4719
J. Phys. Chem. 1992, 96, 4719-4724
Reduction of Pd(I1) in Aqueous Solution: Stabilization and Reactions of an Intermediate Cluster and Pd Colloid Formation Matthias Michaelis and Amim Henglein* Hahn-Meitner-Institut Berlin GmbH, 1000 Berlin 39, FRG (Received: October 28, 1991; In Final Form: January 22, 1992)
In the y-radiolytic reduction of Pd(NH3)4C12in the presence of poly(ethyleneimine), an intermediate cluster is stabilized by the polymer. Pulse radiolysis experiments show that it is slowly formed in the 0.01-1-9 range after the primary reduction of Pd(I1) to Pd(1) by hydrated electrons and organic radicals has occurred in the 0.1-100-ps range. A photochemical method for the preparation of the cluster is also described. The cluster absorbs at 308 and 400 nm and decays within a few days to yield colloidal palladium metal. It can also be transformed into the colloid by freeradical attack (upon prolonged y-irradiation) and by photolysis. The cluster is rapidly oxidized by oxygen and hydrogen peroxide. It reduces methyl viologen, reacts with nitrous oxide to yield Nz, with hydrogen sulfide to yield PdS + Hz, and with cyanide to yield a PdCN complex, and yields colloidal palladium upon reaction with CO. The cluster is not ESR active. The structure of Pd?+ is tentatively assigned to it. Titration with H202yielded an absorption coefficient of 2.2 X lo4 M-' cm-'. Some chemical reactions of colloidal palladium are also described. It reacts with hydrogen sulfide to yield PdS and with cyanide to form Pd(CN)42-;in both reactions hydrogen is generated simultaneously from the aqueous solvent. H2 is also generated during the UV illumination of colloidal palladium in the presence of poly(ethy1eneimine) (quantum yield
Introduction Exposure of aqueous solutions of metal ions to ionizing radiation often leads to the formation of colloidal metals, the reduction of the metal ions being brought about by the hydrated electrons and other radicals formed as the primary reactive species. The first reduction products of the metal ions are atoms in unusual valence states which undergo further reduction, dismutation, and finally agglomeration until larger metallic particles are formed. In order to obtain a stable colloid, the reduction is often carried out in the presence of a polymer to which the colloidal particles attach themselves as they are formed.' One might ask at which point in its growth has a particle first become stabilized by a polymer chain. If clusters consisting of only a small number of metal atoms interacted strongly with the polymer, one would be able to prepare more or less stable solutions of oligomeric metal particles. There is great interest in such particles: they are being studied in molecular beams, in frozen matrices, and on supports such as zeolites. An important question is, how do the physicochemical properties change with particle size in the range where the transition from molecular to metallic behavior takes place? In the case of silver, the search for stabilized nonmetallic clusters has been successful:2 the reduction of Ag+ ions in the presence of a polyanion such as polyphosphate or polyacrylate leads to a cluster solution, the clusters having strong absorption bands in the near-UV region. In the present paper, the reduction of Pd(I1) is investigated, again using a polymer as stabilizer for small intermediates. Whereas polyanions turned out to be not helpful in this respect, poly(ethy1eneimine) was found to efficiently stabilize an early product of Pd(I1) reduction. We describe here the formation and decay of this early product and a few of its chemical reactions. The fmal reduction product is colloidal Pd metal. Its formation by decomposition of the early product is described as well as a few chemical reactions of colloidal palladium. (1) (a) Henglein, A. J . Phys. Chem. 1979, 83, 2209. (b) Henglein, A,; Lilie, J. J. Am. Chem. Soc. 1981, 103, 1059. (c) Henglein, A,; Lilie, J. J. Phys. Chem. 1981, 85, 1246. (d) Buxton, G.; Rhodes, T. J . Chem. Soc., Furuduy Truns. 1 1982, 78, 3341. (e) Marignier, J. L.; Belloni, J.; Delcourt, M. 0.;Chevalier, J. P. Nuture 1985, 317, 344. ( 2 ) (a) Henglein, A. Chem. Phys. Lett. 1989, 154,473. (b) Mulvaney, P.; Henglein, A. J . Phys. Chem. 1990, 94, 4182. (c) Linnert, T.; Mulvaney, P.; Henglein, A.; Weller, H. J . Am. Chem. SOC.1990, 112,4657. (d) Henglein, A.; Linnert, T.; Mulvaney, P. Ber. Bunsen-Ges. Phys. Chem. 1990, 94, 1449. (e) Mostafavi, M.; Keghouche, N.; Delcourt, M. 0.;Belloni, J. Chem. Phys. Lett. 1990, 167, 193.
