SOME GENERAL ASPECTS OF THE INORGANIC CHEMISTRY OF FLU0RI"E By A. G. SHARPE M.A. Ph.D. (UNIVERSITY CHEMICALABORATORY CAMBRIDGE) RECENT advances in the descriptive inorganic chemistry of fluorine and its compounds have been the subject of several reviews,l but their correlation and interpretation in terms of measurable physical properties have rarely been discussed. The aim of this Review is to describe briefly some of the principal properties of the fluorine molecule the fluorine atom and the fluoride ion and to show how a knowledge of these properties can help in the understanding of the chemistry of fluorine and its relationship to the other halogens. The Physical Properties of Fluorhe.-Interatomic Distance. The values obtained by the electron diffraction and the Raman spectroscopic method (1.435 and 1.418 A respectively) are in good agreement and the covalent radius of fluorine (the bond being assumed to be of unit order-see p.50) may be taken as 0.71 A. Since the corresponding radii for chlorine bromine and iodine are 0.99 1.14 and 1-33 8 the volume of a fluorine atom in the combined state is only one-third of that of a chlorine atom and less than one-sixth of that of an atom of iodine. Dissociation Energy. This quantity (the energy absorbed when a gaseous molecule is converted into two atoms in the ground state) has been the subject of much controversy. Fluorine does not show banded absorp- tion in the visible or ultraviolet region (presumably because of the instability of the excited state) and spectroscopic determination of D(F2) is therefore impossible. Older values for the dissociation energy were usually based explicitly or implicitly on extrapolation of those for chlorine bromine and iodine (58.0 46.1 and 36-1 kcal.respectively) and were of the order of 60-70 kcal. In 1950 however Evans Warhurst and Whittle,4 on the basis of recent work on the thermochemistry and absorption spcctrum of chlorine monofluoride suggested that the true value was 37 & 8 kcal. Most later investigations support this lower figure ; the following deter- minations may be taken as representative. ( a ) Simons Editor " Fluorine Chemistry " Academic Press Inc. New York Vol. I 1950 ; Vol. 11 1954 ; ( b ) Haszeldine and Sharpe " Fluorine and its Com- pounds " Methuen and Co. Ltd. London 1951 ; (c) Gutmanil Angew. Ckevn,. 1950 62 312; ( d ) Sharpe Quart. Rev.1950 4 115; (e) Leech Research 1952 5 lOS 449 ; (f) EmelBus J. 1954 2979 ; ( 9 ) Klemm Angew. Chein. 1954 66 470 ; ( h ) Rudge Chem. and Ind. 1956 504 ; (i) Mellor " A Comprehensive Treatise on Inorganic and Theoretical Chemistry " Suppl. 11 Part I Longmans Green and Co. Ltd. London 1956. 2 Rogers Schomaker and Stevenson J . AnTer. Ch,em. SOC. 1941 63 2610. 3 Andrychuk Canad. J . Phys. 1951 29 151. 4 Evans Warhurst and Whittle J. 1050 1524. D 49 50 QUARTERLY REVIEWS (i) Doescher 5 studied the pressure-temperature relationship for fluorine in a ire-treated nickel vessel over the temperature range 759-1115' K by comparing the pressure with that of nitrogen a t the same temperature using a differential manometer containing a fluorocarbon oil. His results indicate a value for D(F,) a t 25" of 37.4 & 0.4 kcal.(ii) From observation of the rate of effusion of fluorine a t low pressures through a small hole (a method which permitted the use of relatively low temperatures 500-650" K) Wise 6 found D(F,) = 39.9 5 0.8 kcal. determined the dissociation energies of the potassium rubidium and caesium halides spectroscopically and combined their results with independently obtained thermochemical data for these substances and the alkali metals (iii) Barrow and Caunt iD(F2) = D(MF) + L(MP) - L(M) - &(MF) where &(MF) is the heat of formation of a solid fluoride MF from solid M and gaseous F, and L(MF) and L(M) are the heats of sublimation of the fluoride and of the alkali metal respectively. They obtained B(F2) = 37.6 5 3.5 kcal. I n this Review the dissociation energy is henceforth taken to be 38 kcal.This low value is generally ascribed to the repulsion of non-bonding electrons in the F molecule but it has also been suggested that for chlorine bromine and iodine hydridisation of p - and d-valence shell orbitals strengthens the b0nding.