This dissertation has been microfilmed exactly as received 66-1851

WALTHER, James Fletcher, 1938- PREPARATION AND PROPERTIES OF T E TRAC HLOROPHOSPHONIUM TETRACHLORODIFLUOROPHOSPHATE.

The Ohio State University, Ph.D., 1965 Chemistry, inorganic

University Microfilms, Inc., Ann Arbor, Michigan PREPARATION AND PROPERTIES OP TETRACHLOROPHOSPHONIUH

TKTRACHLORGDIFLUQROPHOSPHATE

DISSERTATION

Presented In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy In the Graduate School of The Ohio State University

BSP

James Fletcher Walther, B.A.* M.Sc*

The Ohio State University 1965

Department of Chemistry ACKHCMI2DGMENTS

I wish to express ngr sincerest gratitude to my adviser*

Or. Sheldon Q. Shore* Tor suggesting this very interesting and perplexing research problem.

I am dearly grateful to ay wife* Kay* for her faithful assistance in preparing this dissertation and for her enduring

interest and enthusiasm in my work.

Also* I want to express my appreciation to the Chemistry

Department of The Ohio State University for its financial assist­

ance through various teaching assistantships and summer fellow­

ships.

ii VITA

August *** 1 9 3 8 ...... Born* St* Louis* Missouri i960 ••••••••• ••• B*A** Central Methodist College* Fayette* Missouri

1960-1963 ...... Teaching Assistant* Department of Chemistry* The Ohio State University* Columbus* Ohio

1963-196**...... Teaching Assistant* Graduate level* Inorganic Division* Department of Chemistry The Ohio State University* Columbus* (Alio

196*4—1965 ...... Technical Assistant* Department of Chemistry* The Ohio State University* Columbus* Ohio

iii CONTENTS Page

ACKNOWLEDGMENTS ...... 11

VITA ...... Ill

TABLES...... v

ILLUSTRATIONS . . , ...... vi

I. INTRODUCTION ...... 1

II. STATEMENT OF PROBLEM ...... 10

III. EXPERIMENTAL ...... 15

A. Apparatus and Experimental Technique ...... 15 B. Reagents 33 C. Chemical Analysis ...... 4 0 D. Spectroscopic Analysis ...... 45 E. Conductance Measurements ...... 50

IV. DISCUSSION OF RESULTS ...... 52

A. Preparation of Ionic PCl^F by ftrrolysis of [PClZf]+[PF6] - ...... 52

B. Preparation of Ionic PCl^F by Direct Fluorination of PCl

V. S U M M A R Y ...... 100

BIBLIOGRAPHY ...... 103

iv TABLES

Table Page

1* Physical Properties of Phosphorus(V) Chlorofluorid.es . 9

2. Chemical Analysis Data for Ionic PClnF Made from Pyrolysis of [PClit3+[PF^]” ...... 5^

3* Infrared Absorption Frequencies of Ionic PCI4 F • • • 63

1*. Infrared Absorption Frequencies of Pure [p c i ^]+ [p f 6 ]- ...... 61*

5 . X-ray Powder Diffraction,Data for Ionic PCl^F and Pure [FC1^3 [ E F ^ ] * ...... 76

6, Apparent Molecular Weight of Ionic PCl^F in Nitrobenzene by Freezing-Point D e p r e s s i o n ...... 82

?• Conductance Data For Ionic PCl^F in Nitrobenzene Solutions ...... 81*

8* Infrared Absorption Frequencies of Ionic PCljifF in Nitrobenzene ...... 87

v ILLUSTRATIONS

Figure Page

1* Apparatus for Pyrolyzing 18

2. Apparatus for Direct Fluorination of PCl^ with AsF3 ...... 22

3# Powder-Dropping Device ...... 28

4. Assembled Cryoscopy Apparatus ...... 30

5* Apparatus for Filtering Aqueous AgF Solutions . . . . 3?

6 . Apparatus for Evaporating Aqueous AgF Solutions • . • 38

7* Apparatus for Base Hydrolysis of Solid Phosphorus Halides ...... 4-1

8 . Infrared Cell for Corrosive Liquids ...... 46

9* Apparatus for Preparing N.M.R. Sample Solutions . . . 49

10. Infrared Spectrum of Ionic PCI. F by Fluorination of PCl^ with AsF^ in ASCI3 4 ...... 60

11# Infrared Spectrum of Ionic PCl^F by Pyrolysis of [PC1^]+[PF6]- Sample 2. 60

12. Infrared Spectrum of Pure [PC1^]+[PF^]" (Pressed Pellet) ...... 6l

13* Infrared Spectrum of Pure [PCI. ]+ [PFA 3" (Nujol M u l l ) ...... 61

14* Infrared Spectrum of Phosphorus Pentachloride t [PClif]T;PCl6]“ (Pressed Pellet) ...... 62

15* Infrared Spectrum of Phosphorus Pentachloride1 [PC14 ]+[PC16]" (Nujol Mull) ...... 62

16 . Infrared Spectrum of POCl^ in N u j o l ...... 66

17. Infrared Spectrum of Phosphorus Pentachloride Adulterated with P O C l ^ ...... 66

vi ILLUSTRATIONS (Cont'd.)

Figure

18. Infrared Spectrum of Sublimation Condensate from Purification of Ionic PCl^F ...... • • • •

19* Infrared Spectrum of Ionic PCl^F Prepared by Fluorination of PCl^ with AsF- in Nonpolar Solvents ...... •

20* Graphical Comparison of 6-Values of Ionic PCl^F Samples with Values Reported by Kolditz ......

21. Infrared Spectrum of a Saturated Solution of Ionic PCljjF in Nitrobenzene ......

22. Infrared Spectrum of a Saturated Solution of Ionic FCI4.F in Acetonitrile ......

2 3 . Infrared Spectra of a Solution of Ionic PCl^F in Anhydrous Acetic Acid......

2k. Infrared Spectra of Pure Anhydrous Acetic Acid . • ♦

2$. Infrared Spectrum of Acetyl Chloride in Anhydrous Acetic Acid ......

26. Infrared Spectrum of Product from Reaction between AgF and PCl^ in CHjjC^ ......

27* Infrared Spectrum of Product from Reaction between AgF and PCl^ in Benzene I. INTRODUCTION

Phosphorus is one of several non-metalllc elements known to exhibit a coordination number of six toward covalently bonded sub­ stituents • An excellent example of this kind of compound is phos­ phorus pentachloride. This material in its normal crystalline state consists of the complex ions [PCl^J^PCl^]” arranged in a distorted cesium chloride type lattice (1) • For the purpose of later discussions» the dualistic nature of this material should be emphasised.

In contrast to the ionic character of its solid state> the composition of its liquid (2) and gaseous (3) states has been interpreted in terms of PCl^ trigonal bipyramidal molecules. But* the point of interest to this discussion is the octahedral hexachlorophosphate anion in which phosphorus had increased its coordination sphere to accomodate six substituents. Although a wide variety of compounds are known in which phosphorus has a coordination number of six» the actual number of compounds is not very large. A convenient approach to discussing these materials is to consider them all derivatives of phosphorus pentachloride. The various compounds can be placed in one of several classes according to their basic structures.

Class A .— Addition complexes of molecular phosphorus penta­ chloride or its molecular derivatives— L #FClaX^_t.

Class B .— Replacement of the complex [PC1^]+ cation of ionic phosphorus pentachloride by another cation— M’^tPCl^]*'.

1 2

Class C .— Replacement of the [FCl^]* cation by another cation and successive replacement of chlorine atoms on the complex [P&^l anion by other atoms or radicals— ^ [ P C l ^ X ^ ^ 3^.

Class D .— Successive replacement of the chlorine atoms on the

[PC16]“ anion by other atoms or radicals— [PCl^]+[PClaX ^ ^ ] “ .

Class A compounds include a rather extensive list of addition complexes of phosphorus pentafluoride. Phosphorus pentafluorlde has been shewn to be a strong Levis acid forming one-to-one complexes with amines* ethers* nitriles* sulfoxides* and many other organic bases (4-)* Pyridine complexes of the molecular species PCl^ (5)*

PCl^F* PCl^Fg* and PCI2F^ (6) have been described* The older of

Levis acid strengths toward pyridine is as expected; FCl^Fg > FCl^F >

PC15 (?)* The PClgFj complex has the formula 3py*2PCl2F^* The struc­ ture of this complex is hard to visualize unless a seven-coordinate phosphorus species is assumed* A carbon tetrachloride complex of phosphorus pentachloride having the formula CCl^*2PC1^ has been re­ ported (8). The complex vas formed as a precipitate upon cooling a saturated solution of phosphorus pentachloride in carbon tetrachloride*

This complex might involve ionic phosphorus pentachloride instead of the molecular form* such as (CC1^*FC1^ 3+[PC1^ ] % in vhich cationic phosphorus acts as an acceptor for non-bonded p electrons on the carbon tetrachloride molecule* A biphosphorus ccmplex having a bridge type bond involving the chlorines on carbon tetrachloride is another pos­ sibility. The latter is more likely because no evidence has been found for ion pair formation of phosphorus pentachloride in carbon tetrachloride solutions (9)* The species PBr^*2CCl^ vas also 3 reported (8*10). Two other species * PCl^BryCCl^ and FCl^Br^#2CCl^> were isolated from PCl^-Br^-CCl^ systems (11). Whether or not six coordinate phosphorus is involved is an open question* A tensiomstric study in brcmobenzene (12) showed formation of [(CH-^Pjg'FCl^ which was isolated as a white solid* Tertiary amines have not been ob­ served to form addition complexes with phosphorus pentachloride*

Instead* dehydrogenation occurs with resultant formation of the amine hydrochloride (12*13)* Stable pyridine and dimethylfoxmamide com­ plexes of pheoyltetrafluorophosphorane* * have been reported

(14-); but corresponding alkyl phosphorane complexes were too highly dissociated for isolation* The complex anion methylpentafluorophos- phate* [CH^PFcj]” * was described as existing in dimethylsulfoxide; but it was not isolated in a stable salt*

Class B compounds are salts which contain the hexachlorophosphate anion* [PCl^]” . Possibly nine of these compounds are known* including ionic phosphorus pentachloride* and probably there will be very few more ever synthesized* The hexachlorophosphate complex is extremely unstable* as exemplified by its failure to form isolable salts with alkali metal cations and tetramethyl- and tetraethyl nwmonium cations

(15)* Obviously* an important factor involved in the stability of a

[PClg]~ salt is the size or charge density of the cation* However* a more important factor is the combined stabilities of the starting materials relative to the Intended [PCI,]” salt* In the present in- © vestigation it was found that the triphenylcarbonium ion will not form

an Isolable salt with hexachlorophosphate even though it forms stable

salts with hexachloroantlmonate and other similar ions (16)* This * failure definitely expresses the weakness of an octahedral phos­ phorus-chlorine bond since that bond was easily compensated by the weak triphenylcarbon-chlorine bond.

The first member of Class B to be prepared has the formula

AsP&io* In 1873 A. W. Cronander (17) reported that a mixture of phosphorus pentachloride and arsenic trichloride absorbed chlorine gas. Hydrolysis of the resulting product produced arsenic acid. He then reasoned that a compound of formula PCl^»AsCl^ was formed.

Eight decades later V. Gutman (18) confirmed the formation of As FSI^q and assumed its structure to be CAaCl^]+[PCl^]“ . His assumption was based upon observations indicating that arsenic pentachloride does not exist* which in turn indicates that the maximum coordination number of arsenic toward chlorine must be less than five. Also* he observed the formation of AsSbCl^Q. This added support to his argu­ ment because the [SbCl^]” anion is known to be very stable. D. S.

Payne (19) has extended Gutman’s reasoning by observing that promotion energy data suggests that the hd orbitals of arsenic are separated from the ifs and orbitals by an appreciably larger energy difference than the corresponding bonding orbitals of either phosphorus or anti­ mony* This larger energy difference probably accounts for the apparent 3 inability of arsenic to form dsp hybrid bonds.

Two other compounds which contain the hexachlorophosphate anion have been described by M. Becke-Qoehring et al. (20). These materials* formulated P^NCl^ and P ^ ^ C l ^ , were isolated as stable

Intermediates involved in the formation of phosphonitrilic chlorides from reactions between phosphorus pentachloride and ajmnonium chloride. On the basis of phosphorus-31 nuclear magnetic resonance* cryoscopy* and solution conductivity* the structures of the two compounds were characterized to be [Ciy^N-PCl^^PC^ ]" and [Cl^P^N-PCl^N-PCl^]*

[POlg]** respectively*

Another possible member of Class B has the formula C ^ H ^ C l ^ P and was formed during a preparation of t ropy Hu m chloride fay reac­ tion of cycloheptatriene with phosphorus pentachloride (21)* I* R.

Beattie and M. Webster (22) reported that this material had an in­ frared absorption characteristic of the hexachlorophosphate anion in both a Nujol mull and an acetonltrile solution. However* their data showed that a tropylium ion (C^H^+ ) in the compound had an ab­ sorption band at the same frequency where tetrachloro-phosphonium*

[PClif)+ , is known to absorb* Therefore* they could not say that the

CPC1^]+ cation was absent from the compound*

Several other compounds which contain the hexachlorophosphate anion have been reported* They have the general formula [RCClg’FCl^]*

[PC1^]“ * where R is a methyl* ethyl* or propyl group (23)* A similar compound having the formula C^H^P(CC1^)C1^*PC1^ (2h) is probably also a hexachlorophosphate salt.

Class C compounds have the general formula where

M can be any cation except tetrachlozophosphonium* [PC1^]+ . This series includes a long list of salts all having in common the well known octahedral hexafluorophosphate anion* The [PF^]"* ion was probably the first recognized example of octahedral phosphorus. In

1930 W*. Lange (23) reported preparative procedures for a large number of alkali metal and coordinated transition metal salts containing 6 this anion. Among other hexafluorophosphate salts is the interesting compound [PBr^]+[PF^] (26). A complex salt derived from fluorophos- phate and having the formula CC^H^PF(N(CH^)2)2 3 has been described (27).

All of the Class C compounds mentioned above are hexafluorophos­ phate salts and are not necessarily considered derivatives of ionic phosphorus pentachloride. However» one can visualise numerous struc­ tures* some of which will undoubtedly be synthesized. For example* mixed chlorofluoro-anions of octahedral phosphorus would probably be stable. Ten distinct anions from [PC1^]“ to [PF^]", including cis and trans isomers * are reasonable possibilities.

Class D compounds* designated by the general formula [PC1^]+

[PCl^X^^] , font the smallest series in this entire discussion.

Only three compounds have been imported in the literature. L. Kol- ditz (28) prepared the first member of this series* [PC1^]+[PF^]”» by treating phosphorus pentachloride with arsenic trifluoride.

Pyrolysis of this material in arsenic trichloride produced a new compound which Kolditz formulated as [PC1^]*F~ (29). Arguments can be raised concerning the correctness of Kolditz' formulation and conse­ quently this compound became the subject of the present investigation.

Another compound* [PC1^]+[PC1^F]" (30)» was prepared by pyrolyzing

[PC1^]+[PF^]“ in carbon tetrachloride. The third member of Class D» +• M [PCl^] [PCl,jBr] * was prepared by Kolditz (31) by treating phosphorus trichloride with liquid bromine in arsenic trichloride.

Two additional members of this class have been investigated in this laboratory. They probably have the formulas [PC1^]+[PC1^CN]” 7 and [PC1^]'iTPC12|(C)J)2]“; however their characterization was never completed (32).

Probably very few compounds which are known to contain hexa- coordinated phosphorus have been omitted from this discussion# Ex­ cluding the hexafluorophosphate salts and phosphorus pentafluorlde addition complexes» the list is quite short. However, one can extra­ polate from this list and imagine a multitude of structures based upon octahedral phosphorus. Even though many of these postulated materials probably will never be prepared or isolated* the existence of a very interesting and challenging area of chemistry cannot be denied. For instance* in Class D» that is* compounds of general formula [FCl^]*

[PClaX6_aJ~» only three compounds have been reported. And yet* if

fluorine derivatives alone are considered* ten compounds* including cis and trans isomers of species such as [PCl^Fg]”" and [PCl^Fj]"* can be formulated. If fluorine derivatives of the [PC1^]+ cation

are considered also* a total of fifty members can be formulated. If

three different halogens are considered simultaneously* then 810 com­

binations are possible* some of which are optical isomers. Class A

compounds* that is* molecular addition complexes* provide another

series of interesting possibilities. Chelation of phosphorus halides with simple ligands* such as blpyridyl* should be investigated. Syn­

theses of bis-chelate complexes should produce stereoisomers. The

chemist is well equipped with techniques for handling moisture sensi­

tive systems t and one should expect this area of chemistry to became

more popular in the academic laboratory.

The following sections of this thesis are concerned with a study of an ionic compound having the composition PCl^F. Evidence will be presented which Indicates that the compound contains hexacoordinated phosphorus. This compound was discovered by Lothar Kolditz (29) as a pyrolysis product of tetrachlorophosphonium hexafluorophosphate [PC1^]+

[PF^]"» suspended in arsenic trichloride. Conductance measurements of acetonitrile solutions indicated that the new compound was ionic.