0022-365419212096-47 19$03.00/0
Experimental Section The solutions contained Pd(NH3)&I2 (Aldrich), poly(ethy1eneimine) (Sigma), and propanol-2. Ionizing radiation produces hydrated electrons, hydroxyl radicals, and hydrogen atoms in aqueous solutions. In the presence of propanol-2, the OH and H radicals rapidly react with the alcohol to yield l-hydroxyethylmethyl radicals: OH (H)
+ (CH,),CHOH
-
H20 (H2)
+ (CH3)2COH
(1)
The solutions were deaerated by evacuation or bubbling with argon. In these solutions, the hydrated electrons and organic radicals are the reducing agents
+
eaq- Pd2+(c) (CH3)2COH + Pd2+(c)
-
-
Pd+(c)
Pd+(c)
+ (CH3)2C0+ H+
(3)
where (c) designates complexed species. Two reasons can be given for the complexation of Pd2+being caused by the amine groups of the polymer (the complexation with the polymer being stronger than with the original N H 3 ligands): (1) the UV absorption of the solution is slightly weaker than predicted from simple superposition of the spectra of the single components, and (2) the viscosity of a poly(ethy1eneimine) solution is slightly increased upon the addition of Pd(NH,),Cl2; this is explained by the formation of cross-links as Pd2+ions are complexed by amine groups from different chains. It thus seems that the polymer chains are the sites where reactions 2 and 3 take place. In some experiments, the solution was saturated by nitrous oxide, which reacts with the hydrated electrons to yield additional OH radicals: N 2 0 + ea;
+ H20
+
N 2 + OH-
+ OH
(4)
The OH radicals in turn react with the alcohol (eq 1). Thus, organic radicals are the only reducing species in the presence of NzO. The radiation chemical yields (number of molecules formed per 100 eV of absorbed radiation energy) are C(e,,-) = 2.7, G(0H) = 2.7, and G(H) = 0.6. Thus, 6.0 reducing equivalents are produced per 100 eV in the presence of the alcohol. y-irradiation was carried out in the field ofoCa '@ source. Pulse radiolysis was carried out with 3.8-MeV electrons from a Van de Graaff generator, the pulse duration being 0.5 ps. In the photochemical experiments, the light of a 450-W xenon lamp was used (IO-cm water filter; Kratos monochromator). Chemical actinometry was carried out with aberchrome 540. 0 1992 American Chemical Society
4720 The Journal of Physical Chemistry, Vol. 96, No. 11, 1992 1.5 I
Michaelis and Henglein
I
W
c Y
I sulfate? o :none :lo-% A :10%
0.5 c u
+ ID
- 0 d 1.5 m
L
t Id1 Figure 3. Normalized cluster absorption (corrected for the contribution by colloidal Pd) as a function of aging time in the presence of various amounts of sodium sulfate.
n
0
4 1.0
-
1-
0.5
m .E 1.0 +
0
A fnml Figure 1. Absorption spectrum of a solution at various times of irradiM Pd(NH3)& 0.5 ation. The evacuated solution contained 2 X M poly(ethyleneimine),and 6 M propanol-2. Dose rate: 1 X lo5 rad/h. 1.5
I
E
t
I
I
0.2
0.4
I
I
0.6 0.8 fPEI1 [MI
,
1.0
I
1.2
Figure 4. Maximum cluster absorption (corrected for the contribution of colloidal Pd) as a function of polymer concentration. [Pd(II)] = 2 X lo4 M. Inset: maximum cluster absorption as a function of Pd(I1) concentration. [PEI] = 1 M.
7
+
2 1.0
2 Io
:0.5
n
n
"0
100
200
300
400
t Iminl
Figure 2. The 308-nm cluster absorption as a function of irradiation time. Solution and dose rate as in Figure 1.