~9 8 The nice distinction between the bond in fluorine being abnormally weak and the bonds in the other halogens being abnormally strong is essentially a theoretical matter. It should however be pointed out that the N-N and 0-0 bond energies (21 and 35 kcal. in N,H4 and H,O respectively g are also low and that here too repulsion of non-bonding electrons of small atoms may well be the cause. The usual thermodynamic functions for atomic and molecular fluorine have been calculated lo from the interatomic distance,2 the fundamental vibration frequencyY3 and Doescher's value for D(F,).First fluorine is dis- sociated into atoms to a greater extent than chlorine a t the same tempera- ture and since reactions of atomic fluorine are strongly exothermic the great reactivity of the element may be attributed to the weakness of the bond in the F molecule. Secondly since the standard entropies of mole- cular and atomic fluorine (48.6 cal. mole-1 deg.-l and 37.9 cal. g.-atom-l deg.-l) differ but little from those of molecular and atomic chlorine (53.3 and 39.5 entropy units 11) differences between the two halogens are due to heat effects (e.g. the strengths of bonds) rather than to entropy effects. This generalisation holds in fact for all the halogens. The energy required for removal of an electron Thermodynamic Properties.Two important points may be noted. Ionisation PotentiaZ. Doescher J. Ghem. Phys. 1952 20 330. Mulliken J . Amer. C'hem. SOC. 1955 77 884. Cottrell " The Strengths of Chemical Bonds " Rut,terworths Scientific Publications Wise ibid. p. 927. ' Barrow and Caunt Proc. Roy. Soc. 1963 A 219 120. London 1954. lo Cole Farber and Elverurn J. Chem. Phys. 1952 20 586. l1Nat. Bur. Stand. Tables Circular No. 500. SHARPE THE INORGANIC CHEMISTRY OF FLUORINE 51 from atomic fluorine is 401 kcal./g.-atom.12 This figure combined with that for D(F,) leads to a standard heat of formation of the gaseous B'+ ion of 420 lrcal./g.-ion. Such a high value (those for C1+ Br+ and I+ are 327 301 and 268 kcal. respectively 11) suggests that even solvated fluoro- ilium ions are unlikely to be encountered in chemical investigations ; there is in fact a t the present time no evidence of any kind for the existence of " positive fluorine ".Electron Aflnity. The classical method for determining the heat liber- ated E when gaseous halogen atoms combine with electrons giving gaseous halide ions is by the Born cycle where I ( M ) is the first ionisation potential of the alkali metal M and U(MX) the lattice energy of the solid halide MX (the heat liberated when one formula weight is produced from gaseous M+ and X- ions). U may be calculated from the interionic distance T in solid MX (determined by X-ray analysis) and the compressibility of the solid by an expression of the type U = NMz,x2e2 (1 - l / n ) / r where M the Madelung constant is a geometrical constant for a particular type of structure x1 and x2 are the charges on the ions e the electronic charge N the Avogadro number and n a constant (about 9) which takes account of interionic repulsion arising from the finite size of the ions.This method when applied to fluorine yields E(F) = 84 & 2 kcal./g.- atom,13 a value intermediate between those for chlorine and bromine (88 and 82 kcal./g.-atom respectively). This somewhat surprising result is as was first pointed out by Evans Warhurst and Whittle,* an inescapable conse- quence of the low value of D(F,) ; it must however be remembered that additional factors are always involved in determining the stability of com- pounds containing fluoride ions. The Valency of Fluorine.-The electronic configuration of the fluorine atom is ls22s22p5 and expansion of the valency shell beyond 2s22p6 is impossible.It is known from the atomic spectrum of fluorine that pro- motion of an electron to the 38 3p7 or 3d level is a highly endothermic pro- cess ; this of course provides no direct evidence concerning the possibility of promotion in a stable molecule but comparison with other first-row elements strongly suggests that fluorine should be exclusively univalent .14 With the possible exception of the difluorides and trifluorides of the alkali metals l5 (e.g. RbF, obtained by the action of fluorine on rubidium chloride a t 150') this generalisation seems valid for all fluorine compounds. Rubi- dium trifluoride must obviously contain multivalent rubidium or multivalent fluorine ; the structure Rb3+F is inadmissible on energy considerations but a decision between structures such as Rb+F,- and Rb+(RbF,)- is a t present impossible E(X) = &(MX) + L(M) + I(M) + W(XJ - U(MX) l2 Nat.Bur. Stand. Tables Circular No. 467. l3 Pritchard Chem. Rev. 1953 52 529. l4 Gillespie J . 