG. 2.-Rate of oxygen desorption from the corresponding degree of auto-reduction of silver ions in AgA-100, together with the corresponding colour changes in the sample.These colour changes and the concomitant phenomena are typical for the A zeolite structure. Indeed, these phenomena are not observed in Ag chabasite, Ag stilbite or Ag clinoptilolite, zeolites with comparable small pores. They neither are observed in AgA zeolite with higher Si/Al ratio, (AgZK-4 zeolite also does not show the colour changes), nor in germanium substituted faujasite with a &/A1 ratio equal to one. However, they are found on AgNaA zeolite with variable degree of Ag+ exchange. All this allows us to conclude that the A zeolite structure is needed, containing equal amounts of Si and A1 tetrahedra. This structure is typified in that desorp tion temperat ure/K FIG.3.-T.p.d. curve of oxygen from AgA-100: (a) desorption from wet sample; (6) desorption from dry sample, cooled from 673 K in oxygen to ambient temperature ; (c) same as (b) but cooled in labelled oxygen ('*02).60 PROPERTIES OF AgNaA ZEOLITES it contains more cations than sites are available per unit cell. Therefore it has ions that can be considered as zero-coordinated by a distance criterion. This property may well be at the origin of the unusual properties. However, an Ag6+(Agg) cluster 2 s in the sodalite cage is rather improbable, since its formation in low ex- changed forms would result in segregation of cations : parts of the crystal should con- tain only Na+ cations, while aggregation of charged silver should occur at a few other spots. Therefore, the existence of isolated linear species in the cubo-octahedra (Ag+ .. . Ago . . . Ag+) in order to explain X-ray diffraction data l3 has a higher probability. The existence of such clusters will be used further as a working hypo- thesis. There is little doubt that the red colour is due to charged clusters, Me,"+ I d 7' M 0.301 24 Ag+ exchange/ % FIG. 4.-Desorption of oxygen from freshly prepared AgNaA, degassed at 673 K for 1 h. (y > x) since NaA upon sodium vapour treatment at 673 K showed the same colour changes. The red colour of sodium vapour treated NaY could indeed be explained by the presence of Nad+ centred4 It is also well known l5 that, in silver halides, the colour is intensified going from F- to I-. At the same time, there is appreciable increase in the covalent character of the Ag .. . X interactions. In compounds with chain structure (as AgCN) the bond is predominantly covalent. All this supports our model. Fig. 4 shows that the oxygen-donating capability of the A zeolite lattice strongly depends on the degree of Ag+ exchange. The phenomenon is strongly enhanced at the highest degrees of exchange. CHEMISORPTION ON RED AgNaA Upon addition of H2 (at 195 K) and CO (at 293 K) on a red AgA-100, no colour However, increasing amounts of H20 cause gradual dis- changes are visible. appearance of the colour. co CHEMISORPTION Fig. 5 shows carbon monoxide sorption for AgNaA zeolites with different Ag+ content. The amount of CO chemisorbed at 293 K increases with the degree of exchange of Ag+ for Naf. However, the deviation from linearity indicates that both properties are not directly related.The total amount adsorbed also is enhanced by the presence of Ag+ and increases almost linearly with the degree of exchange up to values of 30 %. At higher degrees of exchange, this increase is less pronounced. Fig. 6 shows a linear relation between the amount of CO chemisorbed and the intensity of its i.r. stretching vibration. This constitutes firm proof that supercageP . A . JACOBS, J . B . UYTTERHOEVEN AND H. K . BEYER 61 Ag+ cations are the sites for CO chemisorption. Strong CO sorption has also been observed on AgY l6 and AgZSM-5 zeolites. Using an absorption coefficient determined for AgX, a proportionality factor of 1.10 0.13 was obtained between the amount of CO chemisorbed and Ag+ ions available for interaction.The supple- mentary amount of physisorbed CO most probably corresponds to some organization Ag+ exchange/ % intensity vcolarbitrary units FIG. 5.