The dread and darkness of the mind cannot be dispelled by the sunbeams, by the shining shafts of day, but only by an understanding of the outward form and inner workings of nature. Lucretius, Book 1. THE PREPARATION AND PROPERTIES OF SOME COMPOUNDS OF THE NITROGEN AND SULPHUR GROUPS.
By Robert Howard Nuttall.
A thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy of The University of London by Robert Howard Nuttall.
AUGUST 1962. -1- Ahlman, New complexes have been prenared between trphenyi
amne 5 triphenyistibine, and trl,phenylbismth and some Eilver salts and between azobenzene and some silwi.r and
cuprous salts. The ability of certain group V compounde
10 att ss 1:%.gands is discussed in relavion to their
Oysicai prop(u*ties
A Jtudy of the infrared spoctra of diazonIum salto
showu that there I e -)onding frnw the anitql to the tati,n snipe -)f* Tt 's uake de-N:1j
correlaton between the N-fi s etc/ling freqfl ncy and eloctronc offects dao 4o other substituauts in the arcmatic rthg,. The infrared spectra of the aminclbenencdiazenjwl and p-autin,;:ben2euedionium
ions ?iriv been interpreted ac favourftng ta-uctures In
whic th._ positive charge is transferred fTria the goutp to the amine erouping.
Tri , infrared spectra of a rarge L-01,71- and iAr;1 amontum salts have been studied, particular attention
bcAng iyA:td to the N-H stretching and N-H deformation frequencies, In salts where there is no eation-aniln hydrogen bonding the N-H stretching frequencies occur -1 wr-A,1 above 3100cm , Except for triphonyle,mmonium fluoroborates fluoroborates and tetraphenylborates are not hydrogen-bonded.
The reactions of tungsten hexachloride with some ligands have been examined. Complexes. with triphenyl- phosphine triphenylamine and dipyridyl have been isolated©
The ultraviolet spectra of aromatic hydrocarbons have been examined in anhydrous fluorosulphuric acid and it has been found that fluorosulphuric acid s although a strong sulphonating agent also protonateb fused ring sys= tems such as anthracene. The infrared spectra of solutions of metal carbonyls in fluorosulphuric acid have been examined in the region 1 15002500cm . A new compound 9 manganese c rbonyl fluo- rosulphate, has been isolated from the reaction of man- ganese carbonyl with fluorosulphuric acid. -3- Acknowledggpepts. The work described in this: thesis was carried out in the Inorganic Chemistry Laboratories of the Imperial College of Science and Technology, London,. and the Royal College of Science and Technology, Glasgow under the supervision of Dr. E. R. Roberts to whom I wish to express my thanks. I also wish to thank Dr. D. W. A. Sharp for much help and encouragement. Thanks are due also to Dr. T. A. Dunn* of University College, London for the ultraviolet solid reflectance spectra and to the Microanalytical Departments in both London and Glasgow. I am indebted to Imperial Chemical Industries Ltd. Pisons Ltd. and the United States Navy (Office of Naval Research, Contract No. N 62558-3061 ) for grants. R \-1 tkull.M. Royal College of Science,. Glasgow. Contentp. Abstract. p.l. Acknowledgements. p.3© Introduction. p05. The Formation of Complexes Between Group V Ligands and Some Copper and Silver Salts© p.12. Infrared Spectra of Stable Diazonium Salts. p.37. The Hydrogen Bond. The Nature of The Bond. p.60. Hydrogen Bonding in Alkyl and Aryl Ammonium Salts.p.68_ The Infrared Spectrum of Pyridinium Fluoroborate. p.84. Hydrogen Bonding in Triphenylphosphonium Salts© p.89. Reactions of Tungsten Hexachloride. P.94. The Ultraviolet Spectra of Aromatic Compounds in Fluorosulphuric Acid. p.100. Solutions of Metal Carbonyls in Fluorosulphuric Acid. p.116. Appendix. Note on Infrared Data. p.124. References. po/25. T5- Introdyctiop. The chemical and physical properties of complex fluoroacids may be compared with those of oxyacids and ly the *luence of the electronegative fluorine atoms may be clearly seen. Three strong acids may be considered as characteristic, fluoroboric acid a complex fluoroacid„ perchloric a strong oxyacid9 and fluorosulphuric acid having intermediate properties. Aspects of the chemical properties of these acids will be considerod in this thesis in particular the interaction of anion and cation and its' influence on infrared spectra and structure in various compounds of the acids.
Pluoroboric acid was first prepared in 1J09 by Gay Lussac and Thenard (1) who passed boron trifluoride into water The gas is so soluble that it is essential to pass At into the water through mercury to stop the solution siphoning back. The acid was obtained as a water white liquid which fumed in air. The reaction proceeds by the hydrolysis of the boron trifluoride to give metaboric and fluoroboric acids. 4BF3 3H20 3HBF4 H31303 The acid may also be prepared by the reaction of boric acid with hydrofluoric acid. The reaction appears to go in two stages (Wamser9 2). The first is rapid:- B(OH) 3HF--HBF- OH 4. 2H O0 5 3 ° The second is slow and rate determining:- HBF OH 4- HP # HBP14 - H0.
The last reaction is reversible but slow and in dilute aqueous solution fluoroboric acid may contain appreciable quantities of hydroxyfluoroboric acid (HBP The preparation of anhydrous fluoroboric acid seems inherently unlikely since two improbable alternatives are required for its formation - an isolated proton or divalent fluorine. While fluorine does appear to be a bridging element under some circumstances (to be considered later) Practical examination (MacCauley and Lien9 3) shows that BF and HP do not interact except in the presence of a mole- 3 cule or entity which can be protonated. It thus appears that aromatic systems can be protonated by HP/11F3 mixtures to give cations such as C6117 at low temperatures. On these considerations alone it appears that fluoroboric acid is a strong acid. The measurement of ionisation of the aqueous acid is made difficult by the hydrolysis of the fluoroborate ion but it appears that fluoroboric acid is at least as strong as perchloric acid and stronger than sulphuric acid. Boron trifluoride used in the manufacture of fluoro- boric acid is usually prepared by heating a salt of the general formula Na2(BF3)4 with sulphur trioxide the salt first being prepared by the reaction of borax with hydro- fluoric acid. The boron trifluoride may then be absorbed in water or reacted with organic molecules to give such commercially important entities as boron trifluoride ether ate: Boron trifluoride finds considerable use as an industrial catalyst for the isomerisation of organic mol- ecules for which purpose it may be used as a mixture with hydrogen fluorides as an etherate or as a concentrated aqueous solution (Booth and Martin, 4). Boron trifluoride farina a series of addition compounds at low temperatures with argon. When the system boron tri- fluoride argon is studied thermally (Booth and Martins 4) evidence for compounds ranging from A.BF to A.16BF is found. 3 While the compound BF A.BF3 may be explained on the grounds of co-ordination from argon to boron it is hard to explain the structure of 16BF3 .A without assuming that the compound must have fluorine bridges. In a similar manner with tri- ethylamines Bt3N.(Bl3)2 may be obtained. Brown et al (5) suggest that the structure of these compounds is F. F I Bt3N - B - - - B - F 1 1 F F Again the coMpound is stable only at low temperatures (-50°C.)i In view of the high electronegativiiy of the fluoride ion with its octet of electrons it is surprising that bridged should be formed since the energy for the partial removal of electrons must be very high. Condensed fluorides are known however as the ions Allg and AlFi in which fluorine atoms are apparently common to edges or corners of linked chains (Sharpe, 6). It has been proposed that fluorine bridges occur in 2AsF 0330 and in the mixed fluoride AsF.SbF to 3 3 5 give the bridged structures (Muetterties and Phillips, Muetterties and Coffman, 7):- F F / /As(S03F)F F As SbFL F(S03F)As, \ 4 0‘ /0 sit • 0' The fluoroborate ion does not appear to behave in this manner, no evidence for polyions or for fluorine bridges so far having been found. Compounds such as triphenylmethyl fluoroborate are completely ionic (Kemmitt and Sharp, 8). -9- It is possible that fluoroborates could form hydrogen bonds although the electrostatic interaction would not be very great® In the simplest salt ammonium fluoroborates which is isomorphous with potassium fluoroborates no hydrogen bonding occurs (Sharpe 9). Hydrogen bonding in substituted ammonium fluoroborates will be considered in this thesis. Aqueous fluoroboric acid does not find such extensive use as boron trifluoridee being used in electroplating sol- ution and also in the preparation of diazonium fluoroborates. These last compounds find considerable use in the manufac- ture of fluoro aromatic compounds since, when a diazonium fluoroborate is decomposed by heating, boron trifluoride and a fluoro substituted aromatic result. The diazonium fluoroborates are also of interest because of their stabil- ity as compared to most diazonlum compounds. In contrast to fluoroboric acid, perchloric acid exists as an explosively unstable anhydrous acid HC104© The pro- ton appears to be bonded through a divalent oxygen bridge. The acid forms a much more stakyle monohydiate (m.p.t. 49°) .1. H30 C1OL which exists solely in the form of the oxonium salt (Schumachere 10). An X-ray structural determination (Lee and Carpenters 11) shows that the perchlorate ion is an almost perfect tetrahedron. Simon and Weist (12) have -10 - shown that perchlorie acid is a strong acid, the 77% aqueous solution being 100% dissociated. Perchioric acid is co- valent when anhydrous and in a similar way Hathaway (13) has recently isolated volatile anhydroue cupric perchlorate in which the infrared spectrum of the perchlorate ion shows that it is distorted by covalent bonding. In contrast, the infrared spectrum of silver perehiorate indicates that it is ionic since the spectrum of the perchlorate ion in this compound shows the same features as the spectra of sod- ium and potassium faggates. Perchloric acid and perchlorates find their most exten- sive use in the manufacture of explosives and rocket fuels. Like fluoroboric acid, perchioric acid finds limited use in electroplating baths. In contrast to the noninteraction of hydrogen fluoride and boron trifluoride, hydrogen fluoride and sulphur tri- oxide react to give the non explesive covalent compound fluorosulphuric acid. The acid was first prepared by Thorpe and Kirman (1L) who observed that the acid reacted violently with water and assumed that complete hydrolysis occurred. In fact, the S0 F- ion may exist in solution but 3 it is slowly hydrolysed in a very complex fashion (vide infra). Fluorosulphuric acid may also be prepared by the reaction of potassium bifluoride and sulphuric acid from which mixture it may be distilled. The properties of fluorosulphuric acid seem to be intermediate between those of sulphuric and hydrofluoric acids and in this may lie its future utilisation since the the handling requirements for fluorosulphuric acid are not so stringent as with hydrofluoric acid. The acid is slowly hydrolysed in aqueous solution and the half change occurs over a period of about two days. (Sidgwickp 15) Many metal fluorosulphates are known and covalent organic fluorosuiphates have been prepared (Sharpe, 6). These last compounds, in contrast to the esters of perchloric acid, are non-explosive. They are volatile liquids which etch g14:4s. The aromatic fluorosulphate esters may be pre- pared by the thermal decomposition of benzene diazonium fluo- rosulphates. The fluoroborate perchlorate and fluorosulphate ions are similar in properties. Chemically all give nitrosyl com- pounds and precipitates with nitron, while physically the salts are sometimes isomorphous (Sharpe, 6). -12- Form4ion of Complexes Between GrouviIlgands and Some Copper and Silver Salts in Orgahic Solvents© The ability of group VB triaryls to form complexes with metal ions seems to be greatest for triphenyl phosphine and triphenyl arsine although in recent years a number of tri- phenyl stibine complexes have been prepared (Ahrland et al. 16). No metal complexes of triphenylamine or triphenyl- bismuth are previously known although the amine forms adducts with boron trichloride and tribromide and triphenylammonium salts (Kemmitt et al., 17), whilst triphenylbismuth forms an adduct with sulphur trioxide (Becke--Goehring and Thielemann, 1d). Sharp and Sharpe (19) have shown the solubility of the metal salts of strong acids, such as silver and copper per- chlorates, in diethyl ether. As diethyl ether is a solvent of low basicity it was considered that the weak bases tri- phenylamine and triphenylbismuth might in this solvent form complexes with silver and copper ions. As an additional test of this concept the interaction of the weak base azo- benzene was also examined.
Eaoerimental. All the anhydrous salts were handled in the dry box. -13- Silver perchlorate was obtained commercially and silver fluoroborate was prepared by the action of boron trifluoride on a suspension of silver fluoride in sulphur dioxide. Silver hexafluorophosphate was prepared similarly from phos- phorous pentafluoride and silver fluoride. Phosphorous pentafluoride was prepared by the thermal decomposition of benzene diazonium fluorophosphate (Russell and Sharp, 20). Silver fluoroborate was also prepared by the action of bro- mine trifluoride on silver borate (Sharpe, 21). A solution of cupric perchlorate was prepared by shaking an excess of anhydrous cupric chloride with silver perchlorate (Moss and- Sharp, 22). A solution of cuprous perchlorate was prepared by shaking an excess of copper powder with an ethereal sol- ution of silver perchlorate (Sharp, 23). The ligands tri- phenylamine, triphenylstibine, triphenylbismuth, tribenzyl- amine, diphenylaminel N-dimethylparatolylamine and azoben- zene were obtained commercially. Tri-p-tolylamine was a gift from Dr. D. W. A. Sharp. The complexes were prepared by mixing ethereal solut- ions of the silver salts and solutions of the ligands. If reaction did not occur immediately the solutions were heated and crystallisation occurred at once. The complexes prepared are shown in Table 1.
-14- TWA 1-
Ether 391al, 7)101110 Conditions ygpdyct
.40104. Ph3N heat Ph NpAge10 3' Ph Sb i (Ph Ab)4igC10 3 3 4 Ph331 heat Ph3BioAgC104
Ph2N2 i 3(Ph2N2).2AgC104
(p-CH3004)(CS3)2N. slow blue oxidation
(p-0113C04)3 N slow blue oxidation
PhaNH rapid blue oxidation (PhOH2)3 N i E(PhOH2)344AgC104 Ph3N i Ph314AgB1 4. Ph3Sb i (Phy9b)4AzBF4 Ph3Bi i brown decomposition Ph2N2 3Ph2N2.2AgBFII.
Cu (C104)2 Ph3N slow blue oxidation a Ph3Sb aPpr. (Ph3Sb)2 Ou(nold Ph3Bi slow brown oxidation Ph2N2. no reaction OW104 Ph3N slow blue oxidation Ph3Sb i (Ph3Sb)2 .44.CuC104 Ph3Bi slow brown decomposition Ph2N2 - slow Ph2N2,C104 m immediate, -15- TriPhenylamine 41vey Perchlorate. The complex was obtained as air stable pale green crys- tals by gently warming the mixed ethereal solutions. The crystals were only very slightly soluble in organic solv- ents, much less so than the component silver perchlorate and triphenylamine. The compound in boiling water decom- posed slowly. An ultraviolet spectrum was obtained in ethereal sol- ution and is similar to that of triphenylamine with the add- ition of a band at 430mu.: the solid reflectance spectrum shows the same features with a further weaker band at 510mu. The spectra of the complex and that of triphenylamine are shown in fig. 1. The infrared spectra of triphenylamine, silver per- chlorate and triphenylamine silver perchlorate show the following features in the 2000cm-1 to 650cm-1 region:-
Ph 3N. 1940(w), 1850(w), 1775(w), 1700(w), 15a5(e), 1490(s), 1330(u), 1305(m), 1285(s), 1185(m), 1175(s), 1075(m), 1050(m), (398(w), 833(w), 750(s), 700(s). Ag0104 1080(e), 920(w).
Ph3N.A60104 1590(s), 1480(s), 1330(m), 12,10(153), 1230609 1160(m), 1090(e), (perchlorate), 910(w) (perchlorate), -16 - 900(w), 780(s), 760(s), 695(s). (All values in wave numbers). (Ph = phenyl). (s = strong, m = medium, w = weak). The X-ray photographs of triphenylamine and triphenyl- amine silver perohiorate were obtained in Lindemann glass tubes: the photographs showed that the oonnlex is complet- ely distinct from triphenylamine and silver perohlorate although both are or a low order of symmetry.
Analysis. Ph3NAgC104 Theory C 4708A, H 3.3%, Ag 23.9%, Found C 47.8%, H 3.9%, Ag 24.15$. Trinhe ylbipputh Silver ,perighlora,te. The complex crystallised slowly on gentle heating of the mixed ethereal solutions as a white crystalline solid. It was insoluble in organic solvents and decomposed at once when added to water. It was air stable for two or three days but at the end of that time it was noticeably brown. A powdered sample was much less stable, decomposing in two or three hours. Infrared spectra in the 2000cm-1 to 650cm-1 range: -
(Ph)3Bi 1567(m), 1475(m), 1430(s), 1370(w), 1326(w),
1300 (w)9 1260(w), 1182(w) , 1157(w) 9 1060 (m) 9 1014(m) 9 Fig. r
I Aba
400 450 500 550 Fig. 2
320 350 4.00 450 500 550 -18- 908(m), 723(a), 691(a).
(Ph) i.AgC104 1550(m), 1470(m), 1418(s), 1300(w), 1095(a) (perchlorate), 1025(s) (perchlorate)„ 910(w),, 760(w), 730(e), 720(s)„ 690(m).
Analysis. Theory C 33.14, H 2.3v.4 (Ph)3Bi.AgC104 Found C 33.30, H 2.90.
Ltaaa.9 - The complex precipitated immediately from the mixed solutions as pale yellow microcrystals. It was soluble in organic solvents and in water from which solvent azo- benzene was rapidly precipitated. The compound seemed to be stable indefinitely in air. The solid reflectance ultraviolet spectra shown in fig. 2 have absorptions at 390mu. and 450mu. for azobenzene and at 330mu. with a shoulder at 420mu. for azobenzene sil- ver perchlorate. Infrared spectra 2000-650cm1.
Ph2N2 1595(w), 1495(e), 1455(s), 1385(m), 1310(w), 1225(w), 1155(w), 1075(m), 1020(w), 1005(w), 930(s), 775(s)„ 690(s). -19 3(Ph)2N2.2Age104 1575(w), 1480(m)9 1448 (m)p 1298(w), 1220(w), 1165(w), 1055(e) (perchlorate), 93d(w), 930(w), 920(w), 840(w), 780(m), 768(s), 700(m), 680(m). Ana1ysls. 3(Ph)2Ne.2Age104 Theory 0 45.0%, H 3.1%, Ag 22.5%. Found C 45.5$, H 4°109 Ag 22.10.
It was considered that the compound could be polymeric and the molecular weight in nitrobenzene was obtained. M.W. = 315. This was higher than the molecular weight of either component indicating that there was still some association in solution.
Tetpakistribenzylamirkp Silver Perchlorate. The complex which analysed an approximate 14.01 ratio of ligand to silver was obtained as a white precipitate as soon as the Couponents were mixed. Infrared spectra 2000cm-1 6.50=-10
(PhCH2.)3 N 1600(w), 1495(8), 1450(s), 1365(s), 1305(m), 1245(m), 1205(w)„ 1117(m), 1070(m), 1030(m), 960( ), 915(w), 900(w), 875(w), 790(s), 695(e).
1(PhCH2) 3NVe104 1492(m), 1451(s), 1415(m), 1305(w), 1220,1m)9110060 9 (perchlorate)„ 1055(s) (perchlorate),
1005(If) 9 992(w) 930(1) 9 900 (m) y 856 (W) 755 (s), 737(m) 20- 695(s).
nalysle. ((PhCH2) 3N) 4AgC104 Theory C 64°4-A9 H 5°40,9 Found C 63-3%, H Triparatolylamine„ diphenylamine and N-dimethylparatolyl- amine in ether did not react with silverperchlorate in ether under anhydrous conditions in the dry box. When the sol= utions were exposed to the laboratory atmosphere all three gave dark blue oxidation products. When solutions of anthracene, biphenyl and triphenyl- methane in ether were added to ethereal silver perchlorate9 no product separated.
Triphenvlamine Silver Fluoroborate. The complex crystallised rapidly from the mixed sol- utions as a darker green product than the comparable sil- ver perchlorate complex. It was insoluble in organic sol- vents but decomposed at once in waters it was stable for about twelve hours in the atmosphere at the end of which time the crystals were noticeably darker in colour. Infrared spectra 2000-650cm le 1325(m) 9 (Ph)3 N.AgBF It. 1575(s), 1485(s)T 1475(s), 1435(w)9 1235(m), 1270(0, 1152(w), 1050(s) (fluoroborate)„ -21
970610 (fluorfAorate) p 845601 775(1 )0 745(8)V 6d7(c)
Aga4 1064p 1026. (Sharp and Sharpe 19)
Analypia.' (Ph)3N.A6F4 Theory o 53°09 H 3.7%p Found o 53.2A H 4.2A-
Triphenylbismuth and silver fluoroborate reacted at once to give a white precipitate which immediately turned brown. This appeared to be similar to the decomposition product of triphenylbismuth silver perchlorate.
TrisazobenzerieBissiLxet2iai)robo The complex separated at once as pale yellow micro- crystals. It was air stable but unlike the azobenzene ail- , ver perchlorate decomposed immediately in water. Infrared spectra 2000-650cm-1.
3(11h)2N2.2Agni. 1575(w)9 111.706109 11014(09 1285(m)9 1214(w), 1065(s) (fluoroborate)9 1020(s)(fluoroborate), 757(6)9 71$(m), 680 (m)
Analysis. 303h)2N2.2AgBF4 Theory C 46*2%. H 392A, Found C 46.6% H 3.6A. -22- jt011=11111),mgrjtexaLlaura_allp=t and triphenylamine in ether did not react even when warmed. Triphenylbismuth gave a brown product. Ethereal Cppr/c Perchlorate. Triphenylamine and tri-p-tolylamine both reacted slowly to give a blue oxidised product. Triphenylbismuth gave a brown tar, azobenzene did not react. Triphenylstibine gave a pale blue product of approximate composition ((eh) sb) Cu(C10 ) 3 2 2'
Analysis. Theory C 44°5-A, H 3°2'A9 Found C 46.6-A H Ethe real Uvoric This solution in ether was extremely unstable with respect to the divalent state, the pale yellow solution rap- idly turned green in contact with the atmosphere. Triphenylamine and triphenylbismuth were slowly oxidis- ed to blue and brown products respectively© Triphenylstibine gave a product of an intermediate composition ((Ph)3Sb)2_4CuC104. AlVlifila ((Fh) Sb)2Cu014 Theory C 494 6 , 3 7A9 H 3° 49 ((Ph)3Sb)40uC104 Theory C 53.)9A, H 3.97;. -23- Found 0 51°3 9 H 4.1A.
Azobepzen,e Qpprous Perchlorate. The complex separated slowly as dark red crystals and when warmed it detonated gently. (This was the only explosively unstable compound isolated, all other compounds handled in quantities up to u.25g0 ignited vigourously but did not explode). Infrared Spectra 2000cm-1 - 650cm1. 9 1295(w)9 1095(s)R (Ph)2N2CuCi0 1590(w)9 1466(m)p 1445(m) (perchlorate) D 915(w) Q 750(w)9 715(m)9 683(s). AD41Yaig 4.5.7'A9 (Fh)2N2CuO10 Theory C H 3'3A9 Found C 45.49 H 4'1%.
infrared Spectra. in_the Fotassjup Bromide. Rea' on. The perohlorate ion absorbs at about 600cm-1 in the potassium bromide region. The complexes were examined for splitting of this band. absorbs at 620(s), (Ph)3 NAgC10 617(0 9 K0104 (Ph) BiA010 610(m) ((Ph) Sb)4 AgC10 620(m)9 3 4 9 3 3(Ph)2N2.2AgC104 617(s), No splittings were observed. -~4©
Discussion. Ionic compounds are generally most soluble in solvents of high dielectric constant having lone pairs of electrons available for donation. In such a solvent the loss of energy of the salt as the ionic lattice is destroyed is compensated by the solvation of the component ions by solvent molecules and by the dispersion and consequent shielding of these ions by solvent molecules. Solvation may be by dipole-ion interaction or the solvent molecule may donate electrons to the metal electron system. Themeare many good solvents of low dielectric constant for ionic compounds, typical of which are liquid ammonia, sulphur dioxide, acetone, formamide, and tetrahydrofuran. They are all characterised by having one or more lone pairs of electrons. A small number of ionic compounds are soluble in this type of solvent by virtue of the small size of their component ions and hence their high polarising power which enables them to gain a solvent sheath in solutinn by dipole ion interaction. Typical are the solubilities of sodium iodide and lithium per- chlorate in diethyl ether (Sidgwick, 15)® More compounds are soluble because,the solvent complexes the metal ion, -25- typical are transition metal halides. Here an element of covaleney also lowers the lattice energy. Such an entirely covalent compound as stannic chloride is miscible with etherpand tungsten hexachloride and stannic bromide are soluble in carbon tetrachloride p a solvent of zero dielectric constant and no donor properties. In view of these considerations it is not surprising that silver perchlorate should be soluble in ether since not only is the silver ion relatively small and polar but is capable of accepting donated electrons. What is much more unusual is its solubility in aromatic hydrocarbons Allthough silver perchlorate was first prepared by Serullas(25) in 1331, its solubility in benzene was first noted by Hill (26) who isolated a solid compound benzene silver perchlorate which seemed to be as stable as many normal complexes (Hill127). Since this time this property has been the subject of many papers. Mulliken (28) suggested that the compound is similar to those formed by silver with olefins and predicted that the silver must occupy a position over a carbon-carbon bond. In this position the tTbond system should be capable of interacting with the d-electron system of the silver ion. The structural determination of Rundle and Goring (29) -26- avggiets that this is so and that in the solid Compound silver ions are sandwiched between benzene rings in positions above and below the carbon-carbon bonds. A more refined structural determination (Smith and Rundle, 30) indicates that the benzene ringmaybe distorted by the crystal field so that two angles are of 122.2° and four are of 115-6°. A high degree of thermal motion makes the position of the silver ion uncertain and the average position appears to be midway between a symmetrical position and one above a C®C bond. The symmetry of the perdhlorate ion is apparently diStotted by the polarising effect of the silver ion-so that the. C1-0 distances are
1.44 A and 1.55 As the shorter distance occurring on the side of the chlorine nearest to a silver ion. Based on the use of crystal energies and assuming simple charge transfer, the energy involved in the charge transfer bond is su ested to be 15'7 kocal/Mol. The Raman spectrum of benzene in a saturated solution of silver perchlorate in water shows considerable shifts (Tauffen et al, 31) of the aromatic ring frequency from 1606-1589cm-1 and of the O-H out of plane frequency from 993-980cm-1. In view of this fact and the apparant stability of the complex it appears that more than a simple charge transfer complex is involved. Tildeley and Sharpe (32) have estimated the bond energy to be 25-50 kcals. This value is based on the approximate equality of the lattice energy of silver perchlorate and the energy of hydration of the silver ion. This is of the order of 100 kcals and is about the energy which must result from the interaction of silver with benzene9 since the heat of solution of silver perchlorate is the same in benzene and water© If it is then assumed that in solution each silver ion is associated with two or four benzene rings the energy of this interaction is 25 or 50 kcals. Sharp and Sharpe (19) have shown that the behaviour of silver perchlorate is shadowed by that of silver fluoroborate and also that of many other silver salts of complex fluoroacids which are all soluble in ether and aromatic hydrocarbons. These complex salts appear to be ionic since they are isomorphous with the alkali salts (Sharp 23)© However the possibility that silver perchlor- ate is not truly ionic must be considered since Hathaway (13) has prepared cupric perchlorate which is volatile. The infrared spectrum of the perchlorate ion in this compound is split and may be explained by assuming a -28- coordinating perchlorate group. Silver perchlorate doed not manifeit such behaviour, but in CoMpOunds of such odd stoichiometries as triphenylamine silver perchlorate the possibility shOuld be considered.