1952 1002. l5 Bode and Klesper 2. anorg. Chem. 1951,267,97. For a discussion of the structure of RbF see also Sharpe Ref. l(a) Vol. I1 p. 2. 52 QUARTERLY REMEWS The Physical Properties of the Fluoride Ion.-Ionic Radius. X-Ray structure determinations yield information about atomic positions and for the division of interionic distances into ionic radii the introduction of certain hypotheses is necessary. The simplest of these is that in the lithium halides the anions are in contact; this and a variety of other more elaborate methods l6 lead to the values I? 1.36 ; C1- 1-81 ; Br- 1.95 ; I- 2.16 8.The fluoride and oxide ions are of almost identical size [ r ( 0 2 - ) = 1-40 A] and thence arises the similarity in structure which is often found between oxides and fluorides of the same formula type (e.g. MgO and NaF) ; fluorides and chlorides of the same metals however often have quite different struc- tures (e.g. CdF and HgF both crystallise with the fluorite structure but CdC1 has a layer lattice in which the Cd2f ion has co-ordination number six and HgCl has a molecular lattice). The crystal chemistry of both simple and complex fluorides is admirably discussed by We1ls.l' When an ionic crystal dissolves in water the sum of the heats of hydration of the ions must equal the lattice energy of the solid plus the heat of solution.Two principal difficulties attend the assignment of individual heats of hydration uncertainty over the values for ionic radii in solution [which would enable approximate values to be calculated from Born's expression Ne2(1 - l / D ) / 2 r where D is the dielectric constant of the medium] and the absence of any precise knowledge of the structure of hydrated ions. Differences in heats of hydration how- ever are not subject to the same uncertainty e.g. the heats of solution of sodium fluoride and sodium chloride are both very small and since their lattice energies a t 25" are 215.4 and 183.5 kcal. respectively the heat of hydration of F- must be 32 kcal. greater than that of C1-. The absolute values are estimated l8 to be P- 123 ; C1- 89 kcal. ; these lead to standard heats of formation of the hydrated ions of 188 and 148 kcal.* A similar position exists with regard to entropies of hydration which have to be obtained from entropies of solid salts entropy changes attending dissolution and entropies of gaseous ions calculated by statistical mechanics ; the estimated values l8 for fluoride and chloride ions (I? -29 ; C1- -15 cal.g.-ion-l deg.-l) are however so similar that (as with molecules and atoms) entropy changes in fluoride-chlorine substitutions may almost be neglected in comparison with the big difference in heats of hydration the standard free energies of hydration are F- -114 ; C1- -84 kcal. The Standard Potential of the Fluorine-Fluoride Ion E1e~trode.l~ An Heat and Entropy of Hydration. l6 Pauling " The Nature of the Chemical Bond " 2nd edn.Cornell Univ. Press 17 Wells " Structural Inorganic Chemistry " 2nd edn. Oxford Univ. Press 1950 ; Ithace New Pork 1940. Quart. Rev. 1954 8 380. Latimer Fitzer and Blansky J. C'hent. Phys. 1939 7 108. lQ Latimer " The Oxidation States of the Elements and their Potentials in Aqueous Solutions " 2nd edn. Prentice-Hall Inc. New York 1952. * It should be noted that these are absolute values ; the National Bureau of Standards values l1 are based on a standard heat of formation of the hydrated hydrogen ion of zero. This practice is more coiivenieiit in thermochemistry because of the uncertainty in the magnitude of the hydration energy of the proton SI'IARPE THE INORGANIC CHEMISTRY OF FLUORINE 53 indirect calculation of Eo for the $l?2/F-(aq) electrode from related thermo- chemical data leads to a value of +2.S v (relative to Eo