-Carbon monoxide sorption at 293 K on AgNaA zeolites with different Ag+ content; (a) amount irreversibly sorbed at 293 K, total amount sorbed at equal pressures : (b) 13.3, (c) 26.6, (d) 33.33 kN m-2. FIG. 6.-Relation between the amount of CO sorbed and the amount of available Ag ions, as measured by the i.r. intensity of CO at 2180k 5 cm-’ ; (a) amount chemi- sorbed, (b) amount sorbed at 13.3 kN m-’. in the second layer. The relative decrease of this amount at higher exchange levels could be due to some steric hindrance at the sorption site. This is entirely consistent with our hypothesis that mainly at high degrees of exchange Ag+--AgO-Ag+ species are formed.The location of these Ag+ ions is necessarily at site 11’ inside the cubo- octahedron. Direct coordination with one CO remains possible, while organization in the second layer will be weaker. HYDROGEN CHEMISORPTION Silver metal on support is not known to retain hydrogen at ambient temperature. However, AgNaA zeolite does, as shown in fig. 7. It should be stated that in each of the following cases reduction of Ag+ ions by hydrogen can be excluded, since in no case lattice hydroxyls are formed. The phenomenon becomes important for degrees of silver ion exchange above 70 %. Physisorption of hydrogen (at 195 K) is also observed and shows the same overall but less marked changes.Comparison with fig. 4 shows that these properties are related to the ability of the lattice to release oxygen, and in our mind to the formation of Ag+-Ag-Ag+ species. In fig. 8 it is shown for AgA-100 that the amount of hydrogen retained is strongly dependent on the outgassing temperature of the sample. Hydrogen is only retained when the sample is completely degassed and further increases as more and more lattice oxygen comes off. When water is adsorbed, the hydrogen chemisorption capacity of the AgA-lOO(673) sample decreases linearly, at least for small amounts adsorbed (fig. 9). The slope of the straight line equals 1. This indicates that water is preferentially sorbed at the62 PROPERTIES OF AgNaA ZEOLITES ends of the Agf-Ag-Agf species. Each water molecule is able to replace two hydrogen atoms.This is also strong evidence for the dissociative nature of hydrogen chemisorption on the silver agglomerates, just as was shown for oxygen. The differ- ence in chemisorption capacity in each case (Hz/Oz = 6) may be due to the fact that oxygen molecules cannot enter the six rings of the sodalite unit, while hydrogen can. 0 50 I0 0 Ag+ exchange/ % 13.33 kN m-2, and (b) amount adsorbed at 195 K, not desorbable at 293 K. FIG. 7.-Sorption of hydrogen on AgNaA-(673) ; (a) amount reversibly adsorbed at 195 K and For carbon monoxide, which cannot enter either, the Ag/CO ratio also equals 6. This indicates that only one out of five silver cations are accessible from the supercage. This also may reflect the average availability of the ends of the Agf-Ag-Agf species, although for the moment it seems highly speculative to propose two different locations for this species.I 5 00 6 00 outgassing temperature/][( FIG. 8.-Infiuence of degassing temperature of AgA-100 on the hydrogen retention capability ; (a) and (6) same as in fig. 7.P. A . JACOBS, J . B . UYTTERHOEVEN AND H . K . BEYER 63 0.2 / / b 24 water adsorbedlmol kg-’ FIG. 9.-Decrease of the amount of chemisorged hydrogen when increasing amounts of water are sorbed on AgA-lOO(673). In fig. 10 a straight line relation is shown between the amount of Ago atoms obtained through auto-reduction and the amount of irreversibly held hydrogen molecules. Its slope equals one. Samples with different degree of exchange and outgassed at different temperatures fit this relation.Since hydrogen is dissociatively adsorbed (fig. 9), this relation clearly shows that each Ago atom formed by auto- reduction, is associated with two Ag+ ions, each capable of chemisorbing one atom of hydrogen or one water molecule. 