Assembling a semiconductor quantum supralattice one atom at a time

Bo Brummerstedt Iversen, Susan Latturner, and Galen Stucky ... Susan E. Latturner, Joseph Sachleben, Bo B. Iversen, Jonathan Hanson, and Galen D. Stuc...
0 downloads 0 Views 681KB Size
J . Phys. Chem. 1990, 94, 6943-6948

6943

of site I most probably arising from enhanced Coulombic interactions with the framework oxygens (see above). This effect diminishes in magnitude on passing from NaS6Y to H16NaaY, most likely because of the counterbalancing charge-transfer effects of adsorption of W(CO)6 or (WO3)2 at an a-cage Na+ cation site compared to hydrogen bonding a t an a-cage Bransted acid site. Interpretation of the “sign” of the adsorption induced far-IR a-cage Na+ translatory mode frequency shifts is difficult. Both blue and red shifts have been observed in Na,Y for W(CO)6 and Wo3, respectively (Figure 2A), but only red in H,,NaaY (Figure 2B). Because the frequencies of the Na+ translatory modes depend in a complex way on the precise NaO bond lengths and ONaO bond angles at the framework site of attachment,’ it is not possible with the available data to interrelate adsorption induced alterations in the geometry around the Na+ anchoring site with the magnitude and sign of the far-IR Na+ frequency shifts. This is a fascinating point that deserves further work.

Figure 5. Preferential anchoring of precursors W(CO)6 and photooxidation product (WO,),to a-cage Na+ cations rather than to Bransted acid sites in H8Na48Yand H,6NaQYfor n = 8 and n = 16, respectively.

of W(CO)6 or the terminal tungsten dioxo group of (WO,), to an extraframework a-cage Na+ cation will serve to (a) increase the shielding of those Na+ cations accessible to the act of adsorption and (b) decrease the quadrupole asymmetry parameter. The primary effects of adsorption are then expected to be a shift of the a-cage 23Na N M R resonances to higher fields with a concomitant enhancement of their intensity. This essentially appears to be the situation in practice. A secondary (cooperative) deshielding effect appears to operate on the 23Na NMR resonance

Conclusion This study provides an impressive demonstration of the power of combined 23NaMAS-NMR and Na+/proton FT-far-mid-IR spectroscopy for providing direct information on anchoring sites, distibutions, and agglomeration processes of W(CO)6 precursors and W 0 3 photooxidation products in zeolite Y. The discovery of “preferential” binding of W(CO)6 as well as (W03), dimers to a-cage Na+ cations rather than Bransted protons in zeolite Y is an intriguing observation. The phenomenon most likely originates with the electrostatic fields (polarization effects) associated with “half naked” Na+ site I1 and/or site 111 cation binding sites which are presumed from our observations to be more intense than those around proton anchors; clearly the latter should be considered as an integral part of an a-cage Bransted acid ZOH, site rather than a “naked” proton. Acknowledgment. We are deeply indebted to the Natural Sciences and Engineering Research Council of Canada for generous financial support of this work. S.O. expresses his sincerest gratitude to the Chemistry Department of the Middle East Technical University, Ankara, Turkey, for allowing him an extended leave of absence to conduct his research at the University of Toronto.

Assembling a Semiconductor Quantum Supraiattice One Atom at a Time: Nonstolchlometrlc Silver, Sodium Bromosodalltes Andreas Stein, Peter M. Macdonald, Geoffrey A. Ozin,* Lash Miller Chemical Laboratories, University of Toronto, 80 St. George St.. Toronto, Ontario, Canada M5S I A1

and Galen D. Stucky Department of Chemistry, University of California, Santa Barbara, California 93106 (Received: May 21, 1990)

The synthetic details and structural, vibrational, and optical properties of a series of novel silver, sodium bromohydrosodalites of the type (8 - p - 2n)Na,2nAg, (2 - p)Br-SOD, where p = 0-2, 2n = 0 - 8 are reported. These sodalites form a solid solution of empty @-cagesinterdispersed with Na,,,Ag,Br clusters. Variations in p alter the mean distance between the clusters while changes in n shift the constituents of the clusters from those of an insulator to a semiconductor type. Far-infrared, 23Na MAS-NMR, optical spectroscopy, and Rietveld refinement of high-resolution powder XRD data show that these materials offer the opportunity to manipulate the extent of collective electronic and vibrational interactions between monodispersed clusters in a perfectly crystalline host.

