'•'Addition Products from the Reaction Of

Home , Aluminium chloride, Chlorodiphenylphosphine, Dichlorophenylphosphine, Phosphonite, Phosphorus halide, Phosphorus pentachloride

'•'ADDITION PRODUCTS FROM THE REACTION

OF PHOSPHORUS (III) HALIDES WITH CARBONYL

COMPOUNDS"

By K.L. Freeman, B.Sc.

This Thesis is submitted in fulfilment

of the requirements for the degree of

Master of Science.

.Supervisor : Dr. M.G. Gallaoher,

Department of Orqanic Chemistry,

School of Chemistry,

University of N.S.W. PREFACE

The work described in this thesis was carried out in the Oroanic Chemistry Laboratories at the University of New South Wales from May 1964 to December 1967, under the supervision of Dr. M.J.

Gallagher. It is original except in those parts so indicated. Part of the dissertation (Section i) has been oublished in a condensed form in Tetrahedron Letters and in the Australian Journal of Chemistry under the title "The Aluminium Chloride Catalysed Reactions of Phosphorus

(ill) Halides and Carbonyl Compounds". Sections 2 and 3 have been published in the Australian Journal of Chemistry under the titles

"The Alkaline Fusion of Phosphinic Acids" and "The Reactions of

Tervalent Phosphorus Compounds with Dichlorodiphenylmethane".

This thesis has not been submitted for a higher degree at any other University.

l-'l'ik’ TABLE OF CONTENTS

PAGE‘NO.

SUMMARY - 1

Section 1 Aluminium Chloride Catalysed Reaction of

Phosphorus (ill) Halides and Carbonyl

Compounds.

Introduction 2

Results and Discussion 4

Experimental Section. 25

Section 2 Alkaline Degradation of Phosphinic Acids.

Introduction 43

Results and Discussion 45

Experimental Section 48'

Section 3 Reaction of Phosphorus (ill) Compounds

with Dichlorodiphenylmethane.

Introduction 52

Results and Discussion 56

Exoerimental Section 65

Section 4 Attempted Preparation of 10-Phenylphos-

phacridone.

Introduction 72

Results and Dicussion 76

Experimental Section «2 TABLE OF CONTENTS

PAGE NO

References: 90

Acknowledgements: 96 SUMMARY

The reactions of phosphorus (ill) halides and carbonyl compounds catalysed by partially hydrated aluminium chloride has been studied. It was found that both dichlorophenylphosphine and chlorodiphenylphosphine react with benzophenone to give chlorophenyl- methylphenylphosphinic chloride and benzhydryldiphenylphosphine oxide, respectively. Dichlorophenylphosohine, catalysed by aluminium chloride acted as a deoxygenating agent with phthalic anhydride to give another route to the preparation of biphthalyl. The reaction of phosphorus

(ill) halides and alkyl ketones with aluminium chloride catalyst was complicated by the reaction of aluminium chloride and the ketone. A novel phenyl migration was obtained where benzophenone and dichlorophenyl phosphine reacted under certain conditions to give triphenylmethylphenylphosphinic chloride.

The alkaline fusion of some phosnhinic acids and \ acid chlorides is'described. The degradation of complex phosphinic acids to yield easily isolated products proved useful in structure determination of the original acids.

The site of attack of-phosphines on haloq^n com pounds has been investigated by-reacting a series of phosphines, qraded in order of increasing nucleophilicity with dichlorodiphenyl methane. 2.

SECTION 1.

The Aluminium Chloride Catalysed Reaction of

Phosphorus ^IIl) Halides and Carbonyl Compounds,

INTRODUCTION :

The investigation of the reactions of phosphorus

(TIT) halides and carbonyl compounds in the presence of aluminium chlor

ide. in particular the reaction of benzonhenone and dichlorophenyl-

phosohino, arose when this reaction was studied as a route to the

synthesis of substituted 10-phenylphosohacridone heterocylics (see

Section 4). 22 It has been reported that phosphorus trichloride

and bepzophenone do not react in the presence of aluminium chloride. 2 although Conant etaj., have studied the uncatalysed reaction of saturated

aldehydes and ketones and found, in general, that thet£-hydroxy ohos- phonic acids were formed. With benzonhenone, under forcing conditions 23 and in the presence of benzoic acid they obtained moderate yields of diphenylhydroxymothylphosphonic acid and low yields of another acid to which they were unable to assign a structure.

Ph2C0 + PC13 Ph2C(OH)P(O)(0H}2

In Conant*s study there was no attempt to isolate the acid chloride and the reaction mixture was hydrolysed to the free acid. Kabochrif and 24 Sheleva , in a similar reaction were able to isolate the acid chloride when they reacted benzaldehyde and excess phosphorus trichloride. 3.

3cid ch1''nde obtefn^d as a ]ow mp-1 fdno sol'd was *~honv]chloronethylphos-»

phonic dichloride in 62% yields.

Ph C(0)H r5Cl_/ex'i ~ > PhCH(Cl)p(0)Clo

An interest?no reaction of benzophenone has b^en described by darie ' ''

who in 1903 reacted benzophenone with excess hynoohosphorous acid for

several days at 100°. The acid v/hich was formed was assion^d the

structure o Ph2C(0H)P(0H)2.

With alkyl ketones the reaction with phosphorus (JTT) ha]-ides in the presence of aluminium chloride is complicated by the

ability of aluminium chloride to initiate aldol-type condensations

The resulting -=>c;p-unsaturated ketone is able then to react with the phosphorus (III) halide. A typical example is the reaction of acetone 26, 27. ’ with phosphorus trichloride in the presence of aluminium chloride to produce 2-raethylpentan-2-one-4-nhosphonic acid, and the reaction of 9P, dichl-orophenylphosphine and chalcone ‘ .

The species formed when benzophenone and aluminium ch.lor- og ide react have been investigated and Menshutkin ‘ has been able to isolate a 1:1 mole adduct of m.n.l30(. For this reason two moles aluminium chloride were used in the initial attempt, to r-re --re lO-r-bo-v? • phoschacridone from benzophenone and dichlorophenylphospMne. 4.

RESUITSJWD DTSCUSSIOU :

Benzophonone was reacted with excess dichloroohenylnhos- phine catalysed by "moist" aluminium chloride (lumps which had been

exposed in a loosely stopoered bottle tor 24 hours when a nett increase

in weight of ca 4% had occurred). The reaction was carried out under

nitrooen for 3 hours at 100°. The neutral fraction was obtained by

pourinq the crude reaction mixture into excess, cold sodium hydroxide

solution and extracting with benzene. A number of experiments were

conducted where the ratios of benzophenone and "moist" aluminium chloride were varied. A maximum yield (66%) of chlorodiphenylmethylphenylphos-

phinic chloride (l) was obtained when 1 mole of benzophenone, 2 moles of

"moist" aluminium chloride and excess (4 mole) of dichloronhenylphosrhine were used. (See Table l).

Under identical conditions nhosohorus trichloride and

benzophenone could not be induced to react and benzophenone was recovered

in high yields from the reaction mixture. Recovery of benzophenone in

high yields also occurred when 1 mole of benzochonone an^ excess dichloro-

ohenylphosphine were used or when 4 moles of "moist" aluminium chloride.

1 mole of benzophenone and excess dichloronbenylnhosphine wr' ••-eacted.

A decrease in the yield of acid chloride (l) was noted ? moles ~

anhydrous aluminium chloride, 1 mole of benzoohenone and excess .os' - '-r

were used.

Assioning a structure to the neutral reaction product

from the reaction of dichlorophenylnhosphine and benzophonone ca ta h : 5.

\.ii.n two moles of "moist" aluminium chloride proved difficult. "he

white crystalline produce, n.o.103-104°, obtained by pouring the

reaction mixture into excess aqueous alkali and extractino with benzene

oave an analysis that corresponded to a 1:1 adduct of benzophenone

and the rhosphine. The product was able to be successfully recryst

allized from ethanol and only reacted sluggishly with alcoholic silver

nitrate. This evidence tended to rule out the possibility that the

product was an acid chloride, although it was difficult to otherwise

-lace two chlorine atoms in the molecule. The infrared spectrum

showed a strong (d-0j stretch band and its proton magnetic resonance

spec;.rum demonstrated that only aromatic protons were present. Fusion

of the product (l) with solid sodium hydroxide (s*e Section 2) produced

benzophenone as the only neutral product (57*0 and the acid fraction

yielded benzhydrylphenylphosphinic acid (VI, 39.90. phenylphosnbmic

acid (18x0 and traces of phenylphosnhonic acid.

On this evidence the most likely assumption was that toe product, was chlorodiphenylmethylphenylnhosphinic chloride (.l).

This was confirmed when the acid chloride (l) was reacted with sodium, methoxide to provide the methyl est^r (’ll) with the loss of one halogen atom, and prolonged aqueous alkaline hydrolysis of the .-id chloride (l) afforded the monobasic hydroxy-acid (ill). 6.

Ph Ph ! Phpou2 4-' *ph “ 2Wco A 1C] p^pc(C1)P(°)Cl Ph,,C(Cl)P(0)0:' 3 - (X) ’^Gh " (ii) Ph / \ Ou" , ^ (1) Ph? C(0HJP(0)0H 2 (III)

ihs Si ructure (l) was confirmed by oreoaring the scid

crloride from the hydrolysis product of the Arbusov 40 reaction between

dichlorodiohenylmethane and dimethyl phenylohosohonite. Pot.h

the expected product

~Msrhinate (lu) are r-esur^My formed in Arbusov reaction -

.Ph 1 Ph2CC0 + ^h?(C'\e)2 .____ ^ Ph C H P(0)C-,e - (id

(XV)

^he reaction mixture could onlv bo successfully separated after acid

hvdrolysis which converted (TV) to benzhydrylnhenylohosoMnic acid

'"Vi) and hydrolysed both the ou- chloro and methyl qrouns of (n) to

give hydroxydinhenylnethylnhenylphosphinic acid (Til). t-'-p-r■: c zo

chlorinate (Vt) with M-chlorosuccinimide in the benzylic oc- if ,-n , unsuccessful. 7.

Ph Ph I •; t 1 Ph2C(Cl) P(0)0'".e Php C(0H)P(0)0H —ly> •(1) 2 (II) (HI)

Ph Ph I 1 Ph.C(H) P(o')C.,'e r~v-> Ph.. C(H)P(0)0H ^ h2° z (iv) (VI)

The acid chloride (l) was obtained from the hydroxy-acid

(111)(1 mole) by heating with phosphorus pentachloride (2.4 moles) until evolution of phosphorus oxy-chloride ceased.

Ph I PhoC(0H)p(0)0H + 2PC1 (1) + POCI3 + 2HC1

The unreactivity of the acid chloride (l) tov/ards nucleonhilic reagents must be caused by crowding around the phosphorus and carbon atoms, thus hindering the approach of the attacking species in an S 2 - t.voe reaction, as experienced with neoper.tyl bromide. Horner ^ has described a similar unreactivity to nucleophilic reagents of«c- chlorobenzyldiohenvlnhosrh-’ne oxide which was uneffect*d by heating for 16 hours in rotassi urn acetate and acetic acid, although the bronobenzyldiphenylphosohr.n'' ox id" cou: reduced to benzyldiphenylphosnhine oxide with triphenyl-r'rc .- V: r and

formic acid.

The formation of the<- hydroxy acid instead o* the cxrecteri oc - chloroacid upon alkaline hydrolysis of the

can be more easily explained by an S.,1 mechanism e.o. 8.

Cl 0 OH 0 I H + _-0H If Ph C — PPh Ph22 — PPh + Cl Ph2i - PPh I i i Cl Cl Cl

(I) OH 0 OH 0 l »i IP „ oh" : - 2 , -<--- 2 , OH

From Table 1 it can be se^n that optimum results were obtained in the condensation of benzophenone and dichlorophenylphosohine when two moles of "moist" aluminium chloride, one role of benzonhenone and an excess of the phosphine were used. A mechanism postulating initial attack by the lone-pair of electrons on phosnhorus at the electron deficient carbonyl carbon, made more electrophilic by co-ordination of the carbonyl oxygen with aluminium chloride, could be invisaged, viz.

+ Ph9C - O-.... A1C1 + PhPCl2 ~ Ph2CPCl2Ph L (VIII)

The intermediate (VIIl) could then rearrange to give the<*-chloro acid

chloride (l). ^

(VIII) > Ph-C^P- Ph 2 | , Cl (IX) Ph +-*----- Ph2CClP(0)Cl < ... Ph2CP(0)Cl + C1‘ (!) Ph

This type of mechanism would explain the ne.ed for aluminium chloride, dew

fails to explain the increase in. yield with two moles of aluminium oh1'mum' or the need for "moist" aluminium chloride and why the reaction should I_A _ B_ _L_E_ 1 All reactions, except where indicated, were carried out as for benzophenone in the General Procedure in the experimental section * HALIDE SUBSTRATE catalyst YIELD ~ a .. .ftiole)' (I) (XVII ) XIX PhPCl2 Ph2C0 s

PhPCi9 Ph2CO 1 37 - 2 PhPC12 Ph2C0 - 66 -

PhPClp Ph?CO 4 S

PhPCl2 Ph2CO 2b 7 -

PhPC12 Ph9COC 2 19 41

PhPCl2 ?h9C0 2 s 9e, f PhPCl9 Ph2CO s

PhP(0)Cl Ph9CO 2 s 2 or i P Ph2C0 2 s

Ph9PCl Ph2CC 2 - 35

PnDCl9 Ph,-CClPPh(0)Cl 2 87 12-5

PhPCl2 Ph2CClPPh(0)Cl 2b 28 58

Ph2PCl Ph2CClPPh(0)Cl 2 11# (VL)

PhPCl9 Ph2CO 2 72 -

S Indicates the substrate recovered in high yields. a Reaction time 1 hour. b Anhydrous AICi 0 was used. O c Reaction time 6 hours. d One mole only used. e Boron trifluoride diethyl etherate used, in the place of AlC^ f Addition of water did not affect the result. q Same result in chlorobenzene as solvent. h "Moist" A1C1 (exposed 72 hours) was used. 10.

when equimolar amounts of benzophenone and dichlorophenylphosphine are used. 2 Conant et al in their study of the reaction of phosphorus trichloride with saturated aldehydes and ketones proposed a mechanism similar to the one above; the interned!a+e corresponding to (VITl) de compose no with water +o give theof-hydroxy-phosphor"*u" ralt (x) v-rhich could react with excess phosphine ■'r>d water, viz.

I + HpO PhPOlp + H0° Ph CP 01oPh -C> Ph^CfOH^PCl,-- ■> PhC(Cl)PCl0----> (l • / / ( A l ^ Ph Ph (X)

This mechanism goes further to explain the optimum conditions of 1 mole of benzophenone, excess dichlorophenylphosphine and "moist" aluminium chloride: i.e. it shows why excess phosphine is necessary, why excess water (over the small traces of water needed in Friedel-Crafts reactions" ) and aluminium chloride are required. However, it does not satisfy t’ne n^ed for two moles of "moist" aluminium chloride or why four moles of "moist" aluminium chloride should stop the reaction. Although the exact nature of the phosphine-aluminium chloride adduct is not known, some mechanism would have to be proposed where the nucleophilicity of the tervalent phos phorus is increased. A number of ion-nairs have been proposed for the intermediates of Lewis acids and phosphorus halides, "or example, hi has proposed the following mechanism for the formation of ali'drhr. or : ;- dichlorides from phosphorus trichloride, alkyl chloride and aluminium chloride. _ H2° PC10 + RC1 + A1C13 — > [RPClJ + [aici^J " RP0C12 11.