Moss and Sharp (22) have examined the ultraviolet spcc ® tra of silver and cuprous salts in ether and aromatic hydro- carbons® The solutions exhibit intense Charge transfer bands at about 250mu. The specti'a are similar in both sol- vents.and'it therefore seems possible to say that the type of bonding .donation of electrons. to the silver, is the same in both. Sharp (23) postulates that back bonding in the aromatic complexes will enhance the stability considerably; The triphenylamine complexes- isolated in the present work may be complexes with the nitrOgen donating electrons to the silver as in normal ammines or there may be the type of aromatic-silver interaction which has been discussed or a combination of bothc; That triphenylamine is capable of complexing with suitable acceptors, albeit non metals, has been shown in the introduction and in a later part of this thesis indications of complex formation With tungsten hexa- chloride will be discussed. In the group VB triaryls tri- phenylamine and triphenylbisrauth appear to have the lowest complexing ability. No references to triphenylbismuth 29 complexes are made in the literAture except for that with sulphur trioxide. The poor complexing powers of the two follow the series of dipole moments in benzene due to Berg= mann and Schutz (33) PVT, 0.26: Ph3P, 1.45: Ph3A 1007: Fh3Sb0 0057: Ph3Bi9 0 Debye. A more recent study gives the dipole moment of Ph3N as 0.65 (Klages and Longpapeo 34). In itself the planar natures of Ph3N and Ph3Bi indicated by these dipole moments will make it sterically more difficult for close approach of ligand and acceptor as compared with the implied tetrahedral molecules Ph3P and Ph3As in whicho from one side of approach the donor atom will be more prom- inent. If it is accepted that the low dipole moments in benzene solution indicate flat molecules o it suggests that the lone pair of electrons on the central atom is no longer sterically significant and has become lost by conjugation through the aromatic system and will not he so readily available for donation. This view is supported by Jaffe et al. (35) who considers that the intense band at 300mu in the ultraviolet spectrum of Ph N in ether as compared to 3 the less intense band of Ph P at 261mu indicates that the 3 former is more extensively conjugated. That Ph3P is slightly conjugated is however indicated by the way in which the smooth contour of the 261mu band is broken in the spec© -30-
tr m of Ph P01 a compound in which the lone pair is in- 3 a volved in bonding© Sasaki and Kubo (36) have examined the structure of Ph N vapour by electron diffraction. By com- 3 parison of the results obtained with theoretical models they conclude that the ONO angle is 116°9 The plane of the phehyl groups is twisted at about 45° to the plane of the molecule and the C-N distance is 1.42A. This last result compares with a C-X distance in trimethylamine of 10472A and is another indication of conjugation® The results are con- sidered to be compatible with a planar molecule distorted to a shallow pyramid by thermal bending motions. It appears that if triphenylamine is to form complexes a certain amount of energy must be provided to enable the molecule to assume a tetrahedral configuration and to replace the energy of conjugation. The ultraviolet spectrum of the solution of Ph3N.AgC104 in ether shows a band additional to that in Ph N at 1j30mu9 3 the spectrum in the solid state showing in addition a furth- er additional weaker band at 510mu. These extra bands may be due to modification of the electronic structure of tri- phenylamine when it donates electrons to the silver ion. The green colour of the complex is similar to that of the triphenylammonium ion in triphenylammonium fluoroborate9 and oil in both cases may be due to the necessary absence of con- jugation in these compounds if they have become tetrahedral on forming complexes. The evidence of the ultraviolet spectrum of triphenylamine silver perchlorate in ether in- dicated that the complex was also present in solution. The additional band at 510mu could be due to charge transfer although it is of very low intensity compared to that of the band due to the aromatic system at 300mu. The lack of mod- ification in the 300mu band when the complex is formed is somewhat surprising since donation of the lone pair on the nitrogen or of IT electrons from the aromatic system should cause dislocation of conjugation. It is apparent that the ultraviolet spectrum does not give strong support to either view of bonding. Triphenylbismuth silver perchlorate and triphenylamine silver fluoroborate were insufficiently soluble in ether to obtain solution spectra, and solid reflectance spectra could not be obtained since the compounds were not stable for the long period required for measurement. The infrared spectra of triphenylamine and triphenyl- amine silver perchlorate are similar, the main aromatic frequencies occurring in the same positions namely 15659 14909 and 1330 for triphenylamine and 1590, 14609 and -32-
1330cm-1 for triphenylamine silver perchlorate. The -1 1590cm band in the complex shows a decrease 1nintensity similar in behaviour to the band at this position in the triphenylammonium ion. This change in spectra may be ass- ociated with a change in structure of the amine to tetra- hedral and a decrease in conjugation (Kemmitt et al. 17). The C-H out of plane frequencies occur at 750 in the amine and at 780 and 760cm-1 in the complex© In a planar pro- peller-like molecule of D symmetry there should be two C-H 3 vibrationz, (Sharp and Sheppard, 37); the 750cm-1 band must therefore be degenerate in triphenylamine. It is not sur- prising that in the complex there should be two C-H frequen- cies, since symmetry would be 03 or lower if the rings are rotated for a tetrahedral molecule in which nitrogen donated electrons to the silver. The association of a charged silver ion directly with the aromatic system should result in enhancement of intensities by virtue of the increased change of dipole moment associated with ring vibrations. Since it does not seem likely on the grounds of stoichiom- etry that three silver ions should be associated with one triphenylamine molecule, the consequent lowering of symmetry should lead to greater disparities in the C-C and C-H stretching frequencies in the observed spectra than actually 33 occur. The spectrum of triphenylamine silver fluoroborate is similar to that of the perchlorate and in this complex a triplet is observed l4d59 1475 and 1435cm-1, very similar to that in triphenylammonium fluoroborate. The infrared spectra of the triphenylamine complexes may be interpreted as those of a tetrahedral amine donating electrons from the nitrogen to the silver ion, however, they do not eliminate the possibility that a single silver ion is symmetrically associated in some way with all three phenyl groups. The structure of triphenylbismuth is planar in the solid state (Whetsel„ 3d) and the dipole moment indicates that this symmetry is retained in solution. In spite of this planar, propeller-like structure the spectrum is sim- ilar to that of triphenylphosphine being that of a simple monosubstituted benzene ring (Kross et al., 39) and the selection rules which would predict a different spectrum in a planar as opposed to a tetrahedral molecule appear to break down when the molecule contains as heavy an atom as bismuth. Hence it may only be pointed out that the spec- trum of triphenylbismuth is similar to that of triphenyl bismuth silver perchlorate„ both showing characteristic aromatic frequencies, but that no conclusions can be drawn as to the symmetry attained in the complex. These three complexes show an apparent co-ordination number of one which is hitherto unknown for silver. The co-ordination number could be increased by co-ordination from the perchlorate ion or by the association of a silver ion with thel7 bond system of the phenyl groups of a tri- phenylamine molecule which is bonded to another silver ion through its nitrogen atom. The first possibility can be dismissed since careful examination of the perehiorate bands at 1050 and 600cm-1 shows that no splittings occur. The second possibility can only be elucidated by determination of the structure of the complex. A co-ordination number of one would not be found in solution since a more stable co-ordination sphere could he achieved by solvation by ether. Physical chemical data often appear to indicate an apparent co-ordination number of one in SolUti6n (Bjerrum et al.9 40) but in these cases it is extremely likely that there is further coordination from the solvent to the metal. The 4t1 ligand to metal complexes formed with triphenyl- stibine and tribenzylamine appear to be normal complexes in which the silver is surrounded by four ligands. The form- ation of compounds of this sort is considered by Cotton (24) to be characteristic behaviour where the anion is un- like/y to interact with the metal atom. The nature of the three to two complexes formed with azobenzene is much more difficult to determine. They are similar in stoichiometry to the aromatic silver complexes isolated by Sharp (23) On the other hand of the two azobenzene complexes mentioned in the literature one also is of unusual stoichiometry 3T1014.2Ph2N2 (Dermer and Fernelius9 41) made by the reaction of azobenzene with titanium tetrachloride in carbon tetrachloride. YtCyPh2N2 was prepared by Kharasch and Ashford (42) by the addition of azobenzene to platinic chloride in glacial acetic acid. It is pointed out that this compound is more si ilar to isolated from nitriles and platinic chloride (RON)2Ptel4 which is isolated from the rather than (Qlefin Fte12)2 reaction of an unsaturated compound and platinic chloride. It therefore appears likely that azobenzene in this case acts as a donor of single lone pair of electrons, rather than as an unsaturated compound. Since azomethan readily forms complexes (Brown and Dunitz 43) it is likely that availability for donation of the lone pairs of electrons on the nitrogen atoms in azobenzene is lowered by conjugation The ultraviolet spectrum of azobenzene is typical of a highly conjugated molecule having a smooth contour and high intensity (Fig. 2). The decrease of intensity of the )1J445Inu band and the change of contour in the solid reflectance spectrum on going from azobenzene to trisazo- benzene bis (silver perchlorate) could be attributed to a decrease in conjugation on the formation of the complex® The infrared spectra of azobenzene and of the complexes are similar in that they show characteristic aromatic frequencie0 they do not serve, however, to elucidate the structure of the complexes. The spectra indicate that the anions are free in the solid adducts. It is possible that these complexes are polymeric with the azo groups acting as bridges (Brown and Dunitz, 43). If this is so then the complexes must break down in solution as is indicat d by the low observed molecular weight. It is probable that the azobenzene interacts with the silver ion to form trisazobenzene bis (silver perchlorate and flouroborate) through the nitrogen since the addition of aromatic hydrocarbons such as anthracene, biphenyl and triphenylmethane to ethereal silver perchlorate does not result in the formation of solid complexes. It therefore appears that azobenzene along with the other ligands studied regarded normally to be of low basiscity9 may be persuaded under suitable conditions to act as ligands toward electron deficient atoms or ions. 2.11.911111 112412....1322tae Di o riSaelPile Aroney et al. (44) have examined the spectra of a series of stable diazonium double salts. Their examination of salts such as chloroferrates, tetrachiorolodites and fluoroborates indicates that the diazonium group character- istically absorbs at about 2260cm-1. Whetsel et al® (45) studied the infrared spectra of a similar series of salts in this region and in addition examined the coloured salts of p-dialkylaminobenzene diazonium ions and found that the diazonium groups appear to absorb at about 2250 and 2180cm1 respectively. Whetsel et a10 (45) concluded that for a given cation the position of bands in the compound studied was independent of the anion and, to a limited extent, characteristic of the cation but not in a way linearly dependent on the Hammett constants of the ring sobstituents. However Sharp (46) suggested that since the infrared spectra of fluoroborates regularly showed a peak at higher frequen- cies than the spectra of other ,alts,. the fluoroborates were ionic and that there was some covalent bonding between anion and cation in the rest. Work by Neemeyanov et al. (47) has indicated that fluoroborates tend to decompose by a heterolytic mechanism while others such as the chlorides decompose homolytically. Recent Russian ork published in -38- English translation by kitzitayna et al.(48) supports this view of ionic and covalent character. gxnerimental. Although it is possible to isolate solid diazonium halide compounds from solution in alcohol by diazotisation of the parent amine hydrohalide with alkyl nitrite it was felt that the salts thus isolated would be dangerously unstable and in addition, the instability could well lead to anomalous results if decomposition took place while infrared spectra were obtained. This was a fortunate choice since in many cases the double salts, isolated by the method adopted, which are much more stable than the simple halides, decomposed in solution in nitromethane. It frequently occurs that the aromatic diazonium ion is much more soluble in aqueous solution than is the parent amine. Concentrated diazonium solutions were therefore prepared by diazotising a slurry of the amine in hydrochloric or sulphuric acid when the amine goes into solution (Saunders, 49). To avoid diazo-amine coupling, the amine was cooled externally to 0°C, cracked ice was added to the slurry? and nitrous acid added rapidly® Only in the case of p-nitroaniline was difficulty experienced and in this instance only one in four diazotisations resulted in a colourless solution. Solutions diazotised -39- in hydrochloric acid were added to Saturated solutions of metal chlorides in hydrochloric and the chlorometallate precipitated at once. Fluoroboratee Were isolated by adding a diazotised solution in sulphuric acid to aqueous fluoroboric acid. HexafludrophosPhLtee were prepared by adding the diazotised solution to an aqueous solution Of potassium hexafluorophobphate prepared according to the method of Palmer (50). Ferricyanides were prepared by adding a solution of potassium ±erricyanide to a diazo- tised solution in dilute hydrochloric acid. Paraphenylene- diamine was monodiazotised by adding one equivelent of sodium nitrite to the amine in hyrochloric acid and the fluoroborate was precipitated from solUtion. Spectra were obtained as mulls, in nitromethane solution and in aqueous solution. The spectra of the latter were obtained as thin films pressed between silver chloride windows. The background absorption in this case was abuut 60-70$ and a small number of the diazonium salts proved soluble enough to give weak but clearly distinguishable absorptions. The results obtained are shown in Table 2 Infrared spectra were also obtained in the range 1600- 650cm -1 for benzene diazonium, p-toluene diazonium, and p-dimethylaminobenzenediazonium fluoroborates and the results obtained are presented in Table 3 comparing them with the spectra of cyanobenzene and chlorobenzene. When it became apparent that the diazonium frequencies in solution did not vary methodically with the Hammett function of the benzene ring substituent it was considered possible that the influence of the solvent sheath around the positively charged ion could be to counteract the influence of the substituent. The effect of various solvents on the NN frequency in diazocyclopentadiene was examined in an attempt to isolate solvent and ring substituent effects. Diazocyclopentadiene was prepared according to the method of Daring and de Puy (51) by the addition of p-toluene sulphon- yl azide to lithium cyclopentadienyl. The red liquid ob- tained was distilled under reduced pressure (b.pt.50° at 50mm.) and the infrared spectrum compared with that publish- ed. The spectra were identical in all respects and the specimen was used without further purification. The N-N stretching frequencies were found in three solvents of different dielectric constant. The frequency occurred at 2133cm-1 in nitromethane9 2133cm-1 in acetonitrile9 and 21250m-1 in carbon tetrachloride (dielectric constants 399 399 and 2 respectively). -41 - ttedag_ Mull Nitromethane Aqueous solution solution ,:len.zenediazonium1 2297 2280 2280 2290 2280 n.s. g01 2265 2275 nos. Zn014 2277 d 2280 0d011 2273 2283 2285
BnC16 2276 2288 no So
Pe(ON 2278 aotto 2283 BPDai; 2282. d no eel- 2262 2270 2260 (benzene diazonium ohloroferrate dizeolved in acetone ab- sr2 a.4 2277cm m-Toluone diazonium BF 2294 2274. 2280 P6 2291 2275 n.s. Hel t 2260 22 03 n.s. Zna 2268 2272 2275 0d01 2272 2270 22 70 SzCl6 2269 2263 n. s,: Fs (00 2266 n.s. 2275 -42- Ithlg-Z4_2aatinuale Mull Nitromethane Aqueous solution solution o-Toluene diazoniu BF4 2290 2271 2273 PF6 2289 2275 n.s. Hg0142 - 2253 2260 nos. p-Toluene diazonium BF"4. 2299 2275 2276 PF6 2287 2277 nos° HgC142- 2275 2272 nos. o-Nitrobenzene diazonium BFL 2263 2295 14- 2295 FF6 2293 d n°s© HgC142- 2260 2280 noSo m-Nitrobenzene diazonium BF4 2301 2298 nos° PF6 2305 d n°s© HgC142- 2285 2285 n08° p-Nitrobenzene diazonium BP 2321 2299 nos, PP; 2310 d nos. 40142- 2281 2281 nos. U.112.2.9_2.9.4-Lkimazio
Mull Njtromethane Aqueous solution solution o-Trifluoromethylbenzene d iazoni
2272 no B 0 BF4 2295 2287 2272. nos PF6 HgC142- 22.72. d nos© m-Trifluoromethylbenzene diazonium MIL 2299 noso 4 PF 6 2298 no a0 HgC1 2- 2281 nos. 4 p ethoxybenzene diazonium
BF4 2265 2252 n. 0 2 HgCl4 2214.0 2255 110610 p -Dimethylaminobenzene diazonium BF 2268w 22514w nos,. 4 2181e 21778 2255w nos° PP6 2252w 21788 2175s OdaL2 2250w 2260w nos. 4 21688 2175s
HgC142- 2246w 2255w no S 2170s 21758 -44- Talae ?o cwitiAanesle
Mull NitroMethane Aqueous solution olution p-Diethylaminobenzene diazonium
BA 2260w 2252w nO SO 4 2173s 2168e PF6 2250w 2259w nos. 21708 21638 CdOlL2- 2251w nos. 4 2175s. 21698
HgO142- 221.8w 2256w 11080 2170s 21628 p-Aminobenzene diazonium 2170 dnoso BF4
nos not sufficiently soluble for spectra to be measured. d9 decomposes: 89 strong: wo weak. Tg„ta2_3.2._Igfx4=Ls ectra 1600-650cm-1 (figuras n
06H5CN 06115C1 005N21324 p-MeC H0213114 p-(Me)2NC6He0/4
757 740 750 617 8258 972 965 960 930 985 975 975 983 1000 1003 1026 1026 1015 1070 1068 1075 1093 1096 1100 1115vs 1163 1157 1178 1085 1125 1168 1.125 1192 1174, 1180 1192. 1286 1271 1295 1290 1290 1332 1326 1315 1310 1320 1365
1385 1390 1445 1445 1414 1400 1437 1489 1477 1465 1450 100 1579 1580 1540 1597 1580 1570v3 1590vs 1590v sp strong: vsv very strong Discuerstan. The first diazonium compound re prepareid by Greiss (52) and Ilene (53). Greiss (52) diazotised pi,)ramic acid in alcohol by the action of nitrous fumes deri d from the reduction of nitric acid with starch or arseni.)us acid. His success in diazotisation depended, not (as he thought) on the use of alcohol as a solvent;, but on the stability of the diazotised product and the low temperatur e at which he worked. The use of alcohol as a solvent when the solid diazonium salt is required depends on the low solubilities of such compounds when ether is added to the diazotised sole ution and at present the use of alkyl nitrites, esters of nitrous oxide, to diazotise aminee, in alcohol and to isolate diazonium compounds soluble in water is common practice. Initally Greiss (54) considered that two hydrogen atoms in the aromatic nucleus were replaced by nitrogen atoms and hence chose the name ddiazoc for his new compounds. Greiss appreciated the basi e nature of the diazonium compounds and soon prepared the first complex salts of benzene diazonium chloride with auric chloride and platinum chloride PhN Cl.AuCl 3and ITN9Clatel . However this property of the diazonium ion was little studied and twenty four years elapsed until (PhN2)PtC1 was prepared, Idicheelis and Ruhl, -47- 55) although the compound was formulated as a double salt
(PhN201)2PtOilt. rather than as the diazonium salt of the hexachloroPlatinate ion. The tendency toward double salt formation is so strong that the diazonium ion will form a salt with the bichloride ion HO1% Hantzsch (56) isolated compounds such as (Br3C6H2N2)11012.4H20. The complex salts formed with mercuric chloride are better known than this previous type of salt since they are used in the synthesis of organ mercury compounds. Hantzsch (56) first isolated PhN20HgC130H2O, while Nesmejanov (57) has prepared a series of the type ArNeHg013. By heating this last series of salts in acetone with copper powder and ammonia high yields of diaryl mercury were obtained. The relative ease with which this reaction proceeds caused Nesmejanov and his co- workers to examine the properties of many diazonium double salts, thallium, stannic and auric amongst many others (Zollinger, 56). Chattaway et al. (59) prepared the double salts of the plumbichloride ion and the tetrachiordiedite ion and state unequivocally that the salts are best represented as salts of complex ions. The tetrachloriodite was observed to be extremely stable compared to the simple chloride, and mod- erately stable to the atmosphere and non explosive at room temperature. The stability of the diazonium double salts is an important factor in their industrial use and it has been found that generally the chlorozincate is stable and soluble enough to find extensive use where 'ready madeg diazonium solutions are required (Saunders, 49). Bart (60) found that addition of fluoroboric acid to benzene diazonium chloride solution resulted in the form- ation of the almost insoluble -benzene diazonium fluoroborate. The diazonium fluoroborates are extremely stable, for dia- zonium compounds, and have been the subject of considerable research. The salts are rather insoluble for use in dye baths but such compounds as p-nitrobenzenediazonium fluoro- borate (Nitrazol CP extra) find considerable use technically. The thermal degradation (Schiemann reaction) of fluorobor- ates usually results in the release of boron trifluoride and a fluoroaromatic. The allied hexafluorophosphates decompose to give phosphorous pentafluoride and are a very convenient source for research use of this gas. In general the addition of large anions such as molyb- dates, cobalticyanides, nitroprussides, and complex fluoro= anions (fluorozincates, fluoroaluminates etc.) to aromatic diazonium solutions result in the formation of .a more or lesS stable salt which can usually be isolated and dried safely (Saunders, 49). The structure of the diazoniUM Ion was a topic of some controversy for a munber of years. . Gneiss later accepted that in some way two nitrogen atoms were attached at one point to the aromatic nucleus and speculated as to theie being one pentavalent and one trivalent nitrogen atom in the chains Kekule (62) realised that there was siMilarity between the azo compounds and the diazonium compounds and he postulated the structure kholl:NX. Blomstrand (63) Ultimately it has however suggested the structure Ph-N-X.an been recognised that Blomstrandqs structure involving as it does a triple bond is correct, except that there is no def- inite point of attachment of the anion to an ionic compound. Dilthey et al. (64) postulated a three membered ring struc- ture in which the nitrogen atoms would be equivalent. ,N
Clusius and Hoch (65) have however prepared phenyl hydrazine by the reduction of the benzene diazonium salt made by diazotising aniline with nitrite containing heavy nitrogen NaN1502. On disproportionation the products obtained, -50- NN15 and N15H indicate the reaction: 3 2C MI-NISH C NH + 06H6 + NN15 N15H3 6 2. 6 H5 in which clearly the nitrogen 14 attached to the aromatic nucleus remains in the same position and indicate6 the non- equivalance of the two nitrogen atoms. 