Introduction The present limitations in the fabrication of atomically perfect quantum confined semiconductor materials by physical/engineering methods have led to the development of chemical tech0022-3654/90/2094-6943%02.50/0

niques for synthesizing periodic arrays of monodispersed semiconductor component clusters in solid-state structures.’V2 Recently (1) Stucky, G. D.; MacDougall, J. E. Science 1990, 247, 669-678.

0 1990 American Chemical Society

6944

The Journal of Physical Chemistry, Vol. 94, No. 18, 1990

(a) SODALITE CAGE

(b) SODALITE

CLASS A

CLASS B

Letters single-size Na4-,Ag,Br (n = 0-4) clusters are dispersed within a perfectly periodic array of all-space filling (Federov solid) @-cages (cubo-octahedral cavities formed by a network of Si04and A104 tetrahedra), but in contrast to the previously reported sodalites (class A, class C),3 they are isolated from each other by sodalite cages containing no anion, forming a class B "supralattice", Figure 1. For a solid solution, variations in p are expected to alter the mean distance between Na4-,Ag,Br clusters while changes in n will shift the constituents of the clusters from those of an insulator to a semiconductor type. Such controlled tunable volume filling and compositional alteration of the contents of the sodalite @-cages provides an unprecedented opportunity to adjust the extent of electronic and vibrational coupling between fixed nuclearity Na+Jg,Br clusters. It should therefore be possible to chemically manipulate the electronic band structure of the materials and thus their electronic and optical properties. Experimental Section The parent sodalites 8Na,(2 - p)Br,pOH-SOD, p = 0-2, were hydrothermally synthesized at 95 OC from aqueous silicate (Ludox 30HS) and sodium aluminate solutions, 4M in NaOH, containing substoichiometric amounts of NaBre4 In order to remove intra-@-cagehydroxide ions, the sodalites were Soxhlet extracted with distilled water for 3 days, leaving (8 - p)Na,(2 -p)Br-SOD. For low Br loadings, silver ion exchange was carried out at room temperature by stirring the sodalite for 24 h in an aqueous solution containing stoichiometric amounts of AgN03. Higher halide concentrations required a AgN03 melt at 230 "C for complete silver exchange. All silver exchanges as well as product filtration, washing, and drying were carried out in the dark. The product crystallinity was monitored by powder XRD at each step. The reactions are summarized in the following scheme: 8Na,(2 - p)Br,pOH-SOD

CLASS c

(c) QUANTUM SUPRALATIlCES Figure 1. (a) The sodalite /3-cage exhibiting the imbibed tetrahedral M4Br cluster. (b) The bcc packing arrangement of @-cagesin the sodalite unit cell. (c) Sodalite quantum supralattices. Class A: All cages filled with M4X quantum Class B: Cages containing isolated M4X quantum dots surrounded by cages with M3 "spectator cationic triangles" (this study). Class C: Cages contain mixed MIX and M4Y semiconductor component quantum dots.3b

it was shown that silver,sodium bromosodalites allow the atomically precise assembly of I-VI1 semiconductor components in a sodalite lattice. By tuning the silver concentration it was possible to cover the range from silver bromide exhibiting molecular behaviour at low silver loadings to an "expanded silver bromide" supralattice at complete silver e ~ c h a n g e . ~ In this Letter we report a series of novel silver,sodium bromohydrdalites of the type (8 - p - 2n)Na,2nAg,(2 -p)Br-SOD, where SOD = Si6A16024,p = 0-2, 2n = 0-8. In these materials ( 2 ) (a) Ozin, G.A.; Kuperman, A.; Stein, A. Angew. Chem. 1989, 101, 373-390. (b) The term semiconductor quantum 'supralattice" is intended to describe host-guest composite materials in which one can structurally identify single sire and shape quantized guests, built of the atomic components of bulk semiconductors,organized in periodic arrays within the cavity and/or channel spaces of a perfectly crystalline host lattice, such as a zeolite. The unit cell dimension of the supralattice matches that of the host, which is not true for a crystallographic 'superlattice". (3) (a) Stein, A.; Ozin, G. A.; Stucky, G. D. J . Am. Chem. Soc. 1990, 112, 904-905. (b) Stein, A.; Meszaros, M.; Macdonald, P. M.; Ozin, G. A,; Stucky, G.D. J . Am. Chem. Soc., submitted for publication.