7 Cerrard has obtained infrared evidence for the phosphorus oxy-

chloride - boron trichloride species being £?0C1^|+ 5C14J although

here the high electrophilicity of trivalent boron must overcome the shift

in electron density caused by the electron withdrawing effect of the

phosphoryl group v/hich would tend to hold the chlorine more firmly than

in PCI .

It is well known that aluminium chloride produces

reactions of tervalent phosphorus that do not occur without the salt or in produce unexpected reactions. Pass et al , for example, found that alkyl

halides with phosphine-aluminium chloride adduct gave primary phosphines. The

following equilibria could explain the increased nucleophilicity of the phosphine in the catalysed reaction -

PhPCl2 + PhPCl2 - A1C1„ [PhPClJj jPhPCl * AlCl J| ~ (XI) The anion (Xl) could then attack the carbonvl carbon of the benzophenone

(helped by co-ordination of a further mole of aluminium chloride with

the carbonyl oxygen)

PhoC-0 A1C1 + jPhPCl . AlCl Jl-- :>. Ph2CPPh (XI) ~ Cl 11 The disproportionation of dichloroohenylphosphine by aluminium chlorid; . A1C13 2PhPCl, Ph2PCl + P01o

could be explained by this type of mechanism with a different recombination of the ion pair (Xl),

fphPCl~|~ [PhPCl •AlCil -- v>- Ph2PCl-AlCl3 + PC1Q 12.

A larqe number of reactions involving phosohorus (ill) comoounds have had ion-oairs prooosed as intermediates when they react with "positive" 5 halogen , that is when halogens are transferred with only a sextet of

electrons. A similar reaction, although uncatalysed, has been reported by Frazier et al who reduced phosphorus (III) halides with

tributyl phosphine, viz.

3 (C4H9)3P + 2PC13 --> 3(C4H9)3PC1? + 7?

4 (C4K9)3P + 4PhPC12 - -> 4(C4H9-3PC1-'. + (PhPb

Prolonged heating of benzophenone and an excess of dichlorophenyl^hosphine with "moist" aluminium chloride decreased the yield of the acid chloride (l)

and an acid product was isolated by acidification of the aqueous layer after

alkaline work-up. The acid was partially purified by conversion to the

sodium salt and re-acidifying with hydrochloric acid. The acid obtained

could be re-crystallized from methanol, but the product (n.p.ca 273°) could not again be dissolved in this solvent and was only soluble in dinethyl-

formamide or dimethylsulnhoxide from which solvents it could not be re

crystallized. The equivalent weight of the acid proved that it was dibnsi

and it was converted to its dimethyl ester (XII) whose n.m.r. showed six methyl to nineteen aromatic protons and one benzyl5.c proton, the letter

split by phosohorus (13 c/s). Analysis confirmed that the methyl es:-*r had one of the structures (XII). 13.

"moist1* PhPCl2 Ph2C0 A1C13i_ H0(0)1-/ / 6 hr. (XIII)

Ph then KaOMe Me0(0)P y/Ph \ "CHP(0)0Me (XII) Ph Alkaline degradation (see Section 2) of the dibasic acid gave the acid

(XIV) and diphenylmethane. This acid.was oxidized with permanganate in

acetone to yield the keto-acid (XV). CH2Hl NaOH fh (XIII) ^ H0(0)P ■ ^ i +- Ph2CH2 300° V //

In order to determine the position of substitution, the keto-acid,

(p-benzoylphenyl)phenylrhosphinic acid (XVl) was prepared by the Doak-

Freedman^ reaction between p-benzoylphenyldiazonium fluoroborate and

dichlorodiphenylphosphine,

° r=\ l.HM02 9 /=\ l.PhPClg 0 /=\ fh PhC /)NH9 ---- - PhC-K oSFh------*• PhC 4 ' P (0) OH \ // * 2.NaBF4 \ // 2. H20

(XVI)

For comparison, the ortho acid was a] so prepared by a similar route fror->

o-aminobenzophenone. The infrared spectrum of the keto-acid (AVI;

almost identical to (p-benzoylphenyl)phenylphosphinic acid and r-sMar to the ortho acid* and the mixed m.p. nave no depression with this acid

establishing the structure as ty-hydroxy(phenyl)phosphtnyldirhenylmethy]-

phenylphosphinic acid (XVII). ph /=\ Ph H0(0)p-/ VcHP(0)0H X-- / Ph (XVII) Formation of this dibasic sc-hi could bo explained by the ion-rMr mechanism

used to exolain the formation of the acid chloride (l). Postulating the

initial formation of the acid chloride (l), this could be attacked by

a further mole of dichlororophenylphosphine to form the carbanion (XVTTI) PhpCCIF(0)C1 + PhPCl ------I jPhPCCj jphCPCCIPh Ph 0 —»» A1 Cl■ g (XVIII) By electron distribution around the ohenyl ring, the carbon^’on (XVIII) could

produce mesomeric forms with sites strongly activated for electrophilic

attack, i.e.

/=\ Fh CPhPCl-j H 0-* A1C1 (XViiD^t. - < /-CPCiPh PhP ,rCPClPh 3 \=/ 0*A1C13 Cl

(XVII) — jf2o

hater on the carbanion would produce the dibasic acid (XVII ). needs conditions for Fri odel -Croft s react.’ions, tha! is nnbydro*r. ini urn chloride, it can only occur after the hydrated ale. - hie 15.

has been used. This assumption is strengthened because from Table 1

the yield of the dibasic acid was only 12,5% when the acid chloride (l)

was reacted with excess dichlorophenylphosphine in the presence of

"moist" aluminium chloride, while the yield, under similar conditions was

58£o when anhydrous catalyst was used.

Under conditions-giving optimum yields of the acid

chloride (l) when chlorodiphenylphosphine was substituted for dichloro

phenylphosphine the product obtained was a neutral halogen-free material

(m.p.304-5°). This material showed a (P=0) stretch band in the infrared

and the p.m.r. spectrum showed one benzylic proton to 20 aromatic ^rotcns,

the former being solit (18 c/s) by phosohorus. From the analysis and

the general unreactivity of the product the structure was assumed to be

benzhydryldiphenylphosphine oxide (XIX). The oxide has been preoared by

l . o \ Homer (m.n.303 ) by hydrolysis of the reaction produce of bonzhydryl

chloride and diphenylchlorophosphine,

H20 Ph2CHCl + Ph2°01 ---- Ph2CHP(0)Ph9

(XIX)

A reference material was prepared unambiguously from the reaction of t.’-e

lithium derivative of chlorodiphenylphosphine with benzhydrych]oride to yield benzhydryldiohenylphosphine (XX) which was not isol-1< d but oxidi’et directly to the phosphine oxide. (XIX). 16.

Moist Ph2?Cl + Ph2CO —Ph2P(0)HCPh2

(XIX)

_ + PhoCHCl Ph2PCl + Li Ph2P Li------Ph2PCHPh2

(XX)

When undertaking the reaction of chlorodiphenylphosphine and benzophenone it was thought that the ^-chlorophosphine oxide (XXl) would be the product, by analogy with the reaction to form the acid chloride (l). Assuming that the expectedeC~chlorophosphine oxide is formed first then it could dechlorinate to give the ion-pair (XXII) which would decompose on work un to give (XX), e.g.

H 0 PhoCClP(0)Ph2 + Ph2PCl jPh2PCO+j>h2C P (0) Ph^J _‘2_ >. Ph?CHP(0)Ph2 + OH-

(XXI) (XXII) (XX)

Attemoted preparation of chlorodiphenvlmethyldiphenylphosphin< oxide (XXl) to confirm this mechanism proved fruitless, although the acid chloride (i) could be de-chlorinated by chlorodiphenylphosphine (only in the presence of aluminium chloride) to give the reduced acid, benzhydrylphenylphosphinic acid (Vi). This reduced acid was also prepared by reduction of the acid chloride (i) with zinc amelgan in tetrahydrofuran allowed by hydrolysis. It is presumed that this reaction proceeds through the phosphorane (XXIIl), whereas the reduction of the acid chloride with chlorodiphenylphosphine is facilitated by the co-ordination of aluminium chloride to the phosphoryl oxygen. 17.

Ph l Zn/Hg H20 Ph CC1P(0)C1---- Ph C = P(0)Ph _> Ph CHP(0)0H Z \>h (I) (XXIII) (VI)

Because of the ease of reaction of benzophenone and dic.’nloro-

phenylphosphine and chlorodiphenylphosphine it v.’as decided to extend the

reaction to a wider range of carbonyl compounds. Cyclohexanone reacted

readily to form a neutral product which was difficult to crystallize.

V/hen the oil' was hydrolysed«C-chlorocylohexylphenylphosphinic acid (XXIV)

was readily obtained (42% yield), Ph Cl P(0)0H A1C1. + PhPCl, then

hVh2o (XXIV) Xhe oil that proved difficult to crystallize was Presumably the acid

chloride, -chlorocylohexylphenylphosphinic chloride.

With o-chlorobenzaldehyde a good yield (69%) of (o-chlor- phenyIhydroxynethyl)phenylphosnhinic acid (XXV) from the aqueous alkaline

fraction.

"moist" Ph ------>- CIIOIIP(O) Ph A1C13 (XXV)

The fact that the hydroxy-acid (XXV) was obtained directly from ho aqueous

alkaline extract further'supports the contention that theoC-oh.loro ar.-'d

chloride (I) is hindered. The hydroxy*-.'k. 1 d (XXV) h.»d on l y n v< my w<■■■■■ 13.

(OH) band in the infrarod and the structure was assioned on analysis and n.m.r. data. This weak absorption of the OH band could be due to inter- molecular hydrogen bonding, viz.

as the frequency of the band (3,350cm ') is in the region where hydrogen 31 bonded OH is expected to absorb, although hydroxydiphenylmethylphosphinic acid (ill) with a similar structure shows an extremely sharp band of strong intensity at 3250cm ^.

It Is likely that this reaction is general: it was not pursued further because it holds no advantages over the uncatalysed addition 32 23 of phosphorus (ill) halides to aryl aldehydes 5 . The reaction of phthalic anhydride with dichlorophenylphosphine is of interest since biphthalyl (XXVI) is produced in moderate yields when one mole of phthalic anhydride, two moles of anhydrous aluminium chloride and excess (5 moles) of dichlorophenylohosphine are heated to 100° for three hours (see 30 Table 2). Ramirez etal have prepared biphthalyl in good yields by the

v O reaction of phthalic anhydride and triethylphosnhite at 160 , t’ e ehosr>hi ‘'■ acting as a de-oxygenating agent, viz.

2(C.H,X)) P ^ 2 ■I- 2(C2HjO)3PO

(XXVI) 19.

TABLE 2

Phthalic Anhydride (l mole) reacted at 100° for three

hours with excess Dichlorophenylphosphine (5 ml,ca 4 mole)

and Catalyst (2 mole).

CATALYST YIELD OF TRANS-BIPHTHALYL %

"Moist" A1C13 10

Anhydrous AlCl^ 34

Boron Trifluoride 0.2

\ ' , None Nil 20.

34 Although it has been claimed that Ramirez method produces only trans 33 biphthalyl, other workers have obtained a mixture of ci,s and trans biphthalyl by this method. The biphthalyl preoared by the reaction of dichlorophenylphosphine, phthalic anhydride, and aluminium chloride

is isomerically pure, yielding only trans biphthalyl by comparison with 34 published infrared data . It is likely that the mechanism for the

formation of biphthalyl from dichlorophenylphosphine is similar to the 30 mechanism Ramirez postulated for the triethylphosphite reaction. Attack of the nucleophilic phosphorus at the carbonyl carbon would produce the

carbene (XXVII).

II V +PhP(0)Cl„

H 0 (XXVII)

Dimerization of the carbene with a further mole of carbene would produce biphthalyl.

The infrared spectra of the reaction products from a dialkyl (e.g. n.onan-5-one) and an arylalkyl ketone (e.q. acetophenone)

showed that reaction had taken place but the thin layer chromatogram an:

n.m.r. spectra indicated that there were a large number of products 'as many as six components for the acetophenone reaction). These products were unable to be separated. The large number of products is probably

caused by aldol formation and subsequent reaction of the

ketones (see Introduction).

An anomalous reaction of benzoohenone and dichlorophenylphosphine occurred when one mole of benzophenone was reacted with excess dichlorophenylphosphine, catalysed with three moles of "moist" aluminium

chloride. The neutral product gave a b% yield of triphenylmethylphenylphosphinic chloride (XXVIII) m.p. 208-210°. Acification of the alkaline

extract from the reaction afforded a further 5°o of the hydrolysis product,

triphenylmethylphenylphosphinic acid (XXIX).

3 moles H/F^O PhoC0 + PhP01o----S Ph CP(0)C1 -- Ph CP(C))0H 2 2 Moist 3 I 3 | A1C13 Ph Ph

(XXVIII) (XXIX) The acid chloride (XXVIIl) was identified by comparison with a sample prepared bv reacting triphenyl carbinol with excess dichlorophenylphos phine (49% yield),

Ph3C0H + PhPCl2--- ^ Ph3CP(0)Cl Ph (XXVIIl) Analysis of the mother liquors from this reaction by gas chromotacranhy showed the presence of benzoohenone (2.C£o) trinhenyl methane (l0£o) and unreacted triohenyl carbinol (l3£o)* Phenyl migration in this type of system has not bpen previously described, although thero are r few refer ences to phenyl migration in the alkaline degradation of n -al.kvlt.ri"hen1 1* 69 phosphonium salts. For example, Hellmann and Bader"' have obt~\nod benzyldiohenylphosphine oxide in 10% yield from the alkaline decomposition of triphenylchloromethylphosphonium chloride viz. 22.

OH -Cl [Ph P CH2Cl] Cl' —>- Ph3P0HCHoCl Ph POHCH2

\r -H+ 0 i. Ph2PCH2Ph

They also have obtained benzylchoromethylphenylphosphine oxide from the

reaction of bis{chloromethyl)diphenylphosphonium chloride,

OH [ph2P (ch2ci)2]ci' PhP(0KCH2Ph)CH2Cl by a similar route. A phenyl migration involving a phosphorus (V) 71 compound has been mentioned. Here triphenyloxaphosph(v)olan on

heating to 300° gave phenoxypropyldiphenylphosphine. ,0 300 Ph2P(CH2)30Ph ph3r-?H2 0 /Ch2

ch2

71 The authors here presumed that this reaction was due to an inter-

molecular mechanism similar to that given for the pyrolsis of phosphonium 70 alkoxides by Eyles and Triopett who found that in oeneral n-alkyl-

triphenyl phosohonium salts react with sodium alkoxide at 230° to give

n-alkyldiphenylphosphine and alkylohenyl ether e-g.

+ > 1 280° 1 .(Ph^PPQ X + Ha OR __ >. Ph2R P^, + NaX —PhqPR 4- PhOR ^OR

An interesting example of phenyl migration from an oxygen adjacent to a

phosphorus atom is given by Dannley and Kabre who refluxed bisfdiphenyl-

phosphinic) peroxide in chloroform, carbon tetrachloride or dimethyl formn

ide to obtain the unsymmetrical ester anhydride viz. 23.

00 0 OPh M '« a »1 I Ph^P-O-O-PPh--•>. Ph POPPh 2 2 2 ^ 0

The ester-anhydride was not isolated but was inferred by analysis of the

acids resulting from hydrolysis.

The formation of triphenylmethylphenylphosphinic chloride

(XXVIII) from the reaction of benzophenone (l mole) and dichloroohenylphosphine (excess) catalysed by 3 moles of aluminium chloride could be

explained if it is assumed that the disproportionation of the nhosphine with aluminium chloride observed by Brown and Silver^* occurs to give

chlorodiphenylphosphine, viz.

