Compounds of Alkali Metal Anions By James L. Dye"' Anions of sodium. potassium, rubidium, and cesium are stable both in suitable solvents and in crystalline solids. The latter can be prepared either by cooling a saturated solution or by rapid solvent evaporation. Thermodynamic arguments show that alkali metal anions can probably exist in saturated solutions of the alkali metals in any compatible solvent. but that below saturation, dissociation into the cation and solvated electrons is favored in highly polar solvents such as ammonia. The key to solvent-free salts of the alkali metal anions is stabilization of the cation by incorporation into a suitable crown or cryptand complex. By using such complexes it also appears possible to produce "electride" salts in which the charge of the complexed cation is balanced by a trapped electron. The chemical, electrical, and optical properties of salts of the alkali metal anions and "electrides" could provide useful applications. 1. Introduction Alkali metals are exceptionally good electron donors, tending to form monopositive ions by donation of one electron to an acceptor atom or molecule. The resultant cation has a stable inert gas electron configuration with a filled outer p-shell. Because of this, the history of compounds of the alkali metals has been that of the cations and it was long thought that, except in metals and alloys, only the + 1 oxidation state could exist. Yet the existence of alkali metal anions in the gas phase has been recognized for 30 years['.21.These anions were proposed as major species in metal-ammonia solutions in 19651' 41 and in metal-amine solutions in 1969"'. The isolation and characterization of a crystalline N a salt["-'' in 1974 firmly established that the - 1 oxidation state of alkali metals could exist in condensed phases. This has opened a new field of research on the properties of these highly-reducing species both in solutions and in solids. It is the purpose of this article to describe the current status of research in this field and to speculate on possible future developments. 2. Properties of Alkali Metals in Solution The preparation of salts of the alkali metal anions has its origins in the study o f metal-ammonia solutions and solvated electrons[x ' I 1 . However, the discovqy of alkali metal anions may have actually been delayed by the fact that ammonia is "too good" a solvent so that the dissociation reaction lies far to the right in dilute solutions and most properties are only mildly influenced by the cation. This does not mean that metal-ammonia solutions are simple: quite the contrary; ion-pair formation, electron spin-pairing and the onset of metallic behavior occur as the solutions are concentratedlxl. Because of these complications, alkali metal anions could be present in concentrated metal-ammonia solutions, but there is no specific evidence of such species in this solvent. 2.1. Solutions in Amines and Ethers When any alkali metal except lithium is dissolved in an aliphatic amiae, poly-ether or hexamethyl phosphoric triamide (HMPA), two prominent optical absorption bands are observed. One, in the infrared region at 1200-2000 nm, is the relatively metal-independent band of the solvated electron. The other band, whose position depends upon the metal, solvent and temperature, can be assigned to the alkali metal anion"'. The proof of this assignment has been discussed in detail elsewhere[".7.'2 and will be included here only as needed for the description of specific systems or properties. Progress in identifying the species in metal-amine and metal-ether solutions was hampered by low solubility, rapid decomposition. and the ease of extraction of sodium from borosilicate glassl'xl.The problem of low solubility was overcome by the use of crown ether'"' and cryptand["'' complexing agents. Extraction of sodium from glass presents a major problem only at low concentrations and can be completely eliminated by using a fused silica apparatus. Unfortunately, problems arising from solution decomposition are still with us but can be minimized by working with rigorously clean vessels[*'' and at reduced temperatures. 2.2. Optical Spectra Each alkali metal anion is characterized by a broad, intense, structureless absorption band which depends upon metal, solvent and temperature and is strongly asymmetric on the high-energy side. These features are characteristid5' of the so-called charge-transfer-to-solvent (ctts) bands[22.231of other anions such as the halides, hydroxide, amide. etc. The spectra are also similar in shape and in solvent-'"' and temperature-dependence to those of the solvated electron in various solvents["] but the absorption maxima occur at higher energies. Figure 1 shows the spectra of Na ~,K ~,Rb , Cs and e,,, in ethylenediamine'24'. The spectrum of e,;,l> obtained by pulse-radiolysis"'1 (solid circles) compares well with that obtained by dissolving lithium (solid line) in ethy~ [*I Prof. Dr. J . L. Dye Department of Chemistry. Michigan State University Eaht Lansing. Michigan 48x24 (USA) 16 14 12 - I 8 10 v 07 06 I I I I I I 12 14 I 6 d [pml 10 09 08 CcmC 10.~1 I 1 I 76 18 20 - Excited states of the anion might lie above the ionization limit in the gas phase. yet be observable in solution or in crystals, either because of a strong perturbation of the continuum level or because of low probability for direct excitation to the continuum level of the solvent. In fact. auto-ionization apparently occurs from excited states of the alkali metal anions and other anions in solution[2h"I suggesting that the continuum level lies at lower energies than does the first excited state. It is probably appropriate to consider the intense absorption band of M in solution and in crystals as originating from an s to p transition, modified by the presence of solvent molecules or neighboring ions and including a contribution from direct bound-continuum transitions["I. 05 2.3. Conductivity Fig. 1 . Optical \pectra 01. Li. N d . K. Rh. and C s in cthylrnedianiine 1241 w t h identification of absorption peaka 0 1 N a . K . R h . Cs and e-<,,,.Solid clrcles are for e ~ , , ,produced \ by pul\e radiolysir 1251. lenediamine. The infrared shoulders obtained for solutions of K, Rb and Cs are attributed to the solvated electron and the ratio of this absorbance to that of the anions depends strongly on concentration (see Eq. 1 ). The solvent-dependence of the peak positions is illustrated by Figure 2, which compares the peak positions of Na , K - and e,,,,, in various solvents. The solvent-dependence of the absorption maximum of e,,,, correlates in turn with that of I I2'I. The conductivities of sodium and cesium solutions in ethylenediamine""' and methylamine""' have been measured. The optical spectra show that sodium solutions in these two solvents consist primarily of Na ' and N a - while dilute cesium solutions contain mainly Cs and e,,,,,. The differences in conductivity of these solutions are striking-as shown in Figure 3. The rapid drop in conductivity with in+ I 3h ~ 154 , I 150 , I I -6 OEACD ME O, I y I:"-// 1 30 - o HMPA o I 4kDdA ~ .0 7 u ,1 111,2L6- I UDME U D E A THF DEY' HMPA DIPE Fig. 3. Product of equivalent conductance ( A ) and solvent viscosity (TI) for solutions of Na in methylamine I 3 , and ethylenediamine A , and of Cs in methylamine 0 . ethylenediamine x . and ammonia 0 . creasing concentration for solutions containing Cs and e,,,,, in the amines resembles the behavior in ammonia and can be readily accounted for in terms of ion-pairing between Cs+ and e - . The high limiting conductance of e,,,,,, in both ethylenediamine and methylamine is also similar to the behavior in ammonia especially when the appropriate viscosity corrections are made as in Figure 3. The conductivities and the similarity in optical spectra thus provide strong evidence that metal solutions in methylamine and in ethylenediamine which d o not contain alkali metal anions are similar to metal-ammonia solutions. By contrast, the conductivities of sodium solutions in methylamine and ethylenediamine show that the mobility of Na-- is "normal" and that the extent of + Maialun, Golden and Ottolenghi[51 showed that the wavenumber of maximum absorbance of the alkali metal anions varies inversely with the estimated anion radius, in accord with the predictions of ctts theory. The observed temperature-dependence of the peak position is also predicted We conclude that the absorption spectra by this of M in solution are similar to those of other anions. An examination of ctts theory might lead one to conclude that the excited state bears no relation to excited states of the isolated anion. This does not necessarily follow. however. ~ 588 Anxew. C l e m . In!. Ed. Engl 18. 