p)Br-SOD

2nAg* melt

-pNaOH Soxhlct

(8 - p)Na,(2 -

(8 - p - 2n)Na,2nAg,(2 - p)Br-SOD

Results and Discussion Samples containing a halide in one, two, or three out of eight @-cages on the average have been synthesized with the cavity composition varied over the whole range of 2n = 0-8. Chemical analysis showed that the products obtained from the hydrothermal synthesis consisted of both N a 4 0 H and Na4Br occupied cages, and that the amount of halide included in the hydroxysodalites was generally about 30% lower than the amount added to the reaction vessel. The isotherm for bromosodalite favors the inclusion of bromide over hydroxide or waters4 However, as for most sodalites, excess salt is required to form a product in which all cages are filled with monovalent anions. Otherwise, the sodalites contain cages with imbibed hydroxide or water; or completely different phases such as cancrinite can be ~ r e a t e d . ~ The bromide concentrations were virtually identical for a sodalite before and after Soxhlet extraction. Because of their size (rBr- = 182 pm, C.N. 6)5 bromide ions were not washed out through the sodalite six-rings (r = 110-130 pm),6 and their distribution throughout the sodalite lattice could not change after the hydrothermal synthesis. The sodium aggregates could be progressively exchanged by silver ions to yield at full substitution the corresponding mixture of (2 - p)Ag4Br and pAg3 units. A single sodalite phase was observed for a given bulk composition by powder XRD in fully hydrated samples after each reaction step. In a completely silver-exchanged sample containing Br- in l out of 8 cages, less than 0.9 mol % Na was found by chemical analysis after aqueous exchange. Spectra obtained by mid- and far-IR, MAS-NMR, and powder XRD showed single sets of peaks whose positions depended on (4) Barrer, R. M. Hydrothermal Chemistry of Zeolites; Academic Press: London, 1982. (5) Shannon, R. D. Acta Crystallogr. 1976, A32, 751-767. (6) Breck, D.W . Zeolite Molecular Sieues; R. E . Kieger Publishing Co.: Malabar, 1984. Barrer, R. M. Zeolites and Clay Minerals as Sorbenrs and Molecular Sieoes; Academic Press: London, 1978.

The Journal of Physical Chemistry, Vol. 94, No. 18, 1990 6945

Letters

TABLE I: Separation of tons in M, (2 - p)Br-SOD, M = Na+ ( n = O), Ag' unit cell

M Na Na Na Ag Ag As Ag Ag

dimension u 884.8 896.2 893.1 898.0 895.7 895.4 893.1 891.1

P 2 1.7 0 2 1.7 1.5 0.8 0

M-X

x-x

276 289

776 773

280 268 268 267

776 175 713 772

( n = 4)"

M-M intercage 508 484 474 485 48 1 488 487 486

M-O intracage 376 45 1 472 453 457 438 438 436

(framework oxygen) 247 245 236 24 1 249 238 244

"Values in pm. Standard deviations are smaller than the least significant digit shown. Data for n = 0, p = 2 are taken from ref 7. NO

295,

I

I

I

I

'

I

I

I

8%

-886

.U

; 894

" 0 Y

892

2

-

IIO

-'E

fa

-. .I

Y O - i D P t o K E m A ,

&/ut$

&

cEu

2 e

where p = 0-2, 2n = 0-8.

(7) Felsche, J.; Luger, S. Ber. Bumen-Ges. Phys. Chem. 1986, 90, 731-736. Felsche, J.; Luger, S.; Baerlocher, Ch. Zeolites 1986,6, 367-372.

(c)

IW90-

W z

Figure 2. Unit cell size versus composition diagram for the series of hydrated sodalites of the type (8 - p - 2n)Na,ZnAg,(2 - p)Br-SOD,