0.3 ..( 1 M A4 “0 0.2 E si \ a E g 0.1 s Y 0 0 0.15 0.30 amount Ag+ reduced/equiv. kg-I FIG. 10.-Relation between the degree of auto-reduction (abscissa) of AgNa zeolites and the amount chemisorbed hydrogen (ordinate) ; (a) AgNaA-10(673), (b) AgNaA-30(673), (c) AgNaA-70(673), (d) AgA-100(523), (e) AgA-100(593), cf) AgA-100(673). CONCLUSIONS This work contains strong experimental evidence for the existence of partly reduced silver agglomerates in AgNaY zeolites. The following main observations are on the basis of our hypothesis that these species are isolated and linear Ag+-Ag-Ag+ clusters located in the cubo-octahedra and are responsible for the chemisorption64 PROPERTlES OF AgNaA ZEOLITES properties of the solid : (1) a process of auto-reduction occurs together with intense colouration of the sample, the former implies the appearance of Ago atoms, the latter is indicative of covalent bond formation of silver; (2) comparison with other Ag zeolites different in structure and %/A1 ratio shows that the initial presence of zero coordinated ions requiring A zeolite structure and Si/Al equal to one is related to the formation of coloured centres; (3) the phenomenon occurs at different degrees of Ag+ exchange, making the existence of compact (Agg A$) clusters thermodynamic- ally rather improbable in the present case; (4) oxygen adsorbed dissociatively on the colour centres as evidenced by the tracing with I8O2; (5) there is proportionality between the degree of auto-reduction and the chemisorption of hydrogen, indicating that chemisorption occurs on the colour centres; (6) one molecule of water is able to replace two hydrogen atoms.This shows that hydrogen is dissociatively cheini- sorbed; (7) for every Ago formed, two H atoms are chemisorbed on the colour clusters. Carbon monoxide is chernisorbed in the Ag+ ions available in the supercages, involved in the cluster formation or not. P. A. Jacobs acknowledges a research position as “ Bevoegdverklaard Navorser ” from N.F.W.O. (Belgium). The experimental help of Mrs.I. Szaniszlo (Budapest) and J.-Ph. Linart (Leuven) is appreciated. Stimulating discussions with Dr. W. J. Mortier are also acknowledged. We thank the Belgian Government (Dienten Wetenschapsbeleid) for financial help. M. Ralek, P. Jiru, 0. Grubner and H. Beyer, Coll. Czech. Chem. Comm., 1962, 27, 142. Y . Kim and K. Seff, J. Amer. Chem. SOC., 1977,99,7055. Y. Kim, J. W. Gilje and K. Seff, J. Amer. Chem. Soc., 1977,99,7057. P. A. Jacbos, W. De Wilde, R. A. Schoonheydt, J. B. Uytterhoeven and H. K. Beyer, J.C.S. Furuday I, 1976, 72, 1221. P. A. Jacobs, J. B. Uytterhoeven and H. K. Beyer, J.C.S. Furuduy I, 1977, 73, 1755. P. A. Jacobs, J. B. Uytterhoeven and H. K. Beyer, J.C.S. Chem. Comm., 1977, 128. ’ R. A. Schoonheydt, L. J. Vandamme, P. A. Jacobs and J. B. Uytterhoeven, J. Catalysis, 1976, 43, 292. P. A. Jacobs, M. Tielen, J.-Ph. Linart and J. B. Uytterhoeven, J.C.S. Faraduy I, 1976,72,2793. G. T. Kerr, J. Phys. Chem., 1962, 66,2271. lo G. T. Kerr, US. patent 3,314,752. l1 P. A. Jacobs, Curboniogenic Activity of ZeoZites (Elsevier, Amsterdam, Oxford, New York, l2 R. L. Firor and K. Seff, J. Amer, Chem. SOC., 1977,99, 1112. l3 L. Gellens, W. Mortier and J. €3. Uytterhoeven, to be published. l4 J. A. Rabo, C. L. Angell, P. H. Kasai and V. Schomaker, Disc. Furuduy Soc., 1966,41, 328. 1977), p. 54-55. F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry (Interscience, New York, London, Sydney, 2nd edn, 1966), p. 1041. l6 H. Beyer, P. A. Jacobs and J. B. Uytterhoeven, J.C.S. Fapaday I, 1976, 72,674. l7 J. A. Rabo, J. N. Francis and C. L. Angell, U.S. Patent 4,019,880. (PAPER 8/382)