587-.798 (1979) ion pairing between Na' and Na large interionic contact distance. is small, as expected for a spectra obtained in three solvents[331.The calculated diamagnetic shift of 6 = - 2.6 for Na compared with Na(g)13'.3hlis - I 2.4. The Effect of Crown Ethers and Cryptands One of the major barriers to experimental studies of alkali metal solutions in amines and ethers is the generally low soiubility. Combined with the ever-present tendency towards decomposition, this low solubility has made reproducible measurements difficult. By using crown-ether and cryptand complexes such as those of crown-6 ( l ) ,and C222, we H- 1-1. C r y p t and s C222, m = n = o = 1  crown-6 I 80 were able to dramatically increase metal solubilities in a number of solvents['2.'y~20.3il. For a cation complexing agent (Cry = cryptand) the equilibrium (2) shifts the equilibrium (3) to the right, thus increasing the concentration of M - in M + + Cry 5 M + C r y 2M,,, e M++M ~ , I LO 20 -s 0 I 1 -20 -10 Fig. 4. "Na-NMR spectra of Na ' C222. Na solutions ( ~ I M0) in three m vents . All shifts are referred lo Na,,,. Positive shifts are paramagnetic. The peaks of Na ' C222 at the left are at a = t49.5.T h e peaks of Na at the right dre at a = -2. very close to the measured shifts[331 of 6 = - 1.4, - 1.6, and - 2.3 for Na in methylamine, ethylamine and tetrahydrofuran respectively. This is to be compared with paramagnetic shifts of 60.5, 72.2 and 50.4 for N a + in water, methylamine and the cryptate complex respectively. The absence of an appreciable paramagnetic shift for N a - in solution proves that the solvent is prevented from interacting appreciably with the 2p electrons by the presence of the outer 2s electrons. This, in fact, provides the strongest evidence that Na in solution is truly a n alkali metal anion rather than, say, a solvated cation with two solvated electrons in the vicinity or a solvated cation attached to a solvated electron pair. The NMR spectra[331 of Rb in ethylamine and in tetrahydrofuran (THF) show single lines at 6 = +26.2 and +14.4 ppm respectively relative to Rb(g) while the peak for Cs- in T H F occurs at 6 = + 52.3 ppm relative to Cs(g). These are to be compared with values of 6 = +213 and + 348 for aqueous R b + and Cs' respectively. Exchange effects1"' sometimes prevent the observation of the N MR signal of M and may be the reason that we have not yet been able to observe K by N MR even in solutions whose optical spectra show the existence of high concentrations of this species. - the saturated solution. (This also applies for crown ethers as complexing agent.) When excess metal is present, equilibrium (4) determines the relative concentrations of M - and M I 60 * M,,,+e,,, (4) e,:,,,. In the absence of excess metal, equilibrium (1) governs the relative concentrations of M - and e,,,. The addition of a cation-complexing agent shifts reaction (1) to the right by reducing the concentration of M +. Equilibrium (3) requires that the product of the activities of M - andfree cation cannot exceed the limit imposed by the solubility of metal. These reactions permit a measure of control over the composition tion of the solutions. The maximum value of the ratio [M -]/ [e,,,,] for a given metal is determined by the solvent and the temperature according to reaction (4). If this ratio is large, solutions consisting of primarily complexed M ' and M can be prepared by maintaining an excess of metal in the preparation vessel. Alternatively, if K,[*Iis not too small and K2 is large, the addition of equimolar amounts of complexing agent and metal can yield solutions whose composition is mainly complexed M + and e,,,. ,Section 3 considers the thermodynamics of these reactions In more detail. - - - - 2.6. Other Solution Properties of Alkali Metal Anions The rate of reaction of Na with water in ethylenediamine is much slower than that of eSirvand the reaction order in The ability to prepare concentrated metal-amine solutions water is also different(2'.'71. The rate of formation of Na-. (up to 0.2 M and higher) by using crown ethers and cryptands from solvated electrons and Na' has been studied by pulse opened the door to NMR studies of alkali metal a n i ~ n s [ ~ ' . ~ ~ l . r a d i o l y ~ i s ~ ~the ~ . 'reaction ~~; is second order w.r.t. el,,,,. The It was k n ~ w n ~ from ~ ~ . NMR ~ " studies with salts that the reldependence of the rate on the concentration of Na is comease of Na' from a Na'C222 complex is slow o n the N MR plex and suggests['41that electron-pairing precedes the fortime scale, so that separate "Na-NMR signals were expected mation of N a - . Photolysis of M - in the wavelength region and found for Na Cry and for Na- . Figure 4 shows the of the absorption maximum leads to the production of solvated electron^'^' 471, sometimes in a spin-polarized ['I KI IS the equilibrium constant for reaction ( I ) state[40.46.471 (CIDEP). Photoelectron emission has also been 2.5. Nuclear Magnetic Resonance - + + Angen G e m . Inl Ed. Engl. I d 587-598 (1979) 589 from e,;,,, and from Na in HMPA. These observations generally support the assumption that alkali metal anions are new species and not just a combination of the solvated cation and two solvated electrons. 2.7. Ion pairs and "Monomers" in Solution The electron spin resonance (ESR) spectrum of the solvated electron consists of a single narrow line with a g-value near that of the free electron. When the electron interacts strongly with an alkali cation so that appreciable electron density exists at a metal nucleus with spin I. and when this electron-cation pair has a long enough lifetime (generally microseconds or longer) then a hyperfine pattern results with 21+ 1 nearly equally spaced lines[4x"I. Pulse radiolysis[iX.52.531 and p h o t o l y ~ i s [ ~studies '~ in the presence of alkali cations have shown that the optical spectrum of e,:,,, is blue-shifted by formation of the ion-pair (also referred to as an alkali metal "monomer"). The magnitude of this blueshift increases as the polarity of the solvent decreases and correlates well with the electron contact density at the alkali nucleus as determined by ESR measurements[5". Changes in the cation-electron contact density and in the spectrum with solvent and temperature have been attributed to the formation of solvent-shared and contact ion-pairs"". Although the sensitivity of the ESR technique permits easy detection of the monomer species, the equilibrium concentration of monomers, M, in amines and ethers is generally small compared with M - and e,o,v. Crown ethers or cryptands shift the equilibrium M+M+ t e - (5) for salts of the alkali metal anions (which we refer to as "alkalides") was also based on thermodynamic predictions of their stabilities and we continue to be guided in the search for appropriate metals, solvents and complexing agents by approximate calculations of stability constants. Questions have also arisen about the possible existence of anions of other metals such as the alkaline earths. I t is a simple matter to show thermodynamically that the existence of an ion such as Ca in solution is thermodynamically impossible while the gold anion, Au -, may well be thermodynamically stable. In this section some of the pertinent thermodynamic arguments will be given. Some of the predictions can be made with reasonable accuracy; others are far less certain, either because key measurements have not yet been made or because of limitations of the theory of solvation free energies or lattice energies. In either case, it is often possible to make reasonably accurate relative calculations which at least permit comparisons among the alkali metals. 