for the +H2/H+(,q) electrode as zero).The figure for fluorine expresses the fact that it is the most powerful oxidising agent known and explains why the element can be prepared only by thermal decomposition of a few higher fluorides (such as cobalt trifluoride) or by electrolysis of solutions of fluorides in media (such as hydrogen fluoride) in which no other anion is present. The factors which determine the oxidation potential of a halogen may be seen by considering the following sequence The first stage involves absorption of energy equal to one-half of the dis- sociation energy the second stage the liberation of the electron affinity and the third stage the liberation of the hydration energy. Although therefore the electron affinity of fluorine is lower than that of chlorine the weaker bond in the F molecule and the higher hydration energy of the smaller fluoride ion make fluorine the more powerful oxidising agent.The Electronegativity of Fluorine.163 20-A11 of the quantities mentioned hitherto however difficult their measurenient may be are easily defined. The concept of electronegativity however calls for some comment. The strength of trifluoroacetic acid the absence of basic properties in the fully fluorinated amine (CF,),N and the retardation of attack by electrophilic reagents in aromatic substitution when the CW group of toluene is replaced by CF, all suggest that fluorine in such molecules attracts electrons. This " power of an atom in a molecule to attract electrons to itself'' l6 is what most chemists mean by electronegativity but unfortunately this quantity is not susceptible to direct.experimental measurement. (It is indeed an assumption that each element has a numerically expressible electronega- tivity which remains constant through its compounds.) Three principal methods for the assessment of electronegativity have been suggested. Mulliken proposed taking the mean of ionisation potential and electron affinity ; in this context however it is the ionisation potential of the element in its valency state (which has to be estimated) that is required. Malone related electronegativity to dipole moments ; unfortunately bond dipole moments (as distinct from molecular moments) are not in general measurable since the effect of unshared pairs of electrons on the moment is considerable.This is clearly illustrated by the fact that nitrogen trifluoride although having a pyramidal structure with LFNF = 102" and N-F = 1.37 A has a dipole moment of only 0.2 D. Pauling's scale rests on a number of unproven assumptions but does have the advantage of being related to molecular properties. The energy of a normal single covalent bond between two elements A and B (e.g. H and F,) is taken to be the mean of the Fond energies in A and B,. The difference in electro- negativity of A and B xA - xB is then taken as The corresponding value for chlorine is + l a 3 6 v. QX,(g) -+ X(g) -+ X-M + X-(aq) XA - XB = 0*208[EA - (EAA + EBB)/2]' 2o Pritohard and Skinner CiLern. Rev. 1965 55 745. 54 QUARTERLY REVIEWS where E is the actual bond energy in AB (in kcal.) and the expression in square brackets is regarded as " extra " bond energy arising from the partial ionic nature of the A-B bond.This relationship does lead to a fairly self- consistent set of electronegativities and if xH is arbitrarily taken to be 2.1 (in order to make all values of x positive) xp is found to be 4.0 making fluorine easily the most electronegative element. Bond energies in fluorine compounds are discussed again later. Some Properties of Inorganic Fluorine Compounds The Dissociation of Hydrogen Fluoride in Aqueous Solution.-The large dipole moment of hydrogen fluoride (1-9 D at pressures so low that associa- tion is negligible 21) shows the bond in this compound to be strongly polar and the chain structure in the solid 22 (the other hydrogen halides have close-packed structures) arises from dipole-dipole interaction.I n dilute aqueous solution however hydrogen fluoride is a much weaker acid than the other hydrogen halides. This fact when considered in conjunction with the well-known increase in dissociation constant along the series CH,*CO,H CH,I*CO,H CH,Br*CO,H CH,C1*C02H and CH,F*CO,H a t first seems surprising ; in the carboxylic acids however ionisation always involves the breaking of the same bond whereas in the hydrogen halides the bonds to be broken are all different. The general process of ionisation may be represented as taking place in the following stages 23 24 HX(aq) + HXW -+ H a ( ) + X.