In many experiments, reagents had to be added to the irradiated solution without bringing it into contact with air. The appropriate amount of deaerated reagent solution was sucked into the evacuated irradiation vessel. The vessel was equipped with a side arm carrying an optical cuvette to allow spectrophotometric measurements without opening the vessel. In the case of solutions irradiated under argon, reagents were injected through a septum. Absorption spectra were recorded with a commercial spectrophotometer; a cuvette with water was used in the reference beam.
Results Cluster Formation. Figure 1 shows the optical spectrum at various times of irradiation of a solution containing 2 X M Pd(NHJ4C12, 0.5 M poly(ethyleneimine), and 6 M propanol-2. The solution was deaerated by evacuation. It can be seen from the upper part that absorption bands at 308 and 400 nm develop until maximum absorption is reached after 50 min. Longer irradiation (lower part) leads to a destruction of these bands and a significant absorption increase at longer wavelengths. The spectrum did not change upon prolonged irradiation. The final spectrum resembles the spectrum of colloidal palladium metal, which is usually obtained in the reduction of Pd(I1) with chemical reagents such as sodium borohydride. Assuming that all Pd(I1) was reduced to the metallic state at this point, an absorption coefficient per Pd atom of 4.3 X lo3 M-' cm-I at 308 nm (where the first reduction products absorbs) and 1.2 X lo3 M-' cm-I at 550 nm (where the first product does not absorb) is calculated. Electron microscopic pictures revealed that the colloid had a broad size distribution ranging between 10 and 40 A. The two absorption bands at short irradiation times are attributed to a nonmetallic reduction product of palladium. A palladium cluster consisting of a small number of Pd atoms should be very reactive toward oxygen. In fact, when the irradiated
solution was exposed to air, the absorption bands immediately disappeared and the spectrum of the solution before irradiation was restored. In Figure 2, the 308-nm absorption of the cluster is plotted as a function of irradiation time. The absorption was corrected for the contribution of colloidal Pd. This correction was made by using the above absorption coefficients of palladium a t 308 and 550 nm. Experiments were also carried out with solutions to which 1 X lo-' M sodium sulfate had been added as inert electrolyte. Under these circumstances, the maximum cluster absorption was smaller and colloidal palladium formed a t an earlier stage of reduction than in the absence of sulfate. The cluster which produces the 308- and 400-nm absorption bands disappears thermally within a few days. Its disappearance is accompanied by the appearance of the broad absorption of colloidal palladium. In the experiments of Figure 3, the absorption of the cluster is shown as a function of aging time. In the presence of sodium sulfate, the decay of the cluster is faster. In the experiments of Figure 4, the concentration of the polymer was varied, keeping the concentration of Pd(NH3)4C12constant at 2 X 10-4 M. The figure shows the maximum value of the cluster absorption reached before the decay occurs at longer times of irradiation. It can be seen that appreciable concentrations of the polymer are necessary to ensure a high cluster yield. The inset of Figure 4 presents the results of experiments in which the concentration of Pd(NH3)&12 was changed, keeping the polymer concentration at 1 M. It can be seen that the maximum cluster absorption increases linearly with Pd2+(c)concentration a t low concentrations but increases less strongly at higher concentrations. If one wants to prepare solutions of the cluster which contain as little metallic colloid as possible, one has to use low Pd(I1) concentrations and short irradiation times. A few experiments were also carried out with a solution containing nitrous oxide. It was found that the cluster absorption bands developed at the same rate as in the evacuated solution. Cluster Reactiom. Figure 5 shows the results of an experiment in which various amounts of hydrogen peroxide were added to M a cluster solution. The initial solution contained 1 X Pd(NHJ4C1, and was irradiated for only 20 min, Le., before the maximum cluster absorption was reached, to make sure that it