A1C13 2PhPCl/ __>- Ph2PCl + PC13

Chlorodiphenylrhosphine could then attack benzophenone, oTV 0 + Ph2PCl Ph2C' P+ClPh,

/\ ph c---- PClPiv 2 2

PhoCP(0)Cl' 3 I Ph (XXVIII) However, on testing this mechanism by reacting chlorodiphenylphosnhi.no

and benzoohenone under varying conditions (e.g. boron trifluorido

diethyletherate catalyst; 3 moles of anhydrous aluminium chloride,

one mole of benzophenone and excess phosphine), no trace of the acid

chloride (XXVIIl) could be obtained.

The formation of benzophenone during the preparation of 24. triphenylmethylphenylphosphinic chloride (XXVIII) is of interest. The original reaction, was heated at 100° for one hour. When the reaction was heated for 15 mins, the yield of benzophenone was ca 9%. The reaction of triohenylcarbinol and excess phosphorus trichloride when refluxed for ■

1 hour had produced traces (ca 0.5%) of benzophenone besides the expected ' product, triohenylmethylphosohonic dichloride (62%). Thus the reaction of triphenyl carbinol and dichlorophenylohosphine would seem to be,

Ph i Ph3C0H + PhPCl2 —>- Ph3CP(0)Cl + Ph2C0 + Ph0PCl although no chlorodiphenylphosphine or diphenylphosohinic acid was detected.in the reaction mixture or in the hydrysolate. 25.

E X P E R I M E N T A L

GENERAL :

Unless otherwise stated all compounds are colourless. All reactions involving tervalent phosphorus compounds were carried out under dry, oxvgen-free, nitrogen.

Melting Points are uncorrected and were determined by the capillary method or, in some cases, in a sealed capillary and in these cases the melting point is affixed with the letters (S.T).

Infra-Red Spectra were recorded on a Perkin-Elmer model

137, Perkin-Elmer model 337, or a Hilger & Watts Infrascan spectrophoto meter.

Nuclear Magnetic Resonance Spectra (n.m.r.) were recorded on a Varian A.60 N.M.R. Soectrophotometer. Peak positions were recorded as oarts per million (ppm) downfield from tetramethylsilane, or, for spectra run in Do0, with the sodium salt of 3-(trimethylsilyl)-l-propane sulphonic acid, as internal standard. Positions of doublets were recorded at their mean position.

Gas-Liquid Chromatography (g.l.c.) was carried out on a Perkin-Elmer model

800 using 6 ft. x in. columns packed with SE 30 (5%) domed with Versamid V 900(0.5%) on Chromosorb G (80-100 mesh), using a flame ionization detector.

Solvents were fractionated before use. All anhydrous solvents, except alcohols, were kept over sodium wire. Ethers, when required absolutely dry and peroxide free, were refluxed over lithium aluminium hydride and then distilled. Absolute alcohols were prepared by the standard method of refluxing in a solution of magnesium alkoxido. 26.

Equivalent Weights of phosphorus acids was determined by dissolving or suspending the acid (c_a 50-100 mg), accurately weighed, in aqueous, neutral methanol (5$ aqueous) and titrating the acid slowly to a permanent end-point using methyl red as internal indicator. 27.

GENERAL PROCEDURE USED FOR REACTION OF BENZOPHENONE WITH DICHLORODIPHENYLPHOSPHINE IN THE PRESENCE OF "MOIST" ALUMINIUM CHLORIDE

NOTEt "Moist" aluminium chloride refers to commercial, anhydrous

aluminium chloride in lump form, which had been exrosed overnight in a vessel, loosely plugged with cotton wool, when a net increase in weight of 3-4% had occurred. The aluminium chloride was then finely ground and protected from further contact with moisture.

Procedure:

Benzophenone, m.p 42-44°C and "moist" aluminium chloride was added to dichlorodiphenylphosphine and. the mixture carefully heated in an oil bath to 100° and maintained at this temperature for three hours.

The hot reaction mixture was poured into excess, cold, 2N sodium hydroxide solution. After the addition of benzene the mixture v/as stirred until the phases became clear.. The benzene layer was washed free of alkali with water and the acidic material liberated by the addition of concentrated

MCI, the acid being filtered off, dissolved in excess saturated sodium bicarbonate solution. The solution was filtered through diatamaceous earth and the purified acid liberated by addition of concentrated HC1. The V neutral fraction was obtained by drying the benzene extract with anhydrou sodium sulphate, filtering and removina the solvent under reduced pres n- .

Chlorodinhenylmet.hylphenylnhosmhlnl c ChJ.orlri

Benzoohenone (3.6g., 1 mole) reacted with "moist" aluminium chloride (2.6g., 2 Mole) and excess dichlorophenylphosphine (5 ml., ca 5 rw reacted under similar conditions to the General Procedure, gave, in the 28.

neutral fraction, chlorodiphenylmethylphenylphosphinic chloride (l),

4.7g., {66%), m.o. 103-104° from ethanol, V*(P=0) 1223 cm*

Found - (M-9281, M-9940, M-0887 : C,63.5: H.4.4; 01,20.0; P,8.7?o

Calc, for - CinH, Clo0P : 0,63.2: H,4.2; 01,19.6: P,8.6}£, IV 15 2 n.m.r. spectrum in. CDCl^ showed only aromatic protons, <£7.2 - 7.9.

Benzophenone was the only other detectable constituent of the neutral fraction. Acidification of the acid fraction yielded, besides the expected acid, phenylphosphonic acid from the hydrolysis of excess of the phosphine, the diacid (XVII) in small yields, (70mg.) (See below).

Proof of the Structure.of Chlorodiphenvlmethylphenyl- phosphinic Acid Chloride (l).

A. Preparation of Methyl Ohlorodiohenylmethylnhosphinajhe.

After refluxing the acid chloride (l) in anhydrous methanol for one hour with an excess of standard sodium methoxide in methanol 0.5 mole of chloride ions were liberated. (Found as AgCl : 9.9: calc., 9.3/a) and 0.54 mole of methoxide were consumed (by titration). The reaction mixture was ooured into water, the solid ester was obtained by filtration and recrystallized from 60-80 petroleum, o.5g (79/£) m.p. 100-1° .

Found - (10042) : 0, 67.3; H,5.3% V

Calc, for - C2oH1SC1°2P : C’ 67*3; (P=0) 1230r.m“ ‘

£(CDC1^) 7.15-7.73(20H,ArH); 3.72 (j(P0CH) 11 cps: 3H,OCF3'.

B. Preparation of HydroxyclirhenvImethy lpheny lohos~-hin 1 c r.j d (I X1.)

i. Hydrolysis of the Acid Chloride (l).

A sample of the acid chloride (0.29q, 0.8 mole) was refluxed with a laroe excess of aqueous sodium hydroxide solution for 14 hours. 29

this time the suspended acid chloride had disappeared to leave a

slightly hazy solution. The solution was allowed to cool and the acid

liberated by addition of conc.HCl. The nrecipit.ated, dried acid was

recrystallized from aqueous ethanol to afford the hydroxy acid (ill),

0.26g (98ft) m.p.191-2°.

Found - (10111) : 0,70.45; H, 5.3ft equiv. vrt. 325

Calc, for - : 0,70.4 ; H,5.3ft equiv. wt. 324-3 C19H17°3P -1 v(oh) 3250cm

The S-benzylisothiouronium salt was prepared from the sodium salt of the -

hydroxy-acid and had a m.p. 170-171°.from aqueous ethanol.

Found - (10128) : C, 65.95$ H.5.6 N.5.Sd%

calc, for - C27H2703PSM2 : 0,66.1; H,5.^5r H.5.7ft .

Attempts to prepare the 3,5 - dinitrobenzoate of the hydroxy-acid were

not successful, the only crystalline product that could be isolated was

3.5 - dinitrobenzoic acid. Hydrolysis of the acid chloride (l) (l.3g,

3.6 mole) with ethanolic hydrochloric acid solution (cone. HC1 8ml. water

10ml., and ethanol 30 ml.,) for 18 hours under reflux gave the hydroxy

acid (ill) after reducing the reaction mixture to about 50ft by volume under

reduced pressure, extraction with benzene and sodium bicarbonate solution \ and acidifying the aqueous extract with concentrated hydrochloric acid.

The benzene fraction produced 5lft of the unreacted acid chloride (l).

ii. Synthesis of Hydroxy-dirhenylmethylphenylohosphinic _Acid (I J.

NOTH: Oja the Preparation of Dichlbrodirhenyl Methane

Attempted preparation of dichlorodiohenyl methane by the 10 accepted method " of heating an equimolar mixture of benzophenone and phosphorus pentachloride at 145-150° until the evolution of ^hos; torus

■r-rr*- - * — •!•• 1 ~! r 30. oxychloride ceased and subsequent distillation under vacuum of the reaction product failed to produce dichlorodiphenyl methane free from benzophenone. This method was modified and an 0.2 molar excess of phosphorus pentachloride was used. The reaction mixture was heated as above. When the evolution of phosphorus oxychloride ceased the reaction mixture was checked by infrared'analysis to ensure the absence of the carbonyl band and further 0.2 mole quantities of phosphorus pentachloride were added until this band disappeared. The reaction mixture was heated under vacuum (20 mm Hg) at 140° until traces of any excess phosphorus penta chloride were removed. The residue was distilled under nitrogen and the fraction boiling at 176-178° at 17-18 mm Hg. collected. This method produced dichlorodiphenyl methane of ca 98% purity by g.l.c.

Dichlorodiphenyl methane (l2g, 5 mole) was heated to 100° 25 o and dimethyl phenylphosphonite (5g, 3 mole) (n ^ = 1,5210, E.P.^ 106-110 , lit.nj^ = 1.5186 B.P.^g 94.5°) was added dropwise to the stirred mixture over 0.5 hrs. The solution turned deep red an^ 4‘hc mixture vns kept at 100° for a further 0.5 hr. Ethanolic PCI 'cone. T-Vl, 10ml. ethanol, 60 ml and water, 40 ml was added to the cooled reaction mixture and then the whole was refluxed for 8 hours. The acid fraction afforded. unon ^eoeated rpcrystallization, a crude sample (0.4po) of hydroxydiphenyl - me thylohenylpho sphinic acid (ill) n.o. 184-186° undepressed by admixture with the sample of the acid prepared by hydrolysis of the acid chloride

(l) and with an identical infrared spectrum with this acid. ’Work up of the first bicarbonate wash and the mother liquors gave benzhydrylphenyl-hosphini acid (V1) m.p. 244-246°, o.75g (8/i) see page 49. 31.

Conversion- of Hydroxydiphenylmethylphenylphospiinic yield III) to Chlorodiphenylmethylphenylphosphinic Acid Chloride (l).

Synthetic °C -hydroxy acid (ill) (o.l5g, 0.46 mole) and phosphorus pentachloride (o.22g, 1.1 mole) were heated together at 155° until evolution of phosphorus oxychloride ceased. The cooled residue was extracted with excess 2N MaoH and benzene. After drying over anhydrous sodium sulphate and removal of the benzene under vacuum the acid chloride

(l) was obtained (32mg,20%) m.p. 102-103° which did not depress the melting point of the acid chloride (l) from the reaction of benzophenone, "moist" aluminium chloride, end excess dichlorophenylphosphine. The Rf values of the t.l.c. were identical. The conversion could not be effected even on prolonged refluxing in thionyl chloride.

Reaction of Chlorodiphenylmethylphenylohosnhinic Chloride (i) with Zinc Amalgam.

The acid chloride (l) (0.28g, 0.78 moles) was refluxed with zinc amalgam (0.174g, 2.6 moles), prepared by dipping the zinc in mercuric chloride solution, washing with water, ethanol and drying) in dry tetrahydrofuran (30ml.). The mixture was allowed to reflux for 8 hours. The cooled reaction mixture was poured into excess sodium hydroxide solution, filtered and extracted with diethyl ether. The aqueous extract was acidified with concentrated HC1. The acidic product was recrystallized from aqueous ethanol to yield benzhydryldiohenylphosohinie acid (VI) m.o.244-246°, 0.14g (o9%) which did not depress the m.p. and had identical infrared spectrum to the acid prepared by the alkaline fusion of the acid chloride (l). (See Section 2). 32

Reaction of Chlorodiphenylmethylphenylphosohinic Chloride (i) with Dichlorophenylphosohine and Anhydrous Aluminium Chloride.

Chlorodiphenylmethylphenylphosphenic chloride (l.8g. 0.5 mole)

anhydrous aluminium chloride (l.3g, 1.0 mole) and excess dichlorophenylphosphine (5 ml.) were reacted as under the General Procedure. The

neutral fraction yielded the unchanged acid chloride (l) (0.5g, 28%) and

the acid fraction yielded the dibasic acid (XVII) (l.3g, 58%) m.p. 270-4°

identical infrared spectrum to the acid, liberated from the 6 hour reaction.

Using "moist" aluminium chloride and the same quantities the neutral oroduct yielded the acid chloride (I.13g, 87%) and the acid fraction produced the dibasic acid (XVII) (0.2g, 28%).

Benzhydryldiphenylphosphine Oxide (XX)

Benzophenone (1.8, 1 mole), moist aluminium chloride

(2.6a, 2 mole) and chlorodiphenylphosphine (5 ml.) was heated as under

the General Procedure, except the alkaline hydrolysis mixture was

extracted with benzene, chloroform mixture yielded the oxide (XX),

(0.8g, 35%) m.p. 304-5° from glacial acetic acid (Lit.'*' 304-5°).

Found (M-4627) : C, 81.9; H, 5.9; P,8.3%

C25H21OP requires : C, 81.5; H,5.75; P,8.4% Y ... . £(CF COOH) 5.15 (J(PCH)18c/s; 1H) ; 7.0-7.8 «OH, Aril)

The m.p., mixed m.p. and infrared spectrum were identical with a sample

prepared as follows -

A solution of lithium diphenylnhosnhide was- prepared from diphenylchlorophosphine (2.2g., 1 mole) and lithium (0.14g.; 2 mole) in dry tetrahydro-

furan (35 .ml.). Benzhydryl chloride (2.03g.; 1 mole) in dry T.H.F. (10m-. 33.

was added dropwise until the deep red colour of the phosphide was discharged, after ca 80,6 of the benzhydryl chloride, had been added. The mixture was heated under reflux.(0.5 hr.) The cooled reaction product was poured into a saturated solution of aqueous NaHCO (100 ml.) and 100- volume hydrogen peroxide (4 ml.) The product was allowed to stand over night and the crude oxide filtered off and recrystallized from a large volume of ethyl acetate, m.p.304-305°('1.4g., 39a>).

The Diacid (XVIl)

Increasing the reaction time for the general reaction of benzophenone (l.8g.), dichlorophenylphosohine (5 ml.) and "moist." aluminium chloride (2.6g.) from 3 to 6 hours decreased the yield of acid chloride (l)

" I to 0.67 g. (19%). However, acidification of the aqueous phase gave the diacid (XVIl) which was purified by dissolving in saturated sodium bicarbonate solution, filtering through diatomaceous earth and reprecipi tating the acid with concentrated HC1. Recrystallizat.ion from methanol gave a product of m.p. 278-282° (l.83g., 4C?s) which would not again dissolve in methanol and was only soluble in hot dimethylformamide or hot dimethyl sulnhoxide from which it could not be recovered.

The impure acid gave the following analysis:

\ Found . - (M 6513) : C,64.35; H,5,5: P,12.8%; equiv. wt. 245, 246.

Calc, for - ^25^22^4^9 • ^>66.95; H,4.9; P,13.8Jo equiv. wt. 224.

In order to obtain acceptable analytical figures the diacid was converted to the dimethyl ester (XII).

The partially purified acid (lg..) and PCl^ (0.71g., 2.2 mole) were heated together at 155° until evolution of phosphorus oxychloride had ceased. 34.