Cryoscopy measurements of acetic acid solutions gave apparent molec­ ular weights of less than half the calculated molecular weights of

PCl^F. Consequently, Kolditz assumed that the material was tetrachloro- phosphonium fluoride* [PC1^]*F • Another compound, discovered much earlier (33)* also has the composition PCl^F. However, this material is a molecular modification being a liquid at room temperature • Kolditz demonstrated (3*0 that an isomerization process occurs between the molecular and ionic forms of PCl^F. The activation energy of the conver­ sion process was reported as 10.6 kcal./mole with the ionic form being more stable. Fusion of the ionic solid produced the molecular modifi­ cation accompanied by slight disproportionation into phosphorus penta­ chloride and tetrachlorophosphonium hexafluorophosphate. Upon cooling, the fused mass converted back to a solid only after prolonged standing.

The PCl^F system is analogous to that of phosphorus pentachloride which was mentioned previously. Tetrachlorophosphonium hexafluorophosphate has a molecular modification also (3 5 *3 6 )* Th« remaining molecular forms in the PCltj^F^ series are known, but their corresponding ionic modifi­ cations have not been discovered. A sunnary of the known phosphorus (V) chlorofluorides and their physical properties is given in Table 1.

Hie molecular forms have been shown to exist as trigonal bipyramldal molecules with fluorine atoms occupying apical positions on the molecules at low temperatures (37*38). Near room temperature rapid intramolecular 9 TABLE 1

PHYSICAL PROPERTIES OF PHOSPHORUS(V) CHLOROFLUORIDES

Ionic form Molecular form Compound m.p. Compound b.p. m.p. [PC14 ]+ [PC16 ]- (1*0) subl. 162° p c i 5 Liq. under p. m.p. l6?°under p. or as vapor

[PCl^l+CFC^F]- (30) d. 110° into — — PC15 + PCl^F

[ P C 1 ^ ] V (2 9 ) subl. 175° PC14 F (31*) 67° -63° m.p. 177°under p.

— PC13F2 (7) -63°

[PCl4 ]+[PF6r (28) subl. 135°» d. PC12F3 (36) 7.1° -125 to -130° PC1F4 (37) -1*3.1*° -132° — — PF5 (1*1) -81*. 5° -93.7° exchange of halogen positions occurs (7 >39)* An interesting observation

is that the boiling points of the molecular compounds PCl^F* PClgF^* and

PCIF^ fall on a straight line in a plot of boiling points versus molec­

ular weight. This plot would indicate that the unknown boiling point

of P C l ^ might be 36°.

Nearly a decade has passed since the original publication of the

preparation and properties of ionic PCl^F* and no further studies of

this material have been reported. Either the original work was con­

sidered complete* requiring no additional investigation*, or the PCl^F

system was found to be more complex and confusing than what was indi­

cated in the original publication. Since arguments can be raised con­

cerning the conclusions drawn in the original publication* the present

investigation was undertaken to redetermine the nature of the PCl^F

system. II. STATEMENT OF PROBLffl*

The purpose of this study was to reinvestigate the character

of an ionic phosphorus halide compound which has the composition

FCl^F* This compound was observed by Lothar Kolditz (29) as a pyrolysis product of tetrachlorophosphonium hexafluorophosphate according to the reaction Asd-,

+ PF6 IKS-2*" ^ + ”5 ’

The product yield of ionic PCl^F was about 20 per cent of theory due to loss of molecular PC^F^. Kolditz determined that the new com* pound was ionic and existed as a simple fluoride ion salty that is» tetrachlorophosphonium fluoride > [Pd^]+Ir . His reasoning was based upon the followings 1) chemical analysis which gave the stoichio­ metry PCI. F» 2) conductivity measurements of acetonitrile solutions -1 2 - 1 from which equivalent conductances of 14 to 26 ohm cm. eq» were obtained (compare to 35 ohm"-1- cm. eq."^ for PCI,. (9) and

41-95 ohm"^ cm.^ eq.”^ for CPd^3+[PF^] (28) at comparable concen- trationsy, 3) cryoscopic measurements of anhydrous acetic acid solu­ tions. He obtained an apparent molecular weight of 80 in acetic acid. This value is to becompared to $6 which is one half the

calculated molecular weight of PCl^F. An apparent molecular weight of 96 would be expected if complete ionization into {PC1^]+ + F~

10 11 occurred. The discrepancy was attributed to hydrolysis. Gne might wonder if the discrepancy might be attributed more appropriately to solvolysis. Aside from the results of his cryoscopy measurements, it was perhaps not unreasonable for him to assume that a fluoride ion remained after removing phosphorus pentafluorlde from a hexa­ fluorophosphate anion. However, if one considers that the energy of a phosphorus-fluorine bond is twenty to forty kilocalories greater than a phosphorus-chlorine bond (42), one might question how reason­ able Kolditz* assumption really was.

Two possible structures of ionic PCl^F which involve phosphorus- fluorine bonds are [PC1^F]+C1“ and [PC1^]+[PC1^F2]“. A simple thermo- chemical cycle can be used to estimate the relative stabilities of

Kolditz* formulation and other postulated formulations. In the thermochemical treatment given below, the structure [PC1^F]+F~ was admittedly used because the treatment indicates that it might be more stable than the [PC1^]*F“ structure. Nevertheless, the result of this treatment serves to justify an examination of the validity of

Kolditz* conclusions.

®F - EC1 [pci3*3+ + ci* + f“ [PC13*]+ + Cl” + F*

[PCl^F]* + Cl

[PC14 ]+ F“ ------[PC13F]+C r Qc * heat of conversion between the ionic solid forms

Uq_» U « lattice energies of [PClr]+F* and [PCl-Fl^Cl", 1 2 respectively

Pp ? * P-Cl and P-F bond energies> respectively

electron affinities of Cl and F» respectively

% * uoi - °o2 + Qp-Cl - Op-F + Fp - Bel

The lattice energies can be readily calculated by using the familiar Born expression:

, | i > i n (1 _ l/n'). ° ro

The following values were used:

rQ for [PC1^]+F” = interionic distance = 4.45 £

r0 for [PCl^Fl^Cl^ * 4.90 X

The ions were assumed to be spherical. Pauling’s tetrahedral and crystal radii were used to calculate the respective ion radii. 23 1 N = Avogadro's number = 6.02 x 10 mole*

A = Madelung constant ■ 1.748 for NaCl type crystal structure in each case

e unit of electrical charge = 4.803 * 10"^° e.s.u.

% - highest conmon factor of the charges on the ions = 1

n* « B o m exponent * 9

Thusi UQ^* » 116 kcal./mole and UQ^ « 105 kcal./mole.

Also* £p = 95*3 kcal./mole and * 85*8 kcal./mole (43).

Therefore, Qq = 116 - 105 + 95*3 - 85.8 + Q p ^ - Qp_p

** 20 + Qp_cx - Qp_p kcal./mole. As can be seen* the difference in energies of the phosphorus-chlorine bond and the phosphorus-fluorine bond need only be appreciably greater than 20 kcal./mole in order for the [PCl^F] Cl structure to be more stable than the [PCl^]4fr structure.

A similar treatment Involving the postulated [PC1^]+[PC1^F2] structure shows that the total energy for the gas phase reaction

[FCl^]* ♦ 2F~ - PCl^F + F~ - [PCl^F2r must be exothermic and appreciably greater than 150 kcal. per mole

[PC1^F23" in order for the structure [PCl^]+[PC1^F2]" to be more stable than the [PC1^]"*"F" structure. The heat of the above gas phase reaction might very well be greater than 150 kcal./mole» but an estimation of the actual value would be too crude for a substan­ tial argument. However* an argument in favor of the [PC1^]+[PC1^F2]~ formulation is its similarity to the known compounds [PCI. ]+[PCl>]”* if o [PCl4 f[P015. r . [PC14 ]+[PC1 Br]'. [rci4 ]+[PF6 ]*, and [ S b C l ^ f .

The last mentioned ion was demonstrated by Kolditz to exist in antimony pentachloride solutions (44).

The foregoing discussion was an attempt to show that Kolditz*

formulation [PC1^]+F" might have been incorrect. Svidence has been presented which indicates that further investigation is required be­

fore a truly definitive statement can be made about the correct formulation of ionic PCl^F. Therefore* the object of the present

investigation was to prepare ionic PCI F and establish its ion struc-

ture. The synthetic portion of the investigation was to involve possible discovery of a more convenient method to produce the com­ pound and increase product yields. The establishment of the actual

ion structure was to involve the following physical measurements. l*f

1* Chemical analysis to confirm the FCl^F stoichiometry

2* Infrared spectroscopy to determine whether or not phosphorus-fluorine bonds are present

3* Cryoscopy to distinguish between the ion structures

and [ H ^ } Y

Phosphorus-31 and fluorine-19 nuclear magnetic resonance to observe possible P-F spin-spin coupling

Also included in this study was development of some convenient techniques and apparatus for handling these highly moisture sensi­ tive systems* Conventional apparatus for performing normally simple operations such as filtrations and reagent transfers and additions are not suitable for use in phosphorus-halogen chemistry (as well as in many other growing areas of inorganic chemistry)* Consequently» there is an ever Increasing need for apparatus and techniques which provide complete exclusion of atmospheric moistures but which are still simple enough in design for convenient* routine use* III. EXPERIMENTAL

A. Apparatus and Experimental Techniques

Reaction techniques were developed which Involved relatively

simple and convenient apparatus. The systems were designed to

rigorously exclude atmospheric moisture since the reactions and products were extremely sensitive to hydrolysis. The various methods used for moisture exclusion are discussed below. A dry atmosphere glove box was a necessity for handling products and preparing samples

for various analytical measurements. All of the glass apparatus des~

cribed in the following sections were continually dried in an oven

at 110° and then flushed with anhydrous nitrogen gas for at least one hour immediately before using them in the experiments. The nitrogen gas was dried by passage through magnesium perchlorate drying towers.

Pyrolysis of [PC1^]+[PF^]'*.— According to Kolditz (29)* ionic

PCl^F can be prepared by pyrolyzing a suspension of tetrachlorophos­

phonium hexafluorophosphate* [PC1^]+[PF^]“» in arsenic trichloride.

The suspension is heated at 70-85° until only a few flakes of solid

remain. The decomposition reaction is

AsClo [PC14 ]+[PF6] - 4 - PCl^F + PF5 *

15 The pyrolysis la performed in a small flask fitted with a reflux con­ denser. During the pyrolysis some of the [PCl^D^XPF^]” is converted to its PCI2F3 phosphorane isomer (b.p.* 7 .1°) and escapes with the o PF^ (b.p.* -&+.5 )• Also> some of the PC^F^ isomerizes back to the ionic modification and collects on the walls of the condenser. The resulting PCl^F, in its molecular form (b.p., 64°) * refluxes in the condenser and drops back into the solution. However» some of the molecular PCl^F isomerizes to its ionic form and remains in the condenser where it partially recombines with the evolved PF^ to form

[PCl4 ]+[PF63“ again. The escaping PCl^F-j and PF^ gases are collected* respectively* in -20° and -80° traps. When the reaction is apparently complete (about 2.5 to 3 hr.)* the hot mixture is filtered through a glass frit of medium porosity contained in the reaction flask. Ionic

PCl^F is precipitated from the filtrate by cooling to 0° overnight.

The resulting precipitate is separated from its mother liquor ty an­ other filtration. At this point the ionic PCl^F is contaminated with

[PCl^]+[PFg]" to an extent of about 20 per cent (9 per cent according to Kolditz) and with residual arsenic trichloride solvent. However* according to Kolditz the residual arsenic trichloride is removed at a pressure of 1 mm. at ambient temperatures* and then the [PC1^]+[PF^]" impurity is readily sublimed away by heating the product at 80° at the same pressure. Heating the product at 80° under 1 mm. of pres­ sure for 15 min. was said to be sufficient for complete removal of the [PC1^]+[PF^]- impurity. The reaction is initiated with 23 g.of

[PClif]+[PF6 3", and a yield of 3 g. of ionic PCl^F is obtained. The yield is about 20 per cent of theory. 17 Experiments were run In which the directions recommended by

Kolditz were followed as closely as possible* Hie low product yield observed by Kolditz was found to be largely the result of partial loss of molecular from the reaction* However* the yields could be increased considerably by using a -10° reflux condenser which returned the PCl^F^ to the reaction*

The apparatus indicated in Figure 1 was found to be suitable for the pyrolysis experiments. The pyrolysis apparatus consists of a two neck 100-ml. round-bottom flask* One neck consists of a standard taper 14/20 inner joint which is sealed with a ground glass cap during reaction* The other neck consists of a standard taper 24/40 outer joint to which is connected a short cold finger condenser containing a calcium hydride drying tube* The cold finger is maintained at -10° by the method indicated in Figure 1* The cooling fluid was isopropanol which was circulated by a small rotary pump. The filtrate receiver consists of a 100-ml* round-bottom flask containing a drying tube*

A sintered glass frit of medium porosity is fused within a standard taper 14/20 outer joint on the flask*

A typical reaction was darried out in the following manner*

Twenty-three grams of [PCl^j^tPF^]- was transferred to the reaction flask in the dry box* Hie flask was then stoppered and removed from the dry box. The reaction flask was quickly connected to a receiver adapter on a distillation apparatus. The distillation apparatus was being dried by flushing dry nitrogen gas through a calcium hydride drying tube connected to the receiver adapter. Thirty-eight milli­ liters of arsenic trichloride solvent was then distilled directly 18

Reservoir for Soft drink circulating liquid dispensing pump , Tvaon tubing

.Copper coil

^-4-1 beaker -ice- brine Cold finger mix t ure condenser assembly I® 9-CaH2 drying tube

Reaction flask drying tube 100 ml

r Glass frit,

^ i | 4 m ' P ° ro sity

Reaction flask 100 ml.

1 Teflon sleeves (ASCO)on I 24/40joints

Figure I. Apparatus for pyrolyzing[PC|^+[PF6]" 19 onto the [PC1^]+[PF^]"" in the reaction flask. The pyrolysis con­ denser assembly was dried by passing dry nitrogen through the calcium hydride drying tube on the condenser. While flushing dry nitrogen through the distillation apparatust the reaction flask was quickly removed and connected to the pyrolysis condenser. A water bath was placed around the reaction flask> and the temperature was slowly raised by means of a thermostatic hot plate. At 75-76° the mass began to effervesce and fuming occurred at the drying tube outlet. The temperature was increased and maintained at 62-85° until the fuming had subsided (about 3*25 hr.). The filtrate receiver was quickly connected to the 14/20 inner joint on the reaction flask while gently flushing dry nitrogen through the drying tube on the condenser.

The hot mixture was then filtered by rotating the entire apparatus and simultaneously applying a gentle pressure of dry nitrogen through the drying tube on the condenser. The filtrate receiver was then disconnected from the reaction flask under a dry nitrogen flush and quickly stoppered. The flask was placed in an ice-water bath over­ night. The resulting precipitate was separated from the supernatant liquid by filtration in a manner similar to that described for the first filtration. The original reaction flask served as the filtrate receiver. A vacuum stopcock adapter was quickly connected to the

24/40 outer joint on the receiver flask. The residual solvent was then removed from the precipitate at a pressure of 1-2 mm. on a vacuum line. The product was kept at room temperature for 10-15 min. and then warmed slightly for several minutes. The pressure was regulated by means of a precision leak control connected to the vacuum line. 20

At this point the product was a white* free-flowing powder amounting to approximately 4-5 g« The flask was then warmed to 80° with a water bath in an attempt to remove the [PCl^] [PF^] impurity. The pressure was maintained at 1 mm. A white solid condensate developed immediately in the cool stopcock adapter section. Previous experi­ ence showed that much of the sample would be lost if the hot system were left open to the condensation trap on the vacuum line. There­ fore * the system was sealed by turning the key on the vacuum stopcock adapter before heating the sample. Since this method of removing the

[PCli+]+[PF6r inpurity was unsuccessful* other methods were attempted.

Other attempted methods for removing the [PC1^]+[PF^]" impurity involved placing the Impure product in a jointed sample tube which was connected to a vacuum stopcock adapter. The sanple tube was evacuated on a vacuum line* and the pressure in the tube regulated by means of a precision leak control on the vacuum line. The at­ tempted purification treatments included heating the sample to 65° o in a highly evacuated system or to 65-80 at 1-2 nan. pressure in a sealed system for periods up to 12 hr. The results of these treat­ ments will be discussed in a later section.

Reactions of arsenic trifluoride with phosphorus oentachloride.—

A method was discovered by which ionic PCl^F could be synthesized by directly fluorlnating phosphorus pentachloride with arsenic tri- fluoride. The fluorination proceeded according to the following reaction:

AsClo 3PC15 + AsF^ - 3PC14F + AsCl-j. 21

A diagram of the apparatus used in these experiments is shown in

Figure 2. The apparatus consists of either a 250-ml. or 500-ml. jointed round-bottom reaction flask. The choice of flask size de­ pends upon the reagent quantities. The flask contains a sintered glass frit of medium porosity which is fused within a standard taper

24/4-0 drip-tip joint. The flask also contains standard taper 24/40 and 7/25 outer joints. The three joints are arranged on a circum­ ference of the flask with each joint separated by an arc of approxi­ mately 45 degrees. A 10-ml. micro-buret is connected to the 7/25 outer joint on the reaction flask. The micro-buret is fitted with a Teflon straight bore stopcock and a capillary tip which consists of a standard taper 7/25 drip-tlp joint. The section of the buret immediately below the stopcock is bent at such an angle as to put the buret in a vertical position when connected to the flask. A calcium hydride drying tube is connected to the buret after adding arsenic trifluoride to the buret. An efficient reflux condenser containing a calcium hydride drying tube is connected to the 24/40 outer joint on the reaction flask. A ground glass cap is placed over the filter joint during the reaction. The filtrate receiver consists of a jointed distillation receiver adapter to which is connected a jointed flask of such size as to contain the filtrate. A calcium hydride drying tube is connected to the pressure release tube on the adapter. Standard taper Teflon sleeves (ASCO) were used on all the 24/40 joints. The buret joint was lightly lubricated with stopcock grease.