0 Triple binds typically absorb infrared radiation from -1 -1 about 2100 to 2300cm 9 -CEO- occurs at 2100cm , -ClEN at 2240=-1 and the diazonium group at about 2250cm 1 (Bellamy 66) i very strong evidence for the presence of a triple bond in these compounds. From visual observation simple diazonium salts form two cla6sesp those obtained from p-phenylene diamine and p-N.N dialkyldmino anilines which are Coloured and others which are white. The former will be considered later. The N-N Stretching frequencies of the diazonium salts prepared are shown in Table 2. The spectra were measured as mulls in hexachlorobutadiene or nujol and where possible as solutions in water and nitromethane. The spectra of many of the colourless diazonium salts have been studied by Kazitsyna et a10 (67) and in general our observationt are similar to theirs. No multiple peaks were observed for the colourless salts although on occasion a multiple peak occurr- ed in the spectrum of a mull. The multiple peak always disappeared on regrinding the specimen and was attributed to some form of crystal interaction. The frequencies observed are higher than those observed by the Russian workers since all the spectra obtained were calibrated by means of a standard air sPectruh this discrepancy is inex- plicable. Chloroantimonites were not examined, however, in the Russian work these compounds frequently proved to have a double N-N stretching frequency KazitsYna et al® offer no explanation of this feature. In the diazonium compounds examined the NN stretching frequency was always symmetrical for the fluoroborates and the solutions but were frequently asymmetrical for the chlorometallate® It must be assumed that fluoroborates are completely ionic, and aoldschmidt (68) has shown that in aqueous solution diazonium salts are completely ionised. In nitromethane it seems likely that the compounds are like- wise completely ionised but there is a strong possibility of ion pair formation® Whatever the exact nature of the sol- utions the shape of the spectral bands observed indicates that the lattice is sufficiently disrupted to minimise anion-cation interaction. It therefore must be assumed that in some of solid complex salts the structure is such that diazonium groups are in different positions with relation to the anion. This explanation, based on the assumption that anion-cation interaction occurs, is further supported by the observation that in all the salts prepared, for any one given cation the fluoroborates and fluorophos- phatee always exhibit the highest N-N stretching frequency, whilst in the aqueous solutions of the benzene and m-toluene diazonium ions the frequency is in a position intermediate between those of the solid salts. Clearly, unless one assumes fluorine bridges, the spectra of the fluoroborates is that of the free diazonium cation and the spectra of the chlorometallates show interaction with the cation. This interaction is probably due to partial covalent bonding between anion and cation, the bonding being by direct inter- action between the lone pairs of the halide ion and the vacant orbital of the diazonium group. This does not occur with the salts of the strong complex fluoroaoids. Inter- action between the 11 bends of the diazonium group and the metal atom of the complex anion is possible though neither the NN or NEN bonds have yet been shown to form 11' bond complexes. Any relation between the N-N frequency and the overall crystal field at the site of the NN bond is ruled out by the independence of this frequency of the size of anion and, furthermore, by the approximate constancy of the -53- N-N frequency in solvents of different dielectric constant (e.g. benzene diazonium chloroferrate absorbs at 2260cm 1 in water, at 2277cm-1 in acetone and at 2270cm-1 in nitro- methane, dielectric constants 81-5, 21°1. and 24.7 respec- tively (International Critical Tables, 68a)). The un- charged diazo group in diazocyclopentadieyl likewise shows approximatlly the same variation, from 2125cm-1 in carbon tetrachloride to 2133cm-1 in nitromethane, nothing like so large a variation as, for example, that of the nitrosyl group in NOC1 in which the nitrosyl frequency varies from 1814cm-1 in carbon tetrachloride to 1381cm1 in nitrometh- ane. Whetsel et al. (45) have attempted to relate the N-N stretching frequencies to the electronic effects of other ring substituents as measured by the Hammett function. Their obserVations, which were made on the Solid compounds showed trends in the right direction. They comment that solution spectra might give better results since the cation will be insulated from the aniont the solution measurements made again show these trends but the relation does not hold so well as is observed for the C=O group (Fuson et al. 69) in which a linear correspondence iS observed for the carbon- yl frequency and the Hammett constant with a variation of about 30cm1 fromcr= -006 to OrA= # 0G80 However in nitro methane solution the N-N stretching frequency for p-methoxy- benzene diazonium (r. 0°27) is about 2250cm-1, in benzene ®1 diazonium (0-= 0) about 22d0em and in p'-nitrobenzene diazonium (or +1.27) about 2290cm1 . As variation of the frequency for any one cation in solution is as much as -1 -10cm no great importance can be attached to this result. An additional complication must be that with cations such as the diazonium ions, solvation must be greater than for an uncharged molecule. Since the electronic effect of sol- vation may well oppose the effect of the substituent„ an overall complexity might be expected. The infrared spectra of diazonium salts over the range 1700 - 650cm1 (Table 3) are similar to those of the appro- priately substituted aromatic nucleii save that the aromatic skeletal stretching frequency near 15d0cm-1 has a much greater intensity in the diazonium salts. Association of an aromatic system with a positively charged, electron deficient, entity generally has the effect of intensifying this vibration (Sharp and Sheppard, 37). This is a vibrat- ion of the ring system and in consequence will cause move- ment of the positive charge and hence cause a larger change -55- in dipole moment. The division of diazonium salts into two types on the basis of chemical reactivity (Saunders and Nesmeyanov et al. 499 47) follows the pattern based on the 144 stretching frequencies of the crystalline solids. A high N®N frequen- cy typical of the ionic fluoroborates, is associated with heterolytic decomposition while partial covalent character is associated with homolytic dissociation. The first results in a phenyl cation and the second in phenyl radicals. The relationship of heterolytic decomposition with ionic character will be considered in some detail espdcially as the detection of this type of decomposition has involved the synthesis of some novel compounds. Nesmeyanov (47) has observed that the decomposition of diphenyliodonium fluoro- borate in the presence of pyridine results in the formation of N-phenylpyridinium fluoroborate whilst the decomposition of the chloride yields a and Wphenylpyridines. Unfortun- ately the same test applied to benzenediazonium fluoroborate results in the formation of substituted pyridines and it appears that pyridine, a base, first converts the diazonium cation into a diazo compound which homolytically decomposes. In a similar manner the presence of powdered copper results in the formation Of ortho and para nitro biary16 when an -56- aryldiazoniUm fluoroborate is heated in nitrobenzene. In the absence of copper the meta nitro biaryl results ex- clusively. Holever, when an aryldiazonium fluoroborate is decomposed in the presence of bromobenzene or diphenyl ether the previouSly unknown diphenyl bromonium and triphenyl- oxonium ions result, unequivocable evidence of heterolyio0 Diazonium chiorometallates do not behave in a similar manner lending additional support for pobtOlateS of their covalent. nature. Schtidt and 14aier (70) recorded the synthesis of the coloured diazonium salts obtained from p-dialkylamine anilines, stating that the yellow p-dimethylamino benzene diazonium chloromercurate obtained was stable in solution in the dark for periods up to a year. Some of the compounds of this general type are light sensitive and find use in diazotypes (Saunders, 49). Whetsel et al. (4) have studied the infrared spectra of these compounds and have observed two absorptions in the .N-N frequency region, a weak absorption at about 2250cm-1 and a second.stronger absorption at about 2180cm-1. They • suggest that if the structure of a p-dialkylaminobenzene diazonium salt is represented by two resonance hybrids, one aromatic and one quinonoid as 1- (1)
The influence of the electron releasing dialkylamino group would be to increase the degree of quinonoid structure and decrease the strength of the NI triple bond hence explain- ing the lowered frequency of its infrared absorption. This effect shows itself in the decreased reactivity of the diazonium group which will only couple under a small range of conditions (Saunders9 1i9) Anderson and Steedly (71) have compared the ultraviolet spectrum of the po-aminobenzene diazonium ion with that of diphenyl quinomethane (06115)2C =0 and observe the similarity of their spectra which both show bands at about 2,50mu and 390mu and hence conclude that the diazonium ion is likewise quinonoid. It is worth observing that compounds having azo groups associated with aromatic rings have sim- ilar ultraviolet spectra and it does not appear essential to invoke both quinonoid structure and an azo group to ex plain absorption in the visible region. Zollinger consid- ers that Iconclusions about the structure of diazonium salts drawn from considerations of their ultraviolet and visible spectra can no longer be accepted. The infrared spectrum of p-dialkylathinobenzene diazon- ium fluoroborate is shown in Table 3 compared with that of p-toluene diazoniurn fluoroborate. Essentially the spectra are similar both showing absorptionS characteristic of aromatic compounds. The p-dialkylaMino salt shows an ®1 additional intense band at ill5cm in a position character- istic of C-N frequencies which normally Occur in the range 1000-1200cm-1 (Bellamy v 66). The association of this band with a transferred positive charge would involve the vibrat- ion in an increased change of dipole and hence absorption would be intensified. In addition the overtone of this band would occur at about 2240cm-1 and could then account for the additional frequency observed in the N-N stretching region. It may be noted that the single absorption of p-aminobenzenediazonium fluoroborate in this region is explicable since there are two fewer C-N vibrations in this compound. The second strong N-N absorlition observed is markedly less sensitive to the influence of the anion than the compounds previously studied as would be the case when the charge is transferred. This frequency is 100cm-1 lower than that observed in other diazonium salts and this could be explained on the basis of structure (11) which has a double N=N bond. This structure is however unsatis- factory onon a number of grounds, the first being that it shows three double bonds for which there is no infrared evi- dence, namely the N=N and two C=N which should absorb at about 160Ocm-1© In addition the infrared spectra of quin- ones as determined by Yates and Ardao (72) are completely distinct from those of benzenoid aromatics. In particular it may be noted that the C-H out of plane stretching fre- quency at 825cm-1 in p-dialkylaminobenzene diazonium fluoro- borate, characteristic of two adjacent hydrogen atoms, does not occur in 1-4 benzoquinone nor is a peak recorded any- where in this region in this compound In addition the distribution of C-C stretching frequencies in quinones other than that at 1600cm-1 is completely dissimilar. It therefore appears that to suggest a quinonoid structure for these compounds as has been done in the past is incorrect and that in terms of resonance hybrids the structure is best represented as;-
R2N Nr-m-10 (1) (111) or better still as an extensively conjugated system on which the net positive charge lies more on the dialkyl groups than the diazonium group. -60- The Hydrogen Bonch The Nature of The Bond. A hydrogen bond exists between a group of atoms Y-H and a second group Z when it appears that in some way the groups are chemically bonded through the hydrogen atom. There is no particular limitation on the nature of Y or Z which may be part of the same molecule (intramolecular hydrogen bonding) or different molecules or parts of mole- cules (intermolecular hydrogen bonding). The groups Y and Z are usually terminated by having one of the more electro- negative atoms such as fluorine, oxygen or nitrogen next to the central hydrogen atom of the hydrogen bond. The hydro- gen bond varies in bond strength fram the symmetrical bond in the HP2 ion (Darlow, 73) to bonds which are so weak as to be undetectable® Moore and Winmill (71i-) recognided that the difference between tetramethylammonium hydroxide, which is a strong base, and trimethylammonium hydroxide which is a weak base, may only be explained by assuming that in the latter the hydroxide is more strongly attached to the ammonium group. They therefore postulated a hydrogen bond type of structure with the hydroxide attached through the hydrogen to the central nitrogen atom through an H-0-41-N bond. -61- Latimer and Rodebush (75) surveying the nature of bonding in a wide variety of compounds, considered that water is of such a nature that bonding might occur between individual molecules by the juxtaposition of a hydrogen atom of one molecule to the oxygen atom of another in such a way that a weak bond is formed. they further considered that there is no limit to the nuinber of molecules that might thus be bonded together. Since this time it has been extensively noted that the hydrogen bond exists in a wide variety of compciunds and situations. Hydrogen bonding results in detectable dev- iations from ideal gas and solution behaviour and finally and most important to the work carried out in this thesis, strong shifts in infrared spectra (Pimental and Macelellan, 85).
The Detection of TivdrogLen BondingL. bv Infrape0 Spectroscopy. In a generalised hydrogen bOnd represented by
M1-Y-H Z-M2 where M1 and M2 are the remainders of the molecules between which hydr6gen bonding is taking place, the vibrations of the molecules are as follows:-
-62- o M1 -Y -4- Z-M2 Y-H stretch mi-y H z-m 2 Y-H in plane bending
141-Y-HE)- Z-M Y-H out of plane bending (11/-1--H) (z-M ) -4 2 hydrogen bond stretch 1011-y-Hyt- (z-112)1 hydrogen bond bending®
The Y-H stretch usually occurs from 3500-2500cm1p Y-H in plane bending from 1700-1000cm-19 and'Y-H out of plane bending 900-300cm"4. The hydrogen bond stretch is thought to occur at about 200cm-1 whilst it is likely that the hydrogen bending frequency of the bond will occur at less than 50cm 1. The Y-H stretching frequency has been the most widely studied mode of vibration since usually it is intense and occurs in a part of the spectrum where its detection is reasonably certain. The behaviour of the frequency under the influence of hydrogen bonding is predictable though theoretically rather obscure and the subject of much con- troversy. The three most characteristic results of the formation of a hydrogen bond are as follows (Pimental et al. 76):- 1)A shift of the vibration to lower frequencies. 2)An increase of band width. -63- 3) An increase of overall intensity© The decrease of frequency when a hydrogen bond is formed must be a reflection of the decreased force hond constant of. the Y-H bond® 'Measurements tabulated by Nahamoto et al. (77) indiCate that as the Y-Z distance decreasest'that is the hydrogen bond strength increases, the Y-H distance increases. It has been shown that a decrease in force constant is commensurate with an increase of bond length (Badger, 78)0 This process carried to the limit results in a symmetrical hydrogen bond system as in the HF2 ion in which the hydrogen atom is equidistant from two fluo- rine atoms 2°27A apart (Darlow, 73). The increase of band width on the formation of a hydro- gen bond seems to suffer from having too many explanations. However, it does seem possible that in some cases the diff- erences lie in the crudity or elegance of the theoretical background and the method of stating the problem. The observed phenomena consist of broad absorption bands corres= ponding to the hydrogen bonded Y-H system: some bands are diffuse and smooth in contour and others may be partially resolved. In addition, the spectra may frequently be fur- ther resolved by cooling the specimen to liquid nitrogen or liquid helium temperatures (Bratoz and Hadzi, 79). It -64- would therefore appear that the band consists of series of temperature independent absorptions smoothed out to a great- er or lesser extent by temperature sensitive vibrations. Clearly one of the first places to look for the cause of the band system is within the hydrogen bond system itself since the dependence of the Y-Z distance on the strength of the hydrog-en bond and its relation to the YH distance indicates the extensive interdependence of Y-H and H o - Z. The crudest notion is that in which the (Y-H) Z vibration frequency modulates the Y-H vibration: since, however, the spectra observed do not normally resemble a formal pattern
nV(y_,,H) -Z' this can be dismissed as being too simple. However, the vibration Y-H and the vibration (Y-H)- Z are clearly strongly coupled and it is possible that random thermal vibration of the Y-Z distance would lead to this being an anharmonic coupling with a high variation of absorption position because of the relatively weak nature of the hydrogen bond. By assuming that both vibrations are anharmonic, a band width of 300cm1 has been calculated for ice by Lipincott and Schroeder (80), the higher frequency occurring at long values of the Y-Z distance and vice versa. This is the theory of Batuev (81). The theory of predissociation was first suggested and -65- developed by Stepanov et al. (d2)© This postulates that the Y-H vibration is capable of absorbing a quantum of energy which lifts it to a higher znergy state for which there is a second potential energy curve. This second curve has a minimum Y-Z distance less than that Of the lower potential energy curve. This is a restatement of the fluctuation theory in quantum mechanical terms but it has the advantage that transitions occur from the ground state to the excited state at several points corresponding to V - nV for both curves. The result of ab- Y-H - (Y)- sorbing a quantum of radiation V Will be, even at low temperatures, the fluctuation of the Y-Z distance and a con- sequent residual band width as is frequently observed. It appears Possible that the spectrum of KHF2 can be satis- factorily explained by combination and overtone vibrations based on the theories mentioned (Newman and Badger, 63). An additional complication to the interpretation of the spectra observed is that the Y-H vibration is capable of Fermi resonance with the hydrogen bond deformation vibration. Since this should occur at about 50cm and is in turn capable of reacting with the hydrogen bond stretching vib- ration, overall complexity is not surprising. -66 The increase of intensity in the formation of a hydro- gen bond is not always observed. It appears to occur most readily in charged molecules, and thus in pyridinium chlor- ide there is considerable intensification, resulting presum- ably from a considerable change of dipole of the N-H band on going from a normal NIH vibration to N±H® m Cf. On the other hand there is no increase of intensity in acetyl- acetone (Bratoz and Hadzi, 79). A theoretical study of the change from the H-F to the
HF2 ion indicates that the change in dipole may be as high as from 1°91 to 4.15D (Sigeru Nagahara, 84).
The in plane bonding vibration (V6) varies much less with the formation of hydrogen bonds: little change occurs in width or intensity, while only a small upward shift in frequency is observed (Pimento", 65).
The torsional vibration of the M1Y-1 group in effect about the hydrogen bond has been detected and predicted on a number of occasions. One important application has been in the ammonium salts where the torsional vibration has -1 1 been assigned at 390cm in ammonium chloride and at 319cm for ammonium bromide. Waddington (d6) has pointed out that in those ammonium salts which are isomorphous with the pot- -67- assium and rubidium salts a combination frequency corres- ponding to the ammonium ion deformation frequency (V4) plus the lattice torsion (V6) does not occur. Whereas when nuclear magnetic resonance studies and infrared measurements indicate hydrogen bonding, the salts are not isomorphous with those of the rubidium and potassium salts and a band assignable to V4 + V6 occurs.
The hydrogen bond stretching frequency which is pre- dicted to occur at about 200cm-1 has been observed in the Raman spectra of water (Gross, d7) and in the infrared and Raman spectra of formic acid (Miyazawa and Pitzer, 68). The observation and study of the hydrogen bond stretch is clearly very important and while at present the difficulties of working in this region are considerable, the next few years should see the development of spectral techniques so that the region may be examined more conveniently. Sal 0 The work described in this section has been published as a paper in conjunction with research performed by Dr. D. W. A. Sharp and Dr. T. C. Waddington. The preparations described in this thesis are the writers own work, the res- ults obtained are, however9 made more understandable by com- parison with work which is not the writers own but which has been published under the writers name in conjunction with those above.
Substituted ammonium halides have been known for a long time, though they have to a large extent been regarded as the amine hydrohalides. In recent years a number of papers have been published dealing with the overall spectra of this type of compound. Sheppard and Ebbsworth (90) have examin- ed the spectra of the Me4e„Me3Ne and Me211H iodides (Me = Methyl). The last two ions were deuterated and the posit- ion of the N-H vibration indicated considerable hydrogen bonding. The spectra of substituted ammonium fluoroborates have not previously been examined. Ammonium fluoroborate app- ears not to be hydrogen bonded (Waddington, do) which is not surprising since in this compound the electrostatic term -69- N-H•- - F-13 is not so favourable since presumably the neg- ative charge on the fluoroborate ion is evenly distributed over the ion. Similarly the covalent factor is not favour- ableg .the anhydrous acid HIM, having no independent exist- ance (Maceauley and Lien, 3). Sharp (91) has however pre- pared triphenylammonium fluoroborate and in this compound 1 the N-H frequencies are observed at about 2600cm which suggest hydrogen bonding. In this compound the aromatic system favours the isolation of the positive charge on the nitrogen - hydrogen bond where it may be capable of inducing the formation of a hydrogen bond by polarising the fluoro- borate ion. A systematic series of substituted ammonium fluoroborates were therefore prepared and their spectra examined.
Exnerimental. Ajavlammonium FluorOorates. Trimethylammonium fluoroborate was prepared by the reaction of trimethylamine and aqueous fluoroboric acid. The amine was distilled from a solution of the hydrochloride in caustic soda and condensed in a liquid nitrogen trap con- taining fluoroboric acid. The amine was condensed in excess since it is the more volatile of the components. The solution obtained was evaporated to dryness under an -70- infrared light and the product recrystallised twice from water. The triethyl, dimethyl and methyl salts were prepared by the same method. The dimethylammonium fluoroborate was extremely hygroscopic and was therefore recrystallised twice from absolute alcohol and thereafter handled in the dry box. A satisfactory analysis could not be obtained. Analysis. Me3NHBF Theory 0 24-0, H 6-8A„ (Me = Methyl) Found C 24.7A, H 6°9,49 Et NHEIF 3 4 Theory C 3801*, H (Et = Ethyl) Found 0 38.2A, H 8o4A, Me NH3B1P4 Theory C 11-2A, H 506A, Found C 1104A, II 5°7A.
N-deutero Trimettlylaprownium Fluorobora,g. The deuterocompound was prepared by dissolving tri- methylammonium fluoroborate (0*25g.) in deuterium oxide (1m1.). The solvent was removed under vacuum and the com- pound handled in the dry box. The infrared spectrum indicated that exchange had taken place, the residual N-11 peaks being about 20 of the inten- -71- sity in the hydrofluoroborate. Corresponding to these alkyl salts three phenyl salts were prepared.
Triphenylaptmoylipm F1porobor4te. The compound was first prepared by Sharp (91) by pass- ing boron trifluoride into a solution of triphenylamine in moist benzene when Ph3NHBF4 was isolated as a green crys- talline solid. It was found that in the absence of moist- ure no crystallisation occurred t suggesting that the boron trifluoride was hydrolysed. BF + H2O HBF OH 3 3 2HBF 0H HBF4 + HBF2(OH) etc. 3 2 The infrared spectrum exhibited three moderately inten- se absorptions in the hydrogen bonded N-it stretching region at about 2800cm-1.
N-deptero TylphgnylAmmopiula nuoroboratp. Sodium dried benzene was allowed to stand in contact with heavy water for about twenty four hours in a closed vessel. Triphenylamine was dissolved in this benzene and boron trifluoride bubbled through the solution. The boron trifluoride was scrubbed and dried by passage through a boric oxide/sulphuric acid bubble column. The salt crys® -72- tallised almost at once and was transferred to a des0i4itor and vacuum dried. Examination of the N-H fiequency region showed that the three peaks had decreased considerably in intensity and that new N-0 peaks had arisen.
Diuhenylammonium ?luoroborate. Diphenylamine was dissolved in fluoroboric acid. The solution was evaporated to dryness under an infrared light and the product recrystallised twice from ethanol. The solid was isolated as pale green crystali.
Analy.Ate. Theory C 56.0%, H 4.7%, Found C 5505%, H 4°6%. The solid is soluble in water to yield a solution which decomposes to give the parent amine.
Anilinium Fluoroborate. Fluoroboric acid was added to aniline and the solid obtained filtered and recrystallised from ethanol.
Analysis. Theory C 39.9$, H 5%, Found C 39.8, H 4.7Ao Anilinium fluoroborate forms a stable solution in -73- water. Tri-p-tolylammonium fluoroborate and di-p-tolylammonium fluoroborate could not be prepared by any of the above meth- ods.
13-TO-YlamniolliUMLUVrOoTate was prepared by the same method as for diphenylammonium fluoroborate.
AnDlysls. Theory C 42°4%, H 5°6%9 Pound C 4008%, H 5.5%, The compound forms a stable aqueous solution.
Me3NH p Me2NH2, and MeNH3BPh4 salts were borrowed from Dr. Herringehaw and infrared spectra obtained. Ph2NH2BPh4 and PhNH3B.Ph4 were prepared by the addition of an aqueous solution of sodium tetraphenylborate to acid solution of the amines. The two salts had such complex infrared spectra that their usefulness was invalidated.
Infrared Spectra. Infrared spectra were obtained in the region 4000-400 cm®1 of the compounds as mulls in nujol and hexachlorobuta- diene. Tentative assignments are made for some of these spectra by comparing the results with those of Ebbsworth -74- and Sheppard (90) and from the deutero salts and the tetra - phenylborates. In the case of the aromatic salts assign- ments are made only for the well established features. The ratio of the shift of the N11 to N-D frequencies on deuteration all occur in the region VH/VD =4 1°26 to 1031. The variation from ideal is not surprising in the case of hydrogen bonded systems when it is remembered that there is no reason why the interactions N-H F-B should be similar to N-D F-B in proportion to the mass of 'H and D. In general, if the bond N-1 is not involved in coupling the ideal shift would be 111(.2 in direct proportion to the masses of the hydrogen atom and the substituted deuterium atom. Infrared spectra in the KBr region are compared with those of ionic fluoroborates. The spectra obtained are set out in Tables 4 and S.
Alp9ussiq. Chenon and Sandorfy (92) consider that for an N±H - hydrogen bond the major contribution to the stability of the bond arises from the electrostatic interaction, a simple Coulombic attraction of opposite charges. If X is part of a complex anion the charge on X will be reduced and hence the interaction reduced. Contributions NY X and -75- =31 33:50 (a) -N-H 3200 (a)
29142(a) s, 2975(3) -CH 2850(a) p 2930(0 2680 (w.) 2478 (Iv) 1482 (m) 1475 (8) 114.62(m) 142,5 (ir) 1120 (w) -N-H 14.00 (m) -CH 3 1365 ( ) BP- BP4 1290 (m) 4 1300 (w) 1070 (al)) 1025 (sib ) P 1025 (sb) r 1065 (sb) BA,4 975 (a) BIK7 975 (a) BK 4 4 847(m) CH ? 850 (W) 0H 3 3 80o (n) CH? 807(0 0-N? 765 (m) BP, 767(w) BIN. 4 4
(s strong,. b = broad,. m = medium,. W weak). In Napa% the N-3) stretching frequency is split into a. triplet at 2450p 2.400 and 2375cm'1° The 1420cori dis- appears on deuteration presumably under the strong BFII: absorptions In Me3NHB4a4p N-H occurs at 3185cel. -76 LA131A.12...=tlaugst.
ga NH BP 1%12 244 3257(a) N-H 3257(m) N-H 2935(0)9 3010(a) C-H 2854(m)5, 2926(a) C®H 2860(m) C-H 2772(w) C-H 2790(m) N-H 2560(w) 2450(w) 24.70(w) 1468(s) 1510(w) CH3 N-H CH 1415(m) 1465(m) 3 1335(w) 1375(m) 1303 (w) 1295(w) BF 4 1290W 1265W l025(313), BF 925W, 1060(sb) BF4 1075(sb) 4 693(m) 850(W) 825(s) CH C-N? 3 795(m) BF 756(w) BF 770(w) 4 4 In the system MeNH+3 more N®H frequencies would be ex- + pected than in Me3NH or Me2NHa and it is probable that the three peaks observed in the 1600-1200cm1 region are due to N®H vibrations. Three N-H vibrations are observed in this region in MeNH3I. In MeNyPh4 the N-H frequency occurs at 3175cm-1. In Me2NH2BP114 the N-H frequencies occur at 3410 and 2760cm-1. -77-
oA 3.tg6144.,EM3.13114. X Zh2NR2234 120jbah. N=H 3175 ( 3215 (s) 3195 (a) C-H 3075( =) 3070 (w) 3010 (w) 2925 ( 2923 (s) 292.0 (w) 245 (m) 2850 (m) 2845 ( ) 2583 (w) 2600(w) 2620 (w) 2485 (m) 2325 (w) 2315 (w) ao 1597(w) 1580 (m) C-0 1590 (m) 1587 (w) 1527 (m) C-C 114.92(s) 1491(8) 1507(m) 0.0 3470 (in) 1465 (m) 1394(m) 1373 (w) 1385 (m) 13F 1295 (w) 1290 (m) 1288 (m) 4 1110(s) iaoo (w) BFLi. 1060 (sb) 1060 (sb) 1060 (sb) 1025 ( sb) 1020 (sb) 1027(01) O-H 750 (m) 800 (w) 803 (s) 0-H 744(s) 743(s) 765 (w) C-H 690 (m) 680 (m) 720 (m) 680 (m) 690 (w) -78— TBLE Comoppd lienallEAT. PeaVI Nature of Myle 532(m) 523(s) 532- is shoulder Et3NH+ 532(m) 522(s) 532 is shoulder Me2N14
MeNH3 532-(m) 520( ) doublet or singlet dependent on mull Pyle 520(s) Singlet Ph NH+ 3 53a(s) 520(e) 503(m) Triplet Ph2NHI 517(m) Singlet PhNH 3 520(e) Singlet p-TolNe3 532(m) 520(m) d + Ph3PH 532 (w) 521(m) 511(m) Triplet + K 533(s) 521(s) d + Na 525(s) 550(s) d NII4 525(s) 518(s) 525 is shoulder + MeL4N 532(m) 520(s) d + EtLX 532(m) 520(s) d n-Butyle 532(m) 520(s) - d = well defined ,doublet. -79- NH-X involving H-X bonds will depend on the strength of the acid HX. Complex fluoro and chioro acids are usually so strong that they have no separate existence in the absence of something to protonate. On this basis halides, where both covalent and ionic terms are favourable, should be hydrogen bonded, whereas tetraphenyl borates and fluorobor- ates where the terms are unfavourable should not. However, it is possible that a strongly polarising cation could cause the formation of a hydrogen bond. It is hard to imagine how, under any circumstances, tetraphenyl borates could be hydrogen bonded since free tetraphenyl boric acid decomposes (Wittig et al., 93) and sterically hydrogen bond formation seems unlikely. Where the spectra of the tetraphenyl- borates are simple, the position of the N-H stretching fre- quency is taken to be that of a non hydrogen bonded system. It is intended that the considerations suggested earlier of position, breadth, and intensity of the N-H Vs vibration should be used to diagnose hydrogen bonding.