(,) + H+(g) + X-(g) The stages involving the conversion of a hydrogen atom into a solvated proton are the same for all acids and only four variables have to be con- sidered the energy of solution of the undissociated molecule the dissocia- tion energy of the H-X bond the electron affinity of X and the solvation energy of X-.The first factor is approximately the same for all of the halides the electron affinity of fluorine lies between those of chlorine and bromine and the hydration energy of the F- ion is much larger than those of other halide ions. The decisive factor must therefore be the strength of the bond in hydrogen fluoride (Bond energies HF 135; HCl 103; HBr 8 7 ; HI 71 kcal.). I n more concentrated solutions (5-15~) ionisation into H,O+ and HI?,- H2F3- and H3F4- takes place and hydrogen fluoride becomes a strong acid.25 The formation of these stable acid anions in liquid hydrogen fluoride accounts for the great proton- donating (i.e.acidic) properties of this solvent. Hydrogen Bonding in Fluorine Compounds.-As the most electronegative element fluorine would be expected to take part in hydrogen bond forma- tion and some of the best known instances of this phenomenon do in fact -+ H+laq) + X-(aq) 21 Oriani and Smyth J . Amer. Chem. SOC. 1948 70 125. 22Atmoji and Lipscomb Acta Cryst. 1954 7 173. 23 Bell " Acids and Bases )) Methuen and Co. Ltd. London 1952. 24McCoubrey Trans. Paraday SOC. 1955 51 743. 26 BelI Bascombe and McCoubrey J. 1956 1286. SHARPE THE INORGANIC CHEMISTRY OF FLUORINE 55 involve covalently bonded fluorine or the fluoride ion. The structural difference between hydrogen fluoride and other hydrogen halides has already been mentioned ; the strength of the bonding in the KF,- ion is also remark- able.The fluorine-fluorine distance in this ion is only 2.26 8 and a neutron- diffraction study 26 of potassium hydrogen difluoride shows that the hydrogen is (to within 0-1 8) in the middle of the linear ion. The structure of ammonium fluoride differs completely from those of the other ammonium halides (which crystallise with the sodium chloride or czesium chloride structure) ; in this salt (which has the wurtzite structure) each nitrogen atom forms four N-I€-F bonds of length 2.69 A to the four fluoride ions arranged tetrahedrally around it. 27 The N-H vibration frequency is lowered from its normal value of about 3300 cm.-l to 2820 cm.-l; rather surprisingly nuclear magnetic resonance studies 28 indicate that this reduction in the N-H vibration frequency is not acconipanied by any considerable stretch- ing of the N-H bond (length in NH,F 1.04; in NH,Cl 1.038 A).In hydrnzinium fluoride however (the structure of which is also determined by hydrogen bonding) a slightly greater N-H distance of 1.075 A is reported.29 It should not be thought that all ammonium salts of fluoro-acids exhibit strong hydrogen bonding. In salts of complex acids this is certainly not so ; a wide variety of evidence (X-ray s t ~ d i e s ~ O - ~ ~ infrared spectra,32 33 and nuclear-resonance spectra 34) suggests that in salts such as NH4Bl? and (NH4),TiF there can be no more than very weak hydrogen bonding. This somewhat unexpected conclusion shows that the participation of fluorine in hydrogen-bond formation is not nearly so general as that of nitrogen or oxygen.No satisfactory explantion of this fact has yet been put forward and the recent discovery 35 of the HC1,- ion indicates that it may soon be necessary to modify present ideas about hydrogen bonding and 'its relation to the electronegativities of the halogens. Fluorides of Non-metals Bond Energies and Bond Lengths.-Fluorine often invokes highest covalencies (e.g. in SF, IF,) and although steric factors must be of some importance in this connection a satisfactory dis- cussion of this topic must involve consideration of the energy changes involved. The formation of sulphur hexafluoride may be represented as taking place in the following rdages S (solid) --+ S (gas ; ground state 3523~3~) --+ S (gas ; valency state 3s13p33d2) 3F2 (gas) ________+ 6F (gas) + SF (gas) 26Peterson and Levy J.Chem,. Phys. 1952 20 704. 