A typical reaction was performed in the following manner. Thirty to fifty grams of phosphorus pentachloride» weighed to 0.1 g.> was 22

J Teflon sleeves (ASCO) C aH2 ^ on oil "5 24/4Qjoints drying tubes

24, 40

Drip - tip Micro-buret con denser 10 ml. —►

Receiver adaptor CaH2 drying tube

Teflon 24 /40 stopcock 40

I7> 40

Glass frit Filtrate Reaction flask m. porosity recei ver 500 or 250ml.

CD

Figure 2. Apparatus for direct fluorination of PCI5 with AsF3 . 23 transferred to the reaction flask along with a Teflon coated magnetic stirring bar in a dry atmosphere glove box. Arsenic trichloride sol­ vent was then distilled directly onto the phosphorus pentachlorlde by the same technique as described for the pyrolysis experiments. The volume of arsenic trichloride was equal to approximately twice the value in grams of phosphorus pentachlorlde. Arsenic trifluoride was distilled directly into the micro-buret by connecting the top end of the buret to a receiver adapter on a distillation apparatus. A very short section of Tygon tubing was used for the connection. (Pure arsenic trifluoride does not react with Tygon.) The buret and con­ denser were then connected to the reaction flask. Dry nitrogen flushes were always used In a manner described previously for the pyrolysis experiments. The PCl^-AsCl^ reaction mixture was magnetic­ ally stirred and heated to 73-77°• The temperature was controlled by means of a water bath which was heated by a combination thermo­ static hot plate and magnetic stirrer. Usually only a small amount of phosphorus pentachlorlde remained undissolved at this temperature•

After temperature and solution equilibria had been achieved» the ar­ senic trifluoride was slowly added to the solution. The addition rate was approximately 2 ml. per 30 min. Much of the arsenic tri- fluoride vaporized from the buret tip in the hot chamber and refluxed in the condenser. The reaction was exothermic* and the temperature of the hot water bath would tend to rise even though the control on the hot plate was continually readjusted to lower values. Near the end of the reaction a white precipitate began to form. The end point of the reaction was usually indicated by a sudden increase in fuming at the drying tube outlet on the condenser. However* in order to

insure proper reaction stoichiometry * the exact amount of arsenic

trifluoride required to produce ionic PCl^F was always calculated in advance. A density of 2.60 g./ml. for the arsenic trifluoride was used in the calculation.

The critical temperature range for a successful reaction was

70-80°. Below 70° CPC1^]+[PF^]" was formed; above 80° lower yields

resulted from decomposition of the product.

The hot reaction mixture was stirred several minutes longer and

then allowed to cool slowly to room temperature. A dense white pre­

cipitate quickly formed as the mixture cooled. The buret and con­ denser were quickly replaced by standard taper glass stoppers while flushing dry nitrogen through the drying tube on the condenser.

The mixture was cooled to 0° overnight. A white* crystalline precipi­

tate and a pale yellow solution resulted. The filtrate receiver

assembly was then connected to the filter joint on the reaction flask while flushing dry nitrogen through the drying tube on the receiver

adapter. A large magnesium perchlorate drying tube was connected to

the 24/bO outer joint on the flask in place of the reflux condenser.

The mixture was filtered by slowly rotating the entire apparatus and

simultaneously applying a gentle pressure of dry nitrogen through the

large drying tube. When the precipitate had drained conpletely* the

large drying tubeon the flask was quickly replaced by a vacuum stop­

cock adapter and the receiver assembly replaced by a ground glass

cap. The flask was connected to a vacuum line and evacuated to a

pressure of several millimeters. The residual solvent was then distilled from the precipitate into a -78° trap on the line. A

free-flowing, white powder was produced after 30-*K> min. on the

vacuum line. The evacuated vessel was then transferred to a dry

atmosphere glove box where the product was transferred to other

apparatus for further treatments. The yield of the product was almost

exactly the theoretical amount calculated for PCl^F, either on the

basis of the original amount of phosphorus pentachlorlde or on the

basis of the amount of arsenic trifluoride used in the reaction.

However > an infrared spectrum of the product showed that appreciable

amounts of CPCl^]+[PF^]"t [PC1^]+[PC1^F]*", and phosphorus penta­

chlorlde were present as impurities.

Other experiments (discussed in a later section) showed that

[PC1^]+[PC1^F] readily disproportionstes into phosphorus penta­

chlorlde and molecular PCI F in an evacuated system. Also*, experl- *f ments on Impure ionic PCl^F samples showed that the phosphorus penta­

chlorlde impurity could be removed by almost any method of sublimation

or by extraction with solvents of low polarity. In view of these

facts* the following procedure was found to be best suited for removal

of [PC1^]+[PC1^F]~ and phosphorus pentachlorlde from impure ionic

PCl^F. The precipitate was left in the original reaction flask and

freed from adhering arsenic trichloride solvent as described above.

The container was then fully evacuated and left open to a -196° trap

on the vacuum line for about 2 hr. (An appreciable amount of sample

would have distilled with the arsenic trichloride if the system were

fully evacuated during the drying process.) The precipitate was

then washed several times with 20-30 ml. portions of either dry carbon 2 6 tetrachloride or arsenic trichloride. The solvent was added to the precipitate by first expanding dry nitrogen into the flask and then quickly pouring the solvent onto the precipitate. A very rigorous procedure would involve distilling each wash portion onto the pre­ cipitate on the vacuum line. However, since the precipitate amounted to 30-40 g.» the slight hydrolysis which occurred during momentary exposure to the moist air could be ignored. The precipitate was then dried in the same manner as described for removal of the adhering re­ action solvent.

Reactions between arsenic trifluoride and phosphorus pentachlorlde in solvents other than arsenic trichloride were performed in the same manner as described above. However, the solvents were usually dis­ tilled onto the phosphorus pentachlorlde on a vacuum line. Dry nitro­ gen was then expanded into the evacuated reaction vessel so that the reflux condenser could be attached without exposing the reagents to o atmospheric moisture. A reaction run at -78 in methylene chloride was performed similarly, except a large calcium chloride drying tube was substituted for the reflux condenser.

Reactions of solid fluorinating reagents with phosphorus penta­ chlorlde .--Much attention was given to the possibility of preparing ionic PCljjF by reacting silver monofluoride with phosphorus penta­ chlorlde. These reactions were usually performed in the same type of filtration flask as that shown in Figure 2 . The solid reagents were placed together in the filtration flask in a dry atmosphere glove box.

Solvents which were more volatile than arsenic trichloride were distil­ led onto the reagents on a vacuum line. Dry nitrogen was then expanded into the evacuated reaction system so that a reflux condenser could be attached to the flask without exposing the reagents to atmospheric moisture. The resulting silver halide precipitates were separated from the solutions by filtration in the manner described previously•

The filtrates were then evaporated either on a vacuum line or in a stream of dry nitrogen at room temperature and atmospheric pressure.

Evaporation by the latter method was s uperior to vacuum distillation or distillation at elevated temperatures, because phosphorus penta­ chlorlde and its fluorine derivatives were found to distill along with the solvents.

Powder-dropping device.— Preliminary reactions between silver monofluoride and phosphorus pentachlorlde in methylene chloride produced a mixture containing ionic PCl^F, [PC1^.]+[PF^]">, and possibly other phosphorus chlorofluorides. The reactions proceeded rapidly and went to conpletion in several minutes. Also, the glass walls of the reaction vessels were attacked by the silver fluoride during the reaction. It seemed apparent that the silver fluoride had to be added to the reaction very slowly in order to effect a controlled fluorination of phosphorus pentachlorlde. Also, in order to eliminate siliceous impurities from the product, a cfevlce for adding the reagent had to be inert to fluoride attack. Such a device was developed and is shown in Figure 3« However, it was discovered in later experiments that the rapid reactions with phosphorus penta­ chlorlde were due to incompletely dried silver fluoride. Conse­ quently, the device was not necessary for reactions which were run under vigorously anhydrous conditions. 2 8

Left -handed threads Teflon washer Glass li( Control knob Teflon retaining nut Viton rubber- 0 - ring

Teflon plunger Teflon rod */g diam. Small tolerance friction fitted in between teflon pi unger rod and hole in container. Container made from polypropylene Container section test tube (I6xl0 0 mm.) friction fitted into Powdered reagent glass joint.

24/ *6 fyo inner joint (pyrex)

Teflon restri Teflon wedge friction wedge fitted on end of teflon rod.

^Entire threaded section is a modified commerical high vacuum teflon stopcock (Delmar Scientific La borator i es, Inc. )

Figure 3. Powder dropping device.

4 Even though the powder-dropping device was found to be unneces­

sary for performing controlled reactions between silver monofluoride and phosphorus pentachlorlde* the general utility and simplicity of

the device is significant enough that a brief description is appropri­

ate. The device was designed so that controlled additions of reac­

tive powders to reaction media could be accomplished in a completely

sealed system. The apparatus is completely suitable for high vacuum

systems as well as for systems tinder atmospheric pressure. Also*

solid reagents which react with glass can be used because the entire

container section is constructed from inert plastic materials.

The device is used in the following manner. The device is held

in an inverted position and the solid reagent* either finely divided

or moderately granular* is transferred to the container section of

the device in a dry atmosphere glove box. The Teflon restricting wedge

is then seated in position by turning the control knob at the top of

the device in a counterclockwise direction. The device is then con­

nected in a vertical position to a standard taper 24/40 outer joint

on the reaction vessel. The powder can be dropped into the reaction

at any desired rate by rotating the control knob in a clockwise direc­

tion. Rotation of the knob controls the amount of opening between

the Teflon restricting wedge and the container section.

Cryoscopy measurements .— The apparatus used in the cryoscopy

measurements is shown in Figure 4. The apparatus consists of the

following components:

1. A Beckman thermometer which has a snugly fitted Teflon collar

around the shank. The Teflon collar is accurately machined to fit a

standard taper 24/40 outer joint. 30

Solenoid with _ periodic circuit breocker Variable volt control I" iron bar sealed stirrer •-8 mm ID. Gloss stirrer Beckmon I thermometer 5 2 * * - Drying tube Teflon collar machined to fit I 2% q outer joint and fitted snuggly around thermometer.

1 Z\ q Dewar joint in cell cap

Close fitting evacuated dewar

Cork

Pint dewar for cooling bath

20cm Space between close fitting dewar and freezing-point cell wall filled with isopronol for thermal contact.

2 mm 25mm O.D

Figure 4. Assembled cryoscopy apparatus. 31

2. A cell cap which supports the thermometer and the glass

stirrer housing.

3* The cell body which contains the sample solution.

i*. An all glass stirrer consisting of a cylindrical helix for

stirring the solution. The upper end of the stirrer contains an

incapsulated soft rod for the purpose of magnetic activation.

5. A close fitting* evacuated dewar for insulating the cold

solution. The dewar contains a small amount of isopropanol to provide thermal contact between the cell walls and the dewar in

order to control the heat transfer.

6 . A second* larger dewar around the close fitting dewar. The

larger dewar contains a bath having a temperature appropriately ad­

justed so as to require fifteen to twenty minutes for a freezing-

point determination.

7. A solenoid which has a non-ferrous core. The solenoid is

placed around the stirrer housing near the iron rod contained in

the stirrer* The solenoid is activated at regular intervals of

about two per second causing a jumping action of the stirrer.

Sample solutions for the cryoscopy measurements were prepared

in a dry atmosphere glove box in the following manner. An accurately

weighed amount of the analytical sample (0 .1-0.3 g. weighed by dif­

ference in a small closed vial) was transferred to a small flask.

The combined weight of the flask and sample was then obtained. An

appropriate volume of solvent was then added to the sample to give

approximately the desired concentration. The flask was then re­

weighed. The exact molality <£ the solution could then be calculated from the known weights of sample and solvent. Fifteen milliliters of

* the solution was then drawn into an all glass syringe* The syringe was sealed with a Teflon cap and removed from the dry box* Meanwhile* the assembled cryoscopy apparatus was being dried fay flushing dry nitrogen through the drying tube connected to the stirrer housing*

An escape for the nitrogen flush was provided fay loosening the Beck­ man thermometer in the cell cap joint* A Teflon needle* extending to two inches from the cell bottom* had been previously inserted in the partially separated thermometer joints. The solution was injected into the bottom of the cell through the Teflon needle* The needle was then withdrawn* and the thermometer seated in place* The flushing source was disconnected*

Since nitrobenzene provides a supercooling problem which causes uncertainty in the freezing point value* warming* or melting* curves were used* The stirring action was begun and the solution was frozen into a thick slush by placing a 0° bath around the close fitting dewar. The 0° bath was then replaced by a 12° bath* About five minutes were allowed for the solid and liquid phases to reach equilib­ rium before readings were taken. Time-temperature readings were taken at thirty second intervals during the melting process* After the solid phase was completely gone* a sharp Increase in the warming rate occurred. Readings were then taken at fifteen second intervals. The process was repeated two more times* A plot of time versus temperature for each run produced straight lines for the melting process and for the warming solution after the solid phase had disappeared* The intersection point of the two extrapolated straight lines was taken 33 . o as the freezing point. A precision of .0.003 was obtained for the average of a 0.01 molal solution. A precision of .0.001 was ob­ tained for solutions having concentrations of 0.03 molal or greater.

The thermometer calibration for the freezing point of the pure solvent was accomplished in a manner similar to that described for the solutions.

The apparent molecular weight of the solute was calculated by using the expression

W| 1000 M * — K WjAt where

H = molecular weight of the solute

wl*w2 ~ tre^ht in grams of solute and solvent* respectively

K - molal freezing-point constant

At * freezing-point depression = (f.p. pure solvent)-(f.p. solution)

B. Reagents

Phosphorus pentachlorlde .— Reagent grade phosphorus pentachlorlde*

(J* T. Baker Chemical Co.) was finely pulverized in an agate mortar and evacuated for several hours in a high vacuum system. The reagent in its evacuated container was then placed in a dry atmosphere glove box where transfers to reaction vessels were accomplished.

Arsenic trichloride.— Reagent grade arsenic trichloride (Baker and Adamson* Allied Chemical) was refluxed over activated silica gel in order to remove hydrogen chloride. The solvent was then distilled from the silica gel through a short column packed with glass helixes 3^ directly into reaction vessels containing the reactant materials. The assembled distillation apparatus was flushed with dry nitrogen for one hour before the distillation was begun.

Arsenic trifluoride.— Arsenic trifluoride (Ozark-Mahoning Co.) was refluxed over sodium fluoride in order to remove hydrogen fluor­ ide. It was then distilled from the sodium fluoride through a short colum packed with glass helixes directly into addition burets. The completely assembled distillation apparatus* including the buret* was flushed with dry nitrogen for one hour before beginning the distil­ lation.

Tetrachlorophosphonium hexafluorophosphate .— This reagent was prepared according to a published procedure (28*45). The prepara­ tion involved the reaction

AsClo 2PC15 + 2A s F3 ------[PC14 ]+[PF6]' + 2AsCl^ .

The reaction flask shown in Figure 2 was used. The arsenic trifluoride was added to the PClyAsCl^ mixture by means of a 25-ml. buret. The buret was fitted with a Teflon stopcock and a standard taper 24/40 inner joint having a ring seal drip tip. The buret had a pressure equalizing tube extending from below the ring seal to the top of the buret. A standard taper 24/40 outer joint* containing a calcium hy­ dride drying tube* was fused on the top end of the buret. The buret was connected to the open 24/40 outer joint on the reaction flask.

A dry nitrogen flush was found to be inadequate for drying the product in the system that was used. However* the excess arsenic tri­ chloride solvent could be adequately removed on a vacuum line with 35 only small loss of product. Yields varied from 70 to 80 per cent of theory. The product was identified by its x-ray powder diffraction pattern and its infrared absorption spectrum In the 8-25 micron region.

Silver monofluoride.— Silver monofluoride of high quality was not commercially available. The reagent was prepared according to the procedure of H. Ott (46) • The procedure was modified so that large quantities could be conveniently prepared. The following method is suitable for making one-pound quantities. The reactions involved in the process are:

AgNO^ + NaCH H2° » Ag20 + NaNO-j ,

Ag20 + 2HF n* ► 2AgF + H^O “► Evaporation.