The oroduct was dissolved in dry methanol and an excess of sodium methoxide in methanol was added, the solution then beinq refluxed under nitrogen for 1 hour. The cooled reaction mixture was poured into cold water and extracted with diethyl ether. The ether layer, after washing with water and drying over anhydrous sodium sulphate yielded the diester (XII). The diester was repeatedly recrystallized from ethyl' acetate to yield a product of m.p.195-6°, O.ICq. (%).

Found - (M9187) : C,67.8; H,5.S; P.12.8;o'

Calc, for - Co„H-_0.Po : C,68.1; H,515? P.13.C^. ir(P=0) 1230cm"1;

S(CDC10) 7.05-7.78 (ArH, 19H); 4.5(j(PCH) 18 c/s, 1H),

3.65 (Triplet) (j(P0CH») 14 c/s, 6H). The low yield is not surnrisino o \ $ince four diastereoisomers are possible.

Preparation of (p-Benzoylphenyl)phenylnhosphinic Acid (XVl)

1. From Diacid (XVII).

(p-Benzylphenyl)phenylphosphinic acid (O.lg.) ------from the alkaline degradation of the diacid (XVII) (see Section 2) was refluxed in acetone (10 ml.) with potassium permanganate (0.15g.,) for 2.5 hours.

The cooled reaction mixture was filtered to remove the manganese dioxide and the acetone removed under reduced pressure. The product was dissolved \ in water, filtered through diatomaceous earth and the acid liberated bv the addition of dilute HC1. Yield 70 mg of the acid m.p.103- ° from, benzene.

2. From o-Amlnobenzophcnone

(a) P-benzoylphenyldiazonium fluorobo^a^e

p-Aminobenzophenone (l.94g., 0.0 mole) was cooled in 35.

ice salt bath and to it was added concentrated HCl(2ml) and a filtered

« ** solution of sodium tetrafluoroborate (l.5g., 0.02 mole) in water (6 ml.).

A small quantity of diethyl ether (ca 0.5 ml.) was added to prevent frothing

and the mixture diazotized with sodium nitrite (0.7g., 0.02 mole) in water

(10 ml.), the temperature being kept below 10°. After stirring at 5°

for 0.5 hours, the precipitate (2.4g., 89°o) collected, washed and dried

thoroughly at room temperature. The salt showed an extremely strong,

broad band in the infrared at 1010-1080cm due to BF - J * The salt was 4 used without further purification for the next step.

(b) Doak-Freedman^ Reaction:

The diazonium salt (2.4g.,) was suspended in ethyl'acetate

(60 ml.) and to the suspension was added cuprous bromide (O.lg.,) and

dichlorophenylphosphine (l.6g., 1.1 mole). The mixture was warmed on

a water bath until the reaction commenced and after the reaction had

subsided, the reaction mixture v/as refluxed for 1 hour. The cooled

reaction produce was poured into excess, saturated, aqueous sodium bicarb

onate solution. The separated aqueous phase was extracted with fresh

ethyl acetate (60 ml.), then acidified with concentrated 1 & Cl. The

acid was extracted into CHCl^ and recrystallized from benzene to yield

(p-benzoylpheny])phenylphosphinic acid (XVI) m.p.183-5°

Found - C,70.4; H.5.05; P,9.5%; Equiv. wt., 317

Calc, for CHOP : C,70.8; H,4;7; P,9.6&; Equiv. wt., 322+3, V(C=0) 1670cm*1.

Synthesis of ( o-Benzoylphenyl)phenylphosphinit Acid, From o-Amino-Benzophenone.

(a) (o-Benzoylphenyl)diazonium Fluoroborate:

The diazonium fluoroborate was prepared by an identical 36. method to the method above using o-aminobenzophenone (6g., 3 mole) which was stirred with cone HC1 (10 ml.) and a filtered solution of sodium fluoroborate (4.6g., 4 mole) in water (40 ml.) was added. The solution was cooled to 5°, diethyl ether (l ml.) was added and the salt diazotized with sodium nitrite (2.3g., 3 mole) in water (20 ml.) so that the temoerature of the reaction mixture was kept below 10°. The solid was filtered off, washed well with water and diethyl ether, and air dried to give 4.9g.,

(55J&). The salt was used without further purification for the next step.

(b) Doak-Freedman^ Reaction:

The diazonium salt (4.9a.) was suspended in ethyl acetate

(60 ml.) and cuprous bromide (0.2g.) and dichlorophenylphosphine (3.1g.,

2.1 mole) were added. The reaction mixture was warmed carefully and after the initial reaction subsided, the mixture was heated to reflux for 1 hour.

Work-up of the reaction mixture by the method described above afforded crude (o-benzoylphenyl)phenylphosohinic acid, 3.8a.(34%). Recrystallization from ethanol afforded the pure acid m.p. 213-214,

Found - (M-2619) : C,70.8; H.4.9: P,9.5%; Equiv. wt. 324.

Calc, for - C19Hl503P : C,70.8; H.4.7; P,9.6%; Equiv. wt. 322-3. iTCCssO),

1,660 cm \ Preparation of o-Chlorocyclohezylnhenylphosphinic Acid (XXIV)

Cyclohexanone (l.96g., 0.02 mole), dichlorophenylphosphine

(10 ml.) and "moist” aluminium chloride (o.2g., 0.04 mole' were reacted as under General Procedure. The cooled reaction mixture was worked up by the usual alkaline method with benzene as the extracting solvent. After drying of the benzene extract and removal of the benzene, an oil was obtained

(3.04g.) that showed a weak (C=0) band and a strong (P=0) at 1,220 cm in the infrared. The t.l.c. showed one major produce with two minor oroducts. 37.

Because the oil proved difficult to crystallize, the product was hydrolysed with concentrated HC1 (6 ml), water (15 ml) and ethanol (30 ml) by refluxing the mixture for 6 hours. On cooling, the acid was collected 2.1g., (42%) m.p. 140-143°. Repeated recrystallization from ethyl acetate afforded the cC-chloro acid (XXIV) m.p. 149-150°. r Found - (M-4082) : C,55.85; H,6.3; P)12.2; Cl,12.9%, Equiv. wt. 259

Calc, for - C;l2H16C2^1P : C,55.7; Hj6.2; P,12.0; Cl,13.7%, Equiv. wt. 258.7. v~(P=0)1220 cm"1

Preparation of ( o-Chlorophenylhydroxymethyl)phenylphosphinic Acid(XXV).______’______

o-Chlorobenzaldehyde (l.4g., 0.01 mole), dichlorophenylphosphine (5 ml.) and "moist" aluminium chloride (2.6g., 0.02 mole) v:ere reacted as under General Procedure. At 70° a violent reaction occurred which quickly subsided. Work-up of the reaction product produced ( o-chlorophenylhydroxymethyl)phenylphosphinic Acid (XXV), (l.94g., 69%) m.p.283-284° from ethanol - water (q:l).

Found - (M-408l)(10543) : C,55.5; H,4.1; Cl,12.6%; Equiv. wt. 293

Calc, for C13H1103C1P : C,55.4; H,3.9; Cl,12.6%; Equiv. wt. 282

The n.m.r. in CF3 COOH showed 9 aromatic protons 7.0-7.75) on benzylie proton as a doublet (*>5.54; J PCH 11 c/s), V(0H) 3350 cm 1 (very weak).

Acetophenone jha action :

Acetophenone (l.2g., 0.01 mole), dichlorophenvlehosohine

(5 ml.) and "moist" aluminium chloride (2.6g., 0.02 mole) were reacted as under the General Procedure. The neutral'product (weighing 1.96g.,) showed to be a complex mixture of at le&st 6 products by t.l.c. The infrared spectrum indicated a strong P=0 but no carbonyl band and the xum.r. gave 38.

a complex mixture. ihe products could not be resolved by chromatog

raphy over silica-gel.

Nonan-5- one Reaction :

Nonan-o- one (l.43g., 1 mole), dichlorophenylphosphine

(5 ml.) and "moist" aluminium chloride (2.6a., 2 mole) were reacted as described under the^General Procedure. The neutral fraction yield 2.2g.,< of an oil showing a strong (P=0) band at 1230 cm ^ in the infrared and no carbonyl. The neutral product failed to crystallize and an acid . ! hydrolysis yielded an uncrystallizable oil. 1 Phthalic Anhydride Reaction :

1. Catalysed with "Moist" AlCl^.

Phthalic Anhydride (l.48g., 0.01 mole), excess dichloronhenylphosphine (5 ml.) and "moist" aluminium chloride (2.6g., 0.02 mole) were reacted as described under the General Procedure. In this case, the neutral product was extracted with chloroform. The aqueous alkaline extract when acidified yielded a yellow oil and phthalic acid (0.26a.).

The neutral layer afforded yellow crystals of trans-bipht.halyl (0.12g.,

1Q&) m.p.352-4° (S.T.) from xylene (lit.0"'* m.p.352-354).

Found - (13416) : C,72.75;

Calc, for - C16H804 : C,72.7 ; H,3.0o^, TT(C=0) 1,780 cm"1.

2. Catalysed with Anhydrous AlClq.

The reaction was carried out using identical quantities of reactants as before except anhydrous A1C10 was used. ■ The reaction was worked up by pouring the reaction mixture into excess cold N HaOH solution and filtering the resultant -solution through diatomaceous earth. After 39.

'X 7 XS*. washing and drying the filter cake the biphthalyl was extracted with

hot xylene. This procedure yielded 0.42 q., of transbiphthalyl m.p.

327-329° (S.T.) (34?0-

3. Catalysed with Boron Trifluoride:

Phthalic anhydride (l.48g., 0.01 mole) excess dichloro-

phenylnhosphine (5 ml.) and boron trifuoride, diethyl etherate (2.83g.,

2.5 ml., 0.02 mole) were reacted as described under General Procedure. The

yield of transbiphthalyl in this case was 3.4 mg. (0.2^) m.n.330-332° (S.T.)

4. Without Catalyst:

Phthalic anhydride (l.48g.) and dichlorophenylohosphine

were reacted as described under General Procedure. In this case no

biohthalyl could be obtained.

Attempted Reaction of Benzonhenone and Phosnhorus Trichloride:

Catalysed with "Moist" Aluminium Chloride.

Benzonhenone (l.8g., 1 mole) and "moist" aluminium chloride

(2.6g., 2 mole) were reacted with, or phosphorus trichloride (5 ml.) as

described under General Procedure for benzophenone, except the temperature

was taken to 70°c. Alkaline work-up yielded benzophenone (l.8g., 10C?o)

identified by its infrared spectrum and thin layer chromatography.

Repetition of the reaction with benzonhenone (l.8g., 3. mole)

"moist" aluminium chloride (2.6g., 2 mole), Phosphorus trichloride ('1.4n.,

1 mole) in chlorobenzene (5 ml.) at 100° for 3 hours failed to produce a

reaction and again benzophenone was obtained in a quantitative yield.

Trlnhenylmethylnhenylphosphinic Chioride (XXVI_Il).

Benzophenone (3.64g., 1 mole) and "moist" aluminium chloric-

(7.9g., 3 mole) were reacted as described under the General Procedure. The 40.

neutral fraction (3.2g.) upon recrystallization from ethyl acetate yield triphenylmethylphenylphosphinic chloride (XXVIII) m.p.208-10°

(L.4q, 5%)

Found (12943, It-1459) : C,73.9; H,5.2; P:7.8; Cl,8.8&

Calc, for - C,74.5: H,5.0; P:7.7; Cl,8.8&,^ (P=0) C25H20P0C1 1,235cm

The acid chloride (XXVIIl) was identified by infrared and mixed melting point comparisons with a sample prepared below.

The aqueous alkali fraction was acidified and when recrystallized from glacial acetic acid yielded triphenylmethylphenylphosphinic acid (XXIX)

(0.39g,5/£) m.p. 270-274°. This acid did not depress the melting point of the acid obtained by alkaline hydrolysis of the acid chloride (XXVIII).

Triphenylmethylphenyl^hosnhlnjc Acid XXIX).

The acid chloride (XXVII) (0.105g., 2.6 mole) was hydrolysed by refluxing in a solution of methanol (15 ml.), 2N RaOH (10 ml.) and water (15 ml.) for 5 hours. The cold reaction mixture was extracted with benzene and the aqueous layer acidified with concentrated HC1 to afford the acid (XXIX), 0.085g. (84^o) m.p.275-277°. Recrystallization from glacial acetic acid gave the pure acid. m.p.277-279°.

Found - (12893) : C,78.3; H,5.8%\ Equiv. wt. 379.

Calc, for - P0o : C,78.1; H,5.5%; Equiv. wt. 3P4.

Synthesis of Triphenvlmethylehenylrhos^h \ nj c Ch 3 o p' de

(XXVIII)

Triphenyl carbinol, recrystallized, purity 99% by g.l.c.

(lg.> 318 mole) was added to excess dichlorophenylphosphine (5 ml.). The 41.

reaction mixture was stirred at-100° for 1 hour. The cooled reaction mixture was poured into excess, cold 21! MaOH solution and the neutral material extracted with benzene (160 ml.). The aqueous layer was extracted again with benzene (75 ml.) and the combined benzene extracts, were washed with- water until neutral. Drying and removal of the benzene gave the crude acid chloride (XXVIII) 1.4g.. Repeated recrystallization from ethyl acetate afforded the pure acid chloride (XXVIII) m.n.208-210°.

Found - (12768) : C,74.0; H,5.2^

Calc, for - C25H200PC1 : C,74.5; (0.85g., 44%).

Quantitative and comparative analysis of the ethyl acetate mother liquors gave benzophenone (15 mg., 2%), triphenyl methane (92 mo., 10%), triphenyl carbinol (133 mg., 13%) and a small quantity of material with a retention time less than benzophenone.

Reaction of Trinhenyl'Carbinol and Dichloronhenylphosphine

for 15 mins.

The reaction was carried out as described for the 1 hour reaction using identical quantities of reactants, except, the reaction was stopped after 15 mins. 0.75g., (37,5%) of the acid chloride (XXVIII) m.p. 207-8° was obtained and gas chromatographic analysis of the mother

Y liquors,indicated benzophenone (66mg., 8.8%).

Trirhenylmethylohosphonic Pi chloride^Preparation

Triphenylcarbinol (lg., 3.83 mole) and excess phosphorus trichloride (5 ml.) were refluxed for 1 hour. The reaction mixture turned yellow and a dolid separated. .This solid was separated from the reaction mixture by filtering.' and washing with diethyl ether. Recrystallization 42.

from ethyl-acetate gave triphenylmethylphosphonic dichloride, m.p.

187-8° (lit.7^ 189-190) (0.42g.). After removal of diethyl ether and excess phosphorus trichloride under reduced pressure, the mother liquors afforded a solid which upon recrystallization gave additional dichloride, m.p. 185-6°, total yield 0.845g., (62/0 5 V (P=0) 1250 cm”"''.

Gas chromatography of the combined mother liquors indicated a trace

(ca 1 mg.) of benzophenone, besides triphenylcarbinol and triohenyl methane.

Attempted Reaction of Chlorodinhenylnhosohine and

Benzophenone;

Catalysed by Boron Trifluoride;

Benzophenone (l gm., 0.55 mole) and excess chlorodiphenyl-

(phoschine (5 ml.) were heated to 100° with boron trifluoride diethyletherate (0.2 ml.). A sample of the reaction mixture was taken every

30 mins, hydrolysed and injected into a gas chromatograoh. After 3 hours no triphenyl carbinol was detected. 43.

SECTION 2.

Alkaline Degradation of Phosphinic Acids.