3.1. Stability of Metal Anions in Ammonia as Solvent It is possible to calculate accurate values of AH" and AG" for the process from electron a f f i n i t i e ~ [ ~properties .~~], of the solid and gaseous and statistical thermodynamlcs calculations of entropies['"]. The values of AH',' and AG',' for the alkali metals and some other metallic elements at 25 " C are given in Table lI*]. Table I . Thermodynamic estimates [a] used to Judge the stability of.anions in ammonia to dissociation into metal and solvated electrons Metal Estimated anion radius r Li Na K Rh cs Au AL: cu Ba PI Te Ph 61 TI Sh Sn Electron Affinity A ~! A 6:: AH! A G<., (A) 59.8 52.9 48.4 46.9 45.5 222.7 125.7 118.3 - 52.1 205.3 f83.3 101.3 101.3 4X.2 101.3 120.6 2.35 2.72 3.27 3.39 3.55 2.00 2.00 I .93 3.5 3.55 2.3 2.2 2.2 2.1 2.2 2.2 [a] All values given in kJ mol I: 96.7 53.4 39 I 36.4 30.1 140.9 156.4 217.5 225.4 357.5 I1 0 91.2 103.3 131.5 15X.6 179.0 67 7 30 6 18.2 14.4 10.9 109.0 125.4 1x5 7 202 3 320.6 - 20.x 66.0 12.3 104.6 126.3 152.2 - 65.8 - 18 I 298 K 23X K 1.x 5.6 13.5 15.2 I x.5 5. I 21 4 72.5 207.9 32X 2 -91.4 - 14 1 - 8.4 12.8 45.6 7I .4 - 1I.X - 20.2 - 1.1 - 59.3 56.3 54.9 - 61.5 - 46.0 5.5 137.7 27 I .9 - 156.5 - X6.X - 14.1 5x I - 195 - I.o 0.X 38 - x3 1.9 59 0 193 7 3 16.9 - 104.5 - 29.2 - 21 8 14 32 5 51.3 AH:= 247-535(1/r); Ad:= 334- 51 I ( I / r )evaluated by fitting data for the halide ions (see text). to the right in the absence of excess metal. Although the concentration of monomers is low, they may play important roles in the exchange kinetics of complexed M i and M and in the rate of decomposition of metal solutions. If the solvation free energy of M,, is more negative than that of e;, by an amount equal to or greater than that given in Table 1 then the reaction e , , , + M,., - M,,, (7) 3. Thermodynamic Considerations metal anions in ammonThe proposed existence Of ia[3.41was based on thermodynamic arguments. Our search 590 ['I Many o l t h e thermodynaniic ealimates described in this article have not been previously puhliahed. Details of the calculations arc dvallahle f'rom the author on requebl Angew. Chem. Inr. Ed. Engl. I H . SH7-398 (1979) is feasible and we can expect M - to be stable to dissociation into the (solid) metal and solvated electrons. According to the Born e q ~ a t i o n l " both ~ the Gibbs free energy of solvation and the enthalpy of solvation of an ion are inversely proportional to the ionic radius. Since the Born equation is based on a continuum model and fails to account for specific solvation of the ions, it is not appropriate to use it directly. A better procedure is to use the known free energies and enthalpies of formation of CI-, Br- and 1- in ammonia[5x1together with the thermodynamics of formation of the corresponding gaseous anions to calibrate the dependence of the solvation enthalpy and Gibbs free energy upon radius. Specifically, AC$ and AH'; for are linear functions of the reciprocal of the anion radii of CI-, Br . and I -. These thermodynamic functions were obtained by combining the reactions For ions which do not have the added electron in an s-orbital, the calculations predict stability for T e - , Pb , Bi . and TI- . The first three of these elements do form anions in ammonia'h'.h21which have also recently been isolated in crystalline salts[" "I, but in these cases the anions are polymeric rather than monomeric. Salts which contain polymeric an0ur ions of antimony and tin have also been i ~ o l a t e d ~. ~ '~~'~~~~ calculations indicate instability of monomeric Sb and Sn in ammonia but the addition of metal-metal bonding would stabilize polymeric anions. The existence of detectable concentrations of such ions as Ba- and Pt- in ammonia is clearly impossible from a thermodynamic viewpoint. In the former case, the negative electron affinity of barium shows that Ba- is not even stable in the gas phase. On the other hand, while platinum has a large electron affinity the lattice energy is so great that conversion of Pt,,, to Pt- by the action of solvated electrons is not thermodynamically permitted. - 3.2. Dissociation of M Solution XI,] +el,, + XI,, (11) The values of AG?; and AH': were obtained from the tabulation of Latimer and Jolly[5x1and are based upon AG: = 0 and AH'; = 0 for H in ammonia. The values of AGO for reactions (10) and (1 1) were obtained from known atomization free energies and enthalpies["I of the halogens and their electron affinities[s41.By assuming that the solvation free energy and enthalpy of other anions depend upon radius in the same way as those of C1-, Br- and I - , we can calculate AGA and AH': for any anion if we know its radius. Following the suggestion of Malalon, Golden and Ottolenghi'51 we can estimate the radius of an alkali metal anion by subtracting its ionic radius from the interatomic distance in the metal. The radius of Au- (2.0 A) was obtained by subtracting the radius of Cs' from the cesium-gold distance in the ionic compound CsAulSY1.The values'sxl AG= - 182 kJ mol ' . A M , l = - l 5 9 k J m o l - I f o r N a ' i n N H 3 a t 2 5 " C a n d AH','= - 182 kJ mol-' for eYNH,,combined with A G : 2 ~ 0 . 0 at 25 oC16cr1 for the process + - to M + - and Solvated Electrons in The calculations described above were made for ammonia solutions because we have sufficient data to calculate Gibbs free energies and enthalpies of formation of ions in ammonia, but not in amines and ethers. Similar calculations could be made for amine and ether solvents if sufficient solubility and/or e.m.f. data were available. Yet ironically, we have no XpecQ'ic evidence for the existence of alkali metal anions in ammonia. Let us examine the predicted concentrations for the most favorable case, that of Na -. From the data in Table 1 we can calculate that for the reaction at - 35 "C, K I 4= 3 x lo4. However, in a dilute solution away from metal we must also consider the equilibrium for which K I 5 = 4x lo-' M *. This predicts that at a sodium M (neglecting the effect of activity concentration of coefficients) the ratio [Na-]/[e-] should be about 0.15. If the optical spectrum of N a - were appreciably different from that of e in ammonia as it is in amine and ether solvents, then the presence of this concentration of Na would be easily detectable. Certainly, upon addition of a sodium salt the presence of N a - should be observable. However. no prominent optical band of N a - has been observed either in the presence or absence of sodium salts. We must conclude that either the calculation given for N a - in Table 1 is in error by 15 kJ mol-' or more, or the optical band of Na is shifted into the infrared in ammonia as originally suggested by Golden et al.[31. We conclude that, while such calculations as these cannot be used to determine concentrations, they do demonstrate that the existence of alkali metal anions in ammonia, amines, and ethers is not prohibited on thermodynamic grounds. ~ - gives AH':3= AGy3 = - 182 kJ mol- I for at 25 "C based upon the convention that the enthalpy and free energy of formation of HG,,, are zero. Addition of reactions ( 6 ) , (8) and (13) gives reaction (7). The values of AH'; and Ad: at 25 " C and Ad: at -35 "C for a number of metallic elements are given in Table I. The results show that the reaction of solvated electrons with metal to produce ammoniated metal anions with filled outer s-orbitals may be thermodynamically permitted for all of the alkali metals and for gold1*Iand silver. It appears that Cu would not be stable to dissociation into Cu,,, and e,;,,. Angen: ('hpiit Int. Ed. Engl I X . 