Nine hundred milliliters of a three molar silver nitrate solution» warmed to 40°, was slowly added to 900 ml. of three molar sodium hydroxide also warmed to 40° and contained in a 2-1. Erlenmeyer flask. The base solution was rapidly stirred during the addition and for one half hour afterwards. Silver(I) oxide digests readily but it also peptizes readily. The actual amounts of the reagents were adjusted so that the silver nitrate was in slight excess at the end of the reaction. The warm mixture was decanted through one layer of

Whatman No. 42 filter paper contained in a Buckner funnel. The pre­ cipitate was washed eight times by decantation with 250-rnl. portions of 0.1 per cent silver nitrate solution. The precipitate was then washed twice by decantation with 500-ml. portions of distilled water.

The washing process was rather inefficient due to peptization of the 36 precipitate* The precipitate was transferred to the filter and drained as completely as possible by applying suction from a water pump* The silver otide precipitate was then transferred to a polyethylene beaker which was made from a l6-oz. polyethylene bottle. Forty-eight per cent hydrofluoric acid was added dropwise to the precipitate while stirring with a polyethylene rod and cooling with cold water. Just enough hydrofluoric acid was added to dissolve the silver cocide.

Silver oxide is easily reduced by heat as well as light. Conse­ quently* finely divided metallic silver was present in the resulting aqueous silver fluoride solution. The solution was separated from the metallic silver by filtration as shown in Figure 5 . The fil­ trates were collected in several h—oz. polyethylene bottles.

Aqueous silver fluoride solutions can be stored indefinitely under bright lights without photodecomposition. The clear* colorless fil­ trates were evaporated to dryness in a high vacuum system as shown in Figure 6 . The drying process required about two weeks. During evaporation* the precipitated mass had to be broken up frequently with a polyethylene rod. Light was continuously excluded from the evaporating solution because hydrated forms of silver monofluoride

are photosensitive. Anhydrous silver monofluoride is apparently not photosensitive. The bright yellow to yellow-brown product was

pulverized in an agate mortar in the dry box and transferred to 1-oz. polyethylene screw-cap bottles. The small plastic bottles could

then be placed in convenient evacuation chambers for further drying

of the silver fluoride immediately before transfers to reaction

vessels. Anhydrous silver monofluoride is extremely hygroscopic; Bottomless 4oz. polyethylene screw cap bottle to contain filtering solution. 3 disks Whatman 42 "i filter poper

Thin teflon gasket No. I rubber stopper

Polypropylene funnel fused to screw cap

Pyrex evacuation chamber ------► 4oz. polyethylene bottle to catch filtrate.

Suction applied

Figure 5. Apparatus for filtering aqueous AgF solution. (Partially unassembled to show filters and gasket.) 3 6

To high vacuum source

35 Lightly cross hatched area indicates paraffin coating.

Pyrex evacuation Aq. AgF soln. chamber

No. 12 rubber sto pper

Large capacity condensation trap, cooled with liquid N2 .

Figure 6. Apparatus for evaporatinq aqueous AgF solut i o ns. consequently* all manipulations of the material were accomplished in a dry atmosphere glove bcoc.

The silver monofluorlde was identified fay its characteristic x-ray powder diffraction pattern (46)* Chemical analysis of the acid-soluble silver content indicated a 99*0 per cent purity.

The impurity was presumably metallic silver and a small amount of silver(I) oxide.

Miscellaneous solvents,--Comnonly used solvents such as carbon tetrachloride* methylene chloride* benzene* and acetonitrile were dried over calcium hydride. The solvent-desiccant mixtures were magnetically stirred at room temperature for at least one full day and then distilled onto experimental samples on a vacuum line.

Reagent grade solvents were used whenever possible. The nitro­ benzene used in the cryoscopy measurements was supplied in pure form by courtesy of V, Petro (9)* However* the reagent was dried further fay shaking over and then distilling from **2^5 through a column packed with glass helixes. The middle fraction was taken.

Anhydrous acetic acid was prepared by distilling 99*5 P®** cent acetic acid from triacetyl boride (4-7) • The trlacetyl boride was prepared by heating one part boric acid with five parts (by weight) acetic anhydride to 60°. The triacetyl boride separated upon cooling and was collected fay filtration. 4 0

C. Chemical Analysis

Sample hydrolysis.— All of the phosphorus compounds in this study were extremely reactive toward atmospheric moisture and reacted almost violently with water. Hydrolysis of the compounds; produced orthophosphoric acid* fluorophosphoric acid* and hydrogen halide gases. In order to prevent the hydrogen halides from escaping* samples were hydrolysed in strong base solutions within a closed system. A hydrolysis apparatus suitable for this purpose is shown in Figure 7 • The method of hydrolyzing the samples and preparing the resulting solutions for chemical analysis was the following. The solid compound* 0 .7-1.0 g.* was placed in a dry* tared one-dram Opti- d e a r vial in a dry atmosphere glove box. The vial was then sealed with its plastic cap and removed from the glove box and reweighed.

The plastic cap was then removed from the vial (in the glove box) and the vial* with sample* was carefully placed in the hydrolysis flask. The hydrolysis apparatus was assembled and removed from the glove box. The apparatus was then carefully evacuated through the delivery tube but not left open to the pumping system more than a few seconds. The samples were appreciably volatile and losses would occur if they were evacuated for extended periods of time.

The sample was hydrolyzed in 4-5 ml. of cold 2 Normal potassium hydroxide solution as shown in Figure ?• The base solution must not be wanner than ambient temperatures or else the reaction will be too violent and produce an excess of back pressure and loss of products through the delivery tube.

The analytical solution was rinsed into a 250-ml. nickel 41

Delivery tube

Teflon stopcock straight bore

100ml. beaker Base solution 40

joint

-100ml. bulb moderately evacuated

Weighed sample

Figure 7, Apparatus for base hydolysis of solid phosphorus halides. 42 crucible and evaporated to dryness. The mass was then fused at 500° for thirty minutes in an electric furnace in order to decompose the fluorophosphate salts. The fused mass was then dissolved in distilled water and filtered into a 2^0-ml. volumetric flask and diluted to the mark on the flask. (The filtration removed nickel oxide and would not have been necessary if a silver crucible were used.)

The hydrolysis procedure was verified by using a saxrple of phos­ phorus pentachlorlde. The results were: P» 14.8; Cl, 84.2$; stoi­ chiometry, PC1^^£ ( F d ^ requires; P, 14.9J Cl, 85.I#).

Phosphorus determination.— Phosphorus was determined as ortho­ phosphate by a standard volumetric analysis procedure involving initial precipitation as ammonium phosphomolybdate and titration of the precipitate with standard decinormal sodium hydroxide solution

(48). The following is the general procedure that was used. The analyses were performed in duplicate.

A 10-ml. aliquot portion of the sample solution was placed in a 250-ml. Erlenmeyer flask and diluted to 25 ml. The solution was made neutral to litmus with 1:1 nitric acid and then 3 ml* excess acid was added. Three milliliters of 2 g»/3 ml. ammonium nitrate o solution was added also. The solution was then warmed to 35-^0 o and 75 ml. of molybdate reagent, also wanned to 40 , was added to the solution. The mixture was rapidly stirred for one minute with a glass rod and then stored overnight in a warm place with the glass rod left in the solution. The resulting mixture was fil­ tered by decantation through Whatman No. 42 filter paper and washed by decantation five times with 15-ml. portions of a one per cent potassium nitrate solution* The bulk of the precipitate was left behind in the flask. The filter paper was then washed five times with 15-ml. portions of the wash solution. The filter was allowed to drain conpletely between washings. The filter paper was trans­ ferred to the remaining precipitate in the flask*, and excess standard

N/lO NaOH was added. The filter paper was shredded by use of the original glass rod. The rod was then rinsed and removed from the flask. Six drops of phenolphthalein solution were added and the color discharged with standard N/lO HC1. More standard base was added until a pink end point was observed. The end point was not sharp and care had to be taken so as not to miss the first appear­ ance of a persisting pink color. The phosphorus content was calculated from the net volume of sodium hydroxide required to produce the end point and from the reaction stoichiometry of 23 eq.

NaOH per 1 g. atom phosphorus.

Chlorine determination— Chlorine was determined as chloride by the standard Volhard volumetric procedure (^9)• This method is very accurate and rapid; consequently* the purity of many samples was estimated solely on the basis of chloride content. The following is the general procedure that was used. The analyses were usually done in triplicate.

A 25-ml. aliquot portion of the sample solution was transferred to a 250-ml. Erlenmeyer flask. The solution was made strongly acidic with 1:1 freshly boiled nitric acid and then enough standard N/10 t AgNO^ was added to give 2 ml. in excess of the chloride present.

One milliliter of ferric alum indicator and 3 ml. of nitrobenzene 4 4 were added to the mixture* The flask was then stoppered and shaken vigorously for about one minute* Standard N/lO KSCN solution was carefully added to the mixture until the first appearance of a sal­ mon colored end point. The chloride content was equal to the net equivalents of silver ion used in the silver chloride precipitation*

Fluorine determination*— Fluorine was determined as fluoride by a volumetric procedure involving initial precipitation of lead chlorofluoride (50)* The lead chlorofluoride was then dissolved in nitric aoid and the chloride content of the resulting solution de­ termined by the Volhard method* The following is the general procedure that was used. The analyses were performed in duplicate*

A 50-ml. aliquot portion of the sample solution was transferred to a 400-ml* beaker and diluted to 200 ml* The solution was wanned to 45° and one drop of methyl-orange indicator was added. One-to- one nitric acid was added until the indicator just began to change color and then one more drop of acid was added. Sixteen drops of concentrated hydrochloric acid and 10 drops of glacial acetic acid were added followed by 25 ml. of a warm* clear solution containing

10 per cent lead acetate and one per cent glacial acetic acid* The mixture was rapidly stirred for one minute and then set aside for at least one hour* The mixture was filtered by decantation through

Whatman No. 42 filter paper. The heavy* white precipitate of lead chlorofluoride was washed by decantation once with cold water* four or five times with a saturated solution of lead chlorofluoride* and once more with cold water* The precipitate was quantitatively trans­ ferred to the filter where it was dissolved in hot 10 per cent nitric * 5

acid (about 100 ml*). The filtrate was collected In a 250-ml«

Erlenmeyer flask and the chloride content of the filtrate determined

by a Volhard titration (see chlorine determination above).

Silver determination.— The acid soluble silver ion content of

silver monofluorlde preparations and certain reaction products was determined by a Volhard titration as described for the chlorine

determination. Silver fluoride samples were weighed as described

under the hydrolysis section above and then dissolved in 1*1 nitric

acid in the open air.

Certain reactions produced silver halide precipitates which

approximately corresponded to the composition AgF'AgCl. These

products had to be decomposed by boiling in 1:1 nitric acid for

10 min. in order to completely dissolve the silver fluoride.

D. Spectroscopic Analysis

Infrared spectroscopy.— .All infrared spectra were run on a

Perkin-Elmer Model 337 diffraction grating infrared spectrometer.

Spectra of solutions were obtained by using specially designed liquid

cells shown in Figure 8 . Conventional cells which employed salt

crystal windows and metal delivery ports could not be used because

of the corrosive nature of the solutions. Spectra of solutions in

the 8-14 micron region were obtained by using circular IR-Tran 4

(Eastman Kocak Optical Co.) windows. Each window was 2 mm. in

thickness. One of the windows had ultrasonically drilled delivery

ports. Spectra of solutions in the 14-25 micron region were ob­

tained by using polyethylene windows. Each window was 1 ran. in 4 6

10.

1. Teflon stoppers 2. Teflon delivery ports, threaded into brass plate, flush with rear surface. 3. Brass plate 4. Teflon gasket (not necessary for polyeth. windows) 5. I.R. window 6 . Teflon spacer 7. I. R. window 8. Soft rubber cushion(not necessary for polyet h. win dows ) 9. Threaded holes for clamping screws 10. Brass backing plate

Figure 8 . In fra re d cell for corrosive liqufds. **7

In thickness. The delivery ports in one window were made by means of a hot metal pin. A Teflon spacer of 0.005 in* thickness was used

in each cell. Solvent absorptions were compensated by attenuating

the reference beam on the spectrometer with a cell oontaining the

pure solvent. Potassium bromide windows were used in the compensating

cell. The windows were separated by a Teflon spacer which was pre­

cision pressed to such a thickness that very slight overcompensation

of the solvent absorptions resulted.

Spectra of acetic acid solutions were taken with IR-Tran cell

windows. The windows were separated by a Teflon spacer of 0.001 in.

thickness. The acetic acid absorptions were not compensated in the

reference beam of the spectrometer.

The sample solutions for infrared analysis were prepared in a dry

atmosphere glove box and injected into the cells by means of an all

glass syringe. Metal syringe needles could not be used because of the

corrosive nature of the solutions.

Infrared spectra of Nujol mulls were obtained by using circular

potassium bromide crystals. The mulls were prepared in an agate

mortar in a dry box. Plastic tape (Scotch Magic Mending Tape) was

wrapped securely around the edges of the crystal plates in order to

exclude atmospheric moisture from the mull, hydrolysis of the mull

became quite appreciable during humid weather when the plastic tape

was not used. Nujol mulls did not react with the potassium bromide

crystals.

Pressed pellets consisted of potassium chloride as the dispersion

medium. (Hie samples reacted with potassium bromide in the presence 48 of moisture*) The samples were pulverized with well dried potassium chloride in an agate mortar in a dry box* The mixtures were then placed in dry paper supports between metal dies* The assembled dies were wrapped in heavy plastic bags and pressed outside the dry box*

X-ray powder diffraction— -A Norelco x-ray diffractomer fitted with an 11*45 cm* Debye-Scherrer powder diffraction camera was used.

Samples were loaded into 0*3 mm. x-ray capillaries in the dry box*

The capillaries were then sealed with vacuum grease* removed from the dry box, and sealed with a small flame* X-ray exposure times varied from 16 to 40 hours using CuKq, radiation at 32 kilovolts and

12 milliamperes•

Nuclear magnetic resonance .— Nuclear magnetic resonance measure­ ments were attempted by using a Varian Model HR-60 high-resolution spectrometer equipped with a base line stabilizer unit, Varian Model

V-3521A. A radio frequency of 19*25 Me* was used for attempted ob­ servation of phosphorus-31 resonance* A radio frequency of 6o Me. was used for attempted observation of fluorine-19 resonance*

Even though resonances could not be observed* a description of the technique used for preparing sample solutions is still appropriate*

An apparatus for preparing the solutions is shown in Figure 9* The apparatus was designed so that saturated solutions could be conveni­ ently prepared in a routine manner and still provide complete exclusion of atmospheric moisture* The analytical sample was placed in the as­ sembled apparatus in a dry atmosphere glove box. The apparatus was then evacuated on a vacuum line and the dry solvent was distilled onto the sample. The purpose of the sintered glass frit was to exclude *9

Teflon stopcock " straight bore

0 "-ring joint

Glass frit, /coarse porosity

T 7/

NMR sample tube

Figure 9. Apparatus for preparing NMR sample soluti ons. solid particles from entering the sample tube* After the solution was prepared % the evacuated apparatus was removed from the vacuum line and rotated in a manner to cause the solution to flow into the sample tube* The sample tube could then be either sealed with a flame or else dry nitrogen could be expanded into the system allowing the tube to be removed and quickly stoppered. The latter method en­ ables one to use the same sample tube for later experiments*

E* Conductance Measurements

Conductance measurements were made on nitrobenzene solutions of ionic PCljjF* The apparatus used for the measurements was a Leeds and

Northrup No* 155** capacitance and conductance bridge* Complete de­ tails of the bridge are given in the Leeds and Northrup Std* 21221

3-538* Conductance measurements were made using the grounded point bridge circuit*

A Jones and Bollinger type conductivity cell having a cell con­ stant of 39*5 cm*”'*' was used (9)* The inlet and outlet ports on the cell were sealed with Teflon stopcocks to prevent hydrolysis of the o samples. The cell was dried in an oven at H O overnight and then flushed with dry nitrogen. The solutions were prepared in 50-ml. volumetric flasks in a dry atmosphere glove box. An accurately weighed amount (by difference) of ionic PCl^F was transferred to the volumetric flask. Solvent was then added to the mark on the flask.

The resulting solution was added into the conductivity cell by means of an all glass syringe and a Teflon syringe needle* The cell was then sealed with its Teflon stopcocks* removed from the dry box* and placed in a 2$il0 constant temperature bath* One hour was allowed for thermal equilibrium before readings were taken* Conductance measurements on the pure solvent were performed in the same manner as described for the solutions.

The following formulae were used for the calculations in this section*

Conductance of the sample solutions

R_-R. i C * g 8 ohm*1 Ro R_s where

C = conductance of the sample

Rq = preliminary reading of the resistor in the balancing arm of the bridge

Rg = final reading of the resistor in the balancing arm of the bridge

Specific conductance of the sample solution:

k = Kc ohm” ^ cm*"* t where

k - specific conductance

K = cell constant (cm#*1)

C « measured conductance (ohm“^)

Equivalent conductance of the sample:

1000k 2 i A * — «— ohm cm# equiv.“x '■'e where

A = equivalent conductance

k = specific conductance

CQ = concentration in gram equivalents per liter of solution IV. DISCUSSION OF RESULTS

A. Preparation of Ionic PCl^F by Pyrolysis of [PC1^]+[PF^]

Ionic PCl^F was prepared by pyrolyzing tetrachlorophosphonium hexafluorophosphate*, [PClil_]+[PF^]’*, in arsenic trichloride. The experimental procedure recommended by Kolditz (29) was followed as closely as possible. The pyrolysis reactions proceeded just as des­ cribed by Kolditz except for a minor deviation; the decomposition was found to begin at 76° and not 70° as reported. The resulting pyrolysis product was found to contain[PCl^]+[PF^]~ to an extent of about 20 per cent by weight. Kolditz reported that the impurity occurred to an extent of about 9 per cent by weight* but that it could be easily sublimed from the product under reduced pressure. In fact* heating the product at 80° under 1 inn. pressure for 15 minutes was said to be sufficient for complete removal of the [PCl^] [PF^] impurity.