27 Plumb and Hornig ibid. 1955 23 947. 28Drain Discuss. Faraday Soc. 1955 19 200. 2g Deeley and Richards Trans. Faraday SOC. 1954 50 560. 30 Hoard and Blair J . Amer. Chem. XOG. 1935 57 1985. alCox and Sharpe J. 1953 1783. 321dem J. 1954 1798. 33Cot6 and Thompson Proc. Boy. SOC. 1951 A 210 217. a4Pend.red and Richards Trans. Faraduy SOC. 1955 51 468. 36Herbrandson Dickerson and Weinstein J . Amer. Chem. SOG. 1954 76 4046. 56 QUARTERLY REVIEWS In considering why sulphur forms a hexafluoride but not a hexahydride or a hexachloride the fundamental question to be answered is will the energy liberated by bond formation in the compound compensate for the energy required to raise the sulphur atmom from its ground state to its valency state and to effect dissociation of the halogen (or hydrogen) molecules ? In general (molecular fluorine constitutes an exception and is discussed again below) smaller atoms form stronger bonds a fact which is simply accounted for on modern valency theory by the greater overlapping of orbitals of low principal quantum number.Equally important however is the dissociation energy of the halogen (or hydrogen) and the low value for fluorine (F, 38 ; Cl, 58 ; H, 104 kcal.) is probably the most important factor in this case. Because of the weakness of the bond in molecular fluorine most fluorine compounds are strongly exothermic (this term it will be remembered refers to heats of formation from elements in their standard states) ; conversely because of the strength of the bond in molecular nitrogen (225 kcaL36) most nitrogen compounds containing a high proportion of the element are endothermic.The widely quoted fact that nitrogen trifluoride is an exothermic compound (Qf = +26 kca'1.j whilst the trichloride is endo- thermic (Qf = -55 kcal.) thus represents a case essentially similar to the existence of SF but not of SC16 or SH,. In discussing electronegativity it was mentioned that bonds involving fluorine are usually much stronger than would be expected on the basis of an " arithmetic mean " rule. They are also much shorter than values calculated by adding standard covalent radii the universally accepted carbon-carbon single bond length for example is 1.54 8 and F-F in F is 1.42 8 ; C-F in CF, however is only 1.32 A.m Similarly unexpectedly short bonds have also been found in fluorides of silicon nitrogen phosphorus arsenic oxygen and sulphur.For first-row elements such as carbon multiple bond formation appears to be impossible in these compounds and a suggestion which has met with much favour is that the observed bond length should be less than the sum of the covalent radii by an amount proportional to the difference between the Pauling electronegativity co- efficients of the elements concerned the actual length being given by the empirical Schomaker-Stevenson equation 38 TAB = rA rB - o'Og(x xg) The general applicability of this relation has been severely criticised by Wells 39 but the qualitative conclusion that bonds between fluorine (and to a smaller extent oxygen nitrogen and chlorine) and less electronegative elements are shorter than expected on the basis of a simple additivity rule is unchallenged.These generalisations about the energies and lengths of bonds involving fluorine suggest that perhaps it is once again a property of the reference 36McDoweII PTOC. Roy. SOC. 1956 A 236 278. 37Hoffman and Livingston J . Chem. Phys. 1953 21 565. 3sSchomaker and Stevenson J . Amer. Chem. SOC. 1941 63 37. a9 Wells J . 