Tertiary Ammoi#um Salts. The overall spectra of triethylamine and trimethylamine fluoroborates are similar to those reported earlier for tri- methylammonium iodide (Sheppard and Ebbsworthi 90). In the -ao- 000-2000om-1 region the spectra include bands due to C1i combination and overtone modes. In the two fluoroborates there can be no doubt that the strong bands observed at 3200cm-1 (trimethylammonium fluoroborate) and 3150cm-1) (triethylammonium fluoroborate) are the N-1 stretching vi- brations. On deuteration of trimethylammonium fluoroborate the 3150cm1 band splits and moves down to 2450, 2400, and 2375cm1. A similar splitting was observed in the spectrum of N-deuterotrimethylammonium iodide and was attributed to Fermi interaction between the N-D vibration and overtone and summation bands (Ebbsworth and Sheppard, 90). The Nil frequency in trimethylammonium tetraphenylborate occurs at 3130cm 1 whereas in the bromide and iodide it occurs at 2730cm1. The N-if deformation frequency occurs at about 1420cm1 and seems to be insensitive to hydrogen bonding since the position is similar in the iodide and bromide. In N-deuterotrimethylammonium fluoroborate the band has dis- appeared under the strong fluoroborate absorption at 1050cm-10 The spectrum of the fluoroborate ion in the two fluoroborates studied in the KBr region is similar to that observed for sodium and potassium fluoroborates. There is, therefore, no evidence for hydrogen bonding in trimethyl and triethyl ammonium fluoroborates. -dl- Primary and Secondary Alkyl and Aryl Ammonium, Salle. The infrared spectra over the range 4000-650cm 1 are given in the tables. The isolated ion should give two N-H frequencies near 3000cm 1. Previous work has placed these frequencies in the alkylammonium halides at 30d0cm 1 and 29d0cm-1 in methylammonium chloride (Waldron, 94) at 2965cm1 and 2745cm1 in dimethylammonium chloride (Sheppard and Ebbeworth, 90) and at even lower frequencies for other alkyl and aryl substituted ammonium salts, (Chenon and sand- orfy, 92 Brissette and Sandorfy, 95). By analogy with previous assignments the N-H vibrations in dimethylammonium fluoroborate occur at 3257 and 2790cm-1 (3140 and 2760cm in the tetraphenylborate). (Values in parenthesis refer to tetraphenylborates in this section). The bands at 3010 (3045), 2935(2995) and 2660cm-1 being C-H modes. In the spectrum of methylammonium fluoroborate only the 3275 (3175)cm-1 can be definitely assigned to an N-H vibration. The second N-H vibration may have become accidentally degen- erate with the first or, more probably, is under one of the C-H bands which occur at 2962 and 2d45 (3050 and 2990)0m-1. The spectra of all the salts considered in this section con- tain weak bands below 2800cm-1. These bands are not as strong or as broad as those observed in the spectra of -82- hydrogen-bonded salts, and it is very probable that they are combination and overtone frequencies such as are observed in this region of the spectra of many hydrocarbon derivatives. The spectra of aryl-substituted secondary and primary ammonium salts look more complex because of the presence of broad, moderately strong bands below 2800cm1. The spectra of the fluoroborates show a strong band above 3100cm 1 which we assign to one of the N-H vibrations. The other N-H frequency has not definitely been identified but probably occurs below 2900cm1 and is responsible for exciting com- bination modes and overtones to give moderately strong peaks in this region. The spectrum in the Ur region of the fluoroborate ion in all the compounds examined in this group A6 maw completely normal. Since all the salts have a strong -1 band above 3150cm we conclude that there is no evidence for hydrogen bonding in any of the primary or secondary ammonium fluoroborates.
WialelallingWalalL1122". It is possible that in the preparation of this compound a hydroxy fluoroborate is formed Ph3NHHOBF3 since the pre- paration involves the use of water to hydrolyse the boron trifluoride. Comparison of the spectrum of this compound with those prepared by Sharp by the passage of hydrogen -83- halide into a solution of metal halide and triphenylamine in benzene shows considerable similarity. All exhibit intense bands in the 2700cm1 region which are dependent to some extent on the anion. In the compound N-deuterotriphenyl ammonium fluoroborate, these bands have shifted from 2842,
2738, and 2655cm 1 to 2240, 2165, and 2065cm1 (VH/VD = 1.28, 1.26, and 1°27). The N-H deformation frequency moves under -1 the fluoroborate at 1050cm on deuteration. In the KBr region the fluoroborate band shows considerable splitting (532, 520, and 503cm-1) which may be interpreted as being due to the asymmetry of an 011)31 3 ion or to the effect of hydrogen bonding. Triphenylamine is a weak base and it is possible that the N-H vibration could occur in the 2700cm-1 region without invoking hydrogen bonding, however, the shift in frequency and the broadness of the absorptions which occur indicate that hydrogen bonding probably occurs and comparison of the spectrum of triphenylammonium fluoroborate with that of other triphenylammonium salts indicates that this is probably a hydrogen bonded fluoroborate rather than the salt of an asymmetric hydroxy fluoroborate. The ;nZrared Spectr pi of Pyri.dinj.pm Fluorobqrate. The infrared spectra of deuterated pyridines has been studied by Wilmshurst and Bernstein (96), and assignments have been made. In general the spectra of pyridine com- plexes are similar to that of pyridine and it has therefore been concluded that backbonding in metal complexes results in the stabilisation of electron density over the pyridine ring (Gill et al.9 97). The pyridinium ion is a special case of the pyridine complex in which all the ring substit- uents are of the same atomic weight instead of one substit- uent being much heavier as in the pyridine complexes. In pyridinium fluoroborate the ring C-H and N-H stretching fr4- quencies all occur at about 3000cm-1 and in this region evidence of hydrogen bonding may be sought: the ske3atal stretching frequencies and C-H out-of-plane vibrations occur in the 2000-650cm-1 region and comparison may be made with the spectra of pyridine complexes.
PYris4iUkUM-URQE2h=t1. Boron trifluoride was passed into pyridine in benzene and the boron trifluoride-pyridine adduct isolated. The adduct was hydrolysed by refluxing in 95% ethanol for two hours when the salt pyridinium fluoroborate was obtained. (Method due to van der Meulen and Heller, 98). -85- Analysis. c NH BF 5 6 4 Theory C 36°0A„ H 3.6%, Found C 3600A, H 3.6%. The infrared spectrum of this salt is extremely complex and hence N-deuteropyridinium fluoroborate was prepared in an attempt to find the N-H vibrations. Pyridinium fluoroborate is recrystallisable from water and hence N-deuteropyri„dinium fluoroborate was prepared by recrystallising a small sample (0.25g.) from deuterium oxide (1 ml.). Since much more vigorous conditions (i.e. reduct- ion of C-halogeno pyridines with zinc in D250k Bak and Rastrup-Anderson, 99) are required to replace C-H hydrogens it is unlikely that any bond, other than N-H was affected. The infrared spectra of the two salts are shown in the table following.
Hyd7ogsaBonding izaztjAinlyalliag e. The infrared spectrum of pyridinium fluoroborate shows three vibrations (3250(w), 3160(s)„ and 3120(s)cm-1) which shift on deuteration (2450(s)„ 2405(s)„ and 2325(e)cm-1 (VHAD = 3..32, 1.32, and 1°29). Unlike the spectra of pyridinium halides (due to Waddington, 100) which show strong bands in the 2500cm-1 region there are no frequencies in the hydrogen bonded N-H region. The tetraphenylborate -86-
S t d LL-: gigzsycarj.d inium n.uor9b9r4e (apC1)
V 9,511/41113E4 95115102-14
3250 (w) 3160(s) N-11 N='D 2450(6)9 2405(e) 3120(6) N-11 N-D 2325 (u)
3080 (m) p 2900(s) C-H 3080 (m) 9 2900 (s ) 2827 (w) C-H 2827(w) 1640(8) 1625(s) 1638(8) 9 1613(w) 1597 (w) 1590 (w) 1540 (m) 1510(w) 1525(m) 1492(w) 1477(in) 1460 (w) 1367(w) 1375 (w) 1327(w) 1335 (w) 1308(s) 1295 (w) 1292(w) 1250 (w) 125d (w)
BF 1075(8) 9 1025(a) 1075(0 p 1025(6) 11- 975(s) BP14. 850 (w) 800 (w) 750(m) CH 769(0 9 756 (in) 695 (m) 667 (m) has strong absorptions at 3230, 3178, and 3130cm1 and again no bands at 2600cm-1 which could be attributed to a hydrogen bonded N-H group. It must therefore be concluded that hyd- rogen bonding does not occur in pyridinium fluoroborate. The behaviour of hydrogen bonded pyridinium salts in showing N-H frequencies in two positions can be attributed to one of two causes, either that the N-H bonds occur in two environments or that there is a double-well effect in the potential energy-interatomic distance curves for the hydro- gen bond in these compounds as has been predicted by McKinney and Barrow (101). Bell and Barrow (102) observing the first overtone of the 0-H vibration of ethanol in var- ious solvents found a singlet when ethanol is mixed with carbon tetrachloride but observed a doublet when it was disoolved in solvents such as pyridine with which It could hydrogen bond.
PNXiiint ADA 2yricline CAMDlex.W. Absorptions assignable to C-H stretching frequencies occur at 3080(m) and 2900(8)cm-11 these vibrations are comparable to those at 3080, 3054, and 3036cm-1 observed for pyridine vapour (Wilmshurst and Bernstein, 96). The group -88Q of absorptions seems to have shifted down about 50cm-1 due to the positive charge at one end of the ring system which should withdraw charge density and hence weaken the 011 vibration. The band at 16400M 1 in pyridinium fluoroborute seems to be similar to that observed at 1624cm-1 for pyridine' boron trichloride by Greenwood and Wade (103) the same vibration apparently occurring about 1600cm-1 in pyridine metal complexes and at 1578cm 1 in pyridine. Unlike the 1640cm-1 which is unaffected by deuteration that at 1600cm 1 is split and reduced in intensity, a band appears at 1306cm-1 (VH/VD = 1°23) in the deuterated salt, the low ratio VH/VD indicating that this is probably a combination mode. The bands at 15H.0, 1295 and 1250cm-1 are all reduced in intensity on deuteration indicating that all are involved in some way with N-H deformation and combination modes. The intense band at 1490cm71 appears to be a skeletal vib- ration similar to that observed in pyridine complexes at 1100cm I enhanced in intensity by the charge on the ring. -1 Strong fluoroborate absorptions occur at 1075 and 1025cm . -1 0-H vibrations occur at 7I4 and 677cm 9 positions characteristic of an aromatic system. -89 Hydrogyn_Bondjpo in'TriphenylphooDhonium Saktp. Sheldon and Tyree (104) have reported the preparation of a series of triphenylphosphonium salts. Unlike the triarylammonium salts previously discussed these salts were air stable. All are sulte of complex haloione no simple triphenylphosphonium salts have been reported. Woodward (105) found that V for the P-H vibration in Pe occurred s 4 at 2372cm-19 and the spectra of the triphenylphosphonium compounds were accordingly examined in this region for evidence of hydrogen bonding. Attempts were also made to isolate triphenylphosphonium chloride and fluoroborate.
EXD er ment al . 3 Dilagai6 was prepared according to the method of Sheldon and Tyree (1014.) Stannic chloride and triphenylphosphine were dissolved in a minimum quantity of absolute alcohol. Hydrogen chlo- ride was passed through the solution. White microcrystals of the salt precipitated at once. 7.Zia1211.thaliaalEILAR1-21 11521=43.41 was prepared in a similar manner and obtained as yellow crys- tals.
-90-
T _ _ •e t 11 • • although reported could not be prepared. Triphenylphosphopiuk )3r onvint imonat e _CP113YHSbBr4). was prepared by adding a solution of antimony tribromide in hydrobromic acid to a solution of triphenyiphosphine in absolute alcohol. The salt crystallised at once. This method due to Sheldon and Tyree was used to prepare the previously unreported Triphemlnhowohonlmm Chlorobismuthate and TpiDhenylphosnliopillmLPr9/400tannOe. Triphenypphospingqp Perphloratg was obtained by adding perchloric acid to a solution of triphenylphosphine in hydrochloric acid. TriDhenylmhosphopipm Flyoroborate was prepared by a method analagous to that for triphenylammonium fluoroborate (Sharp, 17). Triphenylphosphine was dissolved in mcist benzene, and boron trifluoride passed into the solvtion. A white precipitate was obtained.
Analysis. Fh...PHBFL Theory C 60°0%, H 407A, 5 4 Found C 58-4%, H 500%. The salt could be recrystallised from a variety of solvents, however, the recrystallised product retained sol- vent and when pumped decomposed. -91- Attempts to prepare triphenylphosphonium sulphate by the addition of sulphuric acid to triphenylphosphine in alcohol were all unsuccessful® Triphenylphosphonium chlo- ride could not be prepared by passing hydrogen chloride into triphenylphosphine in a variety of solvents. Infrared spectra 4000-650cm1. Ph PHBF 3030(w) 3 4 9 2295(w)9 1572(w), 1470(s), 1300(w), 1282(m) 9 1185(m)v 1072(s)(fluoroborate), 1025(s)(fluorobor- ate), 910(m), $85(s)„ 770(m), 775(w), 695(w), 680(w). The reported spectra of triphenylphosphine salts are very similar to this spectrum. Sheldon and Tyree (104) claim that the bands at 1185, 910 and 865cm1 are character- istic of this type of salt. The spectra of the compounds isolated all showed broad 1 diffuse bands in the 2000cm region. Spectra were accord- ingly repeated using thick hexachiorobutadiene mulls under high resolution conditions at slow speeds. The results obtained are shown in the Table:- Compound Band Width(cm-1 ) Ma74ma(cm'4) - 2275 2300, 2385 (Ph3PH)2SnBr6 2400 (Ph3PH)2Sne16 2425 - 2250 2300, 2400 (The above two bands are of identical shape showing slight splitting®) -92- Compound Maxima(cm-1) Ph3PHSbBr4 2375 - 2250 2295 Ph3PHSbC14 2400-- 2250 2290 Fh3PHBiC14 2400 - 2250 2290 Ph3PH0104 2375 - 2250 2295 Ph PHBF 2375 ® 2250 2295. 3 4 URMatsu The spectra of the complex salts studied show no shifts within themselves, indeed, the spectra are remarkable for their similarity. Comparison of their position with the -1 for the P-H vibration in PH indicates a value of 2372cm 4 lowering of frequency© The breadth of the bands indicates that there is some element of anharmonicity which, as has been previously discussed for substituted ammonium salts, may be caused by anharmonic coupling of the P-H vibration with the hydrogen band. If this was the case then the spectra of triphenylphosphonium hexabromostannate and hexa- chlorostannate are explicable as having hydrogen atoms in two environments which give rise to two maxima, one normal at 2385 and2400cm1 and one hydrogen bonded at 2300cm-1. Some support is lent to this postulate by the shape of the bands which are sharp at the low wave numbers -93- side and extended at the other. In the absence of further evidences it seems possible that hydrogen bonding can occur in triphenylphosphonium salts but it is by no means proven. Reactions of Tungsten Hexachloride. Tungsten hexachloride reacts with organic bases. Cooper and Wardlaw (106) reported the formation of compounds' of odd stoichiometries (C5 H6 N)2. WOC15 H2 0 and (C9 H8 N)8. W3Cl17 when tungsten hexachloride in carbon tetrachloride reacted with pyridine and quinoline respectively. The reactions of tungsten hexachloride with a variety of organic ligands were examined.
Experimental. Tungsten hexachloride was prepared by passing chlorine over red hot tungsten in the apparatus shown (fig.1)0 the apparatus having been previously swept with a stream of nitrogen. The tungsten hexachloride was sublimed into the flask and transferred to the dry box and there divided into 0.5g. samples which were sealed off in glass tubes. In a series of experiments it was found that tungsten hexachloride was soluble in dry carbon tetrachloride and that the solution was stable under nitrogen. The reactions were carried out in the apparatus shown (fig.2). Nitrogen was passed continuously through the solution of tungsten hexachloride in the reaction vessel and the solution of base in carbon tetrachloride was added slowly from the dropping -95-
unst en T metal
Fig. 1
Tap funnel
Sintered glass filter Fig© 2
Suction -96- funnel. The following compounds were found to react with tung- sten hexachloridet- Triphenylamine„ triphenylphosphine„ triphenylarsine, tri- phenylbiemuth„ pyridine and acto dipyridyl. When the reaction was complete the nitrogen exit was shut and the carbon tetrachloride and any excess of reagent were blown through the filter and the yellow complexes which precipitated at once in all the reactions studied were re- tained on the filter and dried by continuing the passage of nitrogen. When the reaction vessel was transferred to the dry box and opened, all the products decomposed rapidly, turning blue within a few minutes® The products from the reactions with dipyridyl, tri- phenylamine and triphenylphosphine seemed to be the most stable to the atmosphere and accordingly infrared spectra and analyses were obtained. The infrared spectra showed bands characteristic of the ligands„ but no matter how quick- ly and carefully the samples for spectra were mulled, weak hydroxyl bands at 3000cm 1 were always present. The pyridine complex was also prepared on the vacuum line by distilling an excess of pyridine on to tungsten hexachloride. The appearance of the sample obtained was similar to that prepared by the first method. Again when the sample was opened in the dry box it decomposed. In this experiment in a closed system, no increase of pressure occurred during the reaction confirming that, as in other reactions, no gas evolution took place. Infrared spectra 4000-650cm 1:-
Ph3NWC16 2950(s), 2912(m)„ 1580(w), 1480(s), 1325(w), 1275(e), 1165(e), 1015(w) 9 790(s), 720(w).
C H N W01 10 8 2 6 3015(w), 2915(s), 1595(s), 1578(09 1530(09
1435(w), 1315(w), 1257(m) 9 1085 (w) 9 950(m), 910(w), 760(s) , 720(w). Reaction of tungsten hexachloride and tungsten hexa- carbonyl in carbon tetrachloride resulted in the formation of a small quantity of an unstable green solid. The infra- red spectrum showed three peaks in the carbonyl region at 2095(w), 2025(m) and 1923cm-1(w). Attempts to chromato- graph the green solid which was soluble in methylene chlo- ride, resulted in immediate decomposition. The reaction of molybdenum pentachloride and molybdenum hexacarbonyl in carbon tetrachloride gave a black solid which was very unstable, and infrared examination showed three very weak peaks in the carbonyl region. .98© In both these reactions the halide was in considerable excess and examination of the solution after reaction, showed that all the carbonyl had reacted. In an attempt to prepareanalogous iodides, iodine and the carbonyl were refluxed in a variety of solvents. ' In low boiling solvents such as 40-60° petroleum ether, no reaction occurred while in xylene the carbonyl decomposed to to the metal.
Ph3N.WC16 Theory C 33.5*, H 1L4*, Found C 29.2%, H
Ph3P.W016 Theory C 30.3, H 10109 Found C 30074, H 2.2A.
010H8N2.W016 Theory C 210'4, H Found C 21°1%0 H 1°9%0
Discussion. The absence of gas evolution in these reactions indic- ates that the complexes were formed by addition rather than by substitution, a result confirmed by the analyses obtained. A similar compound WC16.4NH3 was obtained by the passage of ammonia through a solution of tungsten hexachloride in carbon tetrachloride (Fowles and Osbourne 106a). -99- The odd co-ordination number of seven observed'in tri- phenylamine and triphenylphosphine tungsten hexachloride is unusual and is presumably due to steric hindrance since once one ligand molecule is attached there can be very little room for further ligand molecules to approach the central metal atom. That eight co-ordination is possible is indicated by the formation of C10H8N2.WC16 with a biden- tate ligand dipyridyl. Triphenylamine tungsten hexachlo- ride appears to be the first reported triphenylamine metal complex. The infrared spectrum of the complex is similar to that of triphenylamine with no additional N-H frequencies, indicating that this is a true complex and not a triphenyl- ammonium salt. The observation of three carbonyl frequencies in the reaction product of tungsten hexachloride and hexacarbonyl indicates that the product, obviously a mixed carbonyl chloride, probably contains at least three carbonyl groups per molecule. All the complexes formed were unstable with respect to the 5-6 molybdenum blue valency state, whereas tungsten hexachloride is unstable with respect to the hydrated tri- oxide. -100-
The Ultraviolet Spectra of Aromatic Compounds in Fluoro- sulphuric Acid. The physical properties of fluorosulphuric acid have been carefully examined by Woolf (10?) who also examined the properties of metal fluorides in solution in this acid. He concluded that fluorosulphuric acid is a true compound and not a constant boiling mixture and that in the anhyerous acid the predominant ionisation is:- HS 03F = 1-14. + HSO3P- rather than the two alterantives previously proposed i.e. + P -A HSO + P- HS03 3 -1 or 2H6 03F H21 + 206P On electrolysis of the anhydrous acid the only gas released is hydrogen at the cathode. In view of this type of ionisation it is possible that fluorosulphuric could behave in a similar manner to anhydrous mixtures of hydrogen fluoride and boron trifluoride which are capable of proton- ating aromatic compounds (McCauley and Lien 3). Aromatic +HF +BP -A Aromatic 11+ +BF-
The ultraviolet spectra at low temperatures of species of this type have been observed by Reid (108) who found that the formation of a carbonium ion in HP BP3 mixtures is associated with absorptions in the visible region and that the ultraviolet - 101- spectra show quite characteristic absorptions although the individual spectra show little relationship to the spectrum of the parent aromatic compound. Sulphuric acid on the other hand, generally appears not to dissolve aromatic compounds (Sidgwick 15) and also seems more likely to oxidise the solute 6 Gold and Tye (109) have examined the spectra of a limited number of hydrocarbons which are soluble in sulphuric acid and find that the spectra of compounds such as 1:1 diphenylethanol may be interpreted as those of protonated species. However, while triphenyl amine dissolves in anhydrous HF reversibly (Kemmitt et al.17) solution in sulphuric acid appears to be accompanied by the formation of blue oxidised products® The spectra of a number of aromatic compounds in fluorosulphuric acid have therefore been obtained and comparison will be made with the behaviour of similar solutions in HF Bl!mixtures and in sulphuric acid.
Experimental. Fluorosulphuric acid was obtained from 'Consolidated Zinc Ltd.' It was distilled at atmospheric pressure in all glass apparatus before use. Distilled samples were stored in flasks fitted with calcium chloride guard tubes. It was found that ground glass joints were attacked slowly by fluorosulphuric acid vapour and they were protected with -102- teflon sleeves or with perfluoro grease. Triphenylcarbinol, triphenylamine, diphenylamine„ aniline, pyridine, nitrobenzene, benzene, naphthalene phenanthrene and anthracene were obtained commercially and used without further purification. A sample of tri-p- tolylamine was provided by Dr. D. W. A. Sharp. The ultraviolet spectra of the solutions in fluoro- sulphuric acid were obtained in stoppered 1 cm. silica glass cells. At the end of the series of experiments the cells showed only very faint signs of etching. The spectra were obtained on an Optica Ultraviolet Spectrometer. Examination of the ultraviolet spectrum of fluoro- sulphuric acid itself showed that absorption started at 240 mu while at 210 mu absorption was 100%.
Triphenylcarbinol. Triphenylcarbinol dissolved smoothly in fluorosviphuric acid with no apparent reaction to give an intensely yellow solution. There was no apparent change in the colour of the solution over a period of three days. Examination of the ultraviolet spectrum over this period showed that it too was unchanged. The ultraviolet spectrum consists of a doublet occurring at 405 and 432 mu. A solution in sulphuric acid absorbed likewise at 405 and 432 mu. -103-
Triphenylamine. Triphenylamine dissolved ih fluorostilphuric acid to give a pale brown solution. After about ten minutes the solution became pale blue and over a period of about an hour the solution became intensely blue. A weighed quantity Of triphenylamine (1.0 g.) was dissolved in fluorosulphuric acid (10 ml®) and immediately diluted by pouring on to cracked ice (50 g..). A white precipitate was obtained which was recrystallised from ether and identified as triphenylamine by mixed melting point and infrared spectrum. On repeating the experiment it was found that the actual recovery of triphenylamine was 87%. Triphenylamine dissolved in sulphuric acid to give an intensely blue solution at once. The ultraviolet spectrum of this solution showed a peak at 710 mu. The spectrum of the newly prepared solution of triphenylamine in fluorosulphuric acid absorbed at 340 mu whereas the spectrum of two hours old solution showed
peaks at 340 and 630 mu. Triphenylamine in ether absorbs at 297 mu.