This could not be reproduced. Every attempt to remove the impurity^ resulted in failure. Heating the samples at 65° under 1-2 mm. pres­ sure in a system open to a condensation trap and pump only resulted in partial loss of sanple without a decrease in the [PCl^] [PFg] concentration. The same result occurred when the samples were heated at 65° in a highly evacuated system or at 65°* 80°> or 85° under 1-2 mm. pressure in a sealed system for periods up to 12 hours • X-ray powder diffraction patterns of t he sublimates from

52 heating the samples were similar to the patterns reported for ionic

PCl^F. The patterns were definitely not that of [PCl^] [PF^]”.

X-ray powder patterns of the ionic PCl^F samples taken after the purification treatments showed the lines for the reported pattern of ionic PCljjF. However* the measured 9-values were not reproduc­ ible among the various samples and were always 1-2 per cent smaller than the 9-values reported by Kolditz. A few additional* very weak lines were also present. These lines were undoubtedly present in

Kolditz* pattern but were not reported. An x-ray exposure time of

40 hours did not cause lines for [PC1^]+[PF^] tote revealed. It was assumed* therefore* that the samples were solid solutions of ionic PCljijF and [PCl^]+[PFg3“. Ionic PCl^F was the major component and therefore* determined the x-ray structure.

Curiously* a sample of pure [PC1^]+[PF^] was found to be quite volatile; in fact* the pure compound readily sublimed at room tem­ perature in a highly evacuated system. Solid and liquid condensates were produced by subliming [PCI ]+[PF.]“ . The sample was contained 4 6 in a vertical sample tube on a vacuum line. An upper section of the tube was cooled with a dry ice pack in order to trap the solid con­ densate. The liquid condensate was collected in a -19$° cold finger on the vacuum line. The solid condensate was not [PC1^]+[PF^] as might be expected. An x-ray powder diffraction pattern of the solid condensate contained strong lines characteristic of ionic PCl^F and other unidentified lines. At atmospheric pressure pure [PC1^]+[PF^]“ was found to sublime at 83° (743.8 nmu) into an air colled condenser.

This temperature is in contrast to the value of 135° reported by 5 *

Kolditz (28)• The volatility of pure [PC1^]+[PF^]* would suggest

that the compound could be easily removed from ionic PCl^F samples

^ by fractional sublimation; however* such is not the case. Infrared

spectra taken of the ionic PCl^F samples after some of the heating

treatments indicated an unexplained increase in [Pd^]+[PFg]“ con­

centration.

Results of chemical analyses of several pyrolysis products and

the attempted method of purifying the products is shown in Table 2.

The results which Kolditz reported for his purified ionic PCl^F are

reproduced for comparison. The analytical results show a chlorine

TABLE 2

CHEMICAL ANALYSIS DATA FOR IONIC PCl^F SAMPLES MADE FROM PYROLYSIS OF [PC1^]+[PF6]“

Sample Composition ($) stoichiometry Purification treatment Cl F P

1 65*7 17.3 16.9 rcl3.32Fl.6o 65°. 2-1/2 hr., 1-2 mm.' and 80° , 1/2 hr., 1 urn.j m b

2 66.1 18.5 80° * 11-1/2 hr., 1 mm.k — PC1 3.25F1.71 3 67.5 — — — 85°. 30 min., 1 rmnmm. b c Kolditz* 74.3 9.1 16.0 80°, 13 min., 1 mm. results rci^.05F0.93 7^.0 10.9 16.2 rcl3.98F1.10

Theory 73.9 9.9116.2 p c y

a Heated in a system open to a condensation trap and vacuum pump.

^ Heated in a sealed system.

0 Evacuation method unknown. content which is too low and a fluorine content which is too high for the stoichiometry of PCl^F. These results indicate that

[PCl^] [PF^] is present as an impurity to an extent of about 20 per cent by weight (20 per cent on the basis of the chloride determina­ tion; 29 per cent on the basis of atomic ratios)* However* the possible presence of other impurities such as [PC1^]+PC1^F^] should not be excluded even though they have never been identified*

Kolditz reported that his purified ionic PCl^F sublimed at approximately 175° and* under pressure* melted at 177°• The material

remelted at l6l°* The second* lower melting point was due to the presence of phosphorus pentachloride and [PClj, ] [FF6 3~ which occurred as decomposition products. Sample number 1 in Table 2 was found to sublime at 168° in a sealed capillary which was completely submerged in the oil bath. The capillary had a micro-size leak which acted as a pressure release during the heating process. In a small* com­ pletely sealed capillary the sample melted at 169° and remelted at

159-160°. The sample recrystallized very slowly at room temperature; therefore* a full day was allowed before determining the second melting point. Since ionic FCl^F is appreciably volatile* melting points and sublimation points are significantly dependent upon the manner in which the data are determined• The pressure in a capillary which is sealed at atmospheric pressure is increased to 1*5 atmos­ pheres at 175°• Therefore* care must be taken in interpreting melting point data for the products. Nevertheless* the observed data appear to be in good agreement with Kolditz* results* 56

B. Preparation of Ionic PCI. F by Direct Fluorinatlon of Phosphorus Pentachloride

General preparative procedures.— All attempts to prepare ionic

PCl^F by pyrolysis of [PC1^]+[PF^]“ were unsuccessful in obtaining

a pure compound. Therefore* some possible new preparative methods were investigated. A method was discovered by which ionic PCl^F

could be prepared more conveniently and in much higher yields by

directly fluorinating phosphorus pentachloride in arsenic trichloride

at temperatures above 70°. Arsenic trifluoride was the fluorinating

agent. The reaction proceeded according to the equation

AsCla 3PC15 + AsFj --- *4- 3P d ifF + A8 & 2 .

Yields from the reactions were about 90 per cent with 30-40 gm.

product being typical.

Kolditz demonstrated that [PC1^,]+[PF^]*" was apparently produced

in a single step reaction between arsenic trifluoride and a solution

of phosphorus pentachloride in arsenic trichloride. The reaction

was written

,i ^ AsClo * _ [PCl^] [PP6] + 2AsF3 --- 4 . [PCl^]'tTPF6] + 2AsC13#

However* ionic PCl^F can be produced in the same system but at a

temperature where [PC1^]+[PF^]“ is unstable in the solvent. The

critical temperature for the reaction was found to be ?0°. Below

70° [FC1^]+[PF^] was produced; whereas above 70° [PCl^l^CPF^]-

became unstable and PCl^F was formed. The high temperature fluori­

ne t ion did not appear to involve initial formation of [PCl^]+[FFg]~* 57 because the reaction stoichiometry was 3PCl^tlAsF^. If [PC1^]+[PF^]“ were formed initially and then decomposed into PCl^F and PF^» the reaction stoichiometry would have been lPCl^slAsF^. Also* the reac­ tion was noticeably exothermic which would not be expected if an intermediate species such as [PF.]“ were absorbing energy during its 6 decomposition »

The general procedure for the reaction was the following* Ar­ senic trifluoride was slowly added to a solution of phosphorus penta­ chloride in arsenic trichloride* The temperature of the reaction was maintained at 75^2°• The reaction stoichiometry was followed by accurately measuring the amount of arsenic trifluoride added to the solution* When the relative amount of arsenic trifluoride was within one per cent of the theoretical ratio 3PCl^slAsF^* dense white fumes were evolved from the apparatus** The resulting ionic PCl^F was precipitated at 0° overnight and then separated from the solvent by filtration* The weight of the dried product corresponded to within one per cent of the theoretical amount of PCl^F* calculated on the basis of the initial amount of phosphorus pentachloride* When the o reaction temperature was allowed to exceed 80 * yields were reduced because of slight decomposition of PCl^F into phosphorus pentachloride and [PClif]+[PF6]“.

* The fuming which occurred at the reaction end point might have been due to either PFe or a combination of several chlorophosphoranes• If the latter were the case* then the reaction might be used as a con­ venient method for generating various chlorofluorophosphoranea* 58

Purification experiments,— Even though the reagent stoichiometry of the reactions appeared to produce ionic PCl^F quantitatively» the product was contaminated with phosphorus pentachloride,

[Pd^]+[PC1^F] » and [PC1^]+[PF^] . The first two mentioned impuri­ ties were present in relatively small amounts and could be effectively removed by sublimation at 90° and atmospheric pressure over a period of 50 hours. The first two impurities could also be effectively re­ moved by pumping on the product at room temperature for 2 hours and then washing with nonpolar solvents. Sublimation in a fully evacuated system or in a system at 6 mm. pressure resulted In large losses of sample and partial rearrangement into more [PCI. ]+[PF As in the ^ 6 case of ionic PCI F prepared in the pyrolysis experiments,

[PC1^]+[PF^]“ could not be removed.

Chemical analyses of samples after the phosphorus pentachloride and [PC1^]+[PC1^F] impurities were removed showed that the samples contained 10-20 per cent less [PC1^]+[PF^]” than the ionic PCl^F samples obtained in the pyrolysis experiments. Results of chemical analysis gavel Cl, 70.2; F, 16.0; P, 17*6#; stoichiometry,

PCl^^^Fi^ 9 (PCl^F requires Cl, 73*9; F, 9*91; P* 16.256). Analysis of another sample gavet Cl, 70.596* If [PC1^]+[PF^]“ were the only significant impurity in the samples, then the analytical results indicate that the concentration of [PCI ] [PF,] in the samples was h- 6 about 11 per cent (10*6 per cent on the basis of atomic ratios;

12*8 per cent on the basis of chlorine determinations) • These results show significantly favorable changes in composition from the data for the samples given previously in Table 2. 59

Removal of the [PC1^]+[PF^3~ impurity from the various ionic

PCl^F samples was attempted by a variety of methods. Infrared spectra were used to estinate the purity of the samples after each purification treatment. A typical infrared spectrum in the 8-25 micron region taken of a sample of ionic PCl^F after removing a sub­ stantial portion of the phosphorus pentachloride and [PC1^]+[PC1^F]~ impurities is shown in Figure 10. (The impurities were never com­ pletely removed.) Figure 11 shows a typical spectrum of ionic PCl^F obtained by pyrolyzing [PC1^]+[PF^]” . The latter spectrum was taken of sample number 2 in Table 2. The samples were run as Nujol mulls.

The absorption frequencies observed for ionic PCl^F obtained from the direct fluorination experiments are tabulated in Table 3* The general assignments of the various absorption bands will be discussed in the following paragraph. Except for the presence of very weak absorptions due to phosphorus pentachloride and [PC1^3 [PCl^F] impurities* the frequencies are identical to those observed for ionic PCl^F obtained by pyrolyzing [PC1^]+[PF6 3*.

Reference spectra of pure CPC1^]+CPF^]_ are shown in Figure 12 and 13* The absorption frequencies observed for pure [PC3^]+[PF^]"» along with their general assignments* are tabulated in Table Refer­ ence spectra of pure phosphorus pentachloride ([PC1^3+ [Piy") are shown in Figures lb and 15* The characteristic absorptions for the

[FC1^]+ cation are readily discernible in the spectra of ionic PCl^F*

LPC1^3+[PF63~, and phosphorus pentachloride. The frequencies observed for the [PC1^3+ cation in the three materials ares ionic PCl^F*

685 and 6^3 cm."^; pure [PC1^]+[PF^3” » 712 and 662 cm.”^; and MO X(M, .80 200 220 24 0

FiQurt to. Infrorotf spectrum of ionic PCI4F prepared by fluorinotion of PCI3 with AsFj in AtClg. (Nujol mull).

1000 cm' 900 QO

HE

20 20

JO JO

50 .90 60 80 70

ao 8 0 too 180 x(mJ.80 200

Figure it. Infrored spectrum of ionic PCI4F by pyrolysis of [PClJ+[PFg]7 Somplo 2 , (Nujol mull). 61

1300 1000 900 00 00

RHIJSlSSSSSHKSSBKSSS r.rtM - -in.

20 20

9fi 30 ■ Si 30

40 .40

30 .90 JSO j90 70 70

00 9j0 tOO 120 *4.0 ISO ISO too

Figur* 12. Infrared *pectrum of pur* [PC**]*[pp6l 7 (Pr****d p*ll*t)

Figur* 13. In frare d •poetrum of pur*lpcUrLpf*3*’ (Nujol mull). -■

m

m o x(>4, i«o «4J0

Figure 14. Infrared spectrum of phosphorui pentochloride QPCI^PfPCI^ (Pressed pellet).

I

tu * * o

Figure IS. Infrared spectrum [pc^tpcij; (Nu)ot mull). [PCI. ]+[PCl,r, 70^ and 652 cm, \ Those values are in good ^ 6 agreement with the values 707 and 653 cm.”'*' which were reported for [PC1^]+ in ionic phosphorus pentachloride (51)*

TABLE 3

INFRARED ABSORPTION FREQUENCIES OF IONIC PCl^F (NUJOL MULL)

Frequency (cnu-1) I Assignment

8^3 s [pf6 ]-

781 s P-F stretch ([PCl^Fj]") 766

752 vw»sh P-F stretch ([PC1,F]-)

723 vw»sh Nujol

685 w [PC1U ]+

643 s,br

587 w P O d j 580 vWfSh Molecular PCl^

557 vw (pf6]~

535 w,sh C r c v 2]-

525 m [pci4F2r 500 w [PCl^Fj]-

^78 m [PC^Fj,]-

**51* m [P01^F2r

I=intensity, s=strong» ro=mediuin, w*=weak» v=very» br*=broad and sh=shoulder. 64

TABLE 4

INFRARED ABSORPTION FREQUENCIES OF PORE [PCI*. ]+[PF. ] (NUJOL MULL AND PRESSED PELLET) 0

Frequency Assignment (cm#"1) I

881 vw>sh [f f 6]-

837 s*br [pp6r

783 w,sh c w $ r 741 vw cpp6 r

712 w 1 g _

662 s*br 1

573 w c w 6r 556 a o>F6r

528 w [pp6]-

I=intensity> s=strongf w=weak* v=very» sh=shoulder, and br=broad•

The main absorption frequencies for the [PF^]“ anion in pure [PCl^]

[PF^]"* as observed in Figure 13* are 837 and 55^ cm#"1 . These values are in good agreement with the values 845 and 559 cm#"1 which were reported for various metal hexafluorophosphate salts (52)# Two absorption bands occurring at 843 and 557 cm."1 in the spectra of the ionic PCl^F samples are therefore attributed to the [PF^]" ion in the [PC1^]+[PF^]" impurity. The spectrum of ionic PCl^F obtained from direct fluorinatlon of phosphorus pentachloride (Figure

10) shows a significant decrease in the relative intensities of the 6 5 of the [PFg]“ impurity absorptions as compared to the spectrum of ionic PCljjF obtained in the pyrolysis experiments (Figure 11)*

The concentration of phosphorus pentachloride in various ionic

PCljjF samples was estimated in the following manner* The octahedral

[PCI I* anion has a strong* broad absorption band at h49 cm.”^ (51)* 6 but this band cannot be seen in the spectra of impure ionic PCl^F samples. However* a Nujol mull of pure phosphorus pentachloride was found to contain molecular PCI- as well as the ionic fora. The molec- ular fora was observed to have a strong Infrared absorption at 580 cm.

(see Figure 15)* This value compares favorably to the frequencies

580 * 58^* and 591 cm.”-*- which have been reported for a strong funda­ mental vibration of molecular phosphorus pentachloride (51). The

580 cm."^ band observed for molecular phosphorus pentachloride must not be confused with an absorption due to phosphoryl chloride inpurity.

Phosphoiyl chloride has been reported to have an absorption at 58I cm.”’

(53)* The distinction between these two bands was not fully recognized in the past (9*22*51)* The absorption for the phosphoiyl chloride im­ purity can be seen as a shoulder on the high frequency side of the

580 cm.“l band in the spectrum of phosphorus pentachloride (Figure 15)*

The presence of both molecular phosphorus pentachloride and phosphoryl chloride impurity was confirmed by comparing Figure 15 with spectra of pure phosphoiyl chloride in Nujol and a Nujolmmull of phosphorus pentachloride which was adulterated with phosphoiyl chloride. Figure

16 shows a spectrum of phosphoryl chloride in Nujol. The strong ab­ sorption band due to phosphoryl chloride was observed at 587 cm.~^.