1949 55 ; ref. 17 p. 56. SHARPE THE INOR(XAN1C CHEMISTRY OF FLUORINE 57 standard (the F molecule) which lies a t the root of the matter. If the weak bond in the fluorine molecule is due to repulsion of non-bonding electrons i t seems not unreasonable that as the molecule is split into atoms which then combine with elements to form compounds in which there are few or no unshared electrons on the central atom (e.g.CP, SF,) such repulsion should disappear. This it is suggested may be the reason for the " abnor- mal " strength and shortness of bonds in other compounds formed between elements of widely differing electronegativities (e.g. C and 0 P and 0 Si and F). The stability of non-metal fluorides (especially of CF and SFJ is often cited as a remarkable feature of fluorine chemistry and it is not always realised that these compounds are not thermodynamically particularly stable The free energies of the following hydrolytic reactions (neither of which proceeds a t a detectable rate under ordinary experimental conditions) have been calculated from standard thermochemical data 11 3ga CF,(g) + 2H,O(,) = CO,(g) + 4HF(,) SF,(,) + 3H,0(g) = SO,(g) + 6HF(,) AGO = - 36 kcal. AGO = - 72 kcal.They show that the inertness of the fluorides must be due to activation- energy considerations ; these may well involve the failure of a water mole- cule to co-ordinate on to a combined fluorine atom (because of the octet restriction) but it cannot be said that a convincing explanation has yet been given. Fluorides of Metals.-Two general features stand out in the chemistry of metal fluorides first many metals show their highest oxidation states attained in salts in their fluorides (e.g. Co in COP, Ag in AgF, Bi in BiF5 Tb in TbF, Rh in RhF,) ; secondly many fluorides of high oxidation states are salt-like in properties where the corresponding chlorides are not (e.g. AuF3 PbF, TlF3). (The highest fluorides of many transition metals e.g. MoF, UF, OsF8 are volatile and generally resemble the fluorides of non-metallic elements.) Both of these generalisations are illuminated by consideration of a modified Born cycle.Suppose for example the possi- bility of a metal's forming a saline tetrahalide is examined by analysing the stages involved In this context the latent heat of sublimation of the metal will be relatively small ; for all of the halogens the sum of E - D/2 which represents the net energy change in forming a gram-ion of halide ions from half a gram- molecule of molecules is about 60 kcal. The essential question is then will 4(E - D/2) plus the lattice energy of MX compensate for the energy required to remove four electrons from the metal atoms ? I n the absence of a knowledge of the structure (and thence the lattice energy if the calculation 39. Kirkbride and Davidson Nature 1954 174 79.58 QUARTERLY REVIEWS is simple enough) of the halide no precise answer to this question can be given ; but since the lattice energy will depend inversely on the interionic separation it will clearly be a maximum when for a given cation the radius of tJhe anion is a minimum. This condition is fulfilled by the anion’s being fluoride. The other common anion of similar size ( 0 2 - ) involves the absorption of a large amount of energy when it is formed from molecular oxygen ; this is however largely compensated for by the double charge and the conse- quent increase in the electrostatic lattice energy. It is not therefore sur- prising that ionic oxidation states in oxides are often as high as or even higher than those in fluorides (e.g. MnO, Ago Pro,) or that manysalt- like fluorides (e.g.AuF, PbF, COP,) are hydrolysed by water with the formation of very insoluble oxides and fluoride ions ; the high hydration energy of the fluoride ion is also an important factor in bringing about hydrolysis. One further argument which may be developed from the simple electro- static treatment concerns the use of alkali-metal fluorides as halogen- exchange reagents in organic chemistry. If we consider the replacement \ \ / / -C-Cl + MF + -C-F + MC1 where M = Na or K the driving force of the reaction will depend on the free-energy difference (or fairly accurately the difference in lattice energy) between sodium fluoride and sodium chloride on the one hand and potassium fluoride and potassium chloride on the other. Lattice energy being inversely proportional to interionic distance the increase in free energy when sodium chloride is formed from the fluoride will be proportional to and it is easily seen that for a larger cation the amount of free energy which has to be supplied by the C-Cl-+ C-B’ change is less.Pluorine-exchanging ability therefore increases steadily with increasing ionic size among fluorides of metals which form isomorphous compounds.40 In the special case of the use of silver fluoride it is easily shown from inde- pendent thermochemical data that the difference in lattice energy between silver fluoride and silver chloride is very small (owing to the contribution of non-ionic bonding in solid silver chloride) ; hence arises the especial power of silver fluoride as a halogen-exchange reagent. Complex Fluorides and Fluoro-acids.