The Journal of Physical Chemistry, Vol. 96, No. 11, 1992 4121
Reduction of Pd(I1) in Aqueous Solution,
3.0 r
-
IHZOZI IvM1 Figure 5. Titration of a cluster solution with hydrogen peroxide. De-
crease of the cluster absorption as a function of added H202 concentration. contained only the cluster and practically no colloidal palladium. The decrease in cluster absorption is plotted vs the amount of H202 added. After complete oxidation, the spectrum of the initial solution was restored. The decrease in cluster absorption in Figure 5 was corrected for the increase in absorption due to re-formed Pd2+(c). One calculates an absorption coefficient of 1.1 X lo4 M-'cm-' per reduced equivalent in the cluster from the initial slope of the curve in Figure 5. Using this absorption coefficient, one calculates that 60% of the original Pd(I1) had been reduced to form the 308-nm cluster in the maximum of the curve in Figure 2. From the initial slope of the curve in Figure 2 one calculates a 100-eV yield of 2.3 reduced equivalents in the cluster. The question arises whether the cluster consists of monovalent or zerovalent palladium. In the first case, the cluster would probably be Pd22+,as much larger clusters of Pd+ are not conceivable because of strong accumulation of positive charge. Pd?+ would be formed by dimerization of the first reduction product of Pd(I1): 2Pd+(c)
-
(5)
PdZ2+(C)
Under these circumstances the absorption coefficient per Pd+ ion in the cluster is 1.1 X lo4 M-' cm-' (or 2.2 X lo4 M-I cm-l for Pd22+itself). In the second case, the cluster has the composition (PdO),, n being one or a small number. The cluster would be formed by the disproportionation of Pd+(c) 2Pd+(c)
--
Pdo(c) + PdZ+(c)
(6)
(Pdo),(c)
(7)
nPdo(c)
and the absorption coefficient per Pdo atom in the cluster would be 2.2 X lo4 M-I cm-I. A clear distinction between these two cases cannot be made. We favor the first case for the reasons that are discussed below. The cluster is oxidized by methyl viologen. When a solution of this reagent was added to the cluster solution, the blue color of the radical cation of half-reduced methyl viologen developed immediately. However, the full intensity of the color was not reached as the absorption bands of the cation radical increased by about the same amount upon standing for a few days. Possibly, this effect is due to the existence of clusters with different reactivity due to different complexing conditions as will be discussed below. In Figure 6, the solution was irradiated for various times, methyl viologen added, and the amount of the radical cation calculated from its 603-nm absorption reached after 3 days (e = 1.2 X lo4 M-'cm-I). We formulate the reaction of the cluster as PdZ2+(C) + 2MV2+
+
+
2PdZ+(c) 2MV+
(8)
It can be seen from Figure 6 that methyl viologen is still reduced at long irradiation times at which there was almost no more Pdz2+ in the solution (see Figure 1). This shows that the colloidal palladium which is present at the longer irradiation times can also reduce methyl viologen to a certain extent. The cluster decays about 2 times faster under an atmosphere of nitrous oxide than under vacuum, and no colloid is formed under these circumstances. However, nitrogen was detected as a product, which indicates that the cluster is able to reduce N 2 0 . M Pd(I1) Hydrogen sulfide reacts with the cluster: A 1 X solution (containing 0.1 M poly(ethy1eneimine) was irradiated
I
O0
80 120 t Iminl Figure 6. MV+ concentration after addition of methyl viologen function of the irradiation time.