The spectrum of the FCl^-POCl^ mixture is shown in Figure 17* The 66

I I

s i

100 ISA I4J0 WO x(>4) I tO too CtO 2 4 0 Ftgurt 16. Infrorod apoctrum of P0Cl3 in Nujol.

r 600

■SSUbbb*S«ZS3I■

_ _ ilEimi|p^p=IifU^£i^igiiIlii!iH2»Eij|is^^aim======« iHr=I=5SllIij=gliis=i^g£5 5lf=ils=HliI'iil«|giiif:i|iEilil5gg5lfggliliiiEl i i l iliillllfl miiiiiiuiiiminniiniiimiiniifniinniiiinniiiiiiiiiiiiiiiMiiivmiiaiiuiR IOnanninnHmaimiiuoiiiHHniiiHaiHimBJiiiHiHiiiHiaaiiiitM iHniHniikM ■o ^=;|?;=:|5=ra====^:r=5sr??|s|=?f=jr=r:£si=j:55fi^r;=r;:r=:r;==:==rs=r=5t?ss=s-^rr:?s5=: 00 iliiiiiiiaiiiiiiiiiiiiiiiiiiiiuiiiiiiiiiiiiiiiiiiiiiiiaiiiiiiiiiiiiiiiiiiiiiiiiiiitiiii0 0 •o no ito mo no x((t) wo MO tto too Figura (7. Infrorod •poctrum of ptiotphorut por adult orototf with POCt3 . I Nujol mull). 67 absorptions due to molecular phosphorus pentachloride and phosphoryl chloride can be seen to be partially resolved at $80 and $87 cm.”* » respectively. The presence of phosphorus pentachloride in the impure

Ionic PCl^F samples was detected by observing the occurrence of the

$80 cm.”* band in the spectra of Nujol mulls of the samples.

The concentration of [PC1^]+ [PC1^F]” in the ionic PCl^F samples was estimated by comparing the infrared spectra of the aaitples to spectra of other samples which are known to contain [PCl^] [PCl^F]".

Products obtained from reactions between silver monofluoride with phos­ phorus pentachloride were found to be mixtures of [PCI. ]+[PCl,.F]” ** 5 and phosphorus pentachloride • The P-F stretching frequency for these samples was observed to be 752 cm.”* (see figures 26 and 27 in a

later section). In samples of ionic PCl^F which contained very small quantities of [PCI ]+[PCl F]-, the P-F stretching band for the impurity ** 5 could be detected as a weak shoulder on the low frequency side of the

781* 766 cm* ^ doublet in the ionic PCl^F spectra.

+ a* Attempted methods for removing [PC1^3 [PF^] from the samples were the following.

1. [PC1^]+[PF^] is only slightly soluble in arsenic trichloride; whereas ionic PCl^F is appreciably soluble. An impure sample of ionic

PClj^F was mixed with arsenic trichloride for 10 hours at 35°. A

product was precipitated very slowly from the filtrate and was found

to contain [PC1^3+[PF^3 in approximately the same concentration as

in the original sample. The precipitate which remained after the ex­

traction process was unchanged. The remaining sample was treated

again with arsenic trichloride. This time the mixture was heated to 80° for 30 minutes. About half of the precipitate dissolved. The mixture was allowed to cool until precipitation began (about 45°)» and then the mixture was filtered. Both the product isolated from the filtrate and the remaining precipitate appeared to have the same concentration of [PC1^]+[PF^]“ as the original sample. However* an absorption band at 500 em.”^ was missing in the infrared spectrum of the precipitate.

2. A variety of sublimation techniques were tried. A sample of pure [PCI ]+[PF,]“ was observed to sublime readily at room tempera 4 o ture in a highly evacuated system. At atmospheric pressure a sample of pure [PC1^]+[PF^]~ sublimed at 830 (743*8 mn.) into an air cooled condenser; in fact* a 0*3 gram sample completely sublimed at 90° in

1.5 hours. A sample of impure ionic PCl^F was observed to begin subliming at 76° (?43.8 mm.). These values are in contrast to the sublimation temperatures reported by Kolditz; that is* 135° for pure

[PClif]+[PF6] (28) and 175° for pure ionic PCl^F (29)* Sublimation of a sample of impure ionic PCI F at 90° and atmospheric pressure resulted in partial loss of sample without observable change in the

[PC1J+[PFJ concentration. The [PCI ]+[PF/.]"' was distributed be- q- o 4 6 tween the sublimation condensate and the residue. However* the sub­ limation was effective in removing the [PC1^]+[PC1 F]~ and phosphorus pentachloride impurities in the original sample. Figure 18 is an infrared spectrum of the sublimation condensate. The presence of

[PC1^]+[PC1^f3“ is indicated by its characteristic band at 752 cm."1, adjacent to the 781* 766 cm."^ doublet. The presence of phosphorus pentachloride is readily detected by observing the characteristic ab­ sorption of its molecular form at 580 cm."',•• Also* the bands at 478 69

t

i * o M>I) t«0

Figure 18. Infrared spectrum of sublimation condensole from purification of ionic PCI4F. ( Nujol m ull). 70 -1 and 454 cm. have become broadened and more Intense (compare to

Figure 10) which further indicates the presence of [PCl^^CPCl^l* and [PCI ]“ » respectively (see Figures 26 and 27 in a following D section for spectra of mixtures of [PCI, ]+[FClcF]~ and phosphorus 4 5 pentachloride).

Sublimation under reduced pressure produced the same results as

sublimation at atmospheric pressure* except much more sample was lost

in the evacuated systems. Heating the sample under 6 mm. pressure at

room temperature with a condensate receiver at 0° only removed phos­

phorus pentachloride from the sample. Sublimation treatments of the

sauries under 1 mm* pressure at temperatures ranging from 0° to 85°»

with a condensate receiver at -78°* did not change the [PClJ+]+[PF^]

concentration. Treatment of a sample at room temperature for 48 hours

in a fully evacuated apparatus* which had a condensate receiver at o -196 » resulted in a very large increase in concentration of [PCl^]

[PF63 and phosphorus pentachloride. However* the phosphorus penta­

chloride only occurred in the condensate. Infrared spectra of the

residue and condensate showed that the ionic PCl^F components in each

sample were appreciably changed. An absorption band at 500 cm."^

was missing from both samples. Also* the typical 781* 766 cm."^

doublet became a singlet absorption at 781 cm.”\ This observation

might be interpreted to mean that the normally cis octahedral

[PC14F23~ anion had rearranged into the trans isomer. A singlet ab­

sorption would be expected for the P-F stretching frequencies of a

trans [PC1^F23~ octahedral species because only the asymmetrical P-F

vibration mode would be infrared active. For the cis isomer, both 71 of the P-F stretching inodes are asymmetric; therefore both are in­ frared active*

3* The impure ionic PCl^F samples seemed to be insoluble in carbon tetrachloride* A sample was mixed with carbon tetrachloride for 10 hours at 70° with no apparent dissolution of the sample.

However, when the temperature was raised to cause gentle boiling

(about 80°), the sample slowly dissolved and fuming occurred from a drying tube outlet on a reflux condenser attached to the apparatus.

Evaporation of the solution at -26° on a vacuum line left a residue composed only of phosphorus pentachloride and [PC1^]+[PC1^F]"*

This experiment indicates that ionic PCl^F readily disproportionates under relatively mild conditions. Higher fluorine containing species» such as PCI2F3 , derived from the disproportionation would have evapo­ rated from the carbon tetrachloride solution,, thus accounting for the fuming.

The foregoing purification experiments indicated that ionic

PCl^F easily decomposed into phosphorus pentachloride and [PC1^3+

[PFg] • The observed decomposition of ionic PCl^F could involve an

equilibrium such as

3 PClj^F = 2 PCl^ + FCl^F^ (molecular or ionic forms) which could quite possibly occur both in solution and in the vapor phase* Several experiments were designed to test the validity of the proposed d ecomposition equilibrium. Ionic PCl^F was generated In

arsenic trichloride by treating phosphorus pentachloride with arsenic

trifluoride. The usual reaction technique described previously was 7 2 used. Howawer* the reaction was stopped when 95 per cent of the cal­ culated amount of arsenic trifluorlde had been added. Thus> the system contained 5 per cent excess phosphorus pentachloride» which should shift the above equilibrium to the left and thereby decrease the amount of [PC1^]+[PF^]~ in the product* The resulting product was found to contain substantially less [PC1^]+[PF^]" than previous pro­ ducts. However, when the excess phosphorus pentachloride was removed from the product by vacuum sublimation at room temperature* the con­ centration of [PG1^ 3+CPF6 ]“ was again the same as in the previous products. A reaction run to 50 per cent completion (6PC1^.:1AsF^) produced no change in the [PC1^]+[PF^]” concentration in the resulting product. This time the product was not subjected to evacuation. The excess phosphorus pentachloride was removed by washing with arsenic trichloride* and then the product was dried in a stream of dry nitro­ gen at roan temperature. This reaction at least proved that the compound [PC1^]+[PC1^F]“ cannot be prepared by the reaction

AsClq x 6PC1 + AsF- ---- 4. 3[PC1k ] [PClcF3 + AsCl_ . 5 J 750 5 3

Another test of the proposed decomposition equilibrium involved mixing pure phosphorus pentachloride with pure [ PCl^]+[ 3~ in arsenic

trichloride at 4-5° for 6 days. The amounts of the reagents corres­ ponded to the molar ratio 4PCl^rl[PC1^3+[PFg]~. After three days

of stirring* the solid phase had disappeared. No precipitation

occurred after cooling to 0° for three days. Also* no precipitation

occurred after partial evaporation of the solvent by vacuum distil­

lation. The solution was then evaporated to dryness at room temperature 73 In a fully evacuated system. Only a very small amount of unidentified solid residue remained. A large amount of a very volatile substance was observed to distill with the solvent. If a reaction did not occur between phosphorus pentachloride and [PCl^]+[PF^]'^t then some phos­ phorus pentachloride should have remained after evaporating the solvent.

The results of this experiment might be construed to indicate that a reaction occurred between phosphorus pentachloride and [PC1^]+[PF^]“ but that PCljjF was not produced.

Low temperature fluorination of phosphorus pentachloride .— All the attempted physical methods for removing [PC1^]+[PF^3“ from ionic

PCl^F were unsuccessful. Apparently, if pure ionic PCl^F were ever to be obtained, the material would have to be generated initially in pure form. A possible method for generating pure ionic FCl^F would seem to involve a controlled, stepwise fluorination of the octahedral

[PCl6 r anion in phosphorus pentachloride. Such a fluorination was attempted by treating phosphorus pentachloride with arsenic trifluoride at -78° in methylene chloride. However, only [PC1^]+[PF^]“' and excess phosphorus pentachloride were isolated from the reaction.

Fluorination in non-polar solvents.— Reactions between phosphorus pentachloride and arsenic trifluoride (molar ratio 3PCl^:lAsF^) were

run in carbon tetrachloride and 1 ,1 ,2 ,2-tetrachloroethane at 75°.

The products from both reactions were found to contain a large amount of [PC1^]+[PC1^F]" along with ionic PCl^F awl [PCl^l^PF^-. An infrared spectrum (Nujol mull) of the product obtained from the reaction

in carbon tetrachloride is shown in Figure 19. The spectra of both products were substantially identical to the spectra of sublimation QO -4

J-4-U-

*° Xt*> «® Figure 19. Infrored spectrum of ionic PCI4F prepored by fluonnotion of PCI$ with AsF^ in non-polor eolventt. (Nujot mull). 75 condensates (Figure 18) obtained from previous Ionic PCl^F samples.

The products were obtained in about 60 per cent yields based on the expected amount of ionic PCl^F. The product from the reaction in tetrachloroethane precipitated immediately upon cooling the reaction mixture; whereas> the product obtained in carbon tetrachloride pre­ cipitated slowly over a period of one month.

X-rav powder photography.— X-ray powder diffraction patterns of the ionic PCl^F samples showed the lines characteristic of the pattern reported by Kolditz (29) for pure ionic PCl^F. However* the observed

0-values varied widely among the various products and were 2-3 per cent smaller than the reported values. X-ray exposure times of 40 hours did not reveal lines characteristic of the impurities. The variance among the powder patterns and the lack of extra lines due to impuri­ ties indicate that the product mixtures were solid solutions of ionic

PCl^F* [PCl^]+[PFg]"» phosphorus pentachloride* and [PCllf]+[PCl^F]".

The largest 0-values observed for ionic PCl^F are tabulated in Table 5*

The values reported by Kolditz are reproduced for comparison. Theta values observed for pure [PC1^]+[PF^]" are recorded in Table 5 for future reference* since they were not reported in the literature (28).

Normally* d-values would have been tabulated* but other investigators in this field seem to prefer the use of 6-angles. The 0-values in

Table 5 are equal to one half the actual angle diffraction. A graphi­

cal comparison of the 0-values for two ionic PCl^F samples obtained in

the present investigation with the values reported by Kolditz is shown

in Figure 20. The x-ray powder patterns selected for the comparison

chart were those which had the largest 0-values of all patterns ob­ tained in the present investigation. Additional* very weak 76

TABLE 5

X-RAY POWDER DIFFRACTION DATA FOR IONIC FCI4 F AND [PCI^]+[PF6 ]“

(Cu Kq, Radiation* 11*45 cm. Powder Camera)

Ionic p c i */ +[p f 6]- Kolditz This work [PCI*] © (I) 6 (I) 0 (I) 9 (I) 6.5 w 7*28 w 30.29 vvw 7.5 v 8*89 m 30*9^ VW 10.8 vs 10.7 vs 10.30 w s 31.41 vw 12*5 s 12.5 m 12.64 vw 32.3** m 15*3 vvw 13*0 vvw 13.70 vs 33.38 w 15*2 vw 14.65 w I6.5 vs 16.6 s 15.54 m 17.0 vw l6.40 w 17.9 W 17.20 s 19.9 w 19.9 m 16.28 m 21.9 W 21.7 m 20.23 vw 23.4 w 23.1 m 20.91 m 24.4 w w 21.61 w 25.4 vw 22.25 m - 28.3 w 28.2 w 22*90 vw 29.5 s 29.3 m 23.58 w 30.3 W 30*2 w 24.79 vw 32.8 w 32.4 w 26.53 vvw 33.9 W 33*7 m 27.11 m 35*8 vvw 27.63 vvw 38.6 vw 28.19 m 29.76 m

I=visual intensity* s=strong» w=weak, and v=very. 77

Sample

Kolditz 1 1 a 1 _ i 1 1 1 11 11

A 11.1 1 1 1 1I 1 1 1

B - L 4 J L L , ------_ - ^ , 1 1 1 1 MK- 10 15 20 2 5 30 35 4 0 e

Sample A* 0-values from best x-raypowder pattern of ionic PCI4F obtained by pyrolyzing [PC^ + (fFe]“ •

Sample Bs 0-values from best x-raypowder pattern of ionic PCI4F obtained by directly fluorinating PCI5 with As F3

Figure 20. Comparison of x-ray powder diffraction patterns of ionic PCI4F samples prepared in the present investigation with the pattern by Kolditz . reflection lines were omitted from the pattern graphs of samples A and

B so that a visible comparison would not be distracted. Sample A is sanple number 2 in Table 2. Sample B produced the infrared spectrum shown in Figure 10.

The solid solutions of ionic FCl^F and [FClif3"ttP ^ ] “ seemed to behave as azeotropes. The composition of the sublimation condensates and the residues always remained approximately constant. Opposing vapor pressures of ionic PCl^F and [PCl^] [PF^]“ may have caused the total vapor pressure of the solid solution to reach a minimum value when the solution contained [PC1^]+[PF^]" at a concentration between

10 and 20 per cent by weight. Howevert disproportionstion of ionic

PCl^F into phosphorus pentachloride and [PC1^]+[PF^]" in the vapor phase would also account for the occurrence of [PClij.] [PF^]* in the various sublimation products. Both the formation of azeotropes and the occurrence of decomposition equilibria account for the failures in removing [PCl^^CPF^]” from ionic PCl^F.

Structure of ionic PCl^F.— If ionic PCl^F were tetrachlorophos- phoniura fluoride > [PC1^]+F“, as reported by Kolditz; then the infrared spectrum of the material (Figures 10 and 11) should show only absorp­ tions characteristic of the [PC1^]+ cation* There should be no absorp­ tion bands in the P-F stretching frequency region % nor should there be any bands in the P-Cl stretching frequency region for octahedral phosphorus. Figures 10 and 11 plainly show that the infrared spectrum of ionic PCl^F is much more cosplex than what would be expected for the ion structure [Pd^]+F • Two absorption bands occurring at 8*4-3 and 1 557 cm*" in the spectra of the ionic PCl^F samples are attributed to 79 [PF^]" ion in the [PC1^]+[PF^]“ impurity. Even after ignoring the

[PFg]“ absorptions, the occurrence of the remaining bands in the spectrum of ionic PClZj-F cannot be explained in terms of the ionic struc­ ture [PC1^]+F • A strong doublet at 78I and 766 cm.”^ occurs at fre­ quencies which are slightly lower than the normal P-F stretching frequency region for pentavalent phosphorus but still in the region for trivalent phosphorus (5*0* Also, other absorptions are observed between 5^0 and 4-50 cm.-1 which are in the P-Cl stretching frequency region for octahedral phosphorus (51) or neutral phosphorus-chlorine species. A reasonable formulation for ionic PCl^F which adequately accounts for the observed infrared spectra is tetrachlorophosphonium cis-tetrachlorodifluorophosphate, [PClJ+]+[PCl^F2]". The formulation is based upon the following reasoning.