Diphenylamine. Diphenylamine dissolved in fluorosulphuric acid to -104- give a colourless solution. Over a period of three or four days the solution slowly turned pale blue. The ultraviolet spectrum in fluorosulphuric acid of the freshly prepared solution consisted of triplet absorbing at 258, 273, and 280 mu. In ether there was a single broad band at 280 mu.
Aniline. The amine reacted vigorously with fluorosulphuric acid to give a pale brown solution which seemed to be stable indefinitely. The ultraviolet spectrum showed peaks at 242, 256, 260 and 264 mu. Aniline in ether absorbs at 240 mu.
Tri-p-tolylamine. Tri-p-tolylamine dissolved in fluorosulphuric acid to give a solution which became dark blue at once. When the solution was diluted on cracked ice, a brown intractable solid was obtained. The ultraviolet spectrum in fluorosulphuric acid showed peaks at 340 and 787 mu. Tri-p-tolylamine dissolved in sulphuric acid to give a similar blue solution which absorbed at 340 and 790 mu, The solution in ether absorbed at 305 mu. Since an intense blue colour is characteristic of -105-
oxidised tri-p-tolylamine entities the spectrum of the tri-p-tolylaminium ion was examined. Tri-p-tolylamine was oxidised by the method used by Sharp (110). To a solution of tri-p-tolylamine in ether was added first a solution of iodine in ether and secondly, a solution of silver perchiorate in ether. A mixture of silver iodide and tri-p-tolylaminium perchlorate precipitated at once. The precipitate was filtered under nitrogen and extracted with methylene chloride. The ultraviolet spectrum of a portion of this solution was obtained while the remainder was evaporated to dryness and then taken up in fluoro- sulphuric acid and the ultraviolet spectrum again obtained. (p-CH3C04)3N+010; in methylene chloride absorbed at 304, 345, 373 and 687 mu. and in fluorosulphuric acid at 323 and 800 mu.
Benzene. Pluorosulphuric acid can suiphonate aromatic hydro- carbons. Benzene reacts to give G6H5S0.5 (Woolf, 107). It was found that benzene reacted vigorously with anhydrous fluorosulphuric acid. A yellow solution was obtained which on dilution gave a white solid. The spectrum of the solution showed peaks at 300, 340 and 445 mu. -106-
Nitrobenzene. Nitrobenzene mixed with fluorosulphuric acid to give an intensely yellow solution. After 'two hours the solution was diluted and then back extracted first with ether and then with methylene chloride. When the ether extract was dried with sodium sulphate and then evaporated to dryness a pale yellow liquid remained. It was identified as nitro- benzene by its infrared spectrum and was obtained in about 85% yield. The methylene chloride extract gave a small quantity of a white solid which could not be identified but was presumably a sulphonated nitrobenzene. Nitrobenzene in fluorosulphuric acid absorbed at 350 mu.
Naphthalene. Naphthalene dissolved smoothly in benzene to give a brown solution. This solution showed no signs of change over a period of days. On dilution it gave a dirty brown solid, presumably a mixture of sulphonated products. Naphthalene in fluorosulphuric acid absorbed at 3000 310 and 325 mu.
Anthracene. Anthracene dissolved in fluorosulphuric acid to give a dark red solution which was stable for up to half an hour, after which time it had darkened perceptibly. Inspection -107- of the ultraviolet spectrum over this time showed that peaks at 310, 410 and 590 mu slowly diminished in intensity.
Phenanthrene. Phenanthrene dissolved in fluorosulphuric acid to give a dark red unstable solution. The ultraviolet spectrum was obtained as quickly as possible and peaks were observed at 320 , 415 and 585 mu.
Pyridine. Pyridine dissolved in fluorosulphuric acid to .ve a stable colourless solution. Absorption of the dissolved species was under the solvent absorption.
Discussion. The chemistry of fluorosulphuric acid has been reviewed by Lange (111). In general the reaction of fluorosulphuric acid with an aromatic compound results in three types of product, sulphones, suiphonic acids and sulphonyl fluorides. At high temperatures and with a large excess of fluoro- sulphuric acid the sulphonyl fluoride is usually obtained (RS02P) whereas at low temperatures and with smaller quantities of the acid the sulphone is obtained. At higher temperatures the disuiphonyl fluorides may be obtained. The sulphonyl fluorides thus obtained are not hydrolysed -108- by boiling water. Fluorosulphuric acid reacts with many aliphatic compounds ® with ether the fluorosulphate ester is obtained, (0 H ) 2 5 20 + 2HS03F = 2C2H50S02F + H2O while with acetone a coloured product of unknown nature is obtained. The ester may also be obtained by reaction of ethanol with fluorosulphuric acid. These esters are much more reactive than the aromatic suiphonyl fluorides being hydrolysed by water. In order that the behaviour of fluorosulphuric acid and organic compounds could be examined it would have been valuable to be able to dilute the fluorosulphuric acid in a non reactive medium. It was found, however, that fluoro- sulphuric acid was miscible with and reacted with organic solvents or was immiscible in which case reaction occurred at the interface. With solvents such as methylene chloride, chloroform and carbon tetrachloride reaction took place slowly. It therefore proved impossible to study ultra- violet spectra in anything other than anhydrous solution. The oxidising properties of fluorosulphuric acid have not been so widely reported as its catalytic or fluoro- sulphating properties. It is reported however to oxidise oxythionaphthene to thioindigo in a patented industrial -109-
process (Hoffa and Luce, 112). The fluorosulphuric acid is first used to cause ring closure by elimination of water and then to cause ring fusion by elimination of four hydrogen atoms, HS 03F \ 2\1CH2 /0
Triphenylcarbinol. The intense yellow colour of the solution of triphenyl carbinol in fluorosulphuric acid suggest that the formation of the triphenyl methyl cation has occurred. + Ph3COH + HS03P "'""") Ph3C + SO3 P- + H2O A similar reaction takes place when triphenylcarbinol is dissolved in concentrated sulphuric acid (Sharp 23). When triphenylchloromethane reacts with silver per- chlorate in dimethylsulphate a yellow solution is obtained. This solution is of the ionic salt triphenylmethyl perchlor- ate (Anderson, 113). The ultraviolet spectrum of triphenyl- methylsulphate dissolved in sulphuric acid shows a doublet at 408 and 430 mu. The spectrum of triphenylcarbinol in fluorosulphuric acid is a doublet at 405 and 432 mu. There can therefore be little doubt that the species present in solution is the triphenylmethyl cation. Triphenylmethyl fluorosuiphate has been prepared by Sharp (114). The
-110-
infrared spectrum of the fluorosuiphate ion in this compound is similar to that in sodium fluorosulphate indicating the ionic nature of the compound.
Trinhenxlamine The blue colour which slowly develops when triphenyl- amine is dissolved in fluorosuiphuric acid does not appear to be similar to that shown by triphenylamine in sulphuric acid since a peak at 630 mu appears in the former while one at 710 mu occurs in the latter. The blue colour which develops in sulphuric acid has been shown to be due to an entity similar to Wursters Salt
p hN Neph) since when the solution is
diluted tetraphenylbenzidine (Ph2N06114C6H5NPh2) is obtained (Sidgwick, 15). Oxidative attack appears therefore to occur on the p.hydrogen position. The position of the absorption of triphenylamine in fluorosuiphuric acid is closer to that of the absorption observed when triphenyl amine is irradiated in rigid solvent when a peak is observed at 656 mu (Lewis and Lipkin, 115) which suggests that possibly in fluorosulphuric acid the aminium ion is formed (Ph3e). The initially colourless solution suggests -111- that the triphenylamine could simply dissolve in fluoro- sulphuric acid, however there is a shift of about 40 mu from a solution in ether to the acid solution. This difference may be due to protonation aftSr which the proton could slowly be removed by oxidation, Ph3N + a+ --:>Ph3NH+ -7> Ph3N+
Whatever the mechanism of this reaction it is quite clear that triphenylamine is much less rapidly oxidised by fluorosulphuric acid than sulphuric acid.
Diphenylamine. The ultraviolet spectrum of diphenylamine in fluoro- sulphuric acid shows the Waterman effect very strongly in a manner characteristic of a protonated aromatic amine (Ross and Tonnet,116) and there can be no doubt that the spectrum is that of the diphenylammonium ion.
Aniline. The ultraviolet spectrum of aniline in fluoro- sulphuric acid is closely similar to that reported for aniline in hydrochloric acid reported by Ross and Tonnet (116) which shows four peaks in similar positions. There can be no doubt therefore that in anhydrous fluorosulphuric acid aniline exists as the anilinium ion. -112-
Tri -13-tolylamine. The ultraviolet spectra of tri-p-tolyamine in fluorosulphuric and sulphuric acids are identical, both having peaks at 340 and 790 mu. This similarity is not surprising since the paraposition in the amine is blocked to oxidative attack and hence a reaction similar to that which occurs in triphenylamine cannot take place. The ultraviolet spectrum of the tri-p-tolylaminium ion is reported from two sources, the first by the chemical oxidation with lead tetracetate of t ri -p -tolylamine in acetic acid and the second by irradiation of the amine in a rigid medium (Lewis and• Lipkin, 115). Both agree that a band characteristic of the ion occurs at 690 mu. Tri-p- tolylaminium perchlorate in methylene chloride absorbs at 687 mu while when dissokved in fluorosulphuric acid the band shifts to 800 mu. - In view of the non-recovery of tri-p- tolylanine on diluting the acid solution and the considerable shift of position of the absorption band from a position characteristic of the aminium ion it appears likely that the solution of tri-p-tolylanine is accompanied by the formation of the aminium ion but that further more complex reactions also ensue. -113-
Benzene. Anhydrous fluorosulphuric acid and benzene react rapidly to give the sulphonated product. The yellow solution which this compound forms in excess of the acid is presumably a protonated species. The spectrum of benzene in. HP/BP3 has a peak at 420 mu which has been attributed to the CO; absorption (Reid, 108). In the case of the substituted aromatic an absorption occurs at 445 mu and it therefore appears probable that in this case a protonated ion has been formed.
Nitrobengene. In the case of nitrobenzene the aromatic system seems to be sufficiently deactivated to prevent further substitution. It is not at all clear whether protonation occurs since the ultraviolet spectrum consists of a single broad band at 350 mu..
Naphth,lepz. Naphthalene is presumably immediately fluorosulphated in anhydrous fluorosulphuric acid and although the spectrum of the solution was obtained as quickly as possible a similar- ity to the spectrum of naphthalene in HP/B103 could not be detected. -114-
Rhepagthrenp. The ultraviolet spectra obtained from this compound consisted of weak absorptions against a heavy background absorption. The peaks corresponded approximately to the position of those observed in HP/3173 3109 400 and 590 mu in the former and 400 and 525 mu in the latter. The solution in fluorosulphuric acid is probably accompanied by protonation but also presumably by sulphonation and probably oxidative breakdown.
Anthrgoene. The solution of anthracene in fluorosulphuric acid is somewhat more stable than that of phenanthrene and better correspondence is obtained with the spectrum of anthracene .1- observed in HVBF (in H80 320,415 and 05 mu, in 1111/12 H 3 3 3 300 and 410 mu) (Reid, 108). The spectrum of anthracene in sulphuric acid has been observed by Gold and Tye (109). The spectrum they obtained is similar to that in HP/13F3.
It is quite clear that aromatic compounds as a whole are much more soluble in fluorosulphuric acid than in sulphuric acid. While solution is accompanied by protonation, in some casesthe acid is also reactive enough to give suiphonated products at room temperature. There is also some -115- evidence to suggest that oxidation may occur, particularly in large complex molecules or where there is an atom other than carbon. On the whole it may be concluded that fluoro- sulphonic acid is much too reactive to be a good solvent for protonation studies. -116=
Solutions of Metal Carbonyls in Flpor9sylp)auLi. Acj.d. The behaviour of organic compounds in strong acid leads to the expectation that other inorganic compounds might be protonated similarly. At low temperatures the compound CH3C64BP; has been shown to exist and it appears possible that similar protonated inorganic species could be isoixted. Fluorosulphuric acid has been shown in the previous section to be capable of protonating organic compounds although it has the drawback of being a strong sulphonating and oxidising agent. In the case of metal carbonyls neither of these drawbacks will be as great as for organic compounds and it therefore seemed reasonable to examine the reactions and spectra of carbonyls in fluor. sulphuric acid.
Axperimentak. Manganese carbonyl, tungsten carbonyl and molybdenum carbonyl were obtained commercially; cyclopentadienyl iron dicarbonyl dieter and cyclopentadienyl iron dicarbonyl iodide were gifts from Dr. K. Joshi. Nickel dicarbonyl diphosphine was prepared according to the method of Chatt (117) The most precise way of studying protonated carbonyls would be by Nuclear Magnetic Resonance, however, at this time access to a reliable machine was not available and -1170 the only studies that could be made were of the in,:rared spectra. It was found that a thin film of anhydrous fluor° - sulphuric acid pressed between silver chloride plates showed a. background absorption of 55% from 2500 to 1500 cm. This is much more satisfactory than sulphuric acid which under similar conditions showed an absorption of 65% through this region.
Nickel Dicarbgnyl DI.Dhosphine. The complex dissolved in fluorosuiphuric acid to give a red solution which slowly turned black over a period of twenty minutes. At this stage the solution was diluted into water and a brown solid remained. It proved impossible to identify the solid which showed no peaks in the infrared in the carbonyl region. In a variety of experiments it was attempted to isolate the compound causing the red colour since this could have been due to the previously unisolated nickel diphosphine difluoride. The infrared spectrum of the freshly prepared solution showed peaks at 2030(m) and 2075(w) cm. while the spectrum of the solid nickel dicarbonyl diphosphine showed peaks at 1915(w), 2003(s) and 2100(w) cm° -1 118-
Tungsten and Molybdenum Hexacarbonyls. Both carbonyls dissolved in anhydrous fluorosuiphuric acid to give stable yellow solutions. On dilution into cracked ice the carbonyl was re-obtained. Mo(O0)6 in ISO? 2000(m), 2075(e) and 2155(w) cm. -1 W(00)6 in HSO3F '1985(w), 2040(s) and 2135(w) cm.
Iron Cyclopentadiepyl Dicarbonyl Dj.mer. The dimer shows a bridging carbonyl frequency in the solid state; the infrared spectrum of the freshly prepared acid solution showed that this had disappeared. If the solution was diluted at once a brown solid was obtained which the infrared spectrum showed to be a very impure specimen of the dimer. The solution allowed to stand for as short a period as two minutes when diluted gave a black solid which could not be identified. (OpFe(00)2)2 solid 1790(s), 1985(m) and 1998(s) cm. ®% (CpFe(C0)2)2 in HBO? 2103(s) and 2150(m) cm.-1 .
Iron Oyclopentadiegyl DLcarborLyl Iodide. The complex dissolved in fluorosulphuric acid with the evolution of iodine. The solution was diluted at once and the spectrum of the brown solid obtained. No evidence for bridging carbonyls was found. 00 frequencies were found at 1980(s), 2005(s) and 2035(s). -119-
Manganese IlentacarlIonyl Diner. The carbonyl dissolved in cooled fluorosulphuric acid to give an extremely atable yellow solution. On dilution the original carbonyl was re-obtained. Mn (CO) 2 10 in HBO3 2045(m) and 2080(s) cm. The addition of sufficient ammonium fluorosilicate to saturate the fluorosulphuric acid solution of manganepe carbonyl resulted in the precipitation of a small quantity of a yellow solid. In a typical reaction manganese carbonyl (i.0 g.) was dissolved in fluorosulphuric acid (5 ml.) with external cooling and after about half an hour all the manganese carbonyl had dissolved. Ammonium fluoro- silicate was added slowly to the solution until it was in slight excess. The mixture was allowed to stand overnight at the end of which period a small quantity of yellow solid had precipitated. The supernatant liquid was removed and diethyl ether (20 ml.) added to the solid. A yellow solution was obtained which was filtered and evaporated to 10 ml. under vacuum. When the solution was cooled, fine yellow crystals appeared which were filtered off and recrystallised from methylene chloride and petroleum ether. Yield 0,25 g. of yellow crystals of Manganese Pentacarbonyl Fluorosulphate. The compound is moderately -120- stable in the atmosphere and soluble in common organic solvents from which it recrystallises with some decomposition. It is also soluble in hot water from which solution a reineckate may be precipitated. It decomposed to give manganese carbonyl at 500 under vacuum. The addition of triphenylphosphine to a solution of manganese carbonyl fluorosulphate resulted in the precipita- tion of a small quantity of a pale yellow solid. No satisfactory analysis could be obtained.
Infrared Spectra. Mn(00) so F shows peaks in the carbonyl region at 5 3 ®a 2035(s) and 2080(a) cm. (mull) and at 2037(s), 2085(s) cm. (methylene chloride solution). Cr(SCN) (NH ) shows peaks at 1960(m), Mn(C0)5 4 0 2 2035(s) and 2108(w) cm.-- (mull). Mn(00)5803P/Ph3P shows peaks at 1945(m), 2000(s) and 2085(w) cm. _a (mull). The spectrum of the fluorosulphate ion in manganese carbonyl fluorosulphate may be compared with those of fluorosulphates prepared by Sharp (114). -121-
Napo g AgS0 3 31 740(m) 767(s) 760(s) 785(s) 785(s) 975(w) 980 (w) 1095(s) 1057(s) 1062(s) 1275(s) 1235(s) 1235(8) 1295(e) 1282(s) 1355(w) 1631(s) 1642(w) 2370(w) 2273(w)
Analysis.
Mn(CO) 80 F Theory C 20.14 SO4 33.0% 5 3 Found C 20.1591 804 33.2A Mn(C0)50r(SON)4(NH3)2 Theory C 21.2A H 1.1A Found C 23.04% H 1024%
Carbon and hydrogen were obtained by microanalysis and sulphate was found by precipitation of the hydrolysed fluorosuiphate from aqueous solution with barium chloride.
Discussion. The infrared spectra of carbonyls in fluorosuiphuric acid solution seem to be characterised by the shift of the carbonyl frequencies to higher values and an increase -122 in complexity of the spectrum. The attachment of an electron withdrawing entity to a metal carbonyl usually results in a shift of the carbonyl frequency to higher wave numbers. In Pt(C0)28 the carbonyl frequencies occur at 2160 and 2120 cm. g (Sharp 118). The effect of protona- tion would also be to lower the symmetry of the carbonyl in solution. The carbonyls seem to exist therefore as simple protonated species in solution although it would appear that the metal-metal bond in manganese carbonyl is broken possibly according to a reaction of the type. 3F Mn(C0) 30 F + Mn(C0) 11 Mn2(M)10 HSO 5 3 5 The solution of iron cyclopentadienyl dicarbonyl dimer is much less stable and is probably oxidised with the breaking of the metal metal bond. Manganese carbonyl fluoroeulphate is clearly anal- gnus to Mn(C0) c1 prepared by Abel and Wilkinson (119) and 5 Mn(C0)51103 prepared by Addison (120). Comparison of the carbonyl stretching frequencies of these three compounds dhow considerable similarity, 1 Mn(00)..01 2070 and 2016 cm. Mn(CO)5NO 2060 and 2002 cm. 7 3 and Mn(CO) so3 at 2085 and 2037 cm. The infrared spectrum of the fluorosulphate group shows one marked difference from that in ionic compounds, the asymetrical 5-0 stretching frequency which normally occurs -123-
a at 1280-1300 cm._ is shifted to 1355 cm. indicating the possibility of covalent bonding between anion and cation which has been shown to exist in the nitrate. Fluorosulphuric acid appears to be a good solvent for protonation studies being cagier to handle than HPAT and less liable to oxidise the solute than is 3 sulphuric acid. The suitability of this solvent for N.M.R. studies remains to be explored. -1214.-
ADDendix. Note Infrared pata. The following infrared spectrometers were used to obtain the spectra presented in this thesist.- Perkin Elmer Model 21, with CaP2, NaCl or Ur optics, Perkin Elmer Infracord, Grubb Parsons S4, and Grubb Parsons D11 2 (CsI optics). The spectra were obtained as mulls or as solutions. The KBr disc technique was not used since it frequently leads to anomalous results. Nujol and hexa- chlorobutadiene were used as mulling agents, spectra were obtained using plates to suit the particular range studied. Spectra in acid or aqueous solution were obtained as films pressed between silver chloride discs obtained from Research and Industrial Instrument Corporation. -125- REFERENCES.. 1)GAY LUSSAC and THENARD„ ANN. CHID. PHYS., k2, 204, 1809. 2)WANSER„ J. AMER. CHEM. SOC., 2a, 12099 1948. 3)MACCAULEY and LIEN, J. AMER. CHEM, SOC., 21 2013, 1951. 4)BOOTH and MARTIN, CHEM. REVS., 330 579 1943. See also BORON TRIFIUORIDE„ PUB. CHAPMAN AND HALL, 1949. 5)BROWN, STEHM: and TIERNEY, J. AMER. CHEM. SOC. 9, 2021, 1957. 6)SHARPE, FLUORINE CHEMISTRY, ED. SIMON, PUB. ACADEMIC PRESS, VOL. 2, 1950. 7)MUETTERTIES and PHILLIPS, J. AMER. CHEM. SOC., 2a, 36869 1957. MUNITITIES and COFFMANN„ J. AMER. CHEM. SOC., Dap 5914, 1958. 8)KEMMITT and SHARP, Private communication. 9)SHARP, ADVANCES IN FLUORINE CHEMISTRY, ED. STACEY et al. PUB. BUTTERWORTHS, VOL. 1, 97, 1960. 10)SCHUMACHER, PERCHLORATES, A.C.S. MONOGRAPH, PUB. REINHOLD 1960. 11)LEE and CARPENTER, J. PHYS. CHEM., La, 279, 1959.
12)SIMON and WEIST, Z. ANORG. u. ALLGEM. CHEM., g.119 19 1938. 13)HATHAWAY, J., 3091, 1961.
14)THORPE and KIRMAN, Jog 121.9 921, 1892. 126
15)SIDGWICK, ausacAL ELEMENTS AND THEIR COMPOUNDS, PUB. OXFORD UNIVERSITY PRESS, 1950. 16)AHRLAND, CHATT and DAVIES, QUART. REVS., 12, 265, 1956. 17)KEMMITT, NUTTALL and SHARP, J.a 46p 1960© SHARP, CHEM. AND IND., 1235, 1958. 18)BECKE-GOEHRING and THIELEMANN, L. ANORG. CHEM. , _up 308, 1961.
19)SHARP and SHARPE, Jog 1855, 1853, 195do
20)RUSSELL and SHARP, Jog 46d9, 1961.
21)SHARPE, Jo, 4538, 1952. 22)MOSS and SHARP, UNPUBLISHED WORK. 23)SHARP, PH.. D. THESIS, CAMBRIDGE UNIVERSITY, 1957.
24)COTTON and GOuDGAME, Jog 5267, 1960, 25)SERULLAS, ANN. CHIM. PHYS., 144 307, 1831. 26)HILL, J. AMER. CHEM. SOC., 4,1, 256, 1921. 27)HILL, J. AMER. CHEM. SOC., !Alio 1235, 1922. 28)MULLIKEN, J. AMER. CHEM. S00., 216 tali, 1952. 29)RUNDLE and GORING, J: AMER. CHEM. SOO., ig, 5337, 1950. 30)SMITH and RUNDLE, J. AMER. CHEM. SOC., lax 5077, 1958. 31)TAUPPEN, MURRAY and CLEVELAND, J. AMER. CHEM. SOC.,'Llat 3503, 1941. 32)TILDSLY and SHARPE, RESEARCH ColtRESTuNDENCE, h, 515, 1953.