5x7-598 llY7Y) ['I W. J. Peer and J. J. Lagowki [J. Am. Chem. Soc. 100. 6260 (1978)) have recently shown that Au can be formed in liquid ammonia. 591 as good a solvent as water, means that the gas phase complexation reaction 3.3. Stability of Metal Anions in Amine and Ether Solvents Alkali metal solubilitiesl2'I and s p e ~ t r a l * ~ . have ~ ' ] been measured for ethylenediamine solutions at 25 "C. The extinction coefficients of N a - and e5& in ethylenediamine are '~~. 8.2 x lo4 and 2.0 x lo4 M ' c m - r e s p e ~ t i v e l y " ~ -These measurements permit estimation of the equilibrium constants K , , K 3 , and K, for ethylenediamine as solvent (see Table 2). Although (excluding Li-) N a - is most stable and K - is least stable relative to e,;,, in the presence of excess metal, Table 2. Estimation of equilibrium constants K , , K , and Metal Path lenght icm) Solubility (M) Li Na 0.29 2 . 3 9 ~in ' 1.04~10 K Rb 1.31~10 cs 5.4x 10 ' 'M. + M,,,, + M,,,, M :Cry Mi,,, 2 0.09 <0.05 [bj 0.Y I 0.46 [c] 0.6s [c] In the presence of' I (M (241 1691 (17) + 3.4. Stability of Metal Anions in Crystals Attempts to grow crystals containing alkali metal anions were based upon their stability in solution, and the realization that stabilization of the cation by its inclusion in a cryptand cavity might prevent the spontaneous recombination of M' and M - to form the metal. The fact that alkali cations are strongly complexed by the appropriate cryptand, even in K ; [aj '1 (241 1691 1241 1241 K7 (hl I) > 4 x 10 3x10 I.hX10 6 2.3 x 10 '' 1 . 7 ~ 1 0" 1 O X 10 ' 4x10 (691 0.40 0.6K 0.9 x 10 ' M K, Ref. ' t2x10 1 . 4 ~ 1 0" 2.7 x 10 ' 2.7x 10 ' 43x10 ' 4.3 x i n 7x10 - c7 220 41 3.4 50 20 5 added K ' . position into its components Cry, M, and M,. By adding all of the AH" and AG" values for the reactions listed in Table 3 one obtains AHY7 and AG?,. While the absolute values of AH:', and AG:7 given in Table 4 may be in error by as much as t 5 0 kJ because of errors of measurement and the various approximations made, their relative values should be much more reliable. Since we can prepare the crystalline salt N a + C 2 2 2 N a - , even in the presence of an excess of sodium and cryptand, it appears to be thermodynamically stable. The calculated value of AG?, for this salt is +32 kJ mol I which is within the estimated absolute error of the value A G ~ , s O required for thermodynamic stability. It appears safe to assume that any proposed salt with a value of AGO, equal to or more negative than that of the reference salt Na C222 Na is likely to be thermodynamically stable. Of course, these calculations do not preclude decomposition via destruction of the cryptand moiety, a process which we have found to be very common. A critical quantity in these calculations is the interionic distance in the crystalline salt. The sum of the calculated raand rNa is larger than the measured distance in dii rN1 the salt Na C222 Na by 14%.This could be caused by lack of sphericity of the cryptated cation, an error in the estimate of its size or that of N a - , and/or compression of these relatively "soft" ions in the crystal. The cation-anion distances in Na+C2221-[761and K'C222Iare within a few percent of those calculated on the basis of the ionic radius of I - and the radii of Na C222 and K C222 calculated from the ligand thi~knessl'~'.Therefore, the most likely source of error in the estimate of the interionic distance in NaiC222Na- is an error in the estimated radius of N a - or neglect of the compression of this ion and of the cryptated cation in the crystal. In order to compensate for this reduction in interionic distance, all of the calculated anion radii were reduced by 14% in computing the lattice energy and entropy in step (h) of Table 3. + iCr, ~ + + 592 + in which M, and M, may be the same or different. If AG';, is negative then the salt M: Cry M; is stable to decom- A hl the concentration of M in the saturated solution increases regularly from N a - to C S C . The dissociation of M to M' + 2es;,, is least likely for N a - and most likely for K - . This apparently anomalous position of potassium solutions seems to reflect extra stabilization of the solvated electrons in the presence of K +,perhaps oia ion pairing. Such an effect has actually been observed both by the addition of a potassium salt to a solution of sodium in ethylenediaminel"] and by pulse radiolysis The appearance of the spectral peak of the solvated electrons occurs in this case but not when Rb or Cs are added. Solubility and spectral data in other solvents are too sparse to permit evaluation of equilibrium constants. However, it is clear that the values of K 2 increase while K , and K 3 both decrease as the solvent polarity decreases. Thus, all metals except lithium when dissolved in ethylamine show primarily the optical band of M - with little contribution from e , , , . Based upon the effect of C222 on the optical spectrumi701we estimate that K , < iO-*O for sodium in tetrahydrofuran. Therefore, solutions of sodium in this solvent, even in the presence of excess cryptand consist essentially of Na'C222Na- with no detectable solvated electron concentration. By contrast, solutions of potassium in tetrahydrofuran with cryptand contain largely K'C222K- when excess potassium is present, but mainly K' C222e,;,, when an equivalent of cryptand is present[701. + Cry,,, A, 2.0 10 (16) M + Cry,,, for reactions of alkali metal anions in ethylenediamine at 25 "C from optical spectra and solubiltties. 0.2 0.84 0.36 0.6 x + is strongly exothermic and highly favored. There are no direct data for this reaction, but indirect methods can be used to evaluate AH:, and AG:, for the reaction Absorbance Data Used 01 10 I .0 0.01 1 .0 01 0.01 [a] K,= K , KS. [bj Estimated conc. of L1=3 K7 MfLi+ Cry,,, + Angew. Chem. In(. Ed. Engl. I X . 587-598 (1979) Table 3. Steps in thermodynamic cycle used to evaluate AH';, and AS:',. - Redction Ytep Origin of ddta or estimate Typical data source Enthalpies of complexation and equilibrium constants Solubility of cryptand a s a function of temperature. enthalpies of solution of pure cryptand in water At€' and AG" of atomization plus ionization potentials of M,,,. Use Sackur-Tetrode equation for entropies of gaseous speciea. Tabulated hydration enthalpies and free energies Born equation plus estimate of size of M ' Cry from ligdnd thickness Sum of a-e AH" and AG" of atomization plus electron affinity of M,,,, and calculation of entropy as in (c) above Kapustinskii equation far lattice energy; AS" from si7e and comparison with alkali m e t d halides [a] Sum off. g, and h 1711 Ibl [55. 561 [721 [57. 73. 741 154. 551 (74. 751 [a] A plot ol.AS(:for formation of alkali halide lattices from the ga7eoua ions L'S ( r M t + r x - j ' is linear, following the equation A$:= - 47.8 - 29.2 ( r M b + r x ) I . while for tlie alkaline earth5 tlie equation is AS:= -72.2-53.6 ( r M ?I + r x ) ' ( S i n J K ' mol '.r in A). [b] Data not available so AHA=AC,'=O.O was used in preparing Tables 4 and 5 . Recent calorimetric measurements give AH;= - 24.5 kJ mol ' 1 M. H . Ahrohum. University 01. Surrey (England). private communication. (1979). Calculated values of A H o and AG" for steps (0,(h) and (i) of Table 3 are given in Table 4 for a few selected salts. Calculations involving the cryptated alkaline earth cations included the necessary valence corrections but are probably less certain than those for the alkali metals. Nevertheless, modynamic stabilities of salts of the alkaline earth cryptates are particularly striking. In the following sections we will describe our progress in the preparation of "alkalide" and "electride" salts and speculate about possible future developments. Table 4 I:stimated thermodynamic data [a] for assessment of the stability of some crystalline salts of the type M: Cry M , ( M y =M,. M,#M,) and M + C r y e - at 25 "C. The subscripts f, h. and i refer to the corresponding reactions in Table 3. M:CryM. M'Crpe Na ' C222 N a LI ' C21 I LI K ' C222 K K C222 Na Ba' C222(Na K ' C222e Li ' C21 I Au Ba' C?22(e 259 245 23 1 23 1 715 23 I )? 