the sample composition, rather than revealing a superposition of peaks corresponding on the end members of the series (p = 0, p = 2). These results point to a 3-D commensurate, compositionally disordered solid-solution model, instead of one involving ordering (crystallographic superlattice), domains or complete segregation of M4Br and M3 aggregates (M = Na+, Ag+). The (2 -p)M4Br clusters are randomly organized in the sodalite lattice of pM3 "spectator cationic triangles". At intermediate bromide and silver loadings one might expect some segregation of silver ions, if they prefer to bind to halide ions and agglomerate preferentially in those cages containing a halide. If this was the case, the domains formed were too small (50-200 A) to be resolved by powder X-ray diffraction. The response of the sodalite framework to the gradual addition of Br- anions to the unit cell of 6Na-SOD to eventually yield 8Na,2Br-SOD is shown for the fully hydrated samples in Figure 2. The interpretation of the unit cell data requires an appreciation of the role of imbibed water. Neutron diffraction studies on 6Na-SODnH20 (n = 0,8)' demonstrate that the water molecules are both coordinated to the Na+ six-ring cations and hydrogenbonded to the framework oxygens. The first B f entering the Na3 cages to form Na4Br clusters causes the unit cell to expand (partial loss of structural hydrogen-bonding from &cage imbibed H20) while the further influx of Br- up to full loading has a cage contraction effect (electrostatic attraction by Na4Br formation) as seen by the gradual diminution of the sodalite unit cell dimension. Upon dehydration the unit cells of these samples expand, thus providing evidence for the retention of some degree of structural hydrogen bonding in the series of hydrated samples, even in the presence of bromide. In the analogous series of hydrated silver sodalites a slight contraction of the unit cell is observed for higher bromide concentrations (Figure 2). The difference in cell dimensions between corresponding sodium and silver sodalites is small, as these cations have nearly identical radii (4-coordinate rNP+= 113 pm, rAg+= 114 ~ m ) The . ~ exceptional cell expansions after introduction of

I

80-

4 70-

0

WAVENUMBER (cm-')

-

0

&/UNIT

CELL

Figure 3. (a) Far-IR spectra for the series of dehydrated sodalites of the type (8 - p)Na,(2 - p)Br-SOD. (i) p = 2; (ii) p = 1.7; (iii) p = 1.5; (iv) p = 0.8; (v) p = 0. (b) Shift in the far-IR pore opening mode with Brloading. (c) Changes in the far-IR correlation splitting between the high-frequency and low-frequency modes of Na+ (A) and Br- (0)with Br- loading.

*

silver to bromide-free 6Na-SOD indicate that structural hydrogen bonding may be ineffective in 6Ag-SOD. Compared to Na+, Ag+ has a lower tendency to be hydrated (hydration energies: Na+, -406 kJ/mol;8 Ag+, -339 kJ/mo19); one would therefore expect weaker structural hydrogen-bonding involving the Ag+ cations and the framework oxygens. Rietveld refinements of high-resolution powder XRD datai0 were carried out on several samples of the (8 - 2n -p)Na,2nAg,(2 -p)Br-SOD series for n = 0, p = 0, 1.7, 2 and n = 4, p = 0,0.8, 1.5, 1.7, 2. Most samples refined well assuming the space group P43n and a solid solution of cages with or without bromide ions. With the exception of one sample, all R factors were in the ranges wR : 9.8-13.5%, R, = 7.3-9.9% so that good unit cell sizes as welfas the locations and separations between framework and guest atoms could be determined. Refinements were carried out using two types of silver atoms (corresponding to silver tetrahedral and silver triangles with vertices situated on the threefold axes near the sodalite six-rings), as well as one (average) type of silver atom. In the latter case both isotropic and anisotropic temperature factors were applied to the silver atoms. The refinement yielded nearly the same R factors in each case, and data corresponding to the average isotropic silver atoms are presented here. EXAFS experiments are planned to distinguish between the two types of (8) Cotton, F. A,; Wilkinson, G. Advanced Inorganic Chemistry, 4th ed.; Wiley: New York, 1980; p 255. (9) Calculated using data from Weast, R. C. CRC Handbook of Chemistry and Physics, 61st ed.; CRC Press: Boca Raton, FL, 1980. (IO) Rietveld refinement, using the Generalized Structure Analysis System, provided by Larson and Von Dreele: Larson, A. C.; Von Drcele, R. B. LANSCE, Los Alamos National Laboratory.