Introduction: 13 Horner et a.1 have found that fusion of tetiary phosphine oxides with alkali at 200-300° gave an analogous reaction to the alkaline 12 degradation of phosphoniurn salts . The phosphine oxide decomposed to give the phosphinic acid with the loss as the hydrocarbon of the groun that is most stable as the anion*

200-300° R3?0 + HaOH ---- R3P(0)0Nn+ RH

Where the R's are different then the order of ease of loss of the group is

PhCH^- > Ph- > alkyl and -napthyl > CH^Ph-

As Horner and co-workers were able to obtain excellent yields of the phosphinic acid (80-lOQo) these acids would seem to be very stable to further thermal decomposition. There are only a few references to thermal decomposition of arylphosphonic acids when heated to 240° to 14 accord hydrocarbon and inorganic phosphate. Griffin obtained 2,5. diphenylfuran when 3-(2,5-diphenylfuryl)phosphonic acid was heated and Doak 15 and Freedman were able to prove the structure of 2,2'diphosphonodiphenyl-

• ether by a similar method. There, are no references to the thermal decomposition under alkaline conditions of nhosnhimc acids in the literature

While no attempt was made to study the thermal degradation of phosphinic acids in great detail, the technique has helped in the elucidation of the structures of some complex phosphinic acids. The method was used to degrade the' acid into a neutral fraction which could be easily identified by standard techniques and an acid fraction which, if 44.

it v/ere not easy to identify, could be nore readily synthesised than the parent acid. 45.

Results and Discussion:

Chlorodiphenylmethylphosphinic chloride (l) readily

decomposed v.ith sodium hydroxide at 300° to yield benzophenone (57ii),

benzhydrylphenylnhosohonic acid (Vi) (33%) phenylphos-'hinic acid 18%)

and traces of phenylphosphonic acid,

NaOH PhoCCl?(0)Cl PhoC0 + PhoCHP(0)H + PhP(0)H + PhP0(0H)o 2 | 3C0° 2 2 I I 2 ?h Ph OH . (I) (VI)

Assuming thecC-hydroxy acid (ill) is formed initially then the reaction must produce a reducing agent to reduce this acid to thec£-hydro-acid (Vi).

Grayson pronosed attack by the nucleoohile at the hydroxy group when he studied the alkaline decomposition of^C-hydroxy phos^honium salts, viz,

HO^^H^Di^P (CH?OH) OCH P(CH20H)3

and not at the phosphorus atom because of steric affects, Rv analogy,

the decomposition of the-hydroxy acid (ill) would be,

Ph HCfG^H -DP* c H O + Ph9C0 + Ph-P , . « * (III) (XXVII)

As a result of the tautomeric equilibrium.:

0 POH p-OH Ous the dianion of phenylphospnon® acid is a known reducing agent and

is responsible for the thermally induced reduction of mononsubstituted

38 phosphinic acids , 46.

3RP(0)H--- >- RPH + 2RP(0)0H I ^ OH

Therefore (XXVII) is probably responsible for the reduction of the e£-hydroxy acid to the^-hydro acid, and a sample of hydroxydiphenylmethylphenylphosphinic acid (ill) when fused with sodium hydroxide at 300° decomposed in the same way.

A more complex example was the dibasic acid (XVII).

However here the products were very straight forward,

300< H0(0)P-<^__ ^-CHP(0)0H -i- H0(0)P- CH2Ph 1-Ph2CH2 NaOH (XVII)

Determination of the structure of trinhenylmethylohenyl- phosphinic chloride (XXVITl) is another example whore the alkaline degradation facilitated the determination of the vt ruet ur»«. Prom the p.m.r., infrared spectral analysis and elemental analysis, the structure of the neutral product, obtained from the reaction of one mole of benzophenone. excess dichlorophenylphosphine and three mole of "moist." aluminium chloride, could have been (XXVTTl) or(XXTX)

PhoCP(0)Cl PhoCClP(0)Pho o j Z Z Ph (XXVIII) (XXIX) 47.

Comparative g.l.c. of the neutral fraction from the fusion of the acid chloride (XXVIII) showed only triphenyl methane (4]$). This evidence supported the assumption that the product was the acid chloride (XXVIII) and not the isomeric phosphine oxide (XXX). 48.

EXPERIMENTAL

GENERAL PROCEDURE:

The acid or derivative (l mole) was mixed v/ith powdered

NsOH (3 mole) and placed in a long-necked glass bulb in a Wood's metal • bath. The temperature of the bath was raised slowly until reaction occurred, which usually took nlace at the melting point of the acid or derivative. When the reaction had subsided, the temperature was taken to 300° and held at this temperature for 30 minutes. If the neutral fraction was low boiling the tube could be taken out of the bath at 300° and placed on its side until cool and the tube cut so as to isolate the neutral fraction directly. In other cases, the cooled tube was extracted with benzene and water. The benzene layer, after drying and evaporation afforded the neutral products. Acids were isolated by acidification of the aqueous layer.

Chlorodiohenylmethylphenylohosphini.c Chloride (i)

The acid chloride (i) (3g., 0.83 m. moles) afforded benzophenone (0.86g., 57%) from the benzene extract, identified by its infrared spectrum and they melting point and mixed melting points of its 2,4-dinlt.ro- phenylhydrozone.

Acidification of the aqueous extract nroduced benzhydrylphenylphosphinic acid (Vi) m.n.244-246° from aqueous ethanol (0.99g., 38%) identified by m.p., mixed m.p. and infrared spectral connarison with an authentic sample (see below) 49.

Found - (M10805) : C.74.1; H, 5.8; P, 9.9%; equiv. wt., 310

Calc, for - C19H170oP : C,74.0; H.5.6; P,10.05%; equiv. wt., 308

The n.m.r. of the sodium salt in D^O showed 15 aromatic protons

(ST.55 - 7.85) and one benzylic proton (6 4.85, JPCH 18 c/s). Concentration of the aqueous afforded phenylphosphinic acid (0.21g., 18%) m.p. 75-77°, 35 o Lit. 82 . The mother liquors were taken to dryness and extracted with hot ethyl acetate to give traces of phenylphosnhonic acid m.p.149-152°

Lit.36 158°.

These acids were identified by mixed m.p. and spectral comparison, with authentic samples. .

Preparation of BonzhydrvlnhenylphospHnlc Acid (VI)

Dimethyl phenylphosphonite (2.3g., 1.4 mole) was slowly

_ A1 added to stirred benzhydryl bromide (m.p. 39-40 , Lit." 44 , 2.6g., C.8 mole) at 100° over half hour period and the reaction mixture was heated for a further 2.5 hours. The crude mixture was hydrolysed by boiling under reflux with excess aqueous ethanolic sodium hydroxide solution

(0.5N) for 12 hours. The cooled hydrolysate was extracted with benzene and the aqueous layer acidified to give benzhydrylphenylohosphinic acid

(Vi) m.p. 244-6° from aqueous ethanol 1.7g., (47%). V \ ‘V Concentration of the mother liquors afforded the exoected by-product, nethylphenylphosphinic acid, m.p.127-128°, Lit.3 133-134 (0.7g.,) undepressed by an authentic sample.

(p-Phenylnhosnhonyl)diphenylmethylohenylphoschinlc Acid (mi).

This diacid (0.37g., 1.5 mole of impure acid) cave a 50.

neutral product, (directly obtained from the cooler parts of the fusion tube) identified as dinhenylmethane (0.048g., 34?o) by infrared and gas

liquid chromatographic comparison with authentic samples. Purity by g.l.c. was at least 95>a. The acid fraction yield (p-benzylohenyl)phenylphosohinic acid (XVl) (0.l5g., 59%) m.p.155-6° from aqueous methanol.

Found - (?'942l) : C,73.8; H,5.7; P,10.15o; equiv. wt. 314

Calc, for - C19H1702P : C,74.0; H,5*.6; P,10.05/o; equiv. wt. 303

StCDClJ: 3.9(-CH0, 2 protons); 13.1(P0H: 1 proton): 7.05-7.85

(ArH; 14 protons).

sC-Hydroxy-o-Chlorobenzylchenvlrhos^hinlc Acid (XXV)

The neutral products (obtained directly) were identified by comparative gas-liquid chromatography. The g.l.c. showed four peaks - three of which were shown to be chlorobenzene, o-chlorotolune and

o-chlorobenzyl alcohol. A material with a higher retention time could not be identified. The acids were liberated from the fusion mixture, after extracting with benzene, by acidification with concentrated HC1.

The free acids were extracted with diethyl ether and were recovered as an oil after the ether extract had been dried and evaporated. Portion of the oil was dissolved in ethanol and diazomethane add«d until evolution \ of nitrogen ceased. The g.l.c. was run on the methyl esters and me+hyl

salicylic acid and methyl phenylphosnhonic acid identified by cor-" arisen with the methyl esters of authentic samples. Pecrystallization of a

further portion of the acid fraction yielded phenylnhosrhonic acid, m.o. 152-154° Lit.^ 158° identified further by mixed m.p. with an 51. authentic sample. Because of the diversity of products no attempt was made to obtain yields of the products.

Triohenylmethylphenylnhoschinic Chloride (XXVIII)

The acid chloride (O.G46g., 1.15 mole) when reacted with sodium, hydroxide as described under the general procedure produced trephenylmethane as the only neutral product (O.Ollg., 4Y%) identified by comparison of infrared spectrum and retention time by g.l.c. with an authentic sample.

Hydroxydiphenylnethylphenylphosphinlc Acid (III)

The acid (ill) (5.9mg) afforded benzonhenone (4.3 mo. 66%) isolated as its 2 4-dintror>henylhydrazone and identified by its m.p. and mixed melting point with an authentic sample. 52.

SECTION 3:1

Reaction of Phosphorus (ill) Compounds with Dichlorodi-

phenylmethane.

Introduction : 39,40 The Michaelis - Arbusov reaction, where a nhosnhonate

ester is prepared by heating trialkyl phosphite with an alkyl halide, is one of the most useful routes to the formation of new carbon-phosphorus bonds, e.g.,

(R0)3P + R'X —>- (R0)9P(0)R' + RX

35 It has been shown that this is a general reaction and that it can be used for synthesis of phosphinate esters and phosphine oxides. The mechanism of the simple Michaelis - Arbusov reaction has been thoroughly 42 investigated and the generally accented pathway is -

A + ,0-R A \ \ >0R + R'X P p = 0 + RX / \ B . R' B R'

When the. halide (R'-X) has electron-withdrawing substituents

in the molecule, e.g., &Cbromocyclohexanone (XXX) and in polar solvents V the product is not the one expected from the normal flichaelis-Arbusov 41 reaction. For example in the Perkov reaction &c-haloaldehydes react.-,

with trialkyl phosphites to give dialkylvinyl phosphates (XXXI) together

with the exnected phosphorate (XXXIl), viz-. 53.

0 II R’RXCCHO + P(OR)3 —> RjRC = CHOP(OR)2 + RX (XXXI) (R0)2P(0)CR'RCH0

(XXXII)

This type of reaction is best explained by attack of nucleophilic phosphorus (ill) at the "positive" halogen to give the phosphonium salt

(XXXIIl) as intermediate. Attack of the nucleophilic oxygen of the anion leads to the phosphate ester (XXXI).

R’H

(RO)3PX -0!r'RCHC —> (RO)3P X + ~C*C=0 V 0-CH=CR'R R (XXXIV) (XXXIII) + (RO)3P X R x"

RX + (R0)2P(0)0CH=CR’R NPOCH=CR'R / (RO). (XXXI)

Extensive study of a large number' of similar reactions has indicated that attack of phosphorus (ill) on halogenated compounds can proceed through either attack at halogen, reaction (a), or at carbon, reaction (b)

+ “ R.P + R' X —y R'R P X (a)

r3p + R’x —> |r3px| + R, -R"OH R.H + R POR"X_ ^ / 3 / (XXXVI)

R3PO + R" X (b) 54.

42,43 Kinetic evidence ? indicates that in the displacement of halogen by ohosnhorus reaction (a) follows a S^2 mechanism where the substrate involved is a primary aliphatic halide and this indication is strengthened 44 by investigation of the stereochemistry of the reaction by Horner

In equation (2), the presence of the ion-pair (XXXV) has been detected by carrying out the reaction in a protic solvent. An example of this is the use of diethyl malonate in the reaction of triphenylohosphine and

bromocyclohexanone (XXX). With anhydrous acetonitrile as solvent the enol phosphonium salt (XXXVII) is obtained, while with diethyl malonate cyclohexanone is the major product as diethyl malonate protonates the intermediate viz. _

As there is no quid'* in determining whether attack will 46 be at carbon or halogen Hudson has applied the "hard" and "soft" base 48 concent of Pearson to this problem. A "hard" base is one which is \S highly polarizable and a "soft" base is non-polarizable. Because some bases are only partly polarizable, ‘there are bases that are intermediate types and this introduces a problem, although in a number of cases this concept has been proved very useful. For example, it has been used to explain the small reactivity of«fc-chlorcyclohexanone compared to the 46 facile reaction of oc-bromopyclohexanone towards trlphenylphosphine 55.

As the reaction of halogenated compounds with tervalent phosphorus has been largely confined to the study of simple ohosphines

(R^P) or simole phosphites ((RO)gP) it is of interest to study the effect of varying the nucleonhilicity by using mixed phosohines under identical condition. Thus, the reaction of trimethyl phosphite (f.’.eO^P, dimethyl phenylnhosnhonite, PhPCor.'e)^ methyl diphenylphosphinite, Ph^POMgj triphenylnhosphine, Ph^P, and tri(n-butyl)phosphine, Bu^P was studied under identical conditions with dichlorodiohenylmethane. It has been 42 43 shown 5 that the nucleophilicity of this series increases from phosphite to phosphine. Two other compounds also were investigated in this reaction, tris-(dimetb.ylamino)phosphine, (Me^N)^?? and triphenylarsine, Ph^As. 56.

Results and Discussion :

All the reactions, with the exception of the reaction of tributylphosphine and butanol, were carried out under same conditions. An excess (1.2 mole) of dichlorodiphenylmethane was heated to 105° and the, phosphine (l mole) was added over 0.5 hr. The reaction mixture was heated for a further 0.5 hr. at 105° after the addition of all the phosphine. In the reaction of tributylphosphine v/ith the dichloride in the presence of butanol the addition was reversed, the dihalide being added to a mixture of tributylphosphine and butanol in order to avoid side reactions. In all cases the reaction product had a deep red colour which disappeared on addition of protic solvents. In general, ethanol k was added to the cool reaction mixture to isolate-tetraohenylethylene which is only sparingly soluble in this solvent. In cases of low yields of tetraohenylethylene, the reaction mixture was examined by g.l.c. in order to estimate the amount of tetraohenylethylene formed.

The results are summarized on Page 57 and in Table 1. 57

TABLE 3

Formation of Tetraphenyle.thylene (XXXVIIl).

From (l.O mole) and dichlorodiphenylmethane

(1.2 mole) of 105° for 1 hr. unless otherwise

stated.

R3P Reaction Conditions . Yield %

Bu3P 49

1 Mole Ph^Co added 50

2 Mole BuOH added 12

1 Mole Xylene added 6

1 Mole Xylene added 40

(Me2N)3P ?7

2 Mole of Ph2CClCClPh2 used 76 in boiling be'nzene 8

Ph2P0Me Nil

4 hr. reaction 13

15 hr. reaction at 115° 20 58.

P(OMe)3 + Ph2CCl? (c)

PhP(OMe)2 + Ph2CC19 Ke0P(0)PhCHPh (880 + Me0P(0)PhCClPh2(4^). (d) (V)

MeOPPh + Ph2CCl Ph P(0)CHPh (IQ%) (e) (XIX)

Ph3P + Ph2CCl2 Ph^-CHPhgCr (33^) + Ph2C*CPh2(l.4^) (f) (XXXIX) (XXXVIII)

Bu3P + Ph2CCl2 -> Ph2C=CPh2(5C^)' (g)

'(Me2N)3P + Ph2CCl2~^ Ph2C=CPh2(76$) (h)

Ph3As + Ph2CCl2 (i)

The yields (in parenthesis) have been calculated from the phosphorus (ill) compounds and where tetraphenylethylene is formed by C'-QPhn the theoretical equation: R3P + Ph2CCl2 -->» R3PC12 4- ■£- (Rix^SSb^).