1. The presence of the [PCl^f cation in the compound was confirmed by observing its characteristic infrared absorptions at 685 and 643 cm."3-. The presence of the impurity [PC1^]+[PF^]" cannot account for the observed intensities of the [PC1^]+ absorptions. Spectra of pure [PCl^]+[PFg]“ (Figures 12 and 13) show that the absorption band due to at 837 cm." is more intense than the 662 cm.” band due to [PC1^]+ and that the other main band for [PF^]” at 55^ cm."'*’ is equally as intense as the [PC1^]+ band. However, for ionic PCl^F

(Figures 10 and 11) the band attributed to [PF^]“ at 843 cm.-1 is less intense than the band due to [PC1^]+ at 643 cm.-1. The other band for tPFgl"- at 557 cm.“l is seen to be quite weak. Also, the relative in­ tensities of the bands due to [PF^]” and [PC1^]+ in ionic PCl^F were found to be variable. 80

2m The occurrence of absorptions in the P-F stretching region

indicates the presence of P-F bonds in the compound (in addition to the [PF^]“ impurity). Furthermore* these absorptions occur in the low frequency end of the P-F stretching region which indicates that the involved phosphorus has a relatively small electronegativity. A decrease in electronegativity in going from a cationic or neutral phosphorus species to an octahedraljhosphorus anion ([PCI. F0]~) M' 2 would be expected.

3* Phosphorus-chlorine bonds in addition to those in [PCI 4 are present in the compound as indicated by absorptions in the 540 to

450 cm» region* The P-Cl bonds which absorb in this region must be on an anion since the presence of a cation ([PC1^]+ ) was confirmed.

4. The complex octahedral anion probably contains the two fluorine

atoms predominantly in cis positions* as indicated by a doublet ab­

sorption in the P-F stretching region. A singlet absorption would be

expected for trans fluorines.

5* The formulation [PFC1^]+C1** is eliminated on the basis of an

infrared spectrum reported for the compound [BFCl^]+[SbClg]“ (55).

This compound was observed to have the following infrared absorptions

in the 8-25 micron region: 1018 * 986 * 9?4*. and 697 cm.”\ None of

these values is comparable to those observed for ionic PCl^F (Table 3).

6. Ionic PCl^F was observed to form solid solutions with phos­

phorus pentachloride, [PCl^+CPCl^-, as well as with [PCl/f]+[PF6r .

Solid solution formation would indicate similarities in the complex

ion structures of the various components• 81

7* The proposed formulation [PC1^]+[PC1J^F2]* is entirely reasonable because of the precedent established by the series

[PCl4]+[PCl6r , [FCllf]+[FCl5F]“, [PCl4 ]+[PCl5Br]“ and [SbCl^]".

C. Solution Properties of Ionic PCl^F

Cryoscopy and conductance measurements.— The apparent molecular weight of a sample of ionic PCl^F was determined in nitrobenzene by the freezing-point depression method* The measurements were made at concentrations ranging from 0*008 to 0*07 molal. A freezing-point depression constant of 6*79°C./molal was used for the nitrobenzene

solvent* The freezing-point constant was determined experimentally by using solutions of napthalene at concentrations of 0 .054-3> 0.0855*

and 0.1525 molal. The value 6.79°C*/molal is in agreement with the

reported value 6.85°C./molal (56). The accuracy of the value was

checked by determining the apparent molecular weight of j>-dichloro-

benzene at a concentration of 0.1298 molal. The experimental molecular weight was 1*4-8 which is in agreement with its formula weight 1*4-7 •

The sample of ionic PCl^F used for the cryoscopy measurements produced the infrared spectrum (Nujol mull) shown previously in

Figure 10. The chemical composition of the sample was given previously

on page 58. Results of the cryoscopy measurements are shown in Table 6.

Three limiting possibilities exist for the character of ionic PCl^F in

the nitrobenzene solutions. 1) If the material were completely

ionized into the ions [PC1^]+ + [PCl^Fg]** then the expected apparent

molecular weight would be 192. 2) If the material were completely

ionized into the ions [PC1^]+ + F~, then the expected apparent 82

TABLE 6

APPARENT MOLECULAR WEIGHT OF IONIC PCl^F IN NITROBENZENE BY FREEZING-POINT DEPRESSION

Sample Wt. solute Wt. solvent Molality At App. M.W (g.) (g.) <°c)

1 0.2U44 18.79 0.0678 0.433 204 2 0.1099 15.22 0.0376 0.242 203

3 0.1176 30.51 0.0201 0.134 195 4 0.0957 59.95 0.00832 0.059 184

molecular weight would be one half 192 or 96* 3) If the material existed in the form of molecular PCl^F, then an apparent molecular weight of 192 would be ecpected. Intermediate values would arise from ion pair formation and/or equilibria between ionic and molecular forms. The results shown in Table 6 indicate that ionic PCl^F does not ionize into [PCl^J + F™ in nitrobenzene. Except for the very dilute solution (sample 4), the apparent molecular weights are seen to be relatively constant (within 5 P®r cent) over a concentration range of 0*02 to 0.08 molal. The deviation in the value of 184 for the very dilute solution from the expected 192 is within the experi­ mental error in the determination of the freezing-point depression.

The results of the cryoscopy measurements suggest that ionic PCl^F exists as the ions [PC1^]+ + [ P C l ^ ] - or as molecular PCl^F in nitrobenzene. The fact that the apparent molecular weights are con* slatently higher than 192 and also relatively constant over a range of concentrations indicates further that an equilibrium between the molecular and ionic forms might exist. Such an equilibrium could be written as 2PBV_ _ “ + CPcalfF2r.

If the above equilbrium existed* then an apparent molecular weight of 192 would be expected since the same number of particles occur on either side of the equilibrium expression. Values slightly higher than 192 would result from ion pair formation.

Results from conductance measurements on dilute nitroben­ zene solutions of ionic PCl^F are given in Table 7. Equivalent con- -12-1 ductances of 28-30 ohm cm* eq. have been reported for strong electrolytes in nitrobenzene (57)* Since equivalent conductances of 1 2 only 1.6 and 1.9 ohm"1 cm. eq."* were observed* ionic PCl^F might be either a weak electrolyte in nitrobenzene or the concentration of the ions [PC1^]+ + [PCl^F^] is small due to an equilibrium between the ionic and molecular forms. At least some form of ionization must have been present because the specific conductances of the solutions were 10-20 times greater than that of the pure solvent (8.3 x 10"? ohm"'*' cm."1 for the pure solvent compared to 19 x 10"^ and 8.0 x 1 0 ^ ohm"^ cm.-'*' for the solutions.

Infrared spectroscopy of ionic PCl^F solutions— An infrared spectrum in the 8-25 micron region of a saturated nitrobenzene solu­ tion of ionic PCljjF is shown in Figure 21. A comparison of the observed absorption frequencies with frequencies reported for gaseous molecular PCl^F (58) shows that ionic PCl^F converts largely into the molecular form in nitrobenzene# An absorption at 656 cm."*’ which TABLE 7

CONDUCTANCE DATA FOR SOLUTIONS OF IONIC PCl^F IN NITROBENZENE AT 25°

b b c A A Sample Wt. solute Vol. Soln. C0 a R© Rg C K PCl^F (PC1^F)2

C®11*) (eq. /I.) (ohm"^) (ohm"^cm.“^) (ohnf^cm^eq.”1)

Pure solvent — — — 10000.0 9997.9 2.1xl0“8 8.3xl0~7

1 0.0957 50.0 0.00996 10000.0 9950.4 4.97xl0“7 1.88xl0~5 1.89 3*78

2 0.0443 50.0 0.00462 10000.0 9979.8 2.02xl0“7 7.98x1c-6 1.55 3*10

Concentration based upon the formula PCl^F.

^ Bridge readings (see text), c “1 i Specific conductance corrected for the pure solvent. A cell constant of 39*5 c®1* was used. 85

1300 '

LLlLLlU U iiliiU 240 M m ) Infrorod spectrum of osoturoted solution of ionic PCI4 F in nitrobentene .

; 1 :

hr:

c 1 i ? I n 1 ? irn it-rn 100 120 W O * 0 X (M ) >60 200 2 2 0 2 4 0 2 2 . Infrarod spectrum of a saturated solutionof Ionic PCI4F in a c e to n i t r i l a . occurs as a shoulder on a strong absorption band of the solvent indi­ cates the presence of the [PC1^]+ cation* This value is in agreement —1 + with the value 653 cm*” which was reported for [PCl^] in phosphorus pentachloride (51)* Molecular PCl^F does not have an absorption frequency comparable with this value* An absorption occurring at

8t*7 cm*"^ indicates the presence of the [PF^]~ ion which was known to be present as an impurity. This value is in agreement with the value

8*+5 cm.”*- which was reported for various metal [PF^]“ salts (52). An infrared spectrum of a 0.06 molar solution of ionic PCl^F also showed the presence of the[PCl^]+ cation and molecular PCl^F* However* the solution was too dilute to show the strongest band due to [PFg]”*

Since the intensity of the strongest band due to [PF^]^ is greater than the intensity of the strongest absorption band of [PCI. ]+ in pure

[PCl^] [PF^l (Figure 13)» the presence of the [PC1^]+ cation in the spectrum of the ionic PCl^F solution cannot be attributed entirely to the [PCl^] [PF^]“ impurity. The observed absorption frequencies of ionic PCl^F in nitrobenzene are tabulated in Table 8 * The reported values for molecular PCl^F are included for comparison*

The proposed equilibrium between the molecular and ionic forms»

21X3^ - [PClJt]+ + [PCl4F2r is thus supported by the infrared spectra of nitrobenzene solutions of ionic PCl^F. The proposed equilibrium is analogous to the equilibrium

2pci5 * Cpci^]+ + [pci6r TABLE 8

INFRARED ABSORPTION FREQUENCIES OF A SATURATED SOLUTION OF IONIC PCI4 F IN NITROBENZENE AND OF GASEOUS MOLECULAR PCl^F

Ionic PCl^F Reported for molecular PCl^F -1 Frequency (cm.“^) Frequency (cm. ) and intensity and intensity

1200 vw

1020 w 1025 w

937 v

892 w 902 m

870 w*sh 867 w

847 m CCPF6]“

790 m 778 vs

702 s 725 m

676 w 686 w

656 s ([PCl/f]+)

623 s 626 m

588 s 601 vs

548 w 541 w

510 vw,sh C[p c i ^f 2F?)

500 w

488 w ([PC1^F2]“?)

442 m ([PC1^F2]"?)

423 w»sh 427 w s=strongt m=raedium» w=weak, v=very» and sh=shoulder* 88 which has been reported (9) for phosphorus pentachloride in nitro­ benzene*

Kolditz (29) reported an equivalent conductance of 40 ohm"^ 2 «1 i 9 —1 cm* eq* (14 to 26 ohm“ cm* eq.” calculated from his data) for acetonitrile solutions of ionic PCl^F. The high conductance lent support to his argument that ionic PCl^F was indeed ionic* An infra­ red spectrum of a saturated acetonitrile solution of ionic PCl^F is shown in Figure 22* The spectrum is essentially the same as the spectrum of the nitrobenzene shown in Figure 21* The presence of both the ionic and molecular forms are indicated* Also* the presence of the [PF^]” impurity is indicated by its characteristic absorption at 847 cm*“\ Therefore > an equilibrium between ionic and molecular forms also exists in acetonitrile solutions*

The strongest evidence which Kolditz offered in support of his formulation [PCI ]+F ” was the results of his cryoscopy measurements of anhydrous acetic acid solutions* Kolditz obtained an apparent molecular weight of 80 which is somewhat less than 96 or one half the formula weight of PCl^F. His results indicated that ionic PCl^F was completely ionized into [PC1^]+ + F*. Infraredspectra in 2*5-8 micron and 8-25 micron regions of anhydrous acetic acid solutions of ionic PCl^F are shown in Figure 23* The spectrum in the 8-25 micron region clearly lacks an absorption characteristic of the [PCI ]+ cation (expected at about 650 cm*"^)* A strong absorption at 585 cm* indicates the presence of a large amount of phosphoryl chloride* This value is in agreement with the reported frequency of 581 cm*“^ for phosphoryl chloride (53)* Weak absorptions occurring at 847 and 89

4 0 0 0 t 2000 (

1

'* u ri

x (>*)

typo

;

00 MM 120 MjO *0 MO 200 220 240

Figure 23. Infrared spectra of Ionic PCI4F In anhydrous oeotic odd. 9 0

557 cm* 1 indicate the presence of the stable [PF^]“ impurity which was originally present in the sample* Spectra of pure anhydrous acetic acid are shown in Figure 24 for comparison* A strong absorp­ tion at 1810 cm. ^ in the spectrum of the ionic PCl^F solution

(Figure 23) is attributed to acetyl chloride* This absorption is missing in the corresponding spectrum of pure acetic acid. Figure 25 shows a spectrum of a dilute solution of acetyl chloride in anhydrous acetic acid. The occurrence of the absorption band at 1810 cm,"^ confirms the presence of acetyl chloride in the ionic PCl^F solu­ tions* Furthermore* when the ionic PCl^F solutions were prepared* strong effervescence occurred. Therefore* it is assumed that ionic

PCl^F reacts with anhydrous acetic acid according to the equation

PCl^F + CHyiOOH - CHyjQX + POX^ + HX

where X = Cl or F.

The hydrogen halide escaped during the solution formation and was not detected in the infrared spectra. The solvolysis reaction be­ tween ionic PCi^F and acetic acid is consistent with Kolditz * cryos­ copy results. One particle of ionic PCl^F produced two particles in solution. The third product* hydrogen halide* partially escaped but some undoubtedly remained in solution. Thus, the freezing-point depressions which Kolditz observed for his acetic acid solutions corresponded to two moles plus a small fraction of a mole of particles per one mole of PCl^F. Therefore* he obtained an apparent molecular weight of less than half the formula weight of PCl^F. Ateorbonct A btorbonc*

,Ckf

u

i-tt

♦ f M l

II !■■■

VO 40 0 0

I

i 1 I I

Figure 25. Infrared spectrum of ocetyl cMorlda in anhydrous acafie acid. 93 D. Attempts to Obtain Nuclear Magnetic Resonance Spectra

Nuclear magnetic resonance measurements were attempted on saturated acetomitrile and methylene chloride solutions of ionic

PCl^F. Neither phosphorus-31 nor fluorine-19 resonance signals could be observed for the samples• Only the signals due to the external reference samples were observed. Previous experiments on saturated solutions of phosphorus pentachloride in various sol­ vents (polar and nonpolar) were only successful in producing very weak phosphorus-31 resonance signals. Normally, a solution would be expected to give a very intense signal if the solution contained as much phosphorus or fluorine as did the solutions cf ionic PCl^F or phosphorus pentachloride. Apparently, the signals could not be observed because they”were greatly broadened. The reason for the broadening was undoubtedly connected with extremely rapid relaxation processes of the excited nuclei. Staall relaxation times and broad lines are known to be direct results of broadened energy levels (59).

Fluctuating electric fields produced by the neighboring quadrupolar chlorine nuclei could cause splitting in the energy levels and rapid relaxation processes.