6946

The Journal of Physical Chemistry, Vol. 94, No. 18, 1990

silver. Table 1 summarizes the results from the refinements. A comparison of the trend in atomic separations with the corresponding far-IR spectra reveals a dependence of the vibrational coupling on the bromide content of these samples. The far-IR spectra for samples with varying bromide content (Figure 3a) show the effect of dropping the population of Na4Br clusters from two to zero per unit cell of sodalite (a unit cell consists of two cubooctahedral cages). The smoothness of this transformation is nicely demonstrated by the monotonic lowfrequency shift of the pore opening framework mode as less halide is present (Figure 3b). The splitting of the Br- anion correlation couplet" seen at 161,68 cm-l decreases as the average separation between bromides becomes greater (larger p ) (Figure 3c). Its intensity subsequently goes to zero. However, the Na+ cation correlation couplet observed at 200 cm-' (E) and 105 cm-' (A,) remains, and each band broadens (inhomogeneous distribution of Na4Br/Na3). Even at one Br- every fourth cavity (p = 1.54), the collective vibrational coupling between Na4Br quantum dots has been severely suppressed. The absence of the correlation splitting in 6Na-SOD" demonstrates the requirement of Br- for effective coupling between the Na+ ions of adjacent @-cages. The correlation splitting between the sodium translational modes decreases slightly as bromide is added to the sodalite. If the coupling depended solely on the separation between Na+ ions in adjacent cages one would expect the opposite trend (Table I), since the Na-Na separation between adjacent cages was found to decrease at higher Br loadings. However, the Na-X separation increases at the same time. This provides further evidence that the anion plays a significant mediation role in the coupling between Na+ ions. Since hydrosodalite, containing only "cationic triangles", exhibits no correlation coupling, it is likely that at dilute bromide concentrations the Na+ coupling occurs predominantly between cages containing sodium tetrahedra. The overall effect is therefore complicated by the extent of connectivity between MBr4 clusters throughout the lattice. Similar, but less pronounced effects can be discerned in the far-IR spectra of the corresponding fully Ag+-exchanged (8 - p)Ag,(2 - p)Br-SOD. The Ag-X distance decreases at higher Br- loadings. One might expect an increase in Ag-Ag correlation coupling. Unfortunately the correlation couplet A, type partner of the E-type silver translational mode at 91 cm-' was not observed in the experimental range down to 50 cm-I, and the above idea could not be tested with the silver samples. 23Na MAS-NMR spectra of hydrated class B sodium sodalites show two components, a sharp band (full width at half-height (fwhh): 310-380 Hz) ca. 1 ppm upfield from solid NaCl and a broad band (fwhh 1200-1350 Hz) ca. 17 ppm upfield from the reference sample, Figure 4. The relative intensity of the former to the latter absorption increases as more bromide is added. In a sample which contains essentially all Br cages, the band at -17 ppm is virtually absent. (A small remnant is an indicator for some defects: hydroxide or water-containing cages. It may be used as a gauge to estimate the concentration of such defects.) In dehydrated samples only one resonance is observed which increases in intensity and shifts downfield with higher Br- loading. This peak is assigned to the 23Na resonance in cages containing Na4Br. The absence of an observable peak corresponding to Na3 cationic triangles indicates that intensity loss has occurred through line broadening as a result of locating the bare Na+ ions in less symmetric environments than hydrated sodium ions.I2 In the dehydrated samples the shift of the NalBr resonance with increasing Br- loading is attributed mainly to deshielding of sodium due to electron withdrawal by the relatively electronegative bromide ions, thereby providing evidence for electronic coupling between Na4Br clusters. Any effects involving charge transfer by framework oxygens must be negligible as they would result in the opposite trend: as the bromide concentration increases. the unit cell size ( 1 I ) Gcdber. J