245 715 [a] All values in kJ mol I. AM,' and AQl AG,' A q: 255 226 227 227 72 I 227 226 72 I - 323 - 356 - 303 - 322 - 966 - 294 - 373 - - 258 -291 - 238 - 257 - 866 - 229 - 307 - 782 880 + M,., + 10 - 14 - 33 - 38 - 197 63 13 - 165 - 28 2 6 0 -114 - 3 - 27 61 not included (see however footnote [b] of Table 3 ) they indicate that salts such as Ba2+C222(Na-)* should be thermodynamically stable. These calculations also provide estimates of the thermodynamic stability of " e l e ~ t r i d e s " ~ ~T~ol . calculate and AGVx for the process Cry,,, - M Crye,, + (1 8) we assume that the electron occupies the octahedral holes in a closest-packed lattice of cryptated cations. Therefore the effective "radius" of the trapped electron is 0.414rM+Cry and the lattice energy is assumed to be the same as that of a salt in which a hard monovalent anion just fits into the octahedral hole. Compression of the packed cations or delocalization of the electrons would be expected to predict even greater stabilities. The conservative estimates given in Table 4 suggest that "electrides" should be stable. The values of AH:',, AG:)6, AH?8, and AG:, for a number of other possible compounds are given in Table 5 . At the very least, the results given in Tables 4 and 5 for a variety of crystalline salts of "alkalides" and "electrides" suggest that this field is potentially a very rich one for the synthesis of a variety of new compounds. The predicted therAngeu. Chem. Inr. Ed. Engl. 18. 5N7-5YH (1979) Table 5. Selected estimated values of the enthalpy and Gibbs function (in kJ mol ' j at 25 'C for the formation of crystalline salts of type M : C r y M , or MCrye from the metal and the cryptand [a]. : M Cry M; M'Crye A B ' [b] Li C211 Na Li ' CZI 1 K Li ' C211 Cs Li ' C211 e Li ' C211 H Li C211 Ag Li C211 Cu N a C222 Li N a ' C222e Na C221 Na Na C221 e Ba2+C222(Li.j 2 Ba' ' C222(Cs j? Ba' ' C222(Au )? Ba' ' C222(Ag )2 Ba' ' C222(Cu )> C a l i C221 (Na j z Ca' C221 (e j. Cs'C322Cs Rb ' C222 R b - + + + + AG" [b] 41 - 19 35 33 - x 74 19 - - 26 85 19 - 35 - 5 - 33 - 197 - 136 - 197 - 188 - 134 - 177 - 154 - 2 - 28 to 29 58 41 99 51 26 17 II 120 - 51 - 123 - I I3 - - 59 - I15 70 30 I2 [a] Values of AM2 and Ad: not included (however. see footnote [b] i n Table 3). orCry,,,+M,,,-M ' C r y e , ~ , . [bl Reaction: Cry,,,+M,,,,+M,,,,-M:CryM, 593 4. Solid Compounds Containing Alkali Metal Anions By late 1973 the evidence for alkali metal anions in solution was very strong, and thermodynamic estimates indicated potential stability of crystalline ionic salts of the type M 'CryM -. Rapid evaporation of methylamine or ethylamine from concentrated solutions of sodium or potassium in these solvents in the presence of cryptand C222 left deposits of metallic-looking gold-colored films on the glass walls. Upon cooling a solution of Na+C222, N a - in ethylamine ( = 0.1 M) from + 5 " C to dry-ice temperatures, a gold-colored powder precipitated from solution. Slow cooling resulted in the growth of hexagonal crystals of the salt Nat C 2 2 2N a -. Analysis[61and a crystal structure determinationl7]confirmed the existence of this first solid compound containing a n alkali metal anion. The isolation and characterization of other members of this class of compounds has been hampered by difficulties with stabilities and with finding the proper solvents for crystallization, but recently we have made considerable progress in this area. on the metal at 0°C. After pouring the supernatant liquid into E for the last time it is distilled into a waste flask and either diethyl ether or n-pentane is condensed on the salt in order to remove any unreacted cryptand or other organicsoluble impurities. Provision is usually made on vessel G for transfer of the washed salt into a number of tubes which can be individually sealed-off under vacuum. Crystals of Na+C222Na- up to zz 1 mm in diameter can be grown by slowly cooling a solution which is formed by redissolving the salt in ethylamine. 4.1.2. Crystal Structure of N a C222 Na + In its gross features, the crystal str~cturel'~ of this salt may be viewed as hexagonal closest-packing of the large cryptated cations with the smaller N a - ions occupying the octahedral holes. The cryptand forms a cage for Na with threefold symmetry and a n anti-prismatic arrangement of the ether oxygens. Views down the three-fold axis and side-on are shown in Figure 6. The shape of the cryptate moiety and relative positions of N a ' and N a - are similar to those in + 4.1. Preparation and Characterization of Na +C222Na 0 0 0 0 4.1.1. Preparation of Powdered and Crystalline Forms The preparation of this salt is relatively simple provided one pays close attention to solvent purity and cleanliness of the glassware. The type of vessel used is shown in Figure 5. 0 0 0 0 0 0 Na C A / bl a1 I Fig. 6. Views of the structure of Na 'C222Na j?]. (a) Looking down the threefold axis; filled-in circles represent the "top" Na ions: open circles the bottom ones. Triangles represent 0 atoms, squares represent CH-. groups The two N atoms and Na' are co-linear at the center. Solid lines connect atoms above the center; dotted lines below the center. (b) Perspective side-view of the structure. ~ O D D W Fig. 5. Apparatus for the preparation of the crystalline salt Na ' C222Na A lo G see text.  N a C222 I [761. The closest Na to Na distances are 8.83 A in the packing planes perpendicular to the three-fold axis while the closest distance between N a - ions in different planes is 11.O This could result in considerable anisotropy of those properties which depend upon the distance between anions such as the electrical conductivity. - + The sodium, previously transferred under vacuum into small diameter glass tubing is inserted into glass tube A after scoring the metal-containing tube with a glass knife. Heatshrinkable Teflon tubing B is used to connect tube A to sidearm C. Cryptand is then introduced into vessel G and the entire apparatus is evacuated. Then the tube containing sodium is shaken into the flexible connection B between A and C, broken under vacuum and moved to the constriction. A vacuum seal-off is made behind the metal sample and with the vessel re-attached to the vacuum manifold, metal is successively distilled into vessel E through two or three constrictorr). (With volatile cryptands tions while pumping (=. vessel G must be cooled during this operation.) Solvent sufficient to produce about 0 . 0 5 ~cryptand is condensed in G, the cryptand is dissolved and the solution is transferred into E and allowed to contact the metal mirror at 0 ° C for a few hours. The solution is then poured through the frit F into G and cooled to - 78 "C, resulting in the precipitation of finely-divided crystals of Na C222 Na To improve the yield, the supernatant blue solution is poured from the crystals at - 78 " C back through the frit F into E for further agitation + 594 A. 4.1.3. Properties of N a C222 Na + Crystals of this salt frequently grow as thin hexagonal plates with shiny metallic-looking faces. Their color at liquid nitrogen temperature is bright yellow-gold, which darkens reversibly to a yellow-bronze appearance at room temperatures. The crystals melt with decomposition at 83 " C to yield gray particles suspended in a clear fluid. Analysis after cooling shows that most of the cryptand remains intact and apparently sodium metal is released upon melting. By contrast, crystals left in evacuated tubes for several days at room temperature and under ambient light conditions darken irreversibly, apparently with decomposition of the cryptand. When a sample was kept in the dark at room temperature for three weeks, no apparent decomposition occurred, but the apparent photosensitivity has not yet been investigated further. AnKen.. Chrm Int. Ed. EnRl I K . 