40
as a
for a short time to produce the 308-nm absorption band of Pd?+, but without any long-wavelength contribution due to metallic Pd; the solution was now liberated from the gas formed during the irradiation by bubbling with argon, and excess HIS was then injected into the vessel. The cluster absorption bands disappeared immediately, and the solution acquired the brown color of colloidal palladium sulfide. At the same time hydrogen was formed. The overall reaction is formulated as follows: Pd22+(~)+ 2H2S
2PdS
+ Hz + 2H+
(9) The measured yield of hydrogen amounted to 98% of the theoretical yield. Colloidal palladium is also able to react with H2S: A sample of the solution was irradiated for 2 h until the cluster bands had disappeared and colloidal palladium had been formed. Upon addition of hydrogen sulfide, PdS colloid and hydrogen were generated, the yield of H2 being 65% of the stoichiometrically expected one. The reactions of cyanide with the cluster and the colloid were also investigated. The reaction with the colloid is straightforward: Upon addition of 5 X 1W2M KCN, hydrogen gas was generated, the yield being 100% of the calculated one, and the colloid dissolved: Pd:
+ 4nCN- + 2nH20
-+
-
~ I P ~ ( C N+ ) ~nHz ~-
+ 2nOH(10)
The cluster, however, reacted in a more complex manner: Addition of the equivalent amount of cyanide resulted in a shift of the absorption bands to 302 and 374 nm, which indicates that a new complex was formed. This new species was not stable but disappeared within 2 days. We tentatively attribute the structure of (PdCN)2 to the 302-nm species: Pd?+(c)
+ 2CN-
(PdCN),(c)
(11) The absorption bands of the cluster disappeared when carbon monoxide was added to the solution. The spectrum after CO addition had the typical long-wavelength absorption of colloidal palladium. CO does not undergo a redox reaction with the cluster as no carbon dioxide could be detected. One has to conclude that CO causes a detachment of the clusters from the stabilizing polymer chain (as it is a stronger complexing reagent than the amine groups of the polymer), and the clusters then agglomerate and dismutate to yield colloidal metal. pulse Radiolysis Experiments. In the experiments of Figure 7, a 2 X lo4 M Pd(NH3)4C1zsolution containing 0.3 M propanol-2 and 1 X M NH3 was pulsed under argon. Note that stabilizing poly(ethy1eneimine) was not present. The difference spectrum of the solution at various times after the pulse is shown. After 1.5 ps, absorptions are present below 300 nm; this spectrum is attributed to the Pd(1) species from the reaction of the hydrated electron with the Pd(NH3)t+ complex. The rate constant for this reaction was found to be 7 X 1Olo M-I s-I by recording the pseudo-first-order decay of the 700-nm absorption of e, - in a solution containing the Pd salt at lower concentrations. After 90 ps, the absorption has increased at all wavelengths, especially at the longer ones. After 160 ms, a broad absorption extending over +
4722 The Journal of Physical Chemistry, Vol. 96, No. 11, 1992
Michaelis and Henglein
0.8 [ I &
2
I I.... i
"200
400
600
"250 300
400
500
A inml Figure 10. Photolysis of a 2 X lo-' M Pd(NH3)&l2solution containing 0.5 M poly(ethy1eneimine) and 6 M propanol-2. Spectrum at various times of illumination with 308-nm light. Absorbcd photons: 1.4 X lod M s-I.
t lhl
-
-
O .
a
e
e'
0";
I - ;
' 7 '; '
log t Is1
'
2
0
a
Figure 9. Changes in the absorbance at 250 and 310 nm on a logarithmic scale. Solution as in Figure 6 .
the whole visible wavelength range and the UV is observed, which is attributed to colloidal palladium. The solution acquired a dark color upon y-irradiation, and a black precipitate formed after a short time. It seems clear from these observations that, in the absence of a stabilizer, the colloid already starts to form in the 100-ps range. Quite different observations were made when the solution contained 0.1 M poly(ethy1eneimine). After a few microseconds, a spectrum similar to that (1.5 p s ) in Figure 7 was recorded. It is attributed to the Pd+(c) species formed in reaction 2. Within the next 100 ps this spectrum increased, if the dose of the pulse was small. This increase is attributed to additional Pd+(c) formation by free radicals via the slow reaction 3 (k 7 X lo7 M-' sd). At higher doses in the pulse, the organic radicals react with each other and do not contribute to the palladium reduction. The changes in the absorption spectrum at longer times after the pulse are shown in Figure 8. As the 250-nm absorption decreases, a new absorption band around 310 nm is built up. The 310-nm band finally remains constant as in the above 7-irradiation experiments. (In the y-radiolysis experiments, the peak of the absorption was located at slightly shorter wavelength; this might be an indication for slow attainment of the complexation equilibrium of the Pd?+(c) formed.) Figure 9 shows the temporal behavior of the 250- and 310-nm absorption on a logarithmic time scale. In contrast to