It was unfortunate that the nuclear resonance experiments were unsuccessful. Observation of multiplet structure (spin-spin coupling) in either phosphorus-31 or fluorine-19 resonances might have con­ firmed the presence of P-F bonds in solutions of ionic PCl^F. How­ ever, according to Kolditz, ionic PCl^F exists as [PC1^]+ + F“ in acetonitrlle• Therefore* the failure to observe fluorine resonance 9 * might be construed to be indicative of an absence of free fluoride ions in solution*

E* Fluorination of Phosphorus Pentachloride with Silver Monofluorlde

Reactions between phosphorus pentachloride and silver mono­ fluoride were run under a variety of conditions. The purpose of the reactions was to prepare ionic PCl^F according to the equation

Solv* PC15 + AgF ------^ PCl^F + AgCl*

The reactions were performed in the solvents methylene chloride» acetonitrile* carbon tetrachloride* benzene* and arsenic trichloride*

Incompletely dried silver monofluoride was observed to react rapidly^ with phosphorus pentachloride at room temperature* A mixture of various solids containing mostly [PC1^]+[PF^]” was produced* The silver halide precipitate was a solid solution of silver monofluoride and silver chloride. An x-ray powder diffraction pattern of the precipitate consisted of lines which had 8-values intermediate to those for pure silver chloride and pure silver monofluoride* Anhyd­ rous silver monofluoride reacted very sluggishly with phosphorus pentachloride. Extensive reactions in refluxing solvents seemed to be terminated when the molar ratio of AgF:AgCl reached approximately

1.0* However* x-ray powder patterns of the silver halide precipitates in these reactions produced superimposed patterns of silver chloride and silver monofluoride. Since only half of the fluorine in silver monofluoride seemed to be available for fluorination* a molar ratio of lPCl^:2AgF was always employed* 9 5

A reaction was run in refluxing methylene chloride (about 43°) for 73 hours. Hie reaction product, isolated from the solvent by vacuum distillation* produced an x-ray powder pattern which closely resembled patterns of both ionic PCl^F and phosphorus pentachloride*

Chemical analysis of the product gave: Cl, 81.8; F, 5*34-; P* 15*9?&; stoichiometry, P C I ^ ^ F q ^,. (P^Cl^F requires: Cl, 79*8; F, 4-.75*

P, 15.5# and PCl^F requires: Cl, 73*9; F, 9*91; P* 16.250* The x-ray pattern and the analytical data indicated that the product con­ sisted of approximately equimolar amounts of ionic PCl^F and phos­ phorus pentachloride. An infrared spectrum of the product (pressed pellet) is shown in Figure 26. Another product obtained from a simi­ lar reaction after 209 hours (the solvent was removed by vacuum distillation at -78° over 3 days) produced an x-ray powder pattern which very closely resembled a pattern reported (30) for the compound

[PC1^]+[PC1^F]“. The pattern also resembled that of phosphorus pentachloride. An infrared spectrum of the product as a Nujol mull showed the presence of a large amount of phosphorus pentachloride*

The spectrum had the same appearance as that in Figure 26 except for the presence of an additional band due to molecular phosphorus penta­ chloride at 580 cm*”^. Attempted removal of phosphorus pentachloride from the product by vacuum sublimation at 0° (sublimate receiver at

-196°) for 88 hours resulted in decomposition of the product into phosphorus pentachloride, ionic PCl^F, and a large amount of volatile liquid* The liquid gradually converted into a white solid over a period of one week. The white solid was identified as a mixture of ionic PCljjF and [PCl43+[PCl^F]~ by its characteristic infrared 1300 tooo cm 300 40 0 00 00

20 20

30 JO

J 40 4 0

.30 60 TO

00 00 60 100 120 140 n.o 100 200 220 2 4 0

Figure 26. Infrared spectrum of product from reaction between AgF and PCI* in CH2Cl 2 ■ Product is presumed to be a mixture of [PCU) + [PCi5F]- and pC i^PC isj: (Pressed petlet).

cm1 1300 1000 500 400 00 00

.20 20

30 JO

2 40 .40

3 0 .30 60 00 70 7 0

JQO GO 60 too MO no X(M, too 200

Figure 27. Infrared spectrum of product from rea'ction between AgF and PCU in bensene. Product is presumed to be o mixture of [PCI^^CtjF]- and [PCti]*(PCtg]7 (Nujol mull). 97 spectrum. Kennedy and Payne reported (30) that the compound [PC1^]+ •* [PCl^FJ disproportionates into phosphorus pentachloride and molec­ ular PCl^F at 110°. However» it is not unreasonable to assume that

[PCl^]+trci5F]" would decompose slowly in a highly evacuated system.

Since the product from the 209 hour reaction decomposed into phos­ phorus pentachloride and molecular PCljjFt the product was assumed to be [PC1^]+[PC1^F]“ but contaminated with phosphorus pentachloride.

Even though the product from the first mentioned reaction (73 hours) produced an x-ray pattern resembling ionic PCl^F, its infrared sp>ectrum was the same as the product identified as impure [PC1^]+

[PCl^F]". Since infrared spectroscopy was a much more reliable method of identification than x-ray powder photography* and since the chemi­

cal analysis of the first mentioned product gave the result

PCl^ ^F q ^ 5 » the analyzed product was deduced to be a mixture of

[PCl^] [PCl^F]- and phosphorus pentachloride also.

Another product was obtained by treating phosphorus pentachloride with silver monofluoride in benzene at 60° for 100 hours. The product was isolated from the solvent by evaporation at atmospheric pressure

and room temperature in a stream of dry nitrogen. An x-ray powder

pattern of the product was identical to the pattern produced by the

product from the 73 hour reaction in methylene chloride; that is, the

pattern was very similar to those of ionic PCl^F and phosphorus penta­

chloride. An infrared spectrum (Nujol mull) of the product is shown

in Figure 27* The presence of phosphorus pentachloride was detected

by observing the occurrence of the 580 cm."^ absorption band due to

the molecular fora. Absorption frequencies of the [PC1^]+ cation 9 8 were 685 and 645 cm*”\ which agrees with the reported values 707 and 653 cm* . A band at 752 cm* is attributed to theP-F stretching frequency in the [PCI F] anion* This assignment was based upon the same reasoning as given previously on page 79. The two remaining broad absorption bands in the spectrum at 472 and **53 cm*~^ were attributed to P-Cl bonds in the [PC1_F]“ anion and the [FC1*]~ 5 P anion, respectively. The **53 cm.**^ band is in agreement with the value ****9 cm*-'*' reported for [PC1^]“ (51). The absorption bands due to [PC1^]+[PC1^F]” and phosphorus pentachloride shown in Figure 27 can be readily detected in the spectra of impure ionic PCl^F samples shown in Figures 18 and 19*

Phosphorus pentachloride (15 g.) was treated with silver mono- fluoride in gently refluxing arsenic trichloride for 210 hours*

About 5 B* of product gradually collected in the reflux condenser

on the apparatus* An infrared spectrum of the product (Nujol mull) was identical to the spectra of ionic PCl^F samples obtained in the

pyrolysis and direct fluorination experiments (see Figures 10 and 11).

Fluorination of phosphorus pentachloride with silver monofluoride

in acetonitrile at room temperature for 22 hours produced dark colored

oils and a small am cunt of solid. An infrared spectrum of the solid

could not be Identified. However, the presence of [PC1^]“ in the -1 product was indicated by a strong, broad absorption band at **47 cm* •

Absorptions for the [Pd^]+ cation were not present. Not enough

sample was isolated for chemical analysis*

Mixing phosphorus pentachloride with silver monofluoride in re-

fluxing carbon tetrachloride produced very slight fluorination* Reactions between phosphorus pentachloride and potassium bi­ fluoride in refluxing benzene and methylene chloride produced potas­ sium hexafluorophosphate. The products were identified by infrared spectra (Nujol mull). Only absorptions for the [PFg]“ anion at 83^ cm. ^ and 559 cm*“^ were observed. V. SUMMARY

A reinvestigation of an ionic compound reported to be tetra- chlorophosphonium fluoride* [PCI ]+F * was undertaken* Infrared spectra of the solid material provided strong evidence that the compound was tetrachlorophosphonium tetrachlorodifluorophosphate*

[PCl^]+[PCl^Pg]**. The infrared studied indicated further that the fluorine atoms were located predominantly in cis positions on the octahedral anion*

Cryoscopy and conductivity measurements of nitrobenzene solu­ tions and infrared spectra of nitrobenzene and acetonitrlle solutions showed that ionic PCl^F exists in equilibrium with its molecular

FCl^F modification in polar solvents* The following equilibrium was proposed:

2PC14F = [PCl^r + [p c i 4f 2]~ .

Infrared spectra of anhydrous acetic acid solutions showed that ionic PCljjF reacts with acetic acid to form phosphoryl chloride* acetyl halide* and hydrogen halide* Therefore* the cryoscopy measure­ ments reported by Kolditz for anhydrous acetic acid solutions were proven to be invalid.

A method was discovered for synthesizing ionic PCl^F by fluor- lnating phosphorus pentachloride with arsenic trifluoride in arsenic

trichloride* The method was found to be very convenient for preparing

1 0 0 1 0 1 large quantities of ionic PCl^F. Yields from the reaction were about

90 per cent of theory. Impurities in the product were [PC1^]+[PF^]"»

[PCli+]+[rci^F]*"> and phosphorus pentachloride. Similar reactions in nonpolar solvents also produced ionic PCl^F. However* a larger

amount of [PC1^]+[PC1^F]“ impurity was obtained.

Contrary to the findings of Kolditz, ionic PCl^F could not be isolated as a pure compound. The material was invariably obtained

as an inseparable solid solution of ionic PCl^F and [PCl^] [PF

The presence of the impurity was found to result from disproportions-

tion of ionic PCl^F into [PC1^]+[PF^]" and phosphorus pentachloride.

Chemical analyses of ionic PCl^F samples showed that the concentration

of [PCl^] [PFg] impurity ranged from 12 to 20 per cent* depending

upon the method of preparation.

Nuclear magnetic resonance could not be used for investigating

liquid solutions of ionic PCl^F. Resonances could not be observed

because of extreme line broadening due to large quadrupolar inter­

actions with the surrounding chlorine atoms. X-ray powder photog­

raphy was found to be generally inadequate for identifying impure

samples of ionic phosphorus chlorofluorides» because the compounds

usually occurred together as solid solutions# Since the ionic phos­

phorus halides were quite volatile* reporting of melting points and

sublimation points was found to be meaningless unless exact experi­

mental conditions were also given.

A method was developed for preparing relatively pure* anhydrous

silver monofluoride on a large scale. Reactions between silver mono­

fluoride and phosphorus pentachloride in nonpolar solvents produced 1 0 2 mixtures of phosphorus pentachloride and [PCI., ]+[PCl-F]~. The com- ponents in the mixture were identified by combined data from chemical analysis* x-ray powder photography % and Infrared spectroscopy*

Some convenient techniques were developed for performing experi­ ments on highly moisture sensitive systems.

The major result of the present investigation was experimental evidence which showed that ionic PCl^F exists as [PCl^^CPCljijf^]”*

Thus* ionic PCl^F was shown to be a member of the series of known

compounds [PC1^]+PC16]“ , [PCl^CPCl^F]", and [PCl^j^PF^]". A

very important but general result of the Investigation was that

phosphorus(V) halide systems were found to be much more complex

and less well defined than previous investigators had indicated. VI, BIBLIOGRAPHY

1# Clark* D.» PowellH. M., and Wells* A. F,» J* Cham, Soc.» (1942) 642.

2. Mouren*. H.* Kagot* M.* and Wetroff* G.* Compt. rend..* 205* 345 (1937).

3* Brockway, L. 0.* and Beach*. J. W.» J. Am. Chem. Soc.* 6o» 1836 (1938). “

Rouault* M*> Ann. Phys.* 3£* 78 (1940).

4. Huettertles* £• L«* Bit her, T. A.* Farlow, M.» and Coffman* D. D., J. Inorg. Nucl. Chem.* ^6* 52 (i960).

5. Beattie* 1. R.* and Webster* M. * J. chem. Soc. (1961) 1730.

6. Holmes* R. R.* Gallagher* W. R.* and Carter* R. P.* Inorg. Chem., 2* 437 (1963).

7. Holmes* R. R.» and Patrick* W.* Inorg. Chem.* 2* 433 (1963).

6. Krakowiecki, S.* Roczniki Chem.* 1J>* 197 (1930).

9. Petro* V. P.* "Cryoscopic, Spectroscopic and Conductometric Study of Phosphorus Pentachloride in Various Nonaqueous Solvent SystemsUnpulbished dissertation* The Ohio State University* 1964.

10. Papov* A. I.* and Schmorr* E. H.* J. Am. Chem. Soc.* 74, 4672 (1952).

U . Harris* G. S.* and Payne* D. S.* J. Chem. Soc. (1956) 4613*

12. Holmes* R. R.* and Bsrtant* S. P.* J. Am. Chem. Soc.* 80* 2980 (1959).

13. Beattie* I. R.» and Webster, H.* J. Chem. Soc. (1961) 1730.

14. Muetterties* E. L.» and Mahler* W«* Inorg. Chem.* 4* 119 (1965).

15. Gutman, V.*Manat»h.* j£, 583 (1952).

Gutman* V.* and Mairinger* F.» Z. anorg. u. allgem. Chem.» 289 279 (1957). 103 104

15* Failkov, la. A ., Kuzemko, A. A ., and Kostromino, N. A., Ukrain. Khim. Zhur., 21, 556 (1955)*

Spandau, H., and Brunneck, B. Z.» Z. anorg. u. allgem. Chem., 278 197 (1955)

Kolditz, L.» and Hass, D., ibid., 294, 191 (1958).

16• Sharp, D. tf. A ., and Sheppard, N.» J. Chem. Soc. (1957) 6?4.

17* Cronander, A. W«, Ball. Soc. Chlm. France, 12, 499 (1873)*

18. Gutman, V., Manatsh., 82, 473 (1951); Z. anorg. a. allgem* Chem., 264, 151 (195lT.

19. Payne, D. S. Quart. Revs., 1$, 173 (196l).

20. Becke-Goehring, M., and Lehr, W., Chem. Ber.» 2ft» 1591 (1961)*

Becke-Goehring, M., and Fluck, E«, Angenr. Chem. *, 74, 382 (1962).

21. Bryce-Sbith* D., and Perkins, N. A., J. Chem. Soc. (1962) 1339*

22. Beattie, I. R.» and Vebster, M., J. Chem. Soc. (1963) 38.

23. Kirsanov, 0. V., and Federova, Q. K., Dopovidi Akad. Nauk. Ukr. RSR (i960) 1086.

24. Yagupol'skii, L. M.» and Yufa, P. A., Zhur. Obshchei Khim., 30, 1294 (I960).

25. Lange, W., Ber. dtsch. Chem. Ges., 63, 1058 (1930).

26. Kolditz, L.» and Feltz, A., Z. anorg. u. allgem. Chem., 293, 155 (1957).

27. Schmutzler, R., J. Am. Chem. Soc., 86, 4500 (1964).

28. Kolditz, L., Z. anorg. u. allgem. Chem., 284, 144 (1956).

29. Ibid., 286 , 307 (1956).

30. Kennedy, T., and Payne, 0. S., J. Chem. Soc. (i960) 4126.

31* Kolditz, L«, and Feltz, A., Z. anorg. u. allgem. Chem., 293, 286 (1957).

32. Walther, J. F., "Some Metathetic Reactions Between Phosphorus Pentachloride and Various Silver Salts." Unpublished Master's thesis, Department of Chemistry, The Ohio State University, 1963. 105

33* Boothi H* S*» and Bozci'th* A* R* J* An* Chon* Soc*) 6^) 2927 (1939).

3h. Kolditz* L.) Z. anorg* u. allgem. Chem*) 2g2* lh7 (1957)*

35* Poulenc.) C*» C* R. hebd. Seances Acad* Sci*) 113* 75 (1891).

36. Kennedy) T., and Payne* D. S.) J. Chem. Soc* (1959) 1228*

37. Carter) R. P.t Jr.) and Holmes * R. R*( Inorg. Chem.) h» 738 (1965).

38* MuettertieS) E. L. * Mahler) W,* Packer) K* T., and Schmuztlert R*) Inorg. Chem.) 2> 1296 (196h).

39* Berry*) R. S.* J. Chem* Phys., J2, 933 (I960)* ho. Smith) A*) and Calvert) R. P.) J* Am* Chem. Soc.) 36* 1363 (191h). hi* Linke* R.* and Rohrmann* W.) Z* physik* Chem*) B35» 256 (1937). h2* Neale) E«* Williams) L. T. D., and Moores* V* T.* J* Chem* Soc* (1956) h22.

Portner* C., Helv* Chim. Acta* J2, lh38 (I9h9)* h3* Heraberg> G. > '•Atomic Spectra and Atomic Structure Dover Publications) Dev York) 19hh* p. 219* hh. Kolditz* L*». Z. anorg. u* allgem* Chem** 289* 128 (1957)* h5* Brauer* 0.* "Handbook der Prapfirativen Anorganishen Chemie." Vol. I* Ferdinand. Enke Verlag* Stuttgart* i960* p* 186* h6 . Ott, H., Zeitschrift fur Krist.* 63* 222 (1926). h7 * Schall* C*» and Markgraf* H.* Iran* Am* Slectrochem* Soc** !£> 161 (192h). h8 * Hillebrand* W* F.» and lundell* Q* E«» "Applied Inorganic Analysis*" 2d ed.» John Wiley and Sons* Inc.» New York* 1953* PP* 701-708. h9* Any quantitative inorganic analysis text.

50* Scott* W* W.) "Standard Methods of Chemical Analysis*" 5th ed.» Vol. I* D. Van Nostrand Co.* New York* 1939* PP* h05-h08.

51. Carlson* 0* L«* Speotrochimica Acta* j£* 1291 (1963). 106

52. Peacock* R. D.» and Sharp* D. W. A., J. Cham. Soc. (1959) 2762.

53* Nakamoto* K. > "Infrared Spectra of Inorganic and Coozdinatlon Compounds*" John Wiley and Sons* Inc.* New fork* 1963* p. 112.

54. Bellamy* L. J.» "The Infrared Spectra of Cooqplex Molecules*" 2d ed.» John Wiley and Sons* Inc.* New fork* 195&» p. 324*

55* Ruff* J. K.» Inorg. Chem.* 2* 813 (1963).

56. Wei8Sberger» A.* Proskar* E. S.* Riddick* J. A.*, and Toops* S. E.» "Techniques of Organic Chemistry*" Vol. Ill "Organic Solvents." Interscience Publishers* Inc.* New York* 1955* p. 223.

57. Janz* G. J«* Kelley* P. J.» and Vankakasetty* H. V.* A Survey of Non-aqueous Conductance Data (Rensselaer Polytechnic Inst.* Troy*New York* 1962).

58. Griffiths* J. E«* Carter* R. P.* Jr.* and Holmes* R. R.» J. Chem. Phys.* 41* 863 (1964).

59* Pople, J. A.» Schneider* W. G.* and Bemsteln» H. J.* "Hlgh- Resolution Nuclear Magnetic Resonance." McGraw-Hill Book Company* Inc.* New York* 1959* pp. 28-9* 212-16.