With both trimethyl phosphite and trinhenylarsine no reaction could be detected, with the former this was shown by quantitative recovery of benzophenone (from hydrolysis of the dichloride) and the latter gave 49 high yields of unreacted triphenylarsine. Kosolanoff has been able to obtain good yields of esters of trichloromethylohosphonic acids only after reacting the phosphite with carbon tetrachloride for over 24 hours. Burn 50 and Cadogan have investigated the reaction of trimethylnhosphite and carbon tetrachloride in the presence of methanol finding the reaction sluggish. Thus it is possible the reaction between trimethyl phosphite and the dihalide could be effected under more forcing conditions. The results show that the phosphines, (arranged in order of increasing nucleophilicity) give an absolute increase in per cent reaction product and that the type of reaction’product changes from phosphine oxide, through phosphonium salts to tetraphenylethylene. 59..

A probable explanation of the formation of the phosphonium

salt (XXXIX) would be attack of phosnhorus upon the cC halogen of the phosphonium salt to produce the ion pair (XL) which could be destroyed

upon work up with the ethanol, viz. + + + “ Ph3P + Ph3PCClPh2 —>• Ph3P-Cl 4- Ph3PC-?h2

(XL'

Ph3PCl(OE+) + (Ph3P CHPh^jCl

(XXXIX) (3)

A similar mechanism could be used to explain the formation of (XIX) and

(V). A number of possible mechanisms could account for the formation of

tetraphenylethylene in the above reactions.

(a) Dimerization of the carbene (XLl) formed by attack at the

halogen by the phosphorus (ill) compound to give the ion oair (XLll)

which could decompose to give the carbene (XLl) e.g.,

R3P + Ph2CClQ _>. (R3P"ci)(Ph2CCl) ~> PhC: —> Ph2C=CPh2

(XLII) (XLl) (k)

53 Although Seyferth et aP have noted that phosphines act as carbene

traps, this would not detract from this mechanism as reaction of the

/ m" \ 54 carbene with the phosphorane OJJSET-) could produce tetraphenylethylene

e.g.

R3P=CPh0 + Ph2C:--R3P + Ph2C=CPh?

(lags.) (XLl) (1)

(b) Tetranhenylathylene could be produced from symmetrical 60.

dichlorotetraphenylethane (XLIV). As phosphites are known to convert 55 vicinal dihalides to olefins - although this reaction occurs when the

dihalide has electron withdrawing groups in the molecule - attack of the

anion (XI..III) by the following mechanism could lead to the formation of,

tetraphenylethylene, e.g.

V + Ph2cci2 (R3PCl)(Ph2CCl) ™CC1r?- Ph2CC10ClPh2

(XLITI) (XLIV)

R3PC12 + Ph2 C=CPh2 (m)

In this mechanism the dihalide is not activated by electron-withdrawing

groups, however, relief of steric strain may be put forward as the

necessary driving force for the reaction. Some credibility is given

to this mechanism, at least the formation of the saturated dihalide .

(XLIV), by the formation of tetraphenylsuccinonitrile as a by-product 56 from the reaction of triphenylphosphine and chlorocyanodiphenylmethane

the standard reaction conditions the reaction of methyl diphenylnhos- ohinite with the dihalide, the phosphine oxide (ill) is formed, whereas

if the -conditions are more forcing the yield of tetraphenylethylene

increases. This observation supports a third mechanism which supposes

that the formation of tetraphenylethylene proceeds through the phosphine

oxide (ill) or its conjugate base e.g. CClPh2 PhoCP(0)Ph2 + PhCCl0—Ph2P(0)CTPh2 + Cf->Ph2P(0)Cl + Ph2C=CPh2 + Cl

(XLV) (n) 61.

This mechanism found some support by forming the phosphorane (XLV) from

benzhydryldiphenylphosphine oxide (ill) and phenyllithium and reacting

the phosphorane obtained with the dihalide under similar conditions used

for the standard procedure. Tetraphenylethylene {37%) (based on equation

(n)) was obtained and traces of diphenylphosphinic acid were isolated.

If the above equation (n) were operative the yield of ohosohinic acid

should be the same as the yield of the ethylene. If all the phenyl

lithium had not reacted with the ohosphine oxide (ill) the unused lithio

compound could react with the dihalide to give tetraphenylethylene in a

similar fashion to the reaction of Grignard reagent. Therefore it was extremely important to have some idea of the extent of formation of the phosphorane (XLV). This problem was overcome by forming the phosphoranq and reacting it with excess deuterium oxide viz.

(Ph2CP(0)Ph2) Li+ + D20 --->• Ph2CD P(0)Ph2 + LiOD

(XLVI)

The purified deuteriated compound (XLVl) was examined by n.m.r. which showed an absence of the benzylic proton which proved that it had been exchanged for deuterium. To study the supposition that the phos phorane is involved in tetraphenylethylene formation when the intermediate could be a phosphonium salt, the phosphorane (XLVII) was prepared from benzhydyltriphenylphosphonium chloride with phenyllithium and again tetraphenylethylene (33%) was obtained when treated with the dihalide under the standard conditions. 62

Ph3P = CPh2 + Ph2CCl2 —>- Ph2C = CPh2 + Ph3PCl2

(XLVII) Thus there are at least three ways tetraphenylethylene

can be formed when dichlorodiphenylmethane and the phosphines react.

The first two mechanisms go through the ion pair (XLIIl), formed by

initial attack at the halogen. As this ion pair will decompose with

protic solvents to block the formation of tetraphenylethylene, viz.

'OH hPCI] fPh2cc3 —>_ R'Cl + r3P0 Ph-CHCl (XLIIl)

the reaction of tributylphosphine and the dihalide was investigated in

the presence of butanol. If, however, equation (j) is operative, that

is initial attack at the carbon to give phosphonium salt, butanol should

have no effect. One problem encountered in the case of butanol was the

ability of butanol to react with dichlorodiphenylmethane, e.g.,

BuOH + PhoC012 —>- BuCl + Ph2C0 + HC1

However, it was found that tetraphenylethylene was produced from the

reaction of tributylphosphine and the dihalide in the presence of butanol

although the yield was lower (12^) than the yield of the ethylene (4C%) when an equivalent amount of inert solvent, xylene, was added.

In the reaction of tributylphosphine a further piece of evidence for

attack at the carbon atom was the inability to find any tributyldiphenyl- methylphosphonium salt in the mother liquor and that a complete material

balance was obtained. The gas chromatographic study of the reaction 63.

mixture indicated the presence of tributylphosphine oxide and Bu^PCl^

(which gave material a balnace with the tributylphosphine used. Although traces of phosphonium salt would not be found by the above method, it is unlikely that this would make any difference to the course of the reaction

as there was a low yield of tetraphenylethylene in reaction (f) where

large quantities of phosphonium salt were also obtained. Therefore, the most probable mechanism for the formation of tetraphenylethylene seems to

be initial attack at the carbon e.g.

R3P + Ph2CCl2 —^ [R3P+ CClPh^j Cr~RSP R3P=CPh2 + F^PC^ ^Ph2 Ph2 Ph2C-CPh2 + R3PC12 <- R3P+-^C^ C + Cl'

In an attempt to investigate the effect of a "hard" base on dichlorodiphenylmethane, potassium t-butoxide was reacted with the dihalide in t-butanol. With "soft" bases, e.g. iodide ion and zinc" the product is tetraphenylethylene while "harder" bases give substitution

(e.g. alkoxides ). With the reaction of potassium t-butoxide in t-butanol with the dihalide gave a product which was purified by chromat ography to give a sharp melting crystalline material in low yield. This product was shown by mass spectrometry to consist of a mixture of three components whose molecular weights corresponded to tetraphenylethylene

(XXXVIII), an epoxide of tetraphenylethylene and a monochlorinated deriv ative of tetraphenylethylene. The presence of (XXXVII) and its epoxide were supported by gas chromatography while analysis indicated apcroxi- mately 1:1:1 mole ratio of the three components. As potassium t-butoxide is a "hard" base it should attack the carbon atom, but here 64.

this atom is shielded by bulky groups and the attack takes place at the halogen.

The results above indicate that it is very likely that the phosphine attacks the carbon atom of the dichlorodiphenyl methane. Before the halide would be more amenable to attack at the halogen, it would seem that the anion of the resulting ion pair should be stabilized. An example of this type of attack was obtained by

Mark^ who established attack at the halogen by trialkylphosphites on hexachlorocyclopentadiene.

The resulting stability of the anion is due to its aromaticity and it presented a symmetrical halogenat^d molecule for attack by the cation.

The high reactivity of tris-(dimethylamino)phosphine is of interest. As no data are available for nucleophilic reactions of this compound, the reactivity was qualitatively compared with t.T?rhenvl- phosphine towards methyl iodide. Equi-molar quantities of the rhosnhin^-s were dissolved in the same quantity of acetone and equivalent amounts of methyl iodide (in acetone) were added. After two hours standing at room temperature, while 87% yield of tri,s-(d#methylamino)methylphosphonium iodide was obtained, there was no precipitate of triphenylmethylphosphonium iodide. 65.

EXPERIMENT A L

Tetranhenvlethylene.

An authentic sample of tetraphenylethylene was prepared ’

by refluxing dichlorodiphenyl methane (2.4g., 1.0 mole) (see Page 29 )

in diethyl ether (60 ml.) with zinc dust (0.65., 1.0 mole) until all

the zinc had dissolved (5 hours). The colour of the solution changed

from yellow to green and a solid separated out at the end of the 5 hours.

Benzene (50 ml.) was added and the organic layer was washed with dilute

HCl and water until the water washings were neutral. After drying

over anhydrous sodium sulphate the benzene was removed to yield 1.8g

of product. Repeated recrystallization from glacial acid yielded

tetraphenylethylene (0.85g., 51/o) m.p.217-219° (Lit.°* 220-221' ces

of halogen were difficult to remove. One of the samples belc

analysed,

Found - (M-8085) : C,92.8; H,6.05; Cl,0.4#

Calc, for - ^95^20 : 0,93.9; H,6.1^; 51 Mackenzie also had difficulty in obtaining a good analysis when he

obtained the olefin by the action of sodium t-pentyloxide on dichloro-

diphil -methane in boiling xylene for 12 hours.

Reagents.

Triphenyl - and tributyl-phosphine and triohenylarsine

were commercial samples. Trimethylphosnhite was a commercial sample

which showed £a lO^o dimethylphosphonate by p.m.r.. The following

compounds w'ere prepared by literature methods: dimethyl phenylphosohonii> ,

methyl diphenylphosphinite and tris(dimethylamino)phosphine' 66.

G^jjeiiaJ. lethodT

’■ The d'r.ch.lorodiphenylnethane (1.2 mole) v/as added to a flask

ecu it :ed with macnetic stirring and reflux condenser. The ^lask was heated

in a silicone oil fcajth to l(j>5 and the phosphine (1.0 mole) was added i | , slowly through a rubber sep+tum by means of a hypodermic needle over'30 1 ! , minutes at this terp erasure. \fter addition of the chloride the temper

ature was maintained at 105' for a further 30 minutes. . In aenerali the

T k' ' j i I, ' i'1 reaction mixture was cooled and ethano.' added in order to precipitate

te tr anhen v 1 et hy 1 e,n e In cases of low yields of the ethylene thp , : 11 reaction mixture was studied by g.l.c. using the column described on d ! j ij Page 25. h'ith the injector block at 300 , the column at 275, and'?. t > • . ' , flow raf1? of 40 ml. (min., tetranhenylethylene had a retention time of

4,min. 50: sed.J11- ■ i, i ' ' 1 1 ■ ! • |j' ' 1 1 • ,1 ;»rimethy'lmhosphite React.ion.

From dichlorodiphenylmethane (3.75g., 1.2 mole) and

P(0Me)_, (1.8o«. 1-0 role) reacted under the general reaction, benzo- \.p ' ! t phenone (2.,?2g., 99%) was recovered and identified by comparison of its

infrared spectrum.

’’ethyl dlohenylphosphinite Reaction (l hr.)*

Addition ,of ethanol to the product from tie rr met-' -6 of 25 36 p Ph?PQMe (3.64Q., 1.0 mole) B.P.,.140-143 1.6020 lit. P' \51-2 , .5' ! L ;F 1.6C30 and dichlorodiphenylmethane (4.26g., 1.2 nole) 'cpv? rr.rhyd' a*1 - n 1 ' r |-:i-'>henvlrf(opnhine oxide (XIX) (0.53g., ,9.6%) m.p. 590-292 If-';;

1 dehlifi eiii pf infrared and mixed melting point comoari son with on •ufhenJ ' j i1 '; •;! i s (amp 1 e prbmered .Jchlorgdiohenyl phosphine, and benzophenone c.-teiei rod ltd ■ I 1 i v\3t)e. "moistaluminium chloride. (See Page 28 ) 67.

Ji Methyl Di^henylahosr-hinite Reaction 4 hr.

I * Usina identical amounts of phosohine and dihalide as above , ■!' : I and allowing the reaction tojstir at 105 for 4 hrs., the oroducts were i ’ precipitated from the reaction mixture by addition of ethanol. The I ' i ; i crude oroduct thus obtained was recrystallized from glacial AcOH to yield j o tetraphenylethylene1 (0.36g., 13>4), m.p. 213-215 identical to the sample ;

• M ! * ,1 \ , jl I ' orepared from the reaction of1 zinc or> dichlorodiphenylmethane. Concen-

tration of the acetic acid mother liquors afforded benzhydryldiphenyl-

phosphine oxide (XIX) (0.2g., 3%), m.p.292-294°. i i ! - ' ' • j ' | ^ethvl Diohenylnhos">hin1*te Reaction 1*> hr. at 115 J .

Identical quantities of ohosnhine and'dihalidje were: reacted

as described aljove except the'reaction was allowed |to nroceed for 15 hr. 1 o . !;1 .■ - 1 i at 115 . Isolation of the Products, described abcjve, yielded tetra-

' J I i nhenylethylene (0.54g., 20/o) and traces of the oxide (XIX) (0.05g. , ca 1%).

Tr d i m f thy 3_a mi no )phosph ‘ no React i ort ■ i rris(di'nethylamino)phosnhine (2.lPg., 1.0 nple) B.P. 15 7S! 14 62° n^‘ = 1.4648j} lit./o, B.P. 51-52,nj^4 = 1.4655) v/as reacteji with the I ( I ' ' dihalide (4.26g., 1.2 mole) solidified after only half of the amine v/as It i , 1 • added. Addition of ethanol afforded tetrarhenylethyl-jne (],7n„, 77/0 p.

212-214 .

the reaction was repeated using dichlorodiphenyine’hanc • i |4.26g., 2^0 mole) and amine (l.43g., 1.0 mole). No increase ;in yield

of tetrapKenylethylene was noted (1.11a., 76%) •

^enz.ene (15 ml.) was used as a solvent with the dichlorodiphenylmcthcne

(3.83g., l.l2 mole) and the reaction v/as heated to reflux. Addition of 1A ' . i ' asiini (2.9a., 1.0 mole) was carried out as before. At the end -A- !;m 68.

reaction the benzene was' evaporated under reduced pressure and ethanol •'

added to afford tetraohenylethylene m.p. 216-217° (0.17g., 8.3/o)«

Trlphenylohosfhlne Reaction.