-
Figure 11. The 308-nm photolysis of a cluster solution: cluster Concentration as a function of illumination time. Inset: concentration of hydrogen generated. The cluster solution was prepared by y-irradiation of a solution containing 1 X lo4 M Pd(NH3)4C12, 0.1 M poly(ethyleneimine), and 1 M propanol-2. After the irradiation, the solution was evacuated to remove all gas that had formed during the irradiation. the experiments on solutions without a stabilizer, there is no colloid formation up to a few seconds after the pulse (Le., the longest times attainable with the pulse radiolysis equipment). These observations can be explained by the following mechanism: Two Pd+(c) ions react very slowly with each other in the 10-2-l-s range. The reason for this slow reaction is probably the fact that they are bound to the polymer chains. The disappearance of the 250-nm Pd+(c) absorption and the appearance of the 310-nm cluster absorption (Figure 9) follow the same rate law, and no absorption bands due to an intermediate species could be detected. It is therefore concluded that n = 2, if the cluster consists of monovalent palladium (i.e., if reaction 5 occurred), or n = 1, if it consists of zerovalent Pd (i.e., if reaction 6 takes place). Wotolysio Experimeng 1. Photolysis of the P&+(c) Solution. When a solution with the same composition as that shown in Figure 1 was illuminated with 308-nm light, the absorption bands of the Pd:+ cluster developed as can be seen from Figure 10. The quantum yield is 1 X clusters formed per photon absorbed. A comparison with the radiolytic formation in Figure 1 shows that absorptions at longer wavelengths are already produced at short times in the case of photolysis. At longer times, the cluster bands disappear in the background of the absorption of colloidal metal. This is understood in terms of the cluster Pd?+ beiig photosensitive itself; this is described below in more detail. From the preparative point of view, it is not advisable to carry out the photolytic cluster formation with 308-nm light. It was found that the cluster developed without much colloid as side M acetone to the solution and illuminating product by adding at 260 nm. In the case of the 308-nm illumination, the Pd*+(c) is the absorbing species. Pd(1) formation occurs via electron abstraction from the polymer by the excited state formed upon light absorption by Pd2+(c). In the case of 260-nm illumination, organic reducing radicals are formed as in the case of radiolysis in Figure 1 (via the triplet state of acetone and subsequent H atom abstraction from the alcohol).
.
Reduction of Pd(I1) in Aqueous Solution
The Journal of Physical Chemistry, Vol. 96, No. 11, I992 4123
2. Photolysis of the Pd?+ Cluster. To investigate the photolysis of the cluster in greater detail, a cluster solution without colloid contribution was prepared radiolytically and illuminated. Figure 11 shows the concentration of the cluster calculated from the optical density at 308 nm as a function of illumination time. After 7 h, the cluster had been completely destroyed. The quantum yield molecules per for the disappearance of the cluster was 3 X photon absorbed. As is also shown in the inset of the figure, hydrogen is formed during the photolysis. During the decay of the cluster, little hydrogen is formed. Rather, it is the photolysis of the colloid produced in the decay of the cluster that leads to the generation of hydrogen. The quantum yield for hydrogen formation was about molecules per photon absorbed by the colloid (as calculated from the slope of the curve in the inset of Figure 11 at longer times). Note that much more hydrogen is present at longer times than palladium; this shows that the palladium colloid is not consumed but catalyzes the formation of hydrogen from water. Hydrogen was also produced in a solution that contained methanol instead of propanol-2.