. O m ,G

~~~

A J Phys Chem 1988, 92, 2841-2849,

4980-4987 ( I 2) Engelhardt, G High-Resolution Solid-State N M R of Silicates and Zeolites, Wiley New York, 1987, p 346

Letters

(a )

I

I

I

Iu)

IW

50

I

0

I

I

I

-tQ

-100

-150

ppm

0 Hydrated

0 Dehydrated

? -15v

,

)

I

I

0

0.5 Br/UNIT

1.0

CELL

I 1.5

2.0

Figure 4. (a) 23Na MAS-NMR spectra of hydrated (8 - p)Na,(2 p)Br-SOD. (i) p = 2; (ii) p = 1.7; (iii) p = 1.5; (iv) p = 0.8; (v) p = 0. (b) The dependence of the 23Na MAS-NMR shifts of the Na,Br resonance relative to solid NaCI on the Br- concentration for hydrated ( 0 )and dehydrated (0) samples.

and T-O-T angles both decrease,I3 allowing more charge density to relocate from the sodalite-cage lattice six-ring oxygen to Na+(3s), especially as the Na-O separation is reduced at the same time (Table I). As in the case of the unit cell sizes, one has to consider the role of imbibed water to interpret the 23Nachemical shifts of the hydrated bromohydrosodalites. In these samples, as the Br- loading increases the water content correspondingly decreases. The structural water in the cages containing cationic triangles has the effect of deshielding the sodium cations in bromide-containing @-cages. This occurs presumably by withdrawing charge density from the framework oxygens via hydrogen bonding. Optical reflectance data for (8 - p - 2n)Na,2nAg,(2 - p)BrSOD,where p = 0-2, 2n = 0-8, are shown for the hydrated samples in Figure 5. The series p = 2,2n = 0-8 serves as a pivotal control group to pinpoint those absorptions associated with m = 0-3 Na3-,Agm "ionic triangles", which all fall at 220-240, 300 nm (Figure 5A). For one Br- in every eighth sodalite cage, the line widths of the Nab,Ag,Br cluster UV-optical excitations are so narrow compared to the parent case of one Br- per cavity (Figure 5B) that individual components for n = 0, 0.05, 0.5, 1, 4 can be resolved around 230,248,2501265, and 255/210/320 nm, respectively. When these data are compared with those of class A quantum supralattices3 (Figure 5, Ce, De, Ee), one realizes that all of the resolved components of the class B supralattice are contained within the spectral envelope of the former. ( 1 3) Weller, M . T.; Wong, G . J . Chem. Soc.,Chem. Commun. 1988, 1103.

The Journal of Physical Chemistry, Vol. 94, No. 18, 1990 6947

Letters

200

3(30

400

WAVELENGTH (nm)

1

~

,

k

;

w

WAVELENGTH ( n m l

Figure 5. Optical reflectance data for some members of the hydrated sodalite series (8 - 2n -p)Na,2nAg,(2 -p)Br-SOD. (A) p = 2; (B) p 5 1.7; (a) n = 0; (b) n = = 0.05; (c) n = 0.5; (d) n = 1; (e) n = 4. ( C ) n = 0.5; (D) n = 1; (E) n = 4: (a) p = 2; (b) p = 1.7; (c) p = 1.5; (d) p 5 0.8; (e) p = 0. (F) Progression of the UV-visible spectra from the sodalite encapsulated isolated AgBr molecule to the isolated Ag,Br cluster and to the

extended (AhBr), quantum supralattice. At low silver and bromide loadings a very sharp UV spectrum resembles that of an isolated gas-phase silver bromide monomer (Br-(4p),Ag+(4d) Ag+(5s)). The line broadening and red shift of the band edge observed when either the silver or the bromide concentration is increased indicates electronic coupling between Na&,Ag,Br clusters (Figure SF). Whereas sodium does not appear to contribute much to this coupling (the absorption bands are sharpest a t high sodium concentrations), both bromide and silver must be mediating the communication between clusters. The Ag-Br and Ag-Ag separations (within each cage) change most drastically between p = 1.7 and p = 1.5. However, the most significant broadening of absorption bands in the optical spectra occurs only after the p = 0.8 loading level. This is an indication that the major mechanism leading to band broadening is not an increase in orbital overlap of the ions within a cage but rather greater communication between clusters in adjacent cages, as the number of bromide centers rises and their separation decreases. Broadening due to phonon coupling is an additional factor; upon

-

cooling Ag,Br-SOD to 27 K three major components could be resolved completely and further components partially. Even though the framework may contribute to long-range through-bond interaction, the gap corresponding to the O(2p) AI(III)/Si(IV) ligand-to-metal charge-transfer region occurs at higher energy ( 1 90-220 nm), and no direct effects were observed. Luminescence lifetime and time-resolved luminescence spectroscopy experiments are planned to investigate the origin of the UV-optical absorptions, their relaxation dynamics, and the nature of the interactions between the sodalite constituents in more detail.

-

Conclusions These data for the first time provide compelling evidence for the existence of collective electronic and vibrational coupling interactions between Na.+,,Ag,,Br clusters over the full Ag+ and Br- loading ranges, as well as from the isolated silver bromide molecule to the expanded silver bromide quantum supralattice. Furthermore, they reveal the genesis of the Br-(4p),Ag+(4d)

J . Phys. Chem. 1990, 94, 6948-6956

6948

minivalence band and Ag'(5s) miniconduction band on passing from embryonic Na4-,Ag,Br clusters to a quantum supralattice built of Na.+,,Ag,Br clusters, and ultimately to the 'parent" bulk mixed sodium-silver halides.

Acknowledgment. The financial support of the National Science and Engineering Research Council of Canada (G.A.O.,