5X7-598 lIY701 Packed powders of Na C222 Na have the temperaturedependent electrical conductivity expected for a semiconductor. As shown in Figure 7, logp u s l / T gives a straight line over six decades of resistance where p is the specific resistance. The magnitude of the resistance depends upon the packing pressure and decreases with time as the sample anneals. Considering this salt as a semiconductor gives a band + n - lot 1_ I0 5 ? . . .. I . .. . L P ,... .__.. ..’. ’.......” ’ 1L c t t ./ 1 5 10 I I 15 20 5 [crn-’ 10P I / 25 30 Fig. 8. Absorption spectra of a thin, solid solvent-free film of‘ Na ’ C222Na (solid line) 1781 and of a film of Na ’ 18C-6 Na containing some methylamine (dotted line) [Sl]. 4.1.5. Summary of the Properties of Na C222 Na + Fig. 7 Semi-log plot of specific resistance of powdered samples of Na ’ C222 Na wrsu.7 the reciprocal of the temperature. Sample A was relatively loosely-p;icked. The data shown for the tightly-packed sample B were measured at both increasing and decreasing temperatures [SZ]. gap of 2.4 eV. This salt is diamagnetic, as indicated both by ESR studies and by static susceptibility measurements. No magnetic anomalies have been found between 300 and 4.2 K. The ESR spectrum shows only an extremely weak signal probably caused by trapped electrons. The structure and properties of the salt NatC222Na- in the crystalline state and in solution, confirm that N a- is a “genuine” spherical anion with two electrons in the 3 s orbital. Were it not for the tendency of N a - to donate electrons to any available acceptor, a variety of salts could probably be prepared. However, stability requires that the counterion be extremely reluctant to pick up an electron. Any bare metal cation would probably be reduced by Na- and we will probably be forced to continue to use complexed metal cations or perhaps such species as tetraalkylammonium cations. While the latter will probably not pick up the electron by simple attachment, it is known that solvated electrons can react with small tetraalkylammonium ions, breaking the C N bond. Therefore, salts such as R4N Na may not be stable. + 4.2. Other Salts of Alkali Metal Anions 4.1.4. The Optical Na C222 Na + Spectrum of Thin Films of - Powdered or crystalline samples of this compound can be dissolved in a number of amines and ethers to give deep blue solutions. Upon cooling such solutions (except in methylamine and ammonia) the salt precipitates. When solutions in methylamine, ethylamine or T H F are rapidly evaporated, the walls are covered with a thin film which appears gold by reflected light and blue by transmitted light. The absorption spectrum of such a film is shown in Figure 8. In all cases with this compound the three distinctive features shown in Figure 8 were observed[7x1. The major peak at 15400 cm-‘ (650 nm) is characteristic of N a - in solution[”] and probably arises from a 3 s to 3 p transition, modified by crystal-field effects. The origins of the pronounced shoulder at 18 900 cm - - I and the small but distinct peak at 24500 cm-’ are unknown although the latter could arise from a bound-free transition corresponding to the band-gap of 2.4 eV shown by conductivity. Figure 8 also shows that the gold-colored film formed by using [ 1 81crown-6 in the presence of solvent vapor has the major absorption band at 15400 cm-I but shows neither the extra peak at 24500 c m - ‘ nor the shoulder at 18900 cm-I. The spectrum shows clearly why crystals of Na’C222Naappear metallic by reflected light. The high absorption in the near-infrared and sharp cutoff in the visible mimic the plasma absorption of metals. This can lead to a striking metallic appearance even for nonconducting ~ r y s t a l s [ ~ ~ l . The crystalline compound described in the previous section can be prepared either by crystallization from a solution containing largely Na C222 and Na - or by rapid evaporation of solvent from such a solution. By controlling the complexing agent, metal, and solvent, one can prepare a wide variety of concentrated solutions containing M Cry (or M + crown ether), MY and/or e,,,. The metals M, and M ymay be the same or different. In this section we describe some (as yet incompletely characterized) solids prepared either by crystallization from such solutions or by rapid solvent evaporation. + + 4.2.1. Solids Prepared by Crystallization The only crystalline solids which we have prepared to date contain the cryptand C222 with various alkali metals-no crystals have been made with other cryptands or with crown ethers, nor have any “electrides” been crystallized. However, there seems to be no fundamental restriction on the preparation of such compounds if one can find the appropriate solvent and conditions for crystallization. In addition to the salt of Na-, we have prepared other crystalline solids which appear to be K C222 K -, KtC222Na-, Rb’C222Rband C s + C 2 2 2 C s - . The methods of preparation are similar to those used for Na C222 N a - (s) except that isopropylamine was used rather than ethylamine for salts of K -, R b and Cs . All of the + + - ~ 595 crystals have a metallic appearance which ranges in color from light gold for Na C222 Na - to greenish-gold for K C222 Na - to dark copper-bronze for Cs C222 Cs -. Solutions containing K - , Rb-, Cs- and e,;,, are much more subject to irreversible decomposition than are those containing Na -. This makes it difficult to prepare crystalline salts of these anions since all solution manipulations must be done at low temperatures. Once the solvent has been removed, however, all of the crystalline solids prepared so far seem to be reasonably stable and it is likely that we will be able to handle them and characterize them in much the same way as for Na+C222Na-. + + + 4.2.2. Preparation of Thin Films by Solvent Evaporation By adjusting the solution stoichiometry and flash-evaporating the solvent (practical only with highly volatile solvents) it is relatively easy to prepare solvent-free transparent films or powders whose spectral indicate the presence of M and/or e -. Such films have been prepared from solutions of all of the alkali metals in methylamine in the presence of both crown ethers and cryptands. Films which show optical bands of M - appear blue by transmitted light and vary in color from gold to dark-bronze by reflected light, depending on the metal, while films which have only an infrared band (presumably the absorption band of trapped electrons) appear blue by transmitted light and very dark blue-black by reflected light. Attempts to prepare crystals or powders by slow solvent evaporation have been generally unsuccessful, with irreversible decomposition usually occurring instead of precipitation of the desired compound. Solutions and solids that contain crown-6 tend to be more stable to irreversible decomposition than those that contain C222. However, in the former solid films of N a + 18C-6Na- also require the presence of some solvent. Removal of solvent leads to the formation of solid sodium and crown-6, without apparent decomposition of the crown ether. Evidently the axial position(s) in the N a + 18C-6Na- salt must be occupied by solvent in order for the salt to be stable. It is interesting that films prepared by evaporating solutions containing crown-6 and either K, R b or Cs give spectra which do not change when all of the solvent is removed. and films by solvent evaporation from solutions which contained largely M + cryptand (or M + crown ether) and solvated electrons[’]. The ESR spectra of such powders consists of a very intense narrow line at the free-electron g-value, suggesting the presence of trapped electrons and extensive motional narrowing of the ESR signal. Recently Harris and LugowskiIXol observed narrow ESR lines from solids produced by freezing metal-ammonia solutions which contained crown-6. They attribute the signal to an ammoniated electron in the vicinity of a complexed cation. In our case, solvent-free blue solids are obtained so that the electrons can no longer be solvated but may simply occupy the anion sites in the crystal lattice. We refer to these solids as “electrides” although the disposition of the electrons with respect to the complexed cations is unknown. We have measured the absorption spectra of “electride” films with both cryptand C222 and crown-6 as complexing agents (cf. Fig. 9)[R’1.Often the infrared band which we - 5 10 15 3 [cm” 20 25 30 Fig. 9. Spectra of solvent-free thin “electride” films produced by rapid evaporation of methylamine from solutions of K C222e,,, (solid line), Cs’ 18C-6e,,, (dotted line) and K ‘tXC-6K (with e,,>,”?)