monoxide exerts a similar effect to sulfate: it complexes Pd?+ and in this way detaches it from the polymer chain. The polymer consists of a mixture of linear and branched chains, Le., primary, secondary, and tertiary amine groups are present. It might be that complexation sites of different stability are present. Upon formation of Pd+: in the solution, the more stable positions are filled first at lower cluster concentrations and the less stable ones at higher concentrations. This may explain why a saturation value of the cluster concentration is reached in the experiments of Figure 4 and why methyl viologen partly reacts very slowly with PdZ2+(Figure 6 ) . The chemical properties of the Pd22+(c)cluster are not far from what one would expect. The cluster is readily oxidized by O2and H202to re-form Pd(I1). It is also oxidized by hydrogen sulfide; eq 9 describes the overall process. It is conceivable that the reaction occurs in two steps
Discussion The stabilization of a palladium cluster in the reduction of Pd(I1) in aqueous solution is another example of the influence which is exerted by a polymer on the reduction of metal ions. The metal atoms in a cluster are coordinatively unsaturated and interact with the lone electron pair of the nucleophilic groups on the polymer. The fixation of the clusters to the chains decreases the agglomeration rate. In the present study, a rather small early reduction product is stabilized, and there is no stabilization of larger clusters which certainly are formed as intermediates during the final formation of colloidal metal from the Pd22+clusters. In this respect one encounters conditions different from our previous studies on silver clusters: Ag42+was found to be the first cluster that could be stabilized, but larger oligomeric clusters could also be obtained as long-lived intermediates during silver metal formation.2 The existence of the Pd22+cluster is quite conceivable, since many palladium(1) complexes are known in inorganic complex chemistry which are dimeric as a result of the formation of the Pd-Pd bond.3 The metal atom generally has square planar coordination, the fourth coordination site at each palladium atom being occupied by a neighboring palladium atom. Our cluster is believed to be complexed by the amine groups of one or even more than one polymer chain. If the cluster is reduced by hydrated electrons or organic radicals, the mobility is vastly increased because of the weaker binding to the polymer of the neutral cluster formed, and colloidal metal can form by agglomeration of the detached clusters. It was already pointed out that the cluster could be PdZ2+or (PdO),. In both cases, the cluster would not be ESR active. In fact, no ESR signals could be detected in the cluster solutions. The fact that the cluster is reduced to yield Pd metal upon longer irradiation (Figure 1) can only be understood if it contains partly reduced palladium. In addition, the various observations made on the effects of sulfate are more easily understood if the cluster is positively charged. It was found that sulfate anions cause a more rapid formation of colloidal palladium during the buildup of the cluster by y-irradiation. The cluster disappeared more rapidly upon aging in the presence of sulfate. S042-could (1) dethch a Pdz2+cluster from the polymer chain or/and (2) accelerate the dismuation reaction between detached clusters. We believe that the first effect is most important as we have made an observation which indicates that changes in the structure of complexation of a cation in poly(ethy1eneimine) solution can be M sodium brought about by added sulfate: addition of 1 X sulfate to a solution of Pd(NH,)4C12 and poly(ethy1eneimine) leads to a slight decrease in viscosity. This indicates that sulfate destroys to a certain degree the cross-links which are formed by the complexation of Pd2+by the amine groups of the polymer. Carbon
the driving force being the formation of the strong Pd-S bond. The colloidal metal also reacts with H2S to form hydrogen:
(3) Temkin, 0. N.; Bruk, L. G . Russ. Chem. Rev. (Engl. Trans/.) 1983, 52, 117.
+ H2S PdS + H2 + Pd2+(c) Pd2+(c) + H2S PdS + 2H+
Pd?+(c)
Pd,
-+
+ nH2S
-
-
nPdS
+ nH2
(12) (13)
(14)
As mentioned above, the hydrogen yield was lower than expected from the stoichiometry. This is possibly due to incomplete conversion of palladium into PdS, the insoluble sulfide formed on the surface of the particles protecting the metallic interior from further H2S attack. The question can be raised whether a more complex mechanism is operative in the reaction of eq 14 in which the aqueous solvent participates in the hydrogen formation as in reaction 10 of the colloid with cyanide. The mechanism may be discussed in terms of a model postulated recently in studies on the chemisorption on colloidal silver particles:w4 chemisorption of a nucleophilic reagent such as CN- creates a dipolar layer on the surface of a metal particle. As a consequence, the Fermi level in the metal is shifted to a more negative potential, the metal particle becoming more sensitive toward oxidation by electron accepting substances. In the present case, the shift is so strong that even water can act as an electron acceptor: 2H20
+ 2e-
-
H2 + 20H-
(15)
The photolysis of the cluster (Figure 11) consists of a reduction yielding colloidal palladium. The reducing agent is probably the polymer or the alcohol, which donates an electron to the excited state of the cluster that is formed upon light absorption. Most interesting is the fact that the illumination of colloidal palladium leads to hydrogen formation (Figure 11). It is generally believed that metals do not possess any significant photochemistry as the charge carriers generated by light absorption in metals rapidly thermalize. This is certainly true for compact metal electrodes. However, it has recently been shown in the case of colloidal silver particles that photochemical reactions with substantial quantum yields can be initiated by light absorption in these parti~les.~We explain the H2formation by an accumulation of electrons on the illuminated palladium particles as the positive holes are scavenged by the adsorbed amine groups of the polymer. The accumulated electrons reduce water molecules as on a compact palladium electrode a t negative potential. The initial yield of formation of reducing equivalents (Figure 2) is only 2.3 per 100 eV of absorbed radiation energy, Le., smaller than the yield of 6 per 100 eV of reducing radicals formed in the irradiation of solutions containing propanol-2. The reason for this discrepancy is not yet clearly recognized. It could be that the solutions contain a reducible impurity (possibly in the polymer) (4). (a) Henglein, A.; Mulvaney, P.; Linnert, T. Faraday Discuss. Chem. SOC.,in press. (b) Mulvaney, P.; Linnert, T.; Henglein, A. J . Phys. Chem. 1991, 95, 7843. (5) Linnert, T.; Mulvaney, P.; Henglein, A. Eer. Eusen-Ges. Phys. Chem. 1991, 95, 838.