-The factors which govern the stability of complex fluoro-ions will be similar to those concerned with the stabilities of simple fluorides.Among complex fluorides the relatively small size of the anions (thus leading to increased lattice energy) will play an important part; and within recent years especially by the use of ele- mental fluorine a t medium temperature^,^^ and of bromine trifluoride as a 40 Woyski J. Amer. Chem. Xoc. 1950 72 919. 41 Klemm and HUSS 2. anorg. Chem. 1949 258 221 and later papers by Klemm and his co-workers. SHARPE THE INORGANIC CHEMISTRY O F FLUORINE 59 non-aqueous solvent and fluorinating 43 many new complexes of unusual oxidation states have been obtained (e.g. Cs,CoF, K,NiF, K,CuP, K,RhF, KIrF6 AgAuF,). Most of these compounds are hydrolysed to oxides by water.Although for base metals fluorides are the most stable complex halide ions in solution the reverse is true for noble metals such as platinum and gold.44 This is often interpreted as being due to n-bond formation between the noble metal and chlorine bromine or iodine d-electrons of the metal being used for this purpose.45 Fluorine would not be able to accept more electrons and the bonds in complex fluorides would therefore necessarily be devoid of multiple character with its consequent strengthening effect. It must however be pointed out that this is not the only factor involved in an equilibrium such as PtFO2- + 6CI- + YCC1,2- + 6F- and that the larger solvation energy of the fluoride ion must also play a n important part in influencing the stability of the complex.All complex fluoro-acids (and indeed complex halogeno-acids in general) are extremely strong. The univalency of fluorine provides a simple and convincing explanation of this fact in a case such as fluoroboric acid where the formulation of the undissociated molecule HBF is impossible without invoking quinquevalent boron or bivalent fluorine. A study of the HF-BF system has shown the non-existence of a 1 1 compound ; only if a molecule such as NH, H,O or a second molecule of HF is available to combine with the proton (giving NH,+BF4- H,O+BF,- or H,F+BF,-) will the com- pounds combine. Reasonable formulae for other undissociated molecules such as HPF, H,SiP, and H,PtCl (none of which is known in the free state) are also impossible. Because of the impossibility of a fluoroborate's having a covalent struc- ture this ion is very useful in studies in which it is desirable t o be sure of the ionic nature of bonding e.g.in the interaction of silver salts and aromatic hydrocarbons 46 (AgBF is soluble in and forms stable complexes with these substances) and in the investigation of the spectra of organic cations 47 (e.g. Ph,C+ in Ph,C+BF,-). A mixture of hydrogen fluoride and boron trifluoride is indeed the most acidic solvent known and in it even so weak a base as hexamethylbenzene is largely converted into the salt [C6(CH3),H] +BF4-.489 49 Conclusion.-The principal properties which confer on fluorine its remark- able chemical behaviour are the-smallness of the fluorine atom and the fluorine ion the restriction to an octet of electrons and the weakness of 4 2 Sharpe J.1949 2901 and later papers. 43 Hepworth Robinson and Westland J. 1954 4268. 4 4 Sharpe J. 1950 3444 ; Carleson and Irving J. 1954 4390. 45See e.g. Chatt and Leden J. 1955 2936. 46 Sharp and Sharpe J. 1956 1855. 4 7 Sharp and Sheppard J. in press. 48 McCaulay and Lien J . Amer. Chem. Soc. 1951 73 2013. 49 Kilpatrick and Luborsky ibid. 1953 75 577. 60 QUARTERLY REVIEWS the bond in the F molecule. It would be entirely misleading to suggest that our understanding of the chemistry of the element is yet complete but with the aid of physical methods of investigation a deeper insight into its properties is rapidly becoming possible.