33)BERGANN and SCHUTZ, Z. PHYSIC. CHEM., halt 401, 1932. -127- 34)KLAGES and LONGPAPE„ Z. ELECTROCHEM,;„ 0, 533, 19590 35)JAFFE, J. CHEM., aa„ 1430, 1954. 36)SASAKI and KUBO„ J. CHEM. PHYS., 31.0 477, 1959. 37)SHARP and SHEPPARD, J., 674, 1957. 38)WHETSEL, Z. KRIST.„ jaho 305, 1952, 39)KROSS, PASSEL and MARGOSHES, J. AMER. CHEM. SOC., 2,, 1332, 1956: RAO, RAMCHANDRAN and BALASUBRAMANIAN0 CANAD. J. CHEM., 32, 171, 1961. 40)BJERRUA„ BALI/BUN and WRIGHT, ADVANCES IN THE CHEMISTRY OF CO-ORDINATION COMPOUNDS, KIRSHNER, ED., MACMILLAN, NEW YORK, 1961. 41)DEWIER and FERNELIUS, Z. ANORG. CHEM., gap 83, 1934. 42)KHARASCH and ASHFORD, J. AMER. CHEM. S000, 51, 1736, 1936. 43)BROWN and DUNITZ, ACTA. CRYST.„ 11 28, 1960. 44)ARONEY, LE FEVRE and WERNER, Jos, 276, 1955. 45)WHETSEL, HAWKINS and JOHNSON, J. AMER. CHEM. SO00, 3360, 1956. 46)SHARP, ADVANCES IN FLUORINE CHEMISTRY, 14 68, 1960. 47)NESMEYANOV, MAKAROVA and TOLSTAYA, TETRAHEDRON, 14 1459 1957. 48)KAZITSYNA, REUTOY and BUCHKOVSKII, RUSS, J. CHEM, PH/B. (ENGLISH TRANSLATION), 3 404, 1960. ==128 49)SAUNDERS 9 THE AROMATIC DIAZO COMPOUNDS (2nd. Ed.), PUB. ARNOLD, LONDON, 1949. 50)PALMER, EXPERIMENTAL INORGANIC CHEMISTRY, PUB. CAM- BRIDGE UNIVERSITY PRESS, 1954. 51)DORING and DE PUY, J. AMER. CHEM. SOC., 258 5955, 1953. 52)GREISS, ANN., 12k, 123, 1859. 53)MENE, COMPT: REND., Sg, 311, 1861. 54)GREISS, ANN., lug 207, 1860. 55)MICHAELIS and HURL, ANN., gig„ 114, 1692. 56)HANTSCH, ANN., 3g1s, 11534. 1897. HER., ga, 1743, 1895 57)NESMEJAN0V and KAHN, BER0, 1019, 1929. 58)ZOLLINGER, AZO AND DIAZO CHEMISTRY, INTERSCIENCE PUB- LISHERS, 1961. 59)CHATTAWAY, Ja g 1981, 1924. 60)BART, 'WITCHES REICH PATENT, 281,055. 61)GREISS, BER., 1620, 1874. 62)KEKULE, LEHHEBUCH DER ORGANISCHEN CHEMIE, PUB., ERLANGEN, 1866. 63)BLOMSTRAND, HER., ., 51, 1875. 64)DILTHEY, BLANKENBURG, BRANDY and HUTHWELKER, J.,PRAKT. CHEM., 135, 36, 1932. 65)CLUSIUS and HOCH, HELVO CHIM. ACTA., 314 2122, 1950. 66)BELLAMY, THE INFRARED SPECTRA OF COMPLEX MOLECULES, TUB. -129- METHUEN, 1960. 67) KAZITSYNA, REUTOV and BUCHKOVSKII, RUSS. J. PHYS. CHEM.., (ENGLISH TRANSLATION), 316 404, 1960. 6d) GOLDSCMIDT, BER., zap 32209 1890. 6840 INTERNATIONAL CRITICAL TABLES, 12, 83, 1929. 69)pusoN 9 JOSIEN and SHOLTON, J. AMER. CHEM. HOC., 76, 2526, 1954, 70)SOUIDT and MAIER, J. PRAKT. CHEM., (ii), 13g„ 154, 1931. 71)ANDERSON and STEEDLY, J. AaER. CHEM. SOC., 21.9 51114 1954- 72)YATES and ARDAO, J. AMER. CHEM. S6C.- 071, 650, 1956. 73)DARLOW, HYDROGEN BONDING, ED. HADZI, PUB. PERGAMON PRESS, LONDON, 1959. 74)M.)0RE and WINMILL„ J., al, 16359 1912. 75)LATIMER and RODEBUSH, J. AMER. CHEM. SOC., kg, 1431, 1920. 76)PIMENTAL and HIGGINS, J. PHYS. CHEM., W., 1615, 1956. MECKE, DISC. FAR. SOC., 94 161, 19509 BADGER and BAUER, J. CHEM. PHYS., 5:9 8399 1937. 77)NAHAMATO, MARGOSHES and RUNDLE, J. AMER, CHEM. 800., 220, 6840, 1955. ?a) BADGER, J. CHEM. PHYS., 36 710, 1935. 79) BRATOZ and HADZI, J. CHEF. PHYS., az, 991, 1957. See -130- also P. 90 of HYDROGEN BONDING, Ref. 73. 80)LIPINCOT and SCHROEDER, J. CHEM. PHYS., ga„ 1099, 1955. 81)BATUEV, IZVEST. AKAD. U.S.S.R., ser. fiz., 11, 336, 1947. 82)STEPANOV, NATURE, 1520 808, 1946. 83)NEWMAN and BADGER, J. CHEM. PHYS., 9, 1207, 1951. 84)SIGERU NAGAHARAr „REPORT UNIVERSITY OF TOKYO, DEPT. Of PHYSICS, , 17, 1956. 85)PIMENTAL and MACLELWt.„THE HYDROGEN BOND, PUB. FREEMAN, 118, 1960 86)WADDINGTON, J., 4340, 1958. WAGNER and HORNIG, J. CHEM: 87)PHYS., 1A, 296, 1950. GROSS, HYDROGEN BONDING, P. 203, (Ref 73). 88)MIYAZAWA and PITZER, J. AMER. CHEM. SOC., otl., 74, 1959. 90))SHEPPARD and EBBSWORTH0 SPECTROCHIMICA. ACTA., 3, 261, 1959. 91)SHARP, CHEM. AND IND., 1235, 1956. 92)CHENON and SANDORFY, CANAD. J. CHEM., afi,1181, 1958 93)WITTIG, KEICHER, RUCKERT and RAFF, ANN., 5k1, 110, 1949, WITTIG and RAFF, ibid. 5120 195, 1951. 94)WALDRON, J. CHEM. PHYS., ?II 734, 1953. 95)BRISSETTE and SANDORFY„ CANAD. J. CHEM. , 3,34, 1960. 96)WILMSHURST and BERNSTEIN, CLUB. J. CHEM., 35, 1185, 1957. -131- 97)GILL, NUTTALL, SCAIFE and SHARP, J. INORGO NUCL. CHEM., la, 790 1961. 98)VAN DER MEULEN and HELLER, J. AMER. CHEM. SOC., sh, 4404, 1932. 99)BAK and. RASTRUP-ANDERSON, J. CHEM. PHYS., ga,l 2013, 1954. 100)WADDINGTON, NUTTALL and SHARP, Jo, 49659 1960. 101)MoKINNEY and BARROW, J. CHEM. PHYS., 320 298, 1959. 102)BELL and BARROW, J. CHEM. PHYS., 320 300, 1959. 103)GREENWOOD and WADE, J., 366, 1960. 104)SHELDON and TYREE, J. AMER. CHEM. SOC., §.2„ 21170 19580 105)WOODWARD, TRANS. FAR. SOC., 5g„ 1458, 1956. 106)COOPER and WARDLAW, J., 635, 1932. 106a) FOWLES and OSBOURNE, J., 22759 1959.
107)WOOLF, Jog 2840, 1954, ibid, 433, 1955. 108)REID, J. AMER. CHEM. SOC., Ats, 3265, 1954. 109)GOLD and TYE, J., 2169, 1952. 110)SHARP, J. 4804, 1957. 111)LANGE, FLUORINE CHEMISTRY, ED. SIMON, PUB. ACADEMIC PRESS, 20 176, 1950. 112)HOFFE and LUCE, U.S. PAT., 1:73304580 1929. 113)ANDERSON, J. AMER. CHEM. SOC., 5, 4569, 1930. 114)SHARP, J., 3761, 1957. 115)LEWIS and LIPKIN, J. AMER. CHEM. SUC., iho 2803, 1942. 116)ROSS and TONNET9 Jo g 2013, 19590 117)CHATT 9 CHEMO AND IND.9 14749 1958. 118)SHARP, PROC. CHEM. SOLO, 317, 1960. 119)ABEL and WILKINSON, J., 1503, 1959. 120)ADDISON, KILNER and WOJCIKI, PROC. CHEMO SOC., 202© 1961. ?IA ~< C-k-ke-n kC161
Preprinted from the Journal of the Chemical Society, January, 1960, (10), pages 46-50.
1.0. The Basic Properties, Infrared Spectra, and Structures of Triphenylamine and Tri-p-tolylamine. By R. D. W. KEMMITT, R. H. NUTTALL, and D. W. A. SHARP. Complexes are formed between triarylamines and boron trichloride and boron tribromide. The preparations of these adducts and also of some tri- arylammonium salts are described. From a consideration of the infrared spectra of triphenylamine and tri-p-tolylamine it is considered that there is a planar or almost planar arrangement about the central nitrogen atom in these molecules. IT has generally been considered that triarylamines are almost non-basic since, apart from hydrofluoric 1 and perchlorie 2 acids, they dissolve only in acids which oxidise them. In a recent publication s one of us has given preliminary details of the preparation of some triphenylammonium salts from non-aqueous solvents. We now give further details of these preparations and of the preparation of some tri-p-tolylammonium salts, and describe the preparation of some triarylamine complexes. Triarylamine Complexes.—Triphenylamine and tri-p-tolylamine solutions in benzene become slightly coloured when Lewis acids such as boron trifluoride, stannic chloride, and antimony tribromide are added. It seems likely that some form of weak complex is being formed and these are under investigation. On addition of boron trichloride or tribromide to the triarylamine solutions there is immediate precipitation of dark-coloured solids. The triphenylamine-boron trichloride adduct was too unstable to be isolated at room temper- ature, but 1 : 1 adducts were prepared between triphenylamine and boron tribromide and between tri-p-tolylamine and boron trichloride. An adduct in which the molecular proportions are nearly 1 : 2 was formed between tri-p-tolylamine and boron tribromide. The I : 1 adducts are presumably straightforward co-ordination compounds but the 1 : 2 adduct recalls the 1 : 2 adducts formed between ethers and boron trifluoride 4 and between various amines and boron halides.5 It is possible to write structures which contain tetrahalogenoborate ions for these adducts but Brown and his co-workers 5 consider that they probably contain halogen bridges. The infrared spectra of the 1 : 1 boron trichloride-tri-p-tolylamine adduct and the 2 : 1 boron tribromide-tri-p-tolylamine adduct prepared in the present work are both very similar and it must be concluded that the bonding to the amine in the two adducts is very similar. This is not conclusive evidence, however, for the type of bonding present in the rest of the molecule. The order of stability of the boron halide adducts (BF3 < 13C13 < BBr3) is in accordance with the normal order of strengths of the boron halides as Lewis acids.6 Triarylammonium Salts. The addition of traces of moisture to the solution of tri- phenylamine and boron trifluoride in benzene precipitates a pale green oil which rapidly crystallises and which we formulate as triphenylammonium fluoroborate. Hydrolysis of boron trifluoride gives some tetrafluoroboric acid by disproportionation of hydroxytri- fluoroboric acid' or alternatively the reaction could proceed by disproportionation of a precipitate of triphenylammonium hydroxyfluoroborate. The infrared spectrum of the crystalline product shows no hydroxyl groups. Tri-p-tolylamine gave a similar oil but this would not crystallise. Passage of hydrogen halide through a solution of triarylamine and the appropriate metal halide gave other triarylammonium salts. Because of the insolubility of the boron trichloride and boron tribromide adducts in benzene solution it was not possible to prepare the tetrachloroborates or tetrabromoborates; hydrogen halide passed through a solution of tri-p-tolylamine and stannic halide in benzene gave a deep blue precipitate which was probably the aminium salt. The triarylammonium salts are extremely sensitive to moisture and must be handled in the dry-box. They slowly lose hydrogen halide at room temperature, and analysis and examination of physical properties must be carried out immediately after preparation. The salts are soluble in chloroform but they cannot be recovered and it is very doubtful whether they are stable in solution. The formulation of these derivatives as triarylammonium salts is supported by their infrared spectra which, except for the fluoroborate, are almost identical over the range 2000-650 cm.--1. The fluoroborate has the strong bands associated with the fluoroborate ion near 1000 cm.--1 and at about 750 cm.--1.9 The 1000 cm.--1 peak shows more splitting than is normally observed in fluoroborates and it is believed that this splitting and the occurrence of the normally infrared-inactive band at 750 cm.-1 are due to strong hydrogen bonding between cation and anion. Over the range 1600 650 cm.-1 the spectra are very similar to those of related tetrahedral molecules such as triphenylmethane and tri-p-tolyl- chloromethane (see Figure), the spectra providing further evidence in favour of the formul- ation of these salts as triarylammonium derivatives. Between 4000 and 2000 cm.-1 the spectra depend on the anion. The normal aromatic C-H frequencies occur near 3000 cm.-1 but the N-H frequencies are split and appear between 2900 and 2400 cm.-1. The assign- ment of these bands has been confirmed by preparation of N-deuterotriphenylammonium fluoroborate, in which the corresponding N-D bands are observed between 2250 and 2050 cm.-1. The position and appearance of these bands provides additional evidence for the presence of hydrogen bonding between cation and anion in all these salts.9 It would have been useful to have been able to compare the spectra of the present triphenylammonium salts with the spectra of triphenylamine-hydrofluoric acid and triphenylamine-perchloric acid but we could not repeat the preparation of these latter adducts satisfactorily. Triphenylamine dissolves in anhydrous hydrogen fluoride to give a dark-blue solution but on removal of solvent the triphenylamine was recovered un- changed. Only unchanged triphenylamine could be recovered from aqueous hydrofluoric acid. Attempts to prepare the triphenylamine-perchloric acid adducts gave either unchanged triphenylamine or dark-coloured oxidised products. Triphenylamine is soluble in anhydrous hydrogen chloride to give a pale blue solution but once again only triphenyl- amine could be obtained at room temperature after removal of solvent. The blue solutions of triphenylamine in hydrogen fluoride and hydrogen chloride are presumably due to the presence of protonated species but it seems unlikely that the protonation is on the nitrogen only as triphenylammonium salts are colourless or pale green. Infrared Spectra and Structure of Triphenylanzine and Tri-p-tolylamine.--The present demonstration that triarylamines have some basic properties leads to reconsideration of the evidence for the structure of these amines. The dipole moment of triphenylamine is 0. 5 D; 10 such a small value may not be significantly different from zero. From a study of the ultraviolet spectrum Jaffe 11 considers that there is extensive conjugation in the molecule. The chemical properties of triphenylamine are in accordance with a stable cdnjugated system. We have now shown that the ultraviolet spectrum of tri-p-tolylamine is similar to that of triphenylamine and this is also evidence of the extensive conjugation between the lone-pair of electrons on the central nitrogen atom and the aromatic ring systems. Normally, lone pairs make a finite contribution to the stereochemistry of a molecule but to explain the low dipole moment it seems likely that in the triarylamines the lone pair of electrons is absorbed into the overall conjugated system. If the skeletal arrangement about the central nitrogen atom of triphenylamine is planar it would be expected, by analogy with the triphenylmethyl carbonium ion, that the molecule would be propeller-shaped having symmetry D3.12 If the molecule were slightly bent about the central nitrogen atom the highest possible symmetry would be C3. Ideally it should be possible to decide between these two configurations by a simple process of count- ing the number of components into which known vibrations are split but, as discussed previously,12 such a simple assessment is not possible for molecules of this complexity. The Figure shows the spectra of the amines, the corresponding ammonium salts, tri- phenylmethane, and tri-p-tolylchloromethane, the corresponding carbonium salts, and triphenylboron and the tri-p-tolylaminium cation. The differences between the spectra of the amines and the spectra of molecules such as triphenylmethane and the triphenyl- ammonium ion which must be tetrahedral are sufficiently marked, particularly near 1600 and 1300 cm.-1, to make it unlikely that the triarylamines are fully tetrahedral. The strong bands at 1600 and near 1300 cm.-1 are common to the amines, to the related carbonium ions, and to triphenylboron and the tri-p-tolylaminium ion and favour a planar skeletal arrangement in the amines although it could possibly be considered that these bands merely show the effect of the extensive conjugation in the molecules. The Infrared spectra of triarylantines and related compounds. 1,11 kJ:11,j4
(%) n ti_,R,,ILLAAL:4,JudL io t JAA_Ajott,_;lifAi Absorp :1,14A„"&kilk /700 /500 /100 //00 900 700 /700 /500 /300 //00 900 700 A, Triphenylamine; B, Triphenylammonium bromoantimonite; C, Triphenylmethane; D, Triphenyl- methyl chlorostannate; E, Triphenylboron; F, Tri-p-tolylamine; G, Tri-p-tolylammonium chloro- antimonite; H, Tri-p-tolylchloromethane; I, Tri-p-tolylmethyl hexafluoroniobate; J, Tri-p-tolyl- aminium hexafluorotantalate. presence of only one absorption band in the C—H out-of-plane region of the spectrum of triphenylamine is anomalous since, even with a planar propeller-shaped structure, there should be two infrared-active components for this vibration; 12 an all planar configuration would only give one band in this region. It is unlikely for steric reasons that the molecule is all-planar and there are two bands in the corresponding region in tri-p-tolylamine, so it must be assumed that the two vibrations are accidentally degenerate in triphenylamine. Thus, whilst not conclusive, the infrared spectra of triphenylamine and tri-p-tolylamine agree with other evidence in favouring a planar, or almost planar, arrangement about the central nitrogen atom. In having planar or near-planar structures, triarylamines, with conjugation between the lone pair of electrons on the nitrogen and Tc-electron systems of the aromatic rings, are very similar to trisilylamine, where the lone pair of electrons appears to be bonded to unfilled d orbitals on the silicon atoms.13,14,15 [Added in proof.] Recently Sasaki, Kimura, and Kubo 16 described electron diffraction measurements on triphenylamine. In the vapour at 200° the C—N—C bond angle is estimated to be 116 ± 2°. The authors comment that their values could be interpreted as in- volving a fiat trigonally bonded nitrogen at equilibrium, the molecule appearing pyramidal because of thermal bending motions. They prefer, however, a non-planar molecule of C3 symmetry. Since the infrared studies of the present work were carried out on the solid it is possible that a planar configuration is stabilised in that phase. EXPERIMENTAL Triphenylamine and tri-p-tolylamine in benzene gave dark-coloured products with boron trichloride and boron tribromide. These adducts were characterised by condensing excess of the halides on a solution of a known quantity of the amine in benzene or toluene, the whole experiment being carried out in a vacuum line. After allowing the solutions to warm, excess of reactants and solvent were removed in vacuo and the pale brown products weighed. The -adducts are extremely sensitive to moisture and fume even in the dry-box. Analyses were made for halogen, the transfer to solution being made as quickly as possible. Although a solution of triphenylamine in benzene gave a dark-coloured precipitate with boron trichioride the product was unstable under vacuum at room temperature. We prepared triphenylamine— boron tribromide [1 mole of amine absorbs 0.9 mole of BBrs. Found Br, 45.5. (C61-15)3N,BBr3 requires Br, 48.4%] ; and tri-p-tolylamineboron trichloride [1 mole of amine absorbs 1.1 moles of B03. Found: Cl, 23.4. (C.71-17)3N,BC13 requires Cl, 26.3%], and tribromide [1 mole of amine absorbs 1.8 moles of BBr3. Found: Br, 55.0. (C7117)3N,BBr3 requires Br, 44-6. (C7H7)3N,2BBr3 requires Br, 60.8%]. Triphenylammonium salts were prepared as previously described.3 N-Deuterotriphenyl- ammonium fluoroborate was prepared by shaking sodium-dried benzene with heavy water followed by passage of boron trifluoride through a solution of triphenylamine in the moist benzene. The following tri-p-tolylammonium salts were prepared by similar methods to those used for triphenylammonium salts 3 (tri-P-tolylammonium fluoroborate would not crystallise) : chloraantimonite [Found: C, 44.3; H, 4.4; Cl, 24.4. (C71-17)3NHSbC14 requires C, 45.7; H, 4.0; Cl, 25.7%], bromoantimonite [Found: C, 32.7; H, 3.4. (C71-17)3NHSbBr4 requires C, 34.5; H, ' 3.0%]. The passage of hydrogen halide through a benzene solution of tri-p-tolylamine and stannic chloride or stannic bromide gave a deep blue precipitate. Attempts were made to prepare perchlorates and fluorides as described in the literature.1,2 In all cases, after removal of solvent, only unchanged amine or very dark oxidised products remained. Triphenylamine dissolved readily in anhydrous hydrogen fluoride to give a deep blue solution. On evaporation of the solvent in a stream of dry nitrogen unchanged triphenylamine remained. Triphenyl- amine dissolved in liquid hydrogen chloride to give a pale blue solution but again only triphenylamine remained on removal of solvent. Ultraviolet spectra were recorded in methanol on a Unicam SP.500 spectrophotometer. _The spectrum of triphenylamine was very similar to that recorded by Jaffe: 11 )max. 296- 298 nip., %az. 19,900. 'min. 249-250, Emirs, 1600. That of tri-p-tolylamine had Amax. 298 mp., emax. 56,450; Junin, 252 mt.c, emirs. 1470. Infrared spectra were recorded on a Perkin-Elmer Model 21 spectrophotometer with rock- salt or fluorite optics. Mulls were prepared in Nujol or hexachlorobutadiene (in the dry-box if necessary). We thank Borax Consolidated Ltd. for a gift of boron tribromide, Imperial Smelting Corpor- ation for a gift of boron trifluoride, and the Ministry of Education (R. D. W. K.) and the County Borough of Bolton (R. H. N.) for maintenance grants. INORGANIC CHEMISTRY RESEARCH LABORATORIES, IMPERIAL COLLEGE, LONDON, S.W.7. [Received, July 23rd, 1959.]
1 Weinland and Reischle, Ber., 1908, 41, 3617. 2 Hofmann, Metzler, and Hobold, Bee., 1910, 43, 1080. 3 Sharp, Chem. and Ind., 1958, 1235. 4 Wirth, Jackson, and Griffiths, J. Phys. Chem., 1958, 62, 871. 5 Brown, Stehle, and Tierney, J. Amer. Chem. Soc., 1957, 79, 2020. 6 Brown and Holmes, ibid., 1956, 78, 2173. 7 Wamser, ibid., 1951, 73, 409. 8 Cote and Thompson, Proc. Roy. Soc., 1952, A, 210, 217. Nuttall, Sharp, and Waddington, unpublished observations. 10 Klages -and Langpape, Z. Elektrochem., 1959, 63, 533. 11 Jaffe, J. Chem. Phys., 1954, 22, 1430. 12 Sharp and Sheppard, J., 1957, 674. 11 Hedberg, J. Amer. Chem. Soc., 1955, 77, 6491. 14 Ebsworth, Hall, MacKillop, McKean, Sheppard, and Woodward, Spectrochim. Acta, 1958, 13, 202. 13 Kriegsmann and Forster, . anorg. Chem., 1959, 298, 212. 19 Sasaki, Kimura, and Kubo, J. Chem. Phys., 1959, 31, 477.
PRINTED IN GREAT BRITAIN BY RICHARD CLAY AND COMPANY, LTD.; BUNGAY, SUFFOLK. Preprinted from the Journal of the Chemical Society, December, 1960, (963), pages 4965-4970.
963. Hydrogen Bonding in Alkyl- and Aryl-ammonium Salts.* By R. H. NUTTALL, D. W. A. SHARP, and T. C. WADDINGTON. The infrared spectra of a range of alkyl- and aryl-ammonium salts have been studied, particular attention being paid to the N—H stretching and N—H deformation frequencies. In salts where there is no cation—anion hydrogen bonding the N—H stretching frequencies occur well above 3100 cm.-1. Except for triphenylammonium fluoroborate, fluoroborates and tetraphenylborates are not hydrogen-bonded. PREVIOUS infrared studies on alkyl- and aryl-ammonium salts have been largely concerned with their halides.1-7 The N-H stretching frequencies occur below 3100 cm.-1, generally below 2900 cm.-1, and it has been concluded that the shift from the normal value for N-H stretching frequencies of 3300 to 3500 cm.-13 is because of hydrogen bonding between cation and anion, a conclusion supported by structural studies on this type of salt.9 it has been suggested" that in substituted ammonium salts where there is no hydrogen bonding the N-H symmetrical stretching frequencies would occur in the region 3000-3-200 cm.-1; this paper describes attempts to verify this and to investigate the variation in occurrence of hydrogen bonding with change in anion. The factors involved in the formation of hydrogen bonds in ammonium salts have been discussed by Chenon and Sandorfy 2 in terms of the ionic and covalent character of the various contributing structures. For an +N-H • • • X- bond the major contribution is from an electrostatic component, +N-H • • • X. If X is part of a complex anion, e.g., BF4- or BPh4-, the charge on X will be relatively small and this term will be correspond- ingly weak. Contributions such as N PEI • • • X- and N H-X, which involve H-X bonds, depend markedly upon the strength of the acid HX. Complex fluoro- and chloro-acids are generally so strong that they do not exist in the absence of bases to solvate the proton ; this factor again reduces the strength of hydrogen bonds in ammonium salts of complex acids. On this basis, in a series of substituted ammonium salts, halides, where both electrostatic and covalent terms are favourable, would be expected to show hydrogen bonding, but tetraphenylborates where both terms are unfavourable—free tetraphenyl boric acid immediately decomposes 1° would be expected not to be hydrogen bonded. Fluoroborates should generally not be hydrogen bonded because of the absence of the covalent term—HF and BF3 do not interact in normal circumstances 11—but it is conceiv- able that with a strongly polarising cation the anion may be distorted into some sort of bonding. Unsubstituted ammonium halides show evidence of hydrogen bonding; the fluoroborate and tetraphenyborate are not hydrogen bonded.12 It is difficult to be precise when considering hydrogen bonding but we have used the position of the N-H infrared bands for diagnosis. Where possible, we have used the spectra of tetraphenylborates as standard spectra of unassociated salts. We have not considered in any detail the bands which occur near 2000 cm..-1 in the spectra of many of the salts; Chenon and Sandorfy 2 consider that these bands result from a combination between a scissoring mode and an anharmonic vibration near 400 cm.-1. Tertiary Ammonium Salts.—Trimethylam,monium salts. The infrared bands observed for these salts and also for triethylammonium fluoroborate in the range 4000 2000 cm..-1 and also the position of the N-H deformation frequency near 1430 cm.-1 are given in Table 1. In the 4000-2004 cm.-1- region there will be bands due to C-H symmetrical stretching vibrations and also C-H combination and overtone modes in addition to the N-H frequencies. These C-H modes remain at 3030w, 2960s, 2930s, and 2850w cm.-1 for * Presented in part at the International Symposium on Fluorine Chemistry, Birmingham, July, 1959.
TABLE 1. Infrared spectra (cm.-1) of trimethylammonium salts.