(dashed line) [Xl]. ~ attribute to trapped electrons is accompanied by an M - absorption. In some cases the relative intensities of the two bands shifts in favor of M - as the solution is allowed to contact excess metal[7x1(cf. Fig. 10). In addition to the “electride” band at 7400 cm-‘ and the band of M - , intermediate absorption peaks of unknown origin are often observed (cf. Fig. 9). 4.2.3. Absorption Spectra of Films Films prepared by evaporation of methylamine from solutions with stoichiometric amounts of M + , C222 and M (M C222 M --) show strong absorption bands which are characteristic of the anion. The shoulder and extra peak seen with films of Na C222 Na are not observed for any of the other metals. Indeed, when films are prepared from solutions of K C222 Na the peak of Na is a single narrow band. The origin of this change in spectrum and that obtained with crown-6 (see Fig. 8) remains a mystery although we expect a change in crystal structure from Na’C222Na- to K C222 N a - or to Na 18C-6 Na -. Since we are now able to prepare crystalline samples of K+C222Na-, it may be possible to determine the origin of these spectral differences. Even before the isolation and characterization of Na C222 Na -, we had prepared dark blue solid powders + - + + + - - + + 596 5. Speculation about Properties and Future Potential Since we have only limited information about the structure and properties of “alkalide” salts, and even less information about “electrides”, it is difficult to predict in what ways these substances might be useful. Certainly, as examples of a new oxidation state of a whole family of elements, alkali metal anions are fundamentally interesting. The strongly reducing nature of these compounds may prove useful in syntheses, and the solid-state properties may provide useful materials in such areas as semiconductor behavior, photoelectron emission, photoconductivity, etc. In this section we will speculate about the nature of these compounds and indicate some areas of potential usefulness. Angew C‘hem. Inr. Ed Engl. I X . 5117-59H (1979) tion of alkali metal anions-the complexity and cost of cryptands. If the solids which we call “electrides” prove to be stoichiometric compounds with one trapped electron for each cation, then it might be possible to prepare a variety of electride salts such as Bat +C222(e-)2. The increase in electron density in such compounds compared with M C222 e might be enough to give them the properties of “expanded metals”, similar to the metallic compound Ba(NH,),[’’. The optical and ESR properties of uni-univalent “electrides” such as K C222 e suggest that the electrons are largely localized, perhaps in the traps formed by the octahedral holes between closest-packed cations. As the electron density increases, enough electron-electron overlap may occur to give metallic behavior. + + 10 5 i; 20 15 Ccm-’ 10.~1 25 Fig. 10. Spectra of solvent-free thin films produced by evaporation of methylamine from solutions containing cryptand C222 and various relative concentrations of potassium . Spectra for progressively higher concentrations of potassium are in the order, dashed line (“electride”), solid line (mixture), and dotted line K (“potasside”). 5.1. Solution Properties The use of crown-ether and cryptand complexing agents makes it possible to prepare concentrated solutions containing either M or e,;,, in a variety of amine and ether solvents. For example, a solution of 0.1 M sodium in T H F can be prepared by using the cryptand C222. Since this solution contains mostly Na+C222 and N a - and very little e,,, it might be useful for two-electron reductions. If a one-electron reductant in this aprotic solvent were desired, a solution of potassium with an equivalent quantity of the cryptand C222 would produce K+C222 and e,iiV with little interference from K . ~ 5.2. Possible New Compounds of “Alkalides” and “Elec trides” The thermodynamic estimates described in Section 3 suggest that compounds such as Ba+ +C222(Na-)2 should be thermodynamically stable. Similarly, we predict that solid compounds containing Li- should be stable. The present challenge is to find appropriate synthetic routes to such substances. In solution, Li- tends to dissociate into Li’ and e,,,. Perhaps a direct vapor-phase reaction or reaction in an inert matrix would avoid some of the problems associated with the use of a polar solvent to dissolve the reacting species. One of the problems which we have encountered in working with lithium and the alkaline earths is the inability to purify these metals by distillation in glass or quartz. As a result, solutions tend to be unstable. Distillation of these metals directly into the apparatus from a metal container might eliminate some of these difficulties. It should be possible to substitute other cations for cryptated cations provided they are non-reducible. Long chain tetraalkylammonium ions seem attractive at present since they are stable in solution in the presence of Na-. We are attempting to find suitable conditions for metathesis reactions in solution such as M’CryM-+R,N+X- + M+CryX,J+R,N+M- (19) If tetraalkylammonium “alkalides” could be prepared, they might eliminate one of the major barriers to practical utilizaAngew Chem In1 Ed Engl I8 587-598 ( I g 7 9 ) ~~ - 5.3. Possible Uses for the Solid Compounds Whether the semiconducting properties of Na ’ C222 Na and similar salts of the other alkali metal anions will be useful depends primarily on their stability. It now appears that the parent compound is stable at room temperature in uacuo in the dark. Even its reactivity with air is not as great as we first thought. For example, a compressed pellet of powdered Na’C222 N a - remained gold-colored and shiny in air until reaction with moisture in the air began on the surface. Further surface reaction, as observed under a microscope, occurred at water droplets with little or no general surface corrosion. Light absorption by N a - in the solid is very efficient and covers a broad spectral range extending throughout the visible and near infrared (cf. Fig. 8). If solids can be found in which this absorption is followed by auto-ionization as in solution126**], and if the lifetime of the conduction electron so produced is long enough, sodium anions might be efficient for photoelectric energy conversion. Because the electron is only weakly bound to such anions as Cs- and may be even more weakly bound in “electride” salts the photoelectric properties of these substances would be most interesting. At least in principle, such compounds have the potential to extend the wavelength of photoelectron emission farther into the red than is possible with currently available photoemissive surfaces. Of course, such speculation about potential uses for “alkalides” and “electrides” is based upon very little fundamental information about their properties. Before more can be done to utilize these compounds, a great deal of study is needed to determine their properties. Measurements are not easy with such reactive materials even when they can be prepared in pure form. However, we hope that the demonstration that the - 1 oxidation state of the alkali metals is stable will stimulate others to investigate such systems and find uses for them. The author is grateful to the National Science Foundation for current support of this research under Grant No. DMR7722 975 and to the U.S. Department of Energy f o r prior support under Contract No. EY-76-S-02.0958. Thanks are due to the many graduate students and post-doctoral research associates who worked on this project, most of whom may be identified as co-authors of papers from this laboratory. Previously unpub- 591 lished wofk described in this paper was done by M. DaGue, H. Lewis, P. Smith, B. VanEck and M . Yemen. Received. August 9. 