Additions and Corrections
4124 The Journal of Physical Chemistry, Vol. 96, No. 11, 1992
which scavenges part of the reducing radicals. Acknowledgment. We thank Dr. Eberhard Janata for advice in the pulse radiolysis experiments and Dr. Paul Mulvaney for valuable discussions.
Registry NO. Pd(NH,)&12, 13815-17-3; PdCN, 140875-92-9; CN-, 57-12-5; CO, 630-08-0; Pd?', 140875-97-4;Pd(CN)d2-, 1500487-2; Hz, 1333-74-0; HS,7783-06-4; N ~7727-37-9; , ~ ~10024-97.2; 0 , 02,778244-7; H202. 7722-84-1; Pd, 7440-05-3; PdS, 12648-43-0; poly(ethy1eneimine), 9002-98-6; methyl viologen, 1910-42-5.
ADDITIONS AND CORRECTIONS 1990, Volume 94
R. J. Beuhler, Y. Y. Chu, C. Friedlander, L. Friedman,* and W. Kunnmann: Deuteron-Deuteron Fusion by Impact of HeavyWater Clusters on Deuterated Surfaces. Page 7665. Detection of D-D fusion events when accelerated heavy-water cluster ions impact on deuterated targets' immediately raised questions about the possibility that traces of high-velocity beam contaminants (artifacts) could account for the experimental results. Experimental searches for sufficient fluxes of trace contaminants produced what appeared to be negative result^.*^^ These results led us to conclude that energy a m p l i a t i o n processes during and after cluster impact were responsible for the observed fusion events. Further searches for direct experimental evidence for artifact contributions have now convinced us that this conclusion is in error. The new results indicate that the cluster impact fusion rates reported in recent publi~ationsI-~ were overestimated by at least 2 orders of magnitude. Recent experiments using magnetic deflection of accelerated ion beams coupled with electrostatic deflection of these beams provide experimental evidence that artifacts are primarily responsible for events that have been ascribed to cluster impact fusion. When ions with mass less than 25 Da were magnetically deflected out of the beam, the fusion event rate was reduced by at least 2 orders of magnitude, whereas electrostatic energy analysis of the beam exiting the magnet showed transmission of roughly 50% of the cluster ion beam. Details of these experiments will be presented in a future publication. We still have no quantitative model for the formation of artifacts, and further experiments are in progress to study this problem. Experiments are also in progress with magnetically filtered beams to determine either values or upper limits of cluster impact fusion rates. Y. K. Bae, who joined us after our original and his independent confirmation of these results4 were published, made substantial contributions to the experiments reported here.
1991. Volume 95 J. A. Seetula, D.Cutman,* P. D. Lightfoot,* M. T. Rayez, and S. M. senkan: Kinetics of the Reactions of Partially Halogenated Methyl Radicals (CHzC1, CH,Br, CHJ, and CHClZ)with Molecular Chlorine. Page 10688. One of the authors' names was spelled incorrectly. The correct spelling is M. T. Rayez.