The cooled reaction mixture obtained by the General

1 , — Procedure from triohenylohosphine (3.3g., 1.0 mole) and the dihalide

(3.6g., 1.2 mole), nave on treatment with acetone, benzhydryltriphenyl-

ohosphonium chloride (l.95g., 33/o), m.p. 248-250° (lit.00 240-242°), ij 1 ’ ^(CDCl^): 25 aromatic crotons (7.0-8.C) and one benzylic (-3.30;

JPCH 18 c/.s) croton (lit.52 S (CDClJ • 8.23: JPCH 17.9 c/s). (As a z -I ; 1 i further proof to the structure the m.p. and infrared spectrum was compared 1 i • to an authentic sample prepared by heating together benzhvdryl chloride

and t.rinhenylphosphine.) The hydrolysed pother liquors contained

tet;raphenylethylene (3Cmg., 1.4r/) and fcenzochenone i (1.7g., 62°' recovery)

t i ’ 1 i estimated by gas chromatography. I

Pr no a rat yen o f_ C arb a n l_q n _f ron _B onp^vdrvid ‘ phony Imhosch 5 ne I Oxide (XIX) and Reaction with Dichlorod1pheny1metha ne. ! ( 1 The phosphine oxide (XIX) (0.2g., 0.55 mole) ‘m.p. 295-7 lit.' 303 was suspended in dry diethyl ether (20 ml.) in a fl.ask equipped

• ’ | vjjith stirring, reflux condensor and nitrogen bleed. To the mixture a

I 1 i . • • solution of phenyl lithium in ether (0.55 mole) was added bv- a hypodermic

needle through a serum can. The mixture was refluxed ro;r 15 mins.

. -i >, .phosphine oxide1 dissolved to give a deep red solution. Ihe reaction

(mixture vlfes quenched vdth Do0 and the solid that separated was filtered off

i a;nd recrystallized from acetic agid. The o.m.r. of the resultant product

. i i i • showed the absence of the benzylic proton, indicating its replacement cy

deuterium. 69.

The carbanion was prepared as above usino ehosohine oxide

(XTX) (l.Og., 2.6 mole). After refluxing the oxide with an equimolar amount of ohenyllithium in either, dichlorodiphenylmethane (0.77g.,;3.1 mole), the ether was removedl after refluxing for 1 hour and the reaction I mixture yielded tetraphenylethylene (0.32g., 37/o based on the nhosohine oxide). Extraction of the mother liquors with saturated NaHCO^ solution followed by acidification gave dirhenylrhosbhinic acid (15 mg) m.o. 185-186u lit.^ 192-195" identical to an authentic sample.

In order to determine if benzhydryldiphenylphcsphioe oxide (XIX) m.pr

295-7° may have been contaminated with a trace of diphenylohosphin.ic acid, a sample of the oxide (XIX) (0.6Sg.,) was taken, finely ground, dispersed in ethanol (5 ml.) and shaken,for 30 mins, with a saturated NaHCO^ solution (15 ml.!) The mixture was filtered, a further amount of I • . water (15 ml.) was added and the bicarbonate filtrate acidified with cone. HC1. No acid was obtained.

Trlbutylnhosphine Reaction.

The ehosrhine (2.3o., 1.0 mole) and the dichlo‘'odinhenyl- 1 | methane (3.2q., 1.2 mole)'reacted under conditons given in +ho General

Procedure, gave tetraphenylethylene (0.92o., 495'i) m.n. 216-213°. Hydroly-

I of the mother liquors afforded benzophenone. Gas chromatography of the hydrolysed mother liquor afforded tributylphosphine oxide (2.6e.. ].' 'C

! Recovery) f|

Tributylphosphine Reaction with Xyl ene (3. jjolq).

• ! The phosphine (2.3g., 1.0 mole) was heated to 105°, wiJ‘h xylene (1.0 ml. 1.0 mole) was added and the dihalide (3.2g.. I." mole' 70.

v;as added over 30 mins. After the addition of the halide, the mixture was heated at 105° for a further 30 mins. Ethanol was added to the

cooled reaction mixture to afford tetraphenylethylene (0.76g., 40%) 1 ' . . m.p. 215-217°.

\ Tributylr-hosrhine_ Reaction with Butanol (l 1-ole).

The reaction v;as carried out as described above for the j ! ! I reaction in the presence of xylene, except the xylene was replaced with

n-butanol (0.85g., 1.0 mole). This reaction afforded tetraehenylethylene m.p. 214-216° (0.22a., 12.3f0* Gas chromatography of the hydrolysed mother liquors gave benzonhenone (2.Og., total recovery with the ethylene

equivalent to 2.25g., benzophenone, 94;j recovery).

Tributylpho sphi n< Rea_ction_ with_ 2_Mq 1 e Butano 1

The reaction, was carried out as described above, excent

the butanol was increased to 2.0 mole (l.9g.,). Gas chronatooraohy

of the reaction mixture gave tetraphenylethylene (l20mg., 5.7/o).

Reaction' of Dichlo^odi^henyl^ethane and Dry Butanol.

DjLchlorodiphenylmethane (l.lbg., 0.43 mole) and dry

butanol (C.35g., 1.15 mole) were heated to 105° for 45 mins, under

nitrogen. A small samole (60 mg.) v;as taken from the mixture and

added to dry diethyl ether. Gas chromatography of the solution gave

benzonhenone (0,;35g., 96/o yield).

Reaction of Phosohorane from. Benzhydryltrirhervl i '! " " r ~ ~ . / Phosnhoniun Chloride and, Dlchlorodiphenvly^etbane.

The phosohorane was prepared by treating the nhosrhonium

salt (V: 0.19g., 0.4 mole) with a equimolar quantity of phenyl11.thium 71.

■

,v and refluxing in dry ethyl ether, for 15 min. To the resulting deep

red suspension was added an 0.2 molar excess of dichlorodiphenylmethane.

The ether was removed and the reaction mixture heated to 110° for 1 hr.

Gas chromatography of the reaction product in chloroform showed tetra-'-

phenylethylene (90 mg., 33$ calculated on the phosphonium salt).

Potassium t-Butoxide >

Potassium t-butoxide (2.7g., 0.02 mole) in t-fcutanol

(30 ml.) was heated to reflux dichlorodiphenylmethane (4.74g., 0.02 mole) v was added, and the solution refluxed for 1-ip hrs. The solution turned

brovm, the solvent was removed, and the residue triturated with acetic:

acid, yielding 130 mg. of crystals, m.p. 154-156' from acetic acid.

T.l.c. showed two spots. The product was chromatographed on silica gel

and eluted with ibenzehe to give 125 mg. of material showing a sinole

spot on t.l.c. (Found: 0,06.3; H,5.4: 01,6.1$). 1:1 mixture of

(XXXVIII), a monochlorinated derivative, and the epoxide of (XXXVIII) reqimre

0,38.4: H,5.5; Cl,5.0$. Mass spectra and gas chromatography showed

that the material was a mixture of three products. From mas chromat

ography two products were shown to be tetraphenylethylene and tetre- i phenylethylene opcide: the other Product would anpea^ ■‘•o ~ rh1orin-

ated tetranhenylethylene from its mass spectral molecular v,a^v't $366

Parent ion peaks for (Vl) and its oxide were observed in the ee ] spectrum., (

Dimethyl Phenyiphospnonite Reaction.

This reaction has already been described in Section 1.

see Page 30. 72.

SECTION___±

/_t tempted jnrepare ti on _ of_ JLO- Ph^nyl^hosîpcî ône. I Tntroductton:

Jones and Mann^~ have prepared 10-phenylarsacridcne

(XI.VTl) and they deduced by chemical and physical means that there is considerable contribution to the structure from the zwitterion (XLVIIl) similar to that shown by 10-phenylacridone^

(XLVII) (XLVIIl)

Thus it was found^ that the arsine was not able to be quaternized and the carbonyl group did not react with hydroxylamine or 2,4-dinitrophenyl- hydrazine under riormal conditions. Further, the carbonyl stretching frequency in the infrared spectrum was 15 cm ^ lower than the normal frequency for comparable enthrones. Acridone exhibits to a greater extent evidence of this type of charge separation. 62 Gallagher, Kirby and Mann preoared 1-phenyl 1,2,3,4 - tetrahydr>-4- qxophosphinoline (XLIX) and found no evidence for the charge distribution described above. 73.

0 0 l» tl

(LVII) (XLIX)

Therefore it was of interest to prepare 10-phenylphosphacridone (LVIl) to complete this series of compounds for the Group VA elements and to

investigate the effect of the phosohorus atom on the properties of the 63 carbonyl. The routes to phosphorus heterocyclics have been described ' I in detail. A number of methods would seem to be applicable to this case.

10-Phenylphenoxaphosphine (LIIl) (where X-Ph) has been prepared by reacting the diazonium tetrafluoroborate salt (L) and dichlorophenylphosphine with cuprous bromide to yield the phosphonium salt (l.l) (X=Ph) which upon reduction with aluminium gave the phosphine (LIIl) (X=Ph) from the mono- I chloro compound (LII ).- 74.

hnci2

(LI)

A* PA PJfCl XpH (LII) (LIII)

Zinc chloride can be used also to effect the ring closure, particularly

6 Id in the case of the dichlorophosphine, (LIl) where X=C1. It was 66 decided to try the method used by Davis & Mann to prepare phosphanthren ring systems. They used the dichloro derivative (LIV) which they reacted with lithium and then dichloroethylphosphine to afford the heterocyclic (LV) viz.

(LIV) (LV) 75.

The most promising route seemed to be to prepare 2,2 -dichlorobenzophenone

(LVl), protect the carbonyl group by formation of a ketal and then to react this with lithium and dichlorophenylphosphine to give 10-phenylphosphacridone (LVII) viz,

H2°

(LVII) 76.

RESULTS AND DISCUSSION

o,o-Dichlorobenzophenone was prepared by the method of 67 Hal]er etal who obtained it from the oxidation of bis-(-o-chlorophenyl)

-carbinol. The alcohol was nreoared by the reaction of o-chloroohenyl- magnesium bromide and o-chlorobenzaldehyde e.g.

-^XCHO

Tt is interesting to note that the original workers found 1 he alcohol

(LVIIl) as a high boiling oil, whereas in this study the rroduct. was ' o obtained crystalline m.p.86-87 . No other reference to the preparation of the alcohol by this method was found in the literature.

As dimethylsulphoxide and acetic anhydride has recently 63 been described as an oxidizing agent for alcohols, this method was tried with the alcohol (LVIIl). It was extremely difficult to remove traces of solvent and as the resulting oil could not be crystallized, t'-o- original method of oxidizing the alcohol with chromic acid, acc: • <" i d 77.

and water was used. When the water was omitted a high yield of the acetate of bis(o-phenyl)carbinol was obtained.

The ketal from o,o-dichlorobenzophenone and ethyleneglycol proved impossible to prepare. Even under forcing conditions ethylene glycol and the ketone would not react. It is thought that this could be due to steric hindrance around the carbonyl carbon. Some evidence >. 17 to support this is the report by Avoyan etal who found evidence of steric hindrance in o-chlorobenzoic acid when they studied the bond angles in the acid by X-ray analysis.

Because of the difficulty in forming the ketal, the alcohol

(LVIIl) was converted to the tetrahydropyranyl ether.

The reaction of bis.-(o-chlorophenyl) carbinol (LVIIl) and 2,3-dfhydropyran proved facile and the ether (LIX) was readily obtained as a crystalline product m.p.67-68°.

(LIX)

If ring closure is effected then the alcohol could be oxidized to the ketophosphine oxide and then the phosphine oxide reduced with trichloro- 74 silane . An attempt to close the ring by using the dilithio derivative of dichlorophenylphosphine (prepared by reacting phosphorobenzene (LX) w'• lithium)e.q. 78.

(PhP) + lOI.i —>- 5(PhPLi2)

(LX) met with little success. The ether (LIX) was added to a solution of— the lithio compound in tetrahydrofura.n at -70°. After an hour at this temperature the reaction was allowed to come to room temperature. After a further hour at this temperature the reaction was refluxed for three hours. Preparative thin layer chromatography of the neutral reaction product, gave an almost quantitative recovery of the ether. Extension of the period of reflux to 10 hours gave no better result. .

Reaction of the tetrahydropyranyl ether with lithium, viz.

and then attempting to effect ring closure with dichlorophenylphosphine e.a. 79.

was unsuccessful because replacement of the chlorine atoms by lithium proved difficult even in the presence of triethylene diamine which has 7b 79 80 been reported 5 ' 5 - to heln in this type of reaction. This mode of attack may have been successful if the dibromo derivative was used in lieu of the dichloro compound because of the easier displacement of- bromide by lithium. . ’ !-

A further attempt to replace the halogen this time with an excess of n-butyllithium was unsuccessful, also in the presence of an excess of amine. Butyl lithium and the amine were added to benzene and heated to reflux. The dichloride (LIX) was dissolved in benzene and added to the refluxing mixture. After 6 hrs. a sample removed from the mixture and hydrolysed with aqueous, methanolic hydrogen chloride gave an infrared soectra identical with alcohol (LVTTl).

Attempts to nrepare the Grionard of the dihalide (LIX) in totrahydrofuran did not give the required product. No reaction could be induced and after refluxing for 4hours an almost qua^Mtative recovery of the magnesium was obtained. Again this method may have been successfu1 if the dibromo compound had be^n used.

A further attempt to prepare the di-Grignard using a 77 technique published by Baun et al who were able to nrerare the Grim.-.r- of N-ethyl 2,2’dichlorodiphenylamine, 80.

Even though a positive test for Grignard was obtained, after 50 hrs. refluxing in tetrahydrofuran nroduced very little reaction. A different mode of attack was briefly investigated. The acid chloride from •

(d-bonzoylphenyl)phenylphosphinic acid (LXl) was prenared and reacted with anhydrous zinc chloride65 at 130° for 24 hrs. The expected reaction

sequence is given by

0 0

S0C1

P(0)OH P(0)C1

ZnCl

Ph 0

Mo reaction took place ns the orinlnnl acid (I.XT) was ovm '•(! In almost quantitative yield, after hydrolsis of the reaction >. tore.

Briefly trifluoroacetic acid was used to try to obtain r' c'1 v/i + h +he acid (LXl), e.g. 81.

(LXI)

A Quantitative recovery of the acid (LXl) demonstrated that this method

too, was unsuccessful.

It would seem from this study the best route to 10-phenyl phosphcaridone would be through bis(o-bromophenyl) carbinol and the reaction of this dibromo compound with Grignard. However, the t5me necessary to pursue this method was not available. 82.

EXPERIMENTAL SECTION.

Prenarat/on of bl s(o-chloroph°nyl) Clarhino 1 (i.VIII).

Note:

The alcohol was prepared by one of the methods of 6T Haller et al . Initial attempts to prepare the Grignard met with

little success. It is not known if this was due to using diethyl ether,

that had been too rigorously dried as the ether used for the original

preparations was used after distilling from lithium aluminium hydride.

In the following preparation May and Baker's diethyl ether "Dried - Suitable

for Grignard Reactions" was used and -the technique was that of Davis and

Mann^. Magnesium turnings (27g., 1.1 mole) and diethyl ether (50 ml.''

were added to a flask equipped with stirring, reflux condensor and drying

tube. A crystal of iodine was added, o-bromochlorobenzene (210a., 1.1

mole) was dissolved In diethyl ether (420 ml.) and to a portion of this

solution (40 ml.) was added ethyl bromide (3g.). This portion was added

to the magnesium turnings drop-wise. The reaction started immediately

and all the first portion of o-bromochlorobensene and ethyl bromide was

added in 30 minutes. The remainder of the o-bromochlorobenzene was added

in 2 hours and the reaction was allowed to reflux for a further 2 hours.