1978 [A 282 IEf German version: Angew. Chem. 91, 613 (1979) [ I ] V. M. DukelSkii, E. Ya. Zandberg, N . 1. Ionoc, Dokl. Akad. Nauk SSSR 02. 232 (1948). 121 For the most accurate values of the electron affinities of the alkali metal atoms see: T. A . Parrerson. H. Hotop. A. Kusdan, D. W Norcrow W. C. L i neberger. Phys. Rev. Lett. 32. I89 (1974).  S. Golden, C. Gurrman, T R. Tutrle. Jr.. J Am. Chem. Soc. H7. 135 (1965)  S. Golden, C. Gutrman. T. R. 7irtrle. Jr.. J . Chem. Phys. 44. 3791 (1966). 151 S. Moralon, S. Golden. M. Orrolenghi. J. Phys. Chem. 73, 3098 (1969). (61 J. L. Dye. J . M . Ceru.ro. M T I.ok. B. L. Burnett. I.: J. Tehun. J Am. Cheni. Soc. 96. 608 ( 1 974).  F. J . Tehan. B. L. Bornetr. J. L. Dye. J. Am. Chem. Soc. Y6. 7203 (1974).  For a detailed discussion of the properties of metal-ammonia solutions see: J. C. Thompson: Electrons in Liquid Ammonia. Oxford University Press. Oxford 1976. 191 Review articles on metal-ammonia solutions include a ) C. A. Kruu?. J . Chem. Educ. 30.83 (1953); b) W L. JoU),, Prog. Inorg. Chem. 1.235 (1959). c) M C. R. Symo~is.Rev. Chem. Soc. 13. 99 (1959);d ) U. Schinderwul/; Angew. Chem. 80. 165 (1968). Angew. Chem. Int. Ed. Engl. 7. 190(1968);e) T. P. Dos, Adv. Chem Phys. 4. 303 (1962); C) J. 1,. Dye. Sci. Am. 216. 77 (Feb. 19671, g) M. H. Cohen, . I C. Thompson, Adv. Phys. 17. X57 (1968). [lo] A collection of "landmark papers" in the metal-ammonia field has been given by W. L. Joll,~.:Metal-Ammonia Solutions. Dowden, Hutchinson & Ross; Stroudsberg. Pennsylvania 1972. [ I I ] Papers resulting from conferences in the field o f metal-ammonia Solutions and the broader general area of electrons in fluids are: a ) G. Lepuutre. M J. Sienko: Metal-Ammonia Solutions. Colloque Weyl 1. Benjamin. New York 1964: b) R. F. Guuld. Chem. Ser. 50 (1965): c) J. J. Logowski. M J. Sienko: Metal-Ammonia Solutions. Colloque Weyl 11. IUPAC, Buiterworths. London 1970; d ) U. Schindewolf; Ber. Bunsenges. Phys. Chem. 75. (1971); e) J. Jorlner. N . R. Kesrnert Electrons in Fluids. Colloque Weyl 111. Springer. Berlin 1973; f) Colloque Weyl IV. Electrons in Fluids-The Nature of Metal-Ammonia Solutions. J . Phys. Chem 79. No. 26 (1975): g) International Conference on Electrons in Fluids. Banff, Alberta. Canada. Sept. 1976: Can. J. Chem. 55. No. 1 1 . 1796-2277 (1977). 1121 M. T. Lok, F. J . Tehun, J. L. Dre. J. Phys. Chem. 76. 2975 (1972). M. G. DeBacker. J. L. D p ,J. Phys. Chem. 75.3092 (1971). J. L. Dye. M. G. DeBacker. J. A. E v e . L. M. Dorfman. J . Phys. Chem. 76. 839 (1972). J. L. D w . J. Chem. Education 54, 332 (1977). J. L. Dye, Sci. Am. 237, 92 (July. 1977). J. L. D.re in J. J. Christensen. R. M. Ira//: Progress in Macrocyclic Chemistry. Vol 1 . Wiley-lnterscience. New York, 1979. 1. Hurle?, 7. R. Turrle, J r , S. Golden. J. Chem. Phys. 48. 2818 (1968). J. L. Dye. M. G. DeBacker. V. A. Nicelv. J . Am. Chem. Soc. '12. 5226 (1970). J. L. Dye, M. T. L o k, F. J. Tehun, R. B. Coolen, N Papadakis, J. M. Ceraso, M. DeBacker in Ref. [ I I d], p. 659. L. H . Feldman. R. R. D e w l d , J. L. Dye in Ref. [ I 1 b], p. 163. G. Stein, A. Treinin, Trans. Faraday Soc. 55, 10x6. 1091 (1959). M. F. Fox, E. Havon, J . Chem. Soc. Faraday Trans. 1 72. 1990 (1976). R. R. Dewald. J. L D.ve. J . Phyb. Chem (58.121 (1964) J. L. D w . M. G. DeBacker, L. M. Dorfman. J. Chem. Phys. S2. 6251 (1970). H . Aulrch, B. Boron. P. Delahav. J. Chem. Phys. 58. 603 (1973). H. Aiilich. P. Delahay, L. Nemec, J. Chem. Phys. 59. 2354 (1973). H. Aulich, L. Nemec, P. Deluhay, J . Chem. Phys. 61. 4235 (1974). R . R. Dewuld, J. L. D~w. J. Phys. Chem. 6N. 128 (1964). R. R. Dewald. K. W. Browall, J . Phys. Chem. 74. 129 (1970). J. L. D j e in Ref. [ I le]. p. 77 598 1321 J. M Ceraso. J. 1, D y e . J. Chem. Phys. (51. 1585 (1974). 1331 J L. D w . C W. Andr ew . J . M . Cerusu. J . Phys. Chern 70. 3076 (1975). 1341 J. P. Kinrzinger, J.-M. Lehn. J . Am. Chem. Soc. 06. 3313 (1974).  J. M Cerwo, J. L. Div. J Am. Chem. Soc YC, 4432 (1973). 13hJ G.MoJ& S. Frugo. Theor. Chtm. Actn 5, 275 (1966). (371 J . I-. D w . Acc. Chein Res. 1. 306 (1968)  J. W. Fletcher. W. A. Seddun, J . Phys Chem. 79. 3055 (1975). 1391 If. Linschirz. J . Eluranru. J. Chem. Phys. I N . 2214 (1963) (401 S. H C h u m . J. H Marshull. J . Chem. Phys. 52, 5555 (1970) 1411 A. Guurhon. M. Orrolenghi. Isr. J Chem. H. 165 (1970). 1421 D. //upper/. K. H . Bar-Eli, J . Phys. Chem. 74. 3285 (1970). 1431 L. J. Gilling. J. G. Kloosrerhoer. R. P. If. Rertrchnick. J D. W. run Voorsr. Chem. Phys Lett. 8.457. 462 (1971). 1441 G. &bur. K . H. Bar-Eli. J. Phys. Chem. 75. 286 (1971) 1451 M. Fisher. G. Ramme. S. Claeosun. M. Szwarc. Chem. Phya. Lett. 9. 309 (1971). A. Friedenberg, H Lerunon. Chem. Phys. Lett. 41, 84 (1976). A. Friedenberg. H. Leranun. J . Phys Chem. 81, 766 (1977). K . D. Vur. J. L. Die. J . Chem. Phys. 38. 2033 (1963). K . Bar-Eli. T. R Turrle, Jr.. J . Chem. Phys. 40. 2508 (1964). V. A Nicelr. J. L. Dre. J Chem. Phys. 53. I19 (1970). J . L. Dye. Pure Appl. Chem. 49. 3 (1977) gives a review of monomers and ion pairs with references to the extensive literature in this field. See also Ref. [171. B. Bockrarh, L M. Durfmun. J. Phys. Chem. 77. I002 (1973) W. A . Seddon. J. W. Flercher. R. Cairerall. Can. J . Chem. 55. 2017 (1977). and references cited therein. E. C. M. Then, W. E Wenrwurth. J. Chem. Educ. 52, 4X6 (1975): H. Horop. W C. Lineberger, J. Phys. Chem. Ref. Data 4. 539 (1975). J. A. Dean. Lange's Handbook ofChemistry. I Ith Edit McCirdw-Hill. New York 1973. S. Glu.urune: Textbook of Physical Chemistry. 2nd Edit Van Nostrand. New York 1946. p. 874. E. A . Moelyyn-Hughev: Physical Chemistry. 2nd Edit. Pergamon Press. New York 1961, p. 882. W M Larimer, W L. Jol/r.. J . Am. Chem. Soc 75, 4147 (1953) W. E. Sprcer. A. H. Sommer, J. G. W i r e , Phys. Rev. 11s. 57 (1959). J. L. Dye. G. Lepourre, P. R. Marshall. P. Palor in Ref, [ I la]. p. 92. E. Zintl. J. Gouheau. W Dullenkopx Z. Phys Chem. Abt. A 154. I (1931). E. Zinll. A. Hardet-, 2 Phys. Chem. Abt. A 154. 47 (1931). J. D. Corherr. D. G. Adolphson, D. J. Merrvman. P. A. Edwards. F. J. Armai i s . J . Am. Chem Soc. Y7. 6267 (1975). J. D. Corhelt. P. A. Edwurdr. J. Chem. Soc. Chem. Commun. 1975. 984. D. G. Adofpiiron, J. D. Curbetr. D.J Merrvmon. J . Am. Chern. Soc 98. 7234 (1976). A. Cisar, J. D. Corbetl. Inorg. Chem. 16, 632 (1977). J. D. Corherr, P. A Ed'dward.r,J . Am. Chem Soc. 99, 3313 (1977). P. A. Edu'ardr, J. D Corhett. Inorg. Chem. 16, 903 (1977). I. Hur1e.v. T. R. Tutile. Jr., S. Golden. in Ref. [ l lc]. p. 449. M Lacosre, Dissertation, Universite des Sciences et Techniques du Langue- doc. Montpellier, France 1976 E. Kouffmunn, JAM. Lehn. J. P. Sauroge, Helv. Chim. Acta 59, 1099 (1976). S. Goldman. R. G. Bare.7, J . Am. Chem Soc. 94. 1476 (1972). J.-M. Lehn, Struct. Bonding 16. 1 (1973). J. L. Dye, C. W. Andrew. S. E. Marhen.;, J. Phys. Chem. 7Y, 3065 (1975). A. F. Kapusrmskii. Q Rev. Chem. Sac. 10. 283 (1956). D.Moras. B. Merz, R. Weiss, Acta Crystallogr. Sect. B 29. 388 (1973). D. Moros. B Met:. R. Wei.r.5, Acia Crystallogr. Sect. B ZY, 383 (1973). J. L Dye, M. R. Yemen. M G. DaCue, J - M . Lehn, J. Chem. Phys. 621. I665 (1978). H . J. Hesse. W. Fuhs. G. Weiser, L. ron Sienrpaly. Chem. Phys. Lett. 41. 104 (1976). R. L. Harris. J. J. Lagowski. J Phys. Chem. 82, 729 (1978). M. R. Yemen. M. G. DaGue, J. L. Dve, unpublished. M. R. Yemen, J. L. Dye, unpublished. R. R. Dewuld, J. Phys. Chem. 73. 2615 (1969).