AS., P.M.M.), the Ontario Centre for Materials Research (P. M.M.), and the Office of Naval Research (G.D.S.) is gratefully acknowledged.

Supplementary Material Available: Tables of crystallographic data for (8 - p - 2n)Na,2nAg,(2 - p)Br-sodalites (4 pages). Ordering information is given on any current masthead page.

FEATURE ARTICLE Weak Interaction Potentials of Large Clusters Developed from Small Cluster Information Clifford E. Dykstra Department of Chemistry, University of Illinois, Urbana, Illinois 61801 (Received: March 1, 1990)

The task of finding readily calculable potentials for intermediate-sized and very large clusters and molecular assemblies is important in view of their need in surface, droplet, and condensed-phase simulations. However, the most detailed probes of weak interaction, which are gas-phase or matrix isolation experiments, involve small, isolated species and mostly binary complexes. The connection between the information from those experiments and the potentials for many interacting molecules can come about only with a clear, physical understanding of the major elements of weak interaction. Standard ideas about electrical interaction have often been useful for understanding small clusters, especially where the electrical analysis is done at a high level. This seems, therefore, to be a most useful physical basis for developing potentials. A particular aspect, the role of electrical polarization, has been found to be important in certain ways in a number of binary clusters, even though it is usually a small energetic contributor. More significant is that polarization may be the crucial link with clusters of more than two submolecules since it is necessarily a nonpairwise or cooperative interaction. Within this framework, the strategies for extrapolation of small cluster pictures to realistic potentials for large clusters are considered.

I. Introduction A . Weak Interaction. One of the exciting areas of contemporary chemical physics is the study of clusters of atoms and molecules. Even with attention limited to weakly interacting assemblies of light species, meaning atoms or small molecules of the elements hydrogen through argon, we can see tremendous experimental and theoretical effort. (See,for example, the papers collected in ref 1 .) In fact, the cause of investigating these clusters seems to have stimulated the development of a number of new experimental techniques and, at least in the author's experience, of new theoretical approaches. What makes weak clusters so important as to be the focus of many experimental and theoretical programs? Certainly, they are a challenge to the skills and understanding of modem chemical physics. Their floppy nature, fleeting existence, and sometimes unexpected structures make for exquisite complexity in the spectroscopic analysis and in understanding their dynamics. Even the conditions and mechanisms for producing clusters of a desired size or composition are difficult to uncover. Beyond the challenge they present, there is, or they may be, quite fundamental information locked into these clusters. That information is embodied or manifested in the interaction potential surfaces, and extracting that information is an ultimate goal of much of the experimental work in this area. Klemperer was one of the first to recognize this, and with co-workers, carried out one of the classic spectroscopic studies of a weak complex in work on the hydrogen fluoride dimer published in 1972., In the almost two decades since, the ( I ) Weber, A.. Ed. Sfrucrure and Dynamics of Weakly Bound Complexes; NATO AS1 Series C212: Reidel: Dordrecht, Holland, 1987.

study of weak complexes has mushroomed. The fundamental information that might be locked into small complexes is the nature of intermolecular interaction of the sort that is weaker than chemical bond formation. While it is an interaction that leaves the structures of the interacting species largely unaffected, its effects are pervasive in chemistry. It has a role in the structures of solids and liquids, and it is important in surface physisorption and for bringing about crucial, small energetic differences in biomolecular processes (e.g., via solvation of biomolecules). One could easily get carried away and suggest that in our world weak bonding is equal in importance to chemical bonding; however, it may be more acceptable to just say that we need to understand weak bonding as well as we understand chemical bonding. There is big step between a complex such as (HF),and a protein in water or between (H,O)*and liquid water, and so we must be concerned whether what can be learned about small cluster interactions applies only to small clusters. There are two complications that could interfere with the connection between the small clusters and the large, extended systems. First is the possibility of a number of distinctly different kinds of weak interaction, and then, the net features might result from a composite of effects whose balance is somehow different between the large and small cluster regimes. Coulson brought this out in an early effort to understand the water-water p ~ t e n t i a l and , ~ he concluded that, in fact, the net interaction was a near balance of several competing (2) Dyke, T. R.; Howard, B. J.; Klemperer, W. J . Chem. Phys. 1972,56, 2442. ( 3 ) Coulson, C. A. Research 1957, 10, 149.

0022-3654/90/2094-6948%02.50/0 0 1990 American Chemical Societv