Anion VN-H Vc-H PN-H SN-H t-- -, Br- 3010 2965s 2930 2730s 1430 I- — — 3020 2960s 2930 2730s 1420 BP114- 3130s - 3060* 3030 3000s — 2720 — 1420 PF,- 3230s — — — 2930s 2860 1423 BF4 3200s — 2930s , 2850 — 1042 C10,- 3150s -- 2930s 2800 1418 Et31\THBF4 3150s — 2975s 2942 — — 1425 MeaNDBF, — — 2930s — 2450, 2400, 2375 s = strong. * This peak also contains contributions from the spectrum of the anion. salts with different anions and also on N-deuteration of trimethylammonium fluoroborate; not all these bands were resolved in all the salts. In salts in which hydrogen bonding is unlikely there is a very strong band at about 3200 cm.-1 which we assign to the N—H symmetrical stretching mode. The position of this band varies slightly from one salt to another, but these small shifts are probably due to minor variations in the crystal field at the site of the N—H bond (cf. changes in the N-0 frequency of nitrosyl chloride when dissolved in solvents of different dielectric constant 8). The peak is sharp and well defined, but on deuteration of the fluoroborate the band splits and moves down to about 2400 cm.-1 (vii/vD = 1.33). A similar splitting has been observed in the spectrum of N-deuterotri- methylammonium iodide and was attributed to Fermi interaction between the N—D vibration and overtone and summation bands.' The spectra of trimethylammonium bromide and iodide show no peaks at 3200 cm.-1, but instead there is a strong peak at 2730 cm.-1; this band has been recorded at even lower frequencies in the spectra of trimethylammonium chloride 13 and other trialkylammonium chlorides.2 There is a very weak band at 2720 cm.-1 in the spectrum of trimethylammon- ium tetraphenylborate but this is probably a combination band. The N—H deformation frequency occurs near 1420 cm.-1 in the spectra of all the salts studied; it does not appear to be sensitive to the presence of hydrogen bonding. On deuteration of trimethylammonium fluoroborate this band moves under the strong fluoro- borate absorption at 1050 cm.-1. We conclude that these spectra are in agreement with the presence of strong hydrogen bonding in the trimethylammonium halides but not in the other trimethylammonium salts studied. Pyridinium salts. The infrared peaks which occur between 4000 and 2000 cm.-1 in the spectra of the salts studied are given in Table 2. The peaks between 3080 and 2850 cm.-1 TABLE 2. Infrared spectra (cm.-1) of pyridinium salts.
Anion PN-H PC-H PN-H .,._ A. Cl- 3200w — 3120sh 3045m 2925s 2840s 2450s Br- 3210w 3180w — 3040m 2930s 2284860w 2650b 1- 3210w 3160w 3100w 3060m 3020s 2950s 2870b NO,- 3230w 3170w 3100w 3070m 2960s 2730b HF,- 3230w 3180w — 3080m _ 2800b C116 -S06- 3220w 3160sh — 3080m 2920s 2850w 2700b PbC162- 3230s 3160s 3110s 3080m C104 3255s 3180s 3110s 3080m 2980ss 2930s ____. BPh,- 3230s 3178s 3130s 3058m 3010s 2930s 2860w BF,- 3250w 3160s 3120s 3080m 2900s 2827w BO,- 3240s 3130s 3100s 3090m 2960s 2930s 2800w PF6- 3230s 3190s 3110s — 2920s 2870w pyDBF4 — — 3080m 2900s 2827w 2450s, 2405s, 2325s w = weak, m = medium, s = strong, sh = shoulder, b = broad, py = pyridine. are due to C—H modes.8 Three peaks due to N—H vibrations occur above 3100 cm.-1 in the spectra of all salts studied, whether hydrogen bonded or not; these peaks are of very low intensity in salts which might be expected to be hydrogen bonded but are of high
intensity in the tetraphenylborate and other salts in which the tendency to hydrogen bonding should be small. On deuteration of pyridinium fluoroborate these bands shift to about 2400 cm.-1 (vEr/vD = 1.32, 1.32, and 1.29 for the 3250, 3160, and 3120 cm.-1 band, respectively). In the halides and other salts where hydrogen bonding can be expected there is an additional, intense, broad band well below 3000 cm.-1. The infrared spectra of pyridinium salts below 2000 cm.-1 are complex and will be discussed separately." From the spectra, the presence of the strong band below 3000 cm.-1 being taken as diagnostic of hydrogen bonding, we conclude that pyridinium halides, nitrate, bifluoride, and methanesuiphonate are hydrogen bonded but that the other salts that we have studied do not have this type of cation-anion interaction. The presence of N-H peaks above 3100 cm.-1 in the spectra of salts in which there is hydrogen bonding suggests that either there are some non-associated cations in these salts or, more likely, that there is a double well effect in the potential energy-interatomic distance curve for the hydrogen bonds in these compounds (cf. ref. 15). Triarylammonium salts. The infrared spectra from 4000 to 2000 cm.-1 of the salts studied are recorded in Table 3. There is a series of peaks near 3000 cm.-1 which are assigned TABLE 3. Infrared spectra (cm.-1) of triphenylammonium salts.
Anion VC-H VN-H 8N-11 BF4- 3070 3015 2930 2842 2738 2655 1391 SbC14 3062 3038 2888 2795 10 3380 SnC162- 3085 3060 3038 2920 2850 2805' 2710 260225402702 A1C14- 3047 2883 2800 2705 2483 1390,1375 Snr13 5- 3030 2855 2790 2700 25800 2503 1378 SbBr4 3034 2882 2811 2710 2592 2522 1385 Ph3NDBF4 3050 3045 2910 2240 2165 2065 All peaks of approximately equal intensity. The most intense peak in the N-H region is italicised. to C-H vibrations and the N-H peaks are widely split and occur between 2850 and 2400 cm.-1. Since triarylamines are such weak bases,16 it is possible that the N-H bond in triarylammonium salts is excessively weak and that the frequencies fall in the observed region without the necessity of invoking hydrogen bonding. Triarylammonium tetra- phenylborates have not been prepared, so a direct test of this cannot be made but, since these spectra are broadened in the manner of known hydrogen-bonded ammonium salts, and since the frequency of the vibrations is influenced by the anion, we consider that these spectra imply the existence of strong hydrogen bonds in all the triarylammonium salts that we have studied. If the position of the principal band is taken as a measure of the strength of the hydrogen bonding, the fluoroborate and the chloroantimonite are the least strongly associated. In N-deuterotriphenylammonium fluoroborate the N-H peaks have shifted to about 2100 cm.-1 (vEr/vD = F28, 1.26, and 1.27 for the 2842, 2738, and 2655 cm.-1 peak, respectively). The N-H deformation frequency at 1391 cm.-1 moves under the fluoroborate band at 1050 cm.-1 on deuteration. Secondary and Primary Ammonium Salts.-The infrared spectra over the range 4000- 2000 cm.-1 of the salts studied are recorded in Table 4. Anilinium and fi-toluidinium TABLE 4. Infrared spectra (cm.-1) of primary and secondary ammonium salts. Me2NH2BF4 3257s, 3010, 2935, 2860, 2790, 2450 Me2NH2BPh4 3140s, 3045, 2995, 2760 MeN1-1313F4 3275, 2926, 2854, 2772, 2560, 2470 MeNH3BPh4 3175, 3130, 3043, 2955, 2915, 2875, 2775, 2556 Ph2NH2BF4 3175, 3075, 2925, 2845, 2583, 2485 PhNH3BF4 3215, 2923, 2850, 2600, 2325 C4114Me•NH2BF4 3195, 3070, 3010, 2920, 2845, 2620 tetraphenylborates were precipitated from solution as hydrates and, as there would certainly be hydrogen bonding between the ammonium ions and water, these cannot be used as spectra of non-hydrogen bonded salts. The isolated ions should give two N-H frequencies near 3000 cm.-"." Previous work on the alkylammonium halides has placed these frequencies at 3080 and 2980 cm.--1 in methylammonium chloride,4 at 2965 and 2745 cm.-' in dimethylammonium chloride,' and at even lower frequencies for other alkyl- and aryl-ammonium halides.2,3 By analogy with previous assignments we consider that the N—H vibrations in dimethylammonium fluoroborate absorb at 3257 and 2790 cm.-1- (3140 and 2760 em.-1 in the tetraphenylborate) (values in parentheses throughout this section refer to the tetraphenylborates), the bands at 3010 (3045), 2935 (2995), and 2860 cm:" being C—H modes. In the spectrum of methylammonium fluoroborate only the 3275 (3175) cm.-1 band has been definitely assigned to an N—H vibration; the second N—H vibration may have become accidentally degenerate with the first or, more probably, is under one of the C—H bands which occur at 2926 and 2854 (3050, 2990) cm.-". The spectra of all the salts considered in this section contain weak bands below 2800 cm.-1. These bands are not as strong or as broad as those observed in the spectra of hydrogen-bonded salts, and it is very probable that they are combination and overtone frequencies as are observed in this region of the spectra of many hydrocarbon derivatives.1,8 The spectra of aryl-substituted secondary and primary ammonium salts look more complex because of the presence of broad, moderately strong bands below 2800 cm.-4. The spectra of the fluoroborates show a strong band above 3100 cm:" which we assign to one of the N—H vibrations. The other N—H frequency has not definitely been identified but probably occurs below 2900 cm.-1 and is responsible for exciting combination modes and overtones to give moderately strong peaks in this region. Since all the salts have a strong band above 3150 crn.-1 we conclude that there is no evidence for hydrogen bonding in any of the primary or secondary ammonium fluoroborates.
EXPERIMENTAL Infrared spectra were measured in hexachlorobutadiene or Nujol mulls by using a Perkin— Elmer model 21 spectrophotometer fitted with rock-salt or fluorite optics. Fluoroborates were generally prepared froth the parent amine and aqueous fluoroboric acid, followed by recrystallisation from ethanol (Found, for trimethylammonium fluoroborate: C, 24.7; H, 6.9. Calc. for C31-110NBF4: C, 24.5; H, 6.8. Found, for triethylammonium fluoro- borate: C, 38.2; H, 8.4. Calc. for C6H16NBF4: C, 38.1; H, 8.5. Found, for diphenylammon• ium fluoroborate: C, 55.5; H, 4.6. Calc. for C12H12NBF4: C, 56.0; H, 4-7. Found, for anilinium fluoroborate: C, 39.8; H, 4.7. Cale. for C6H8NBF4 : C, 39.9; H, 4.5. Found for p-tolylammonium fluoroborate: C, 40.8; H, 5.5. Calc. for C7H,0NBF4 : C, 42.4; H, 5.1%). Deuterations were effected by crystallisation from 99% D20. Dimethylammonium fluoroborate was prepared from dimethylamine and fluoroboric acid. It was recrystallised from dry ethanol and subsequently handled in the " dry-box," Since it is extremely deliquescent, analysis was not attempted; the infrared spectrum showed no traces of water or other impurites. Pyridinium fluoroborate was prepared by hydrolysis of pyridine—boron trifluoride in 95% ethanol 17 (Found: C, 36.0; H, 3-6. Cale. for C51-16NBF6: C, 36.0; H, 3.6%). Triphenyl- • ammonium fluoroborate and all other triarylammonium salts were the samples prepared and analysed preyiously.16,18 Tetraphenylborates were prepared by precipitation from aqueous solutions of the amine hydrochloride and sodium tetraphenylborate (Found, for trimethylammonium tetraphenyl- borate: C, 85.8; H, 8.2. Calc. for Me2NHBP114: C, 85.6; H, 8-0. Found, for pyridinium tetraphenylborate: C, 87.2; H, 6.9. Cale. for C,H,NHBPh4: C, 87.3; H, 6.6. Found, for dimethylammonium tetraphenylborate: C, 86.0; H, 8.0. Calc. for Me2NH2BP114: C, 85.7; H, 7.8. Found, for methylammonium tetraphenylborate: C, 86.0; H, 7.9. Calc. for MeNH3BP114: C, 85.7; H, 7.5%). Anilinium and p-toluidinium tetraphenylborates were precipitated as hydrates and were not studied further, Di- and tri-phenylammonium tetra- phenylborates could not be prepared. Perchlorates were prepared from solutions of the amine in aqueous perchloric acid. After decomposition with sodium hydroxide, perchlorate was estimated by precipitation as the tetraphenylarsonium salt (Found, for trimethylammon- ium perchlorate; C104, 61.8. Cale. for C21-110NC104: C104, 62.2. Found, for pyridinium perchlorate: C104, 55.6. Calc. for C,H8NC104: C104, 55.4%). An aqueous solution of hexa- fluorophosphoric acid was prepared by passing a solution of the sodium salt over a cation- exchange resin. Hexafluorophosphates were prepared from solutions of the amine in the acid (Found, for trimethylammonium hexafluorophosphate: C, 18.0; H, 5.2; N, 6.5. Calc. for C1li10NPF6: C, 17.6; H, 4.9; N, 6.8. Found, for pyridinium hexafluorophosphate: C, 27.5; H, 2.65; N, 6.5. Calc. for C5H6NPF6: C, 26.7; H, 2.7; N, 6-2%). Trimethylammonium chloride and bromide were prepared by freeze-drying aqueous solutions which had been made froin the amine and free acid [Found, for trimethylammonium chloride: Cl, 37.4. Calc. for C31-140NC1: Cl, 37.1. Found, for trimethylammonium bromide: Br, 60-3. Calc. for C31-140NBr: Br, 56.8% (this sample contained a slight excess of hydrogen bromide)]. The trimethylammon- ium iodide was a commercial sample. Pyridinium nitrate was prepared by freeze-drying an aqueous solution of the salt (Found: N, 19-0. Calc. for C51-1,N203: N, 19.7%). Pyridinium hydrogen fluoride was obtained by dissolving pyridine in an excess of aqueous hydrofluoric acid, evaporation, and drying at 1200. Fluorine was estimated gravimetrically as PbC1F (Found: F, 32.4. Calc. for C5H,NHF2: F, 31.8%). Pyridinium methanesulphonate was prepared from a solution of pyridine in the acid (Found: C, 40.7; H, 5.5; N, 7.6. Calc. for C5H6NSO3CH3: C, 414; H, 5.2; N, 8.0%). Pyridinium hexachloroplumbate was made as described by Palmer 19 [Found: C, 20.6; H, 2.3; N, 4.9%; M, 578. Calc. for (C5H6N)2PbC18: C, 20.7; H, 2.1; N, 4.8%; M, 580]. Pyridinium chloride, bromide, and iodide were made by condensing excess of the dry hydrogen halide on to anhydrous pyridine at liquid-nitrogen temperatures and allowing the solids to warm very slowly. In the case of the chloride the pyridinium hydrogen dichloride formed initially was decomposed by heating in vacuo to 100° (Found, for pyridinium chloride : Cl, 30.4. Calc. for C5I-I0NC1: Cl, 30.7. Found, for pyridinium bromide: Br, 49.3. Calc. for C5H,NBr: Br, 49%8%. Found, for pyridinium iodide: I, 61-9. Calc. for C51-19NI: I, 61.4%). Pyridinium tetrachloroborate was prepared from pyridine, hydrogen chloride, and boron trichloride, the hydrogen chloride being used as a solvent 29 (Found: Cl, 60.1. Calc. for C5H6NBC14 : Cl, 60.9%). We thank the Imperial Smelting Corporation for a gift, of boron trifluoride. INORGANIC CHEMISTRY RESEARCH LABORATORIES, IMPERIAL COLLEGE, LONDON, S.W.7. [R. H. N. and D. W. A. S.] UNIVERSITY CHEMICAL LABORATORIES, LENSFIELD ROAD, CAMBRIDGE. [T. C. W.] [Received, April 19th, 1960.]
1 Ebsworth and Sheppard, Spectrochim. Acta, 1959, 13, 261. 2 Chenon and Sandorfy, Canad. J. Chem., 1958, 36, 1181. 3 Brissette and Sandorfy, Canad. J. Chem., 1960, 38, 34. • Waldron, J. Chem. Phys., 1953, 21, 734. 5 Stone, Cymerman-Craig, and Thompson, J., 1958, 52. OBellanato and Barcelo, Anales real Soc. espaci. Fis. Quim., 1956, 52, B, 469. 7 Heacock and Marion, Canad. J. Chem., 1956, 34, 1782. • Bellamy, " The Infrared Spectra of Complex Molecules," 2nd edn., Methuen, London, 1958. 9 Donohue, J. Phys. Chem., 1952, 56, 502. 10 Wittig, Keicher, Ruckert, and Raff, Annalen, 1949, 563, 110; Wittig and Raff, ibid., 1951, 573, 195. • 1, McCaulay and Lien, J. Amer. Chem. Soc., 1951, 73, 2013. 12 Waddington, J., 1958, 4340. 13 Lord and Merrifield, J. Chem. Ploys., 1953, 21, 166. 14 Gill, Nuttall, Scaife, and Sharp, J. Inorg. Nuclear Chem., in the press. 15 McKinney and Barrow, J. Chem. Phys., 1959, 31, 294; Bell and Barrow, ibid., pp. 300, 1158. " Kemmitt, Nuttall, and Sharp, J., 1960, 46. 17 Van der Meulen and Heller, J. Amer, Chem. Soc., 1932, 54, 4404. 19 Sharp, Chem. and Ind., 1958, 1235. 19 Palmer, " Experimental Inorganic Chemistry," Cambridge Univ. Press, London, 1954. 20 Klanberg and Waddington, Naturwiss., 1959, 46, 578.
PRINTED IN GREAT BRITAIN BY RICHARD CLAY AND COMPANY, LTD., BUNGAY, SUFFOLK, J. Inorg. Nucl. Chem., 1961, Vol. 18 pp. 79 to 87. Pergamon Press Ltd. Printed in Poland
THE INFRA-RED SPECTRA OF PYRIDINE COMPLEXES AND PYRIDINIUM SALTS
N. S. GILL f, R. H. NUTTALL t, D. E. SCAIFE* and D. W. A. SHARP t t Inorganic Chemistry Research Laboratories, Imperial College, London
* William Ramsay and Ralph Forster Laboratories, University College, Gower Street, London, W. C. 1. (Present address: Chemical Research Laboratory, C.S.I.R.O., Melbourne, Australia).
(Received 17 March 1960; in revised form 3 May 1960)
Abstract —The main qualitative differences between the spectra of pyridine, co-ordinated pyridine and the pyridine ion are discussed and an assignment is made for all of the bands observed in the spectra of co-ordinated pyridine. In view of the small shifts observed when comparing the spectra of pyridine complexes, it is concluded that the electron density over the pyridine ring remains almost constant whatever the acceptor atom and it is suggested that back-bonding plays a part in the bonding of these complexes. Some of the N-H bands in the spectra of the pyridinium ion have been identified by deuteration and the spectra of this ion are discussed.
THE infra-red spectrum of pyridine has been extensively studied and unequivocal assignments have been made for most of the observed bands (1,2,3) but the infra-red spectra of pyridine complexes—with which we include pyridinium salts—have received comparatively little systematic attention although o-phenanthroline and a, a'-dipyridyl complexes have been studied in some detail.(4) Pyridine complexes for which spectra have been recorded and discussed include indium, silver, nickel, and zinc salts ;(5) boron and other light non-metallic acceptors ;(6) pyridine-boron trichloride and two pyridinium salts.(7) Accounts of parts of the infra-red spectra of other pyridine complexes have also been published.(8-13) BICELLI(5) considers that, apart from the splitting of some bands, the infra -red spectrum of co-ordinated
( I) C. H. KLINE and J. TURKEVICH, J. Chem. Phys. 12, 300 (1944). (2) L. CORSSIN, B. J. FAX and R. C. LORD, J. Chem. Phys. 21, 1170 (1953). (3) J. K. WILMSHURST and H. J. BERNSTEIN, Canad. J. Chem. 35, 1185 (1957). (4) A. A. SCHILT and R. C. TAYLOR, J. Inorg. Nucl. Chem. 9, 211 (1959). (5) L. BICELLI, Nuovo Cim. 9, 184 (1958); Istituto Lombardo, Rend. Class. Sci. (A) 92, 536 (1958); Ann. Chim. (Rome) 48, 749 (1958). (2) A. R. KATRITKZY, J. Chem. Soc. 2049 (1959). (2) N. N. GREENWOOD and K. WADE, J. Chein. Soc. 1130 (1960). I" B. CHENON and C. SANDORFY, Canad. J. Chem. 36, 1181 (1958). 491 N. S. HAM, A. L. G. REES and A. WALSH, Nature, Lond. 169, 110 (1952). (20) J. P. COLLMAN and H. F. HOLTZCLAW, J. Amer. Chem. Soc. 80, 2054 (1958). ( u) R. A. ZINGARO and P. J. TOLBERG, J. Amer. Chem. Soc. 81, 1353 (1959). Its) G. S. RAO, Naturwissenschaften 46, 556 (1959). (") R. HULME, G. J. LEIGH and I. R. BEATTIE, J. Chem. Soc. 366 (1960). [79] 80 N. S. GILL et ca. pyridine is very little different from the spectrum of pyridine itself, but GREENWOOD and WADE(7) record fairly major changes in the positions of some of the bands when comparing the spectrum of co-ordinated pyridine with that of the free base. We have recorded the spectra of a number of pyridine complexes and pyridinium salts with the aim of examining them for systematic changes on co-ordination and also for deciding whether there is any relation between the splitting of the bands and the structures of the complexes. We have found that there is a sharp difference between the spectrum of pyridine co-ordinated to elements other than hydrogen and the spectra of the pyridinium ion. Qualitatively, co-ordinated pyridine is usually readily distinguished from the free base by the presence of a weak band between 1235 and 1250 cm-1; by a shift in the strong 1578 cm-1 band to 1600 cm-1 (this band appears with a shoulder in the spectrum of the free base); and by shifts of the 601 and 403 cm-1 bands to 625 and 420 cm-1, respectively. The ca. 1240 cm-1 band occurs also in the spectrum of the pyridinium ion but there are major changes when comparing pyridinium spectra with the spectra of other pyridine complexes. Intense bands occur at 3200 or 2800, 1640, 1530, 1327, and 1250 cm-1 in the spectra of pyridinium salts—there are no strong bands in these positions in the spectrum of pyridine.
Pyridine complexes to metals The positions of the principal infra-red bands observed in the spectra (2000 to 400 cm-1) of pyridine complexes are given in Table 1. Each band in the spectrum of pyridine is faithfully reproduced with only minor shifts or splittings in the spectra'
JiJ
0 a.
d 700 1500 1300 1100 900 700 500 Frequency, ern'
FIG. 1.—Spectra of pyridine complexes and pyridinium salts. (A) Pyridine (as film); (B) MnPY2C12; (C) (PYF1)2CoC14; (D) (pyD)BF4. * Peaks due to fluoroborate ion of the complexes. The assignment of the various bands has been made by a straight- forward comparison (Fig. 1) between the spectrum of co-ordinated pyridine and the spectrum of the free base. To a first approximation the vibrations of pyridine co- The infra-red spectra of pyridine complexes 81 ordinated to a heavy atom will not involve the vibrations of the metal-nitrogen bond and this one-to-one correspondence is, therefore, to be expected and it should be possible to describe the observed bands in terms of the corresponding vibrations for free pyridine using the notation of KLINE and TuRKEvicx.(11 The appearance of the spectra of pyridine coordinated to light atoms' are similar to those recorded in the present investigation although rather different assignments have been made. The main details of the assignments are given in Table 1 but a few points, particu- larly where we disagree with the assignments of GREENWOOD and WADE(7) call for special comment. Between 990 and 1217 cm-1 there is a group of five strong bands in the spectrum of pyridine (see figure); there is then a gap, with only very weak bands, to 1436 cm-1. In the spectra of co-ordinated pyridine there is this same group of five bands, occasionally with slight splitting, and with an extra moderate to weak band between 1235 and 1250 cm-1. In view of the general appearance of this main group of five bands we assign them on the basis of a one-to-one correspondence between the bands in co-ordinated pyridine and in the free base. On the assignment of GREENWOOD and WADE(7) this group of bands is shifted to higher frequencies when pyridine is acting as a ligand. The ca. 1240 em-1 band which occurs in the spectra of pyridine complexes cannot be a metal-ligand frequency as it would not be expected near this value. It seems most likely that vibration 3, which is normally degenerate with vibration 9(a), has been altered in frequency so as to appear near to 1240 cm-1 in the spectra of the complexes, or it is possible that it is either an overtone (2 x 6a; 2 x 620=1240) or a combination band (6a + 6b ; 620+650=1270) which bas become activated. There are weak bands at 1350 and 1372 cm-1 in the spectrum of pyridine and these are often observed, although with weak intensity, in the spectrum of co-ordinated pyridine. The assignment of the bands between 1570 and 1650 cm-1 has offered some difficulty. In the spectrum of pyridine there are four bands at 1570(w), 1578(s), 1593(sh), and 1627 (w) cm-1. The 1593 cm-1 band is a combination band (1 + 6a) and is enhanced in intensity by Fermi resonance with vibration 8a.(1) We agree with GREENWOOD and WADE(7) in observing that this band either does not appear in the spectrum of co- ordinated pyridine or has become degenerate with the 1625 cm -1 band. The band at 1627 cm -1 in the spectrum of pyridine is also a combination band (6a + 12 or 1 + 6b); this band appears, although sometimes with reduced intensity, in the spec- trum of co-ordinated pyridine. The weak band at 1570 cm -1 in the spectrum of pyridine occurs in almost the same position in the spectra of co-ordinated pyridine but the strong 1578 em-1 band is shifted to 1600 cm-1 when pyridine is co-ordinated to a metal. GREENWOOD and WADE(7) record that this band has moved to an even higher wave number in the spectrum of pyridine-boron trichloride and in this compound it seems to have become degenerate with or obscured the 1627 cm-1 combination band. There are few systematic changes 'n the positions of the bands with changes in mass, electronegativity, or valency of the central atom or with changes in the other ligands. For the series of octahedral-bridged(14) Mpy,X, complexes there is a regular (Jo N. S. GILL and R. S. NYHOLM, T. Inorg. Nucl. Chem. 6 82 N. S. GILL et al.
TABLE 1. — INFRA-RED SPECTRA OF PYRIDINE AND PYRIDINE COMPLEXES* Assignment
1+6b Complex 1 +6a 8a 8b 19a 19b 14 6a+ or 1013 6a +12
1. Pyridine 1627 1593 1578 1570 1478 1436 1372 1350 2. Mnpy2Cl2 1632 1597 1573 1471 1442 1390 1362 1237
3. Mnpy2Br2 1631 1599 n.o. 1486 1442 n.o. 1358 1235
4. MnPY41.2 11626 1592 n.o. 1480 1442 1374 11.0. 1237
5. Fepy4(NCS)2 1632 1596 1569 1482 1440 11.0. 1355 1233
6. CopY2C12 1645 1604 1575 1487 1449 11.0. 1360 1242
7. Copy2Br2 v.b. 1602 17.0. 1481 1442 1370 11.0. 1237
8. Copy212 v.b. 1600 1562 1478 1442 1370 11.0. 1238
9. Copy2(NCS)2 1637 1600 1567 1486 1442 1375 n.o. v.b.
10. CoPY4(NCS)2 1633 1599 1572 1481 1441 1375 1353 1235
11. Nipy2C12 n.o. 1603 1573 1487 1447 1400 1360 1243
12. Nipy4(NCS)2 1633 1603 1570 1485 1440 n.o. 1356 1240
13. Nipy4Br2 n.o. 1600 n.o. 1483 1446 71.0. 11.0. 1237
14. Nipy4I2 n.o. 1599 n.o. 1485 1444 1374 11.0. 1240 1434
15. Cupy2C12 1638 1600 1570 1484 1447 1372 71.0. 1239
16. Cupy213r2 n.o. 1600 1568 1489 1449 1363 71.0. 1241 17. Cupy2(NCS)2 1632 1604 1572 1487 1447 1383 1353 1238
18. CupY4(BF4)2 1650 1601 1576 1487 1446 1362 n.o. 1238
19. Agpy213F4 1642 1601 1571 1480 1437 11.0. /2.0. n.o. 20. Znpy2Cl2 1660 1610 1575 1483 1452 1400 1360 1248 1378 21. Znpy2I2 n.o. 1602 n.o. 1482 1444 1380 1360 1245 22. ZnPY2(NCS)2 n.o. 1611 n.o. 1487 1447 1374 n.o. 1245
23. Cdpy2(NCS)2 1633 1598 1570 1483 1442 1363 n.o. 1235
24. CdP374(BF4)2 1625 1600 1572 1485 1443 1391 1357 1237 25. Hgpy2(BF4)2 1639 1615 1573 1490 1461 71.0. 11.0. 1252 * Assignments are based on the work of WILMSHURST and BERNSTEIN (3) using the notation of KLINE and TURKEVICH'ii n. o. not observed; v. b. very broad; sh shoulder. Thiocyanate and fluoroborate bands are not recorded. All figures in cm -1.
The infra—red spectra of pyridine complexes 83
Assignment
9a 15 18a 12 1 5 4 11 6b 6a 16b
1217 1145 1067 1031 991 942 747 700 650 601 403 1223 1152 1076 1005 750 691sh 625 417 685 1221 1165sh 1076 1038 1009 970 749 691sh 628 424 1150 688 419 1219 1170sh 1069 1036 1004 840 764 701 631 423 1213 1144 1064 754 698 622 418 1217sh 1146 1066 1038sh 1007 762 707 650 622 426 1213 1035 752 696 416 1222 1152 1082 1043 1013 757 695sh 628 427 1062 688 1214 1152 1064 1042 1011 752 685 640 418 743 680 1214 1151 1064 1042 1012 758 687 638 423 749 415 1217 1150 1072 1041 1011 753 692 627 422 418 1218sh 1148 1069 1043sh 1007 766 710 645 620 426 1214 1038 755 698 417 1220 1152 1082 1042 1017sh 755 684 632 437 1062 1013 1215 1151 1069 1042 1010 769 713 647 641 440 758 700 1223sh 1145 1068sh 1040 1008 773 703 624 427 1216 1073 767 698 758 1225 1168 1081 1040 1011 965 758 704 628 430 1215 1157 1074 750 694 417 1060 1219 1150 1077 1041 1015 946 767sh 692sh 641 440 1061 818 756 682 1223 1150 1075 1038 1015 930 751 680 640 439 1217 1152 1072 1042 1016 943 755 692sh 637 416 687 1222 BF4 absorbs here 762 698 633 433 741 1218 BF4 absorbs here 750 693 636 421 1221 1164 1071 1046 1017 887 762 697 637 417 1156sh 752 688 1219 1153 1067 1042 1014 751 690 635 420 1218 1154 1069 1047 1022sh 755 697 641 422 1214sh 1043sh 1017 752 687 410 1216 1149 1070 1036 1008 753 692 625 416 413 1220 Ea774 absorbs here 753 695 622 412 1228 Ea-74 absorbs here 758 686 640 425 84 N. S. GILL et al. trend in the position of the 625 and 420 cm -1 bands (Table 2). The frequencies of these bands increase with decrease in ionic radius of the metal; the latter factor is proportional to the polarizing power of the metal ion. The other bands show no trends at all. The great similarity in the spectra when pyridine is co-ordinated to a wide variety of metal ions and non-metals is in favour of a similar electronic density over the ring system in all these complexes since, for benzene, changes in
TABLE 2.—SHIFTS IN 420 AND 620 cm -1 BANDS WITH CHANGE IN CENTRAL METAL ION*
Metal Mn2 + Fez} Co' + 1 Ni2 + Cu2 + I Zn2 +
MPY2X2 octahedral Cl - 625 628 632 641 417 427 437 440 Br - 628 640 422 439
* All figures in cm-, the electron density in the aromatic ring produce wide differences in the observed spectra.(15) Such a constant electron level over the pyridine ring can result from back bonding from the metal atom and would be in agreement with the great stabil- ity of a, a'-dipyridyl and o-phenanthroline complexes which is believed to be due to similar back bonding.(16) These is no positive evidence for back bonding in the pyridine complexes(17) but we believe that the observed infra-red spectra are best explained on the bas's of this model. We have, rejected the alternative model in which the electron density is dispersed or increased by way of the metal ion from the other ligands by examining a series of complexes (Nos. 18, 19, 24, 25) in which pyridine is the only ligand. The spectra of these salts are once again similar to those of the other pyridine complexes. From a comparison with the values found for other metal-ligand bonds (MF in complex fluorides 400 cm-1,(18) M—C in cyanides 500 cm-1,1191 Pt —N in amines. 540 cm-1,(2()) Co—N in [Co(NH3)6]3+ 575 em-1 (21)) it would be expected that metal- nitrogen vibrations in pyridine complexes would occur well above 400 cm-1. We have not observed any bands which cannot be assigned to pyridine vibrations above 400 cm- and BscELLT,(5) from an empirical relationship involving the C—N bonds in the pyridine ring, has predicted that the M—N frequencies would occur in the range 150 to 250 cm-1. Double bonding of the type mentioned in the last paragraph
(") L. J. BELLAMY, J. Chem. Soc. 2818 (1955); R. D. KROSS, V. A. FASSELL and M. MARG OSHES, J. Amer. Chem. Soc. 78, 1332 (1956). (16) R. S. NYHOLM, J. Chem. Soc. 3245 (1951); F. BURSTALL and R. S. NYHOLM, Ibid. 3570 (1952). (") W. W. BRANDT, F. P. DWYER and E. C. GYARFAS, Chem. Rev. 56, 959 (1954). us> R. D. PEACOCK and D. W. A. SHARP, J. Chem. Soc. 2762 (1959). (10) V. CAGLIOTI, G. SARTORI and M. SCROCCO, J. Inorg. Nucl. Chem. 8, 87 (1958). (20) D. B. POWELL and N. SHEPPARD, J. Chem. Soc. 3089 (1959). (21) H. BLOCK, Trans. Faraday Soc. 55, 867 (1959). The infra-red spectra of pyridine complexes 85 would be expected to strengthen these bonds but, if present, it does not appear to strengthen them sufficiently to give infra-red peaks above 400 cm-1. Well defined splitting of the 420, 620, 702, 748, 987, 1029, 1067, 1144 and 1215 cm-1 bands has been observed for certain of the complexes. This splitting has been related to the overall configuration of the complex(12) but the splitting is as likely to be due to interaction between molecules in the unit cell, low lattice site symmetry for the complex, or slight rotations of the co-ordinated pyridine about the metal-nitrogen bond. The slight rotations are probably necessary to relieve steric strains in some complexes. In view of these other possible causes of splitting we consider that it is likely to be very difficult to relate splitting to the overall configura- tion of the complex.
Pyridinium salts The spectra of pyridinium salts are very different from those of other pyridine complexes. Although the ion retains the same (C„) symmetry as pyridine the presence of an extra hydrogen atom on the nitrogen will increase the number of vibrational modes. Details of the infra-red spectra of the complexes studied in the present work are given in Table 3. Pyridinium salts have symmetrical N—H stretching frequencies near 3200 cm-1 in non-hydrogen bonded salts and near 2800cm -1 in hydrogen-bonded salts.122) Below 2000 cm -1 the major differences between the spectra of pyridinium and pyridine derivatives are as follows. The bands at 1540, 1327, 1295 and 1250 cm -1 in the spectrum of pyridinium fluoroborate do not appear in the spectra of pyridine complexes. The intensities of these bands are much reduced on N-deuteration (Fig. 1) and they must result from N—H deformation and combination modes involving N—H deformation frequencies. The wide splitting of these bands is similar to that observed for the symmetrical stretching frequencies.(22) The strong band at 1640 cm -1 in the spectrum of pyridinium fluoroborate does not alter in intensity on deuteration but the 1600 cm -1 band is split and reduced in intensity and must result from an N—H mode. We conclude that the 1640 cm -1 band results from a vibration similar to that which produces a band at 1578 cm -1 in pyridine (8a) and ca. 1600 cm -1 in the spectra of pyridine-metal complexes. GREENWOOD and WADE record that this band occurs at 1624 cm -1 when pyridine is joined to the light acceptor atom boron in pyridine-boron trichloride. The only major band that appears on deuteration is that at 1306 cm -1 which may be the same band as that which occurs at 1600 cm -1 in the undeuterated salts (vH/vr, = 1.23). All of the other N—H bands would be expected to move under the fluoroborate peaks on deuteration. The band at 1490 cm -1 in the spectra of pyridinium salts is much stronger than the 1480 cm -1 band in the spectra of other pyridine complexes but this appears to be only an enhancement in the intensity of the 1480 cm -1 band in pyridine complexes. Between 1100 and 650 cm -1 the spectra of pyridinium salts are more complicated than those of pyridine complexes. This is to be expected since this is the region of the spectrum
(22) R. H. NUTTALL, D. W. A. SHARP and T. C. WADDINGTON. J, Chem. Soc. in print 00
TABLE 3. — INFRA-RED SPECTRA OF PYRIDINIUM SALTS* pyH BF, 1625 — 1597 1525 1477 1367 1327 1250 BF4 absorbs here 744 677 608
1281 1. .1%
(py14),MnCI, 1631 1615sh 1603 1528 1495 1480 1418 1365 1325 1230 1187 1151 1073 1045 1025 969 888 845 750 668 606 528 413 'S 517 710
(PY1-1)2G:C!4 1636 .615sh 1603 1528 1478 1422 1364 1323 1238 1188 1153 1074 1047 1025 1005 930 852 750 668 605 505 414 a
981 0 l 905 1 888 pyD BF, 1638 1613w 1590w 1540w 1492 1460 1375 1335w 1308s 1292w 1258w BF4 absorbs here 800 769 756 677 608
* All figures in cm-. The infra-red spectra of pyridine complexes 87
where C—H out-of-plane and in-plane frequencies occur (1,2,3) in the spectra of pyridine and benzene derivatives, and the presence of an N—H bond will increase the number of vibrations which will occur in this region. A more detailed assignment for the bands observed in the spectra of pyridinium salts is not possible at this stage but, in general, with the exception of the N—H vibrations, we believe that the various groups of vibration—ring stretching and breathing, C—H out-of-plane, C—H in plane— occur with very similar frequencies in the spectra of pyridine and of pyridinium derivatives.
EXPERIMENTAL All complexes were prepared by standard literature methods and were analysed for metal, car- bon, hydrogen and nitrogen. Pyridinium fluoroborate was prepared by hydrolysis of pyridine- boron trifluoride in 95 per cent ethanol 23). N-deuteration was effected by recrystallization of pyrid- inium fluoroborate from D20. Since reduction of C-halogeno pyridines with zinc in D2SO4 under much more vigorous conditions gives specific replacement of halogen by deuterium" it is unlikely that the present deuteration will affect the C-hydrogen atoms. Infra-red spectra were measured in Nujol or hexachlorobutadiene mulls using a Perkin—Elmer Model 21 spectrophotometer fitted with rock-salt or potassium-bromide optics. Acknowledgements—We thank Dr. N. N. GREENWOOD for allowing us to see the manuscript of his paper prior to publication. One of us (D. E. S.) thanks C. S. I. R. 0. for leave of absence and a maintenance grant.
(23) P. A. VAN' DER MEULEN and H. A. HELLER, J. Amer. Chem. Soc. 54, 4404 (1932). (20 B. BAK, L. HANSEN and J. RASTRUP-ANDERSEN, J. Chem. Phys. 22, 2013 (1954). Speetrochimica Acta, 1961, Vol. 17, pp. 947 to 952. Pergamon Press Ltd. Printed in Northern Ireland
The infra-red spectra of some stable diazonium salts
R. H. NITTTALL, E. R. ROBERTS and D. W. A. SHARP Inorganic Chemistry Research Laboratories, Imperial College, London, S.W.7
(Received 10 March 1961)
Abstract—A study of the infra-red spectra of a number of diazonium salts shows that there is bonding from the anion to the cation in some of the salts. It is impossible to make a definite correlation between the N—N stretching frequency and electronic effects due to other sub- stituents in the aromatic ring. The infra-red spectra of the p-dialkylaminobenzenediazonium and p-aminobenzenediazonium cations have been interpreted as favouring structures in which the positive charge is transferred from the diazonium group to the amine grouping.
Introduction THE infra-red spectra of diazonium salts show a characteristic band between 2100 and 2300 cm-' which has been attributed to the N—N stretching frequency [1-3]. Previous workers [1, 2] concluded that, for a given cation, the position of this band is independent of the anion but one of the present authors [4] suggested that, since the infra-red spectra of fluoroborates showed a peak at higher frequencies than the spectra of other salts, the fluoroborates were ionic and that there was some covalent bonding between anion and cation in the other salts. Recent Russian work [3] has supported this conclusion. Experimental Solutions of diazonium chlorides were prepared [5] by diazotising a slurry of the amine hydrochloride in 50 per cent hydrochloric acid with the requisite quantity of sodium nitrite. Chlorometallates were formed by adding the solution of a dia- zonium chloride to a solution of the appropriate metal chloride in hydrochloric acid. Fluoroborates were precipitated by addition of the solution of diazonium chloride to 40 per cent aqueous fluoroboric acid. Hexafluorophosphates were precipitated from an aqueous solution of potassium hexafluorophosphate prepared by the method of PALMER [6]. p-Aminobenzene diazonium chloride was prepared by addition of solid sodium nitrite to a slurry of the diamine hydrochloride until the hydrochloride was in solution; the fluoroborate was precipitated from solution. Infra-red spectra were obtained using a Perkin-Elmer model 21 spectrophotom- eter fitted with rock-salt or fluorite optics. The spectra of the crystalline solids were obtained as mulls in Nujol or hexachlorobutadiene; the spectra of aqueous solutions were obtained from films pressed between silver chloride plates.
[1] M. ARONEY, R. J. W. LE FEVRE and R. L. WERNER, J. Chem. Soc. 276 (1955). [2] K. B. WRETSEL, G. F. HAWKINS and F. E. JOHNSON, J. Am. Chem. Soc. 78, 3360 (1956). [3] L. A. KAZITSYNA, 0. A. REUTOV and Z. F. BUCHKOVSKII, Russ. J. _Phys. Chem. (English translation) 34, 404 (1960). [4] D. W. A. SHARP, Advances in Fluorine Chem. 1, 68, (1960). [5] K. H. SAUNDERS, The Aromatic Diazo Compounds (2nd Ed.). Arnold, London (1949). [6] W. G. PALMER, Experimental Inorganic Chemistry. Cambridge University Press (1954).
947 R. H. NUTTALL, E. R. ROBERTS and D. W. A. SRARP
Table 1. N—N stretching frequencies (cm-1)
Nitromethane Aqueous lull ➢ solution solution
Benzenediazonitun BF4- 2297 2280 2280 PF6- 2290 2280 n.s. HgC142- 2265 2275 n.s. ZnC142- 2277 d 2280 CdC142- 2273 2283 2285 SnC162- 2276 2288 n.s. Fe(CN)63- 2278 n.s. 2283 BPh4- 2282 d n.s. FeC14 2262 2270 2260 (benzene diazonium chloroferrate dissolved in acetone absorbs at 2277 cm-1) m-Toluene diazonium BF4- 2294 2274 2280 PF6- 2291 2275 n.s. 1=4.042- 2260 2273 n.s. ZnO142- 2268 2272 2275 CdC142- 2272 2270 2270 SnC162- 2269 2263 n.s. Fe(CN)63- 2266 n.s. 2275 o-Toluene diazonium BF4 2290 2271 2273 PF6- 2289 2275 n.s. HgC142- 2253 2260 n.s. p-Toluene diazonium 15F4 2299 2275 2276 PF6- 2287 2277 n.s. HgC142- 2275 2272 n.s. o-Nitrobenzene diazonium 13F4 2295 2283 2295 PF6- 2293 d n.s. HgC142- 2280 2280 n.s. m-Nitrobenzene diazonium BF,- 2301 2298 n.s. PF6- 2305 d n.s. HgC142- 2285 2285 11.S. p-Nitrobenzene diazonium 13F4- 2321 2299 n.s. PF6- 2310 n.s. 1-1gC142- 2281 2281 n.s. o-Trifluoromethylbenzene diazonium BF4- 2295 2272 n.s. PF6- 2287 2272 n.s. HgC142- 2272 n.s. m-Trifluoromethylbenzene diazonium BF4- 2299 n.s. PF6- 2298 n.s. HgC142- 2281 d n.s.
948 The infra-red spectra of some stable diazonium salts
Table 1—continued
Nitromethane Aqueous Mull solution solution
p-Methoxybenzene diazonium BFr 2265 2252 n.s. HgC142— 2240 2255 n.s. p-Dimethylaminobenzene diazonium BF4 2268 w 2254 w n.s. 2181 s 2177 s PF 6- 2252 w 2255 w n.s. 2178 s 2175 s CdC142— 2250 w 2260 w n.s. 2168 s 2175 s HgC142— 2246 w 2255 w n.s. 2170 s 2175 s p-Diethylaminobenzene diazonium BF4— 2260 w 2252 w n.s. 2173 s 2168 s PF6— 2250 w 2259 w n.s. 2170 s 2163 s CdC142— 2251 w n.s. 2175 s 2169 s HgC142— 2248 w 2256 w n.s. 2170 s 2162 s p-Aminobenzene diazonium BF4— 2170 n.s.
n.s., not sufficiently soluble for spectra to be measured. d, decomposes s, strong. w, weak. Discussion From visual observation, simple diazonium salts fall into two classes, those in which the cation is colourless and those derived from p-phenylenediamine and p-N:N-dialkylamino anilines in which the cation is coloured. These two types will be considered separately. The N—N stretching frequencies of a variety of diazonium salts are given in Table 1. Spectra were measured on mulls of the crystalline solids and, wherever possible, on solutions in nitromethane and water. The spectra of many of these salts have been studied by KAZITSYNA et al. [3] and, in general, our observations are similar to theirs. We did not observe multiple peaks for any of the colourless salts. Occasionally a multiple peak would occur in the spectrum of a mull but this always disappeared on regrinding the mull. We therefore attribute these peaks to some form of interaction in the crystalline solid. The frequencies which we observed are all higher than those observed by the Russian workers. Our spectra were cali- brated by means of a standard air spectrum and we cannot explain this discrepancy. We agree with the Russian workers in explaining the lower frequencies observed in some salts as being due to partial covalent bonding between the diazonium cation and the anion, this bonding being by direct interaction between the lone pairs of the halide ions and the vacant orbital of the diazonium group. This interaction
949 R. R. NUTTALL, E. R. ROBERTS and D. W. A. SHARP
does not occur with the salts of the very strong complex fluoro acids. Interaction between the Tr-bond of the N—N group and the metal atom of the complex anion is also possible although neither the N=N nor bonds have yet been shown to form 7T-bond complexes. Any relation between the N—N frequency and the overall crystal field at the site of the N—N bond is ruled out by the lack of a dependence of this frequency on the size of the anion, and, furthermore, by an approximate constancy of the N—N frequency in solvents of different dielectric constant (e.g. benzenediazonium chloroferrate absorbs at 2260 cm-1 in water, at 2277 cm-1 in acetone, and at 2270 cm-1 in nitromethane, dielectric constants 81.5, 21.4, and 24.7, respectively [7]. The division of diazonium salts into two types of the basis of chemical reactivity [5, 8] follows the pattern based on the N—N stretching frequencies of the crystal- line solids. A high N—N frequency is associated with decomposition by a hetero- lytic mechanism, whilst partial covalent character in the solid state is associated with homolytic dissociation. WHETSEL et at. [2] attempted to relate the N—N stretching frequency to the electonic effects of other ring substituents—as measured by the Hammett function. Their observations, which were made on the solids, showed trends in the right direction and our solution measurements again show these trends but the relation does not hold as well as is observed for the C=0 [9] and N—H [10] groups. Such observations should ideally be made on solutions and a complicating factor with cations such as the diazonium ions is that solvation must be much greater than for uncharged molecules. Since the electronic effects of solvation may well be in an opposite direction to those of the substituent groups an overall complexity is not unexpected. The infra-red spectra of diazonium salts over the range from 650 to 1700 cm-1 (Table 2) are similar to those of the appropriately substituted aromatic nuclei save that the aromatic skeletal stretching frequency near 1580 cm-1 has a much greater intensity in the diazonium salts. Association of an aromatic system with a posi- tively charged, electron-deficient, entity generally has the effect of intensifying this vibration [11]. Diazonium salts of p-dialkylaminoanilines show a number of features which are different from other diazonium salts. WHETSEL et at. [2] observed two infra-red bands in the N—N stretching region and the present work has confirmed this observation both in the solid state and in solution. The infra-red spectra of p-dialkylaminodiazonium salts show an intense band near 1100 cm-1 which is not observed in the spectra of other diazonium salts. The ultra-violet spectra of these salts are also very different from those of other diazonium salts, there being two strong absorption bands at about 3800 A and 2500 A compared with the single band observed in the spectra of other diazonium salts [12, 13].
[7] International Critical Tables (1st Ed.) 6, 82 (1929). [8] A. N. NESMEYANOV, L. G. IYIARA.nova and T. P. TOLSTAYA, Tetrahedron 1, 145 (1957). [9] N. FusoN, M.-L. JOSIEN and E. M. SHOLTON, J. Am. Chem. Soc. 76, 2526 (1954). [10] S. CALIFANO and R. MocciA, Gazzetta 86, 1014 (1956). [11] D. W. A. SHARP and N. SHEPPARD, J. Chem. Soc. 674 (1957). [12] L. C. ANDERSON and J. W. STEDLY, J. Am. Chem. Soc. 76, 5144 (1954). [13] A. Wour, Bull. soc. chim. France 1319 (1939).
950 The infra-red spectra of some stable diazonium salts
Table 2. Infra-red spectra 1600-650 cm--1 (figures in cm-1)
C6H5ON C611501 C6H5N2BF4
757 740 750 972 965 960 985 975 1000 1003 1026 1026 1015 1070 1068 1075 1096 1100 1163 1157 1178 1085 1125 1192 1174 1180 1286 1271 1295 1332 1326 1315 1445 1445 1414 1489 1477 1465 1579 1580 1540 1597 1580 1570 vs
p-CH3C6114N2BF4 p-(CH 3)2NC6H4N 213F 4
817 825 s 975 93.0 983 1093 1168 1115 vs 1192 1125 1290 1290 1310 1320 1365 1385 1390 1400 1437 1450 1480 1590 vs 1590 vs
The predominant electronic structure of an aryl diazonium salt is generally considered [2] to be (I) and on the basis of ultra-violet and infra-red studies of the N—N stretching frequencies it has been considered that a quinonoid structure (II) makes a finite contribution to the structure of the p-dialkylaminodiazonium cations [12]. We consider that our results support this formulation of charge trans- fer and suggest further that the structure (III), where complete charge transfer
R2N— —N=N:R2N—
(I)
R2N—
951 R. H. NE TTALL, E. R. ROBERTS and D. W. A. SHARP from the diazonium group to the amino group has taken place, is of major import- ance in these cations. The infra-red spectra of quinones are completely distinct froM those of benzenoid aromatics [14] and our spectra show the presence of an aromatic skeleton in these compounds. We assign the intense infra-red band at about 1100 cm-1. to C N vibrations involving the dialkylamino group—these vibrations normally occur [15] between 1200 and 1000 cm--1- and the charge would be expected to increase the intensity of absorption. The two bands in the N—N stretching region of the spectra of these salts are of unequal intensity; that at 2170 cm-1 is the most intense and we assign this to the N—N stretch, the fre- quency being lowered from that observed in other diazonium salts because of an increase in the contribution of structures (II) and (III). The other, weaker, band at ,2250 cm--1 appears to be the first overtone of the 1120 cm-1 band. p-Amino- benzene diazonium fluoroborate shows one absorption in this region and the overtone is not observed. The frequency, 2170 cm1 , is low and this indicates that charge transfer is taking place in this cation also. The N—N frequency in the spectra of these salts does not depend markedly upon the anion and we conclude that with the charge transfer the interaction between the N--N group and the complex anions has been reduced.
Acknowledgement—We thank Fisons Ltd. and Imperial Chemical Industries Ltd. for financial support.
[14] P. YATES, M. I. ARDAO and L. F. FIESER, J. Am. Chem. Soc. 78, 650 (1956). [15] L. J. BELLAMY, The Infrared Spectra of Complex Molecules (2nd Ed.). Methuen, London (1958)
952