At the end of this time only a small amount of magnesium was left. rne

Grignard was cooled and a solution of freshly distilled o-chl-■ obonze1

hyde (l54g., 1.1 mole) in diethyl ether (200 ml.) was added so the mix'ore

refluxed gently. Towards the end of the addition of the aldehyde a

sludge layer separated out having a clear yellow layer. After the

addition of the aldehyde-, the solution was refluxed for 30 minuses and

hydrolysed with 5N HC1 (300 ml.). The ether layer was washed th - - 83.

until neutral and the organic layer was dried over anhydrous Na„SO^.

After filtration and removal of the solvent under reduced pressure over a water bath the residue was distilled. The fraction B.P.O-5 140-150°

(mainly 144°) was collected and recrystallized from 60-80 petrol to afford bis(o-chlorophenyl) carbinol (LVIl) m.p. 86-87° (46/£).

Found - (13289) : 0,61.5; H,4.15# \

Calc, for - C13H100C12 : 0,61.7; H,4.<$, V* (C-0H)3200cm"1.

& (CDC13) 7.0-7.4(ArH,8H); 6.42(J(HC0H)4 c/s; 1H); 2.91 (j(HOCH)

4 c/s; 1H), the latter disappeared on exchange with D^O.

Preparation of ietrahydropyranyl F.ther of Bis- (o -chloronhenyl

Carbinol (LIX)

Bis-(P-chlorophenyl) carbinol (lOg., 0.04 mole), and excess

2,3 - dehydropyran (lOg., 0.12 mole), and p-toluene sulphonic acid (O.lg.) were added to dry benzene (100 ml.). The solution was poured into b% potassium carbonate solution (50ml.) and the organic layer was washed twice with more K0CO„ solution (10 ml.). The benzene solution was dried over anhydrous K^CO.^, filtered and the benzene removed under reduced vacuum on a water bath. Recrystallization from ethanol gave the ether (LIX) m.p.

63-4° (l2.2g., 9C|%). Repeated recrystallization from ethanol afforded the pure ether m.p. 67-3°,

Found - (13684) : 0,64.4; H,5.5%

Calc, for - C18H1802C12 : 0,64.1; H,5.4%; 6(01X1.) 1.3 -1.8

unresolved multiplex (6H, SH); 3.3 - 4.1, unresolved multiplex

(2H,2fH); 4.79, unresolved triplex (lH,^Ii); 6.58 (1H, venzylic

proton) 7.1 - 7.65 (8H, ArH)

Preparatlon of o,o'-r)ichlorobenzophenone ^(LXII) 84.

) '

(a) Oxidation with Chromic Acid, Acetic Acid and Water

The method used is a variation of the method described 67 by Haller et al . These authors used a mixture of chromic acid and acetic acid and made no mention of water. When the literature method was tried, small yields of the required ketone were detained and a large quantity of the acetate of bis-/o-chlorophenyl)carbinol m.p. 135-6°, lit.°7 137-8° was found.

Bis-(o-chlorophenyl) carbinol, m.p. 86-87°, (5.0g., 2x10 mole) was refluxed with a solution of sodium dichromatic (7q.), cone.

(5 ml.), glacial acetic acid (20 ml.) and water (50 ml.) for 2 hours. The cooled solution was poured into excess saturated sodium bicarbonate solution and extracted with diethyl ether (50 ml.).

The neutral ether extract was dried over anhydrous potassium carbonate solution, filtered and the solvent removed under reduced pressure over a water bath. Recrystailization from ethanol gave the ketone m.p.

50-51° lit.3' 45.4 - 46.6°, V(C=0) 1670 cm"1.

Attempted Preparation of 2,2_ - Dichlorobenzoohenone

Ethyleneketal

I* In Benzene

Ethylene glycol, freshly distilled, 0.7/1 water by varl r:? titration (0.52.g; 1.3 mole), the ketone (LXIl) (l.6g: 1.0 mo]p.' ->nd p-toluenesulphonic acid (0.036g) were refluxed in dry benzene (so ml.'' •V- ' with azeotropic removal of water. After'6 hours refluxing tl - cooled reaction mixture was poured intd 2N NaOH (10 ml.) The oceanic leye^ 85.

was washed with water until neutral. After drying over anhydrous

and filtering, the solvent was removed under reduced pressure.

The infrared spectrum of the product was identical with the ketone

(LXIl) and the relative intensity of the carbonyl band, relative to

the other bands in the spectrum, was identical.

2. In Toluol . •

Ethylene glycol, as above, (0.72g., 1.8 mole), the ketone

(LXIl) (l.6g., 1.0 mole) and p-toluenesulphonic (0.036g.) were refluxed

as above for 4 hours. A sample was removed and treated as described

for the experiment in benzene. The'infrared of this sample was identical

with the starting material.

3. In Ethylene Glycol•

Similar quantities of ketone (LXIl) and n-toluenesulnhonic

acid were heated on a water bath with a large excess of ethylene glycol

(20 ml.). After 4 hours a sample was taken, dissolved in diethyl ether

(10 ml.) and worked-up as described above. Again the infrared spectrum was identical with the starting ketone.

Heating the reaction mixture to 190° for 1 hour did not produc

any detectable reaction. Anhydrous CaS04(2g.) was added and the reaction mixture heated over a water-bath for 10 hours. The cooled product

dissolved in diethyl ether (30 ml.) and was worked up as above to give

the ketone (LXIl) 1.6g., lOCfo recovery), identified by comparison of its

* I infrared spectrum with the spectrum of the authentic 2,2 -dichlorobonzo Ten-

one. 86.

Attempted Reaction of the Tetrahydropyranyl Ether of the

Atlcohol (LIX) With Lit hi31m and Then Diehlorpohenyl-

rhosnhine.

The ether (LIX) (3.38g., 1 mole) was dissolved in dry T.H.F

(20 ml.). The reaction mixture, under nitrogen, was cooled to -45° when lithium ribbon (0.28g., 4 mole) was added. Stirring the mixture

for 6 hours nroduced a dark red solution with a large quantity of undissolved lithium. Dichlorophenylphosphine (2.1q., 1.16 mole) in

T.H.F. (20 ml.) was added drop-wise to the reaction mixture at -70° over

30 minutes. After stirring the mixture at -70° for one hour it was allowed to warm to room temperature (over a further hour). The reaction mixture was poured into excess 2N NaOH and extracted with 5^:50 mixture of benzene and diethyl ether (50 ml.). The organic layer was washed well with water, dried over anhydrous Na9S0^ and the solvent removed under reduced pressure over a water bath to yield a neutral fraction

(4.0g). Part of the neutral fraction (l.5g) was chromatographed on

Silica Gel G (llerck) (40g) to give the ether (LIX) (0.85o, (flyi recovery), bis(c-chloronhenyl) carbinol (O.lng, 6%) identified by infrared srec+ral comparison and three fractions which contained phosphorus hu-f- their structures could, not be inferred by n.m.r. analysis.

Preparetion of Di 111hiumphenylphosphfdp and ijttem i ed

■Reac+i on

With the Ether (LIX).

o *7 q Phosohorobenzene, m.p. 148-150 S.T. lit. “m.p.

149-152°(2.2g., 1 mole) was dissolved in dry T.H.F. (50 ml) and Imthium 87.

(0.285g, 2 mole), finely divided, was added at room temperature. The solution slowly turned red (after 1 hour) and was left .stirring for

24 hours. The unreacted lithium was filtered off under nitrogen

(O.OSg., 28/o recovery). The phosnhide was cooled to -70° and the etjer (LIX) (7g, 1 mble) in T.H.F. (50 ml.) was added over 20 minutes ! maintaining the temperature at -70°. After stirring at -70° for one hour the reaction was allowed to warm to room temperature for one hour.

Then the reaction was heated to reflux for a further hour. 5N HC1 (10 ml) was added to the refluxing mixture and the heating continued for an hour.

To the cooled reaction mixture was added diethyl ether (50 ml) and the organic layer washed with alkali, and water until neutral. ~'he ether layer, after drying and removal of the solvent gave the neutral r.-roduct

(8.0g.). Chromatography of part of the product (l.5g.) o" Silica Gel

G (Terek) (40 g) yielded the ether (LIX) (0.9g., 68.5% recovery) and the alcohol (LVIIl) (0.28g). The total recovery of the ether (LTX) was

96%

Attenuated Reaction of the Ether _.(LIXj_with n-Butyl

Lith_ium and_ Triethvlene Hi amine.

Excess n-butyl lithium (0.19g, 4 mole) and triethvlene diamine (l.4g, 4 mole) was added to dry benzene (40 ml. '. > ;iis solution was added the ether (LIX) (l.Og, 1 mole) in benzene (1C ml).

The solution was refluxed for 6 hours and a sample was -withdrawn and hydrolysed with excess, aqueous,, methanolic HC1 by refluxing for one hour. The cooled hydrosylate was poured into water, washed with alkali- and then water until neutral. The infrared spectrum and thin lever 88.

chromatograph of the dried reaction product indicated that the only product was bis-(o-chlorophenyl)carbinol.

Attempted Preparation of Grlcnarri of the Ether (LIX*) 77 The method used was after Baum et al whose method was closely followed. The ether (LIX) (6.7g., 1 mole), dried magnesium turnings (1.09), 2 mole), a few crystals.of iodine, isopropyl alcohol

(0.25 ml.) and dry T.H.F. (7 ml.) were heated to 55° for 30 minutes and then to 120° to remove the T.H.F. The reaction mixture was cooled to

55° and a further amount of isopropyl alcohol (0.25 ml.), T.H.F. (7 ml.) and a few crystals of iodine added. After the initial frothing subsided

T.H.F. (30 ml.) was added drop-wise. A positive test for Grignard was obtained after 4 hours - however a large amount of magnesium remained unreacted. Dichlorophenylphosphine (3.6g., 1 mole) in T.H.F. (25 ml.) was added slowly at 55 ■ and the reaction stirred at this temperature for

2 hours. The reaction was quenched with methanol (5 ml.) and poured into water and extracted with benzene. After washing t'-e org-ntc layer with saturated sodium bicarbonate solution and water untf l ne^tm-'l, it was drived over anhydrous Ha 30 and the benzene removed to yield the neutral fraction (6.5g, 97% recovery,), identified as the eth-wr (rIX) infrared spectral comparison.

Preparation of (o-Benzoylphenyl)phenylphosphinic -

and_ Attempted Cyclization with Zinc. Chloride.

(o-Benzoylphenyl)phenylphosphinic acid (i.Xl'' (l.5o.,

1 mole) was refluxed with excess thionyl chloride (20 r>i.' for ° hours. 89.

The excess thionyl chloride was removed under reduced oressure over a water bath, dry benzene (10 ml.) was added and again the solvent was removed under reduced pressure. The treatment with benzene was repeated until the small of thionyl chloride had disappeared. To the acid chloride was added anhydrous zinc chloride (99% by zinc assay)

(Q.7g., 1.1 mole) and the melt heated at 130° for 24 hours. The purple ! reaction product was extracted with CHCl^ (20 ml.) and excess 2N NaOH solution. The aqueous alkaline extract was made acid with cone. HC1. to yield the starting acid (LXl), 1.45g. (97%) m.p. 210 - 212° v/hich did not depress the m.n. of (o-benzoyl;:heny.l)phenylrhos"hinic acid and' which had an infrared spectrum identical with this acid.

Attempted Cyclization of (_o-8_enzoy_lp_henyj )phegi^lnhos-

phinic Acid (LXl) T’ith Tri fInoroacetic Anhydride

The acid (LXl) (2g) was heated with excess trifluor^- acetic anhydride (1 ml.) at 100° for 6 hours. The reaction rroduce was poured into excess 2N NaOH in which it was connletely soluble. The alkaline solution was extracted with CHCl^ (10 ml.). Acidification of the aqueous alkaline extract afforded the starting acid (LXl), 1.9'ag

(97.5% recoverv), identified by comparison of its infrared ■> or •-rum • the spectrum of the authentic acid. 90.

REFERENCES:

1. Horner, L., Hoffmann, H., Klahre, G.

Toscaro, V.G., and Ertel, H., Chem. Ber.• 94, 1987 (1961)

2. Conant, J.B., and Wallingford, V.H. J. Am. chem. Soc., 46 192

(1994) ! 3. Olah, G.A., "Friedel-Crsfts and Related Reactions". Vol. 1.

(interscience : New York 1964).

4. Frazier, S.E., Neilsen, R.P., and Sisler, H.H., Inoro.Chem.,

3. 292 (1964).

5. Hiller, B., "Tooics in Phosr.horus Chemistry". Vol. ]] (John

Wiley: New York 1965)

6. Doak, G.O., and Freedman, L.D., J. Am. chem. Soc., 74, 2884

(1956).

7. Gerrard, W., Mooney, E.F., and Willis, H.A., J. chem. Soc., 4255

(1961).

8. Gibson and Johnson J. chem. Soc., 92 (1928).

9. Van Wazer, J.R., "Phosphorus and Its Comoounds". Vol. 1

(interscience: New York 1958).

10. - Pass, F., Steininger, E., Zorn, H • ’ | h. Chem. CO _ 930 (1962)

11. Brown, M.P., and Silver, H.B. Cherny. Ind., 94 (1961).

12. Horner, L., Hoffmann, H., Wippel, H.G., and Hassel. G.,

Chem. Ber., 91. 52 (1958) ~ — ------zrr-z 13. Horner, L., Hoffmann, Hi, and Wiopel H.G.

Chem. Ber, 91 (1958). 91.

14. Griffin, G.E., and Brown, J.T. J. Org. Chern., 26, 853 (1961).

15. Doak, G.O., Freedman, L.D., and Pettit, E.L.

J. O-go. Chern.. 25, 140 (i960).

16. Horner, L., Hoffmann, H., Ertel, H., and Klahre, G.,

Tetrahedron betters^, JL, 9 (1961).

17. Avovan, R.L., Struchkov, Y.T.,-and Dashevskii, V.G.

J. Structural Chern., 7, 283 (1967)

18. Kinnear, A.M., and Perren, E.A. J. chrm. So_o., 3437 (1952)

19. Blott, A.H., "Organic Synthesis - Collective Volume II" p. 575

John Wiley and Sons: New York, 1943)

20. Arbusov, A.E.. and Razumov, A.I., Bull. Acad. sci. URSS,

Classe sci. chin., 167 (1945)

(Chern. Abs.tr.. 40, 3411 (1946)

21. Courtot, C., Ann. Chen., (9), 5, 80 (1916)

22. Thomas, C.A., "Anhydrous Aluminium Chloride in Organic Chemistry"

P. 168, (Reinhold: Mew York, 1941).

23. Conant, J.B. MacDonald, A.D., and Kinney, A.McB.

J. Am. chern. Sop.. /3. 1°2B (1921)

24. Kabochrif and She'pelve, ' .Iz vest. Akad. Mauk. V C .

Otdil. Khin. Mauk. ?'• (IP, )

(CJhem. Abstr. 44. *70 ' lc50)

25. Marie, M.C., Comet. Rendu, 136, 508 (]c03^

26. Michaelis, A., . Chen. Ber., 17. 1273 (1884^ 92.

27. Drake, L.R., and Marvel, C.S. J. Orn. Chem»> 2, 387 (1937).

28. Conant, J.B. J. Am. Chen. Son 43, 1705.

(1921).

29. Menshutkin, B., Zh. russk. fig. - Khim. Obshep., 42, 1298 (1910).

Chem. Abstr., 5, 1434 (1911).

30. Ramirez F., Yamanaka, H., Basidow, O.H. i

J. Am. chem. S_oc., 73, 173 (1961)

31. Cross, A.D., "Introduction to Practical Infrared Spectroscopy".

(Butterworths: London I960).

32. Mic.haelis, A., Ann., 293, 193., (1896).

33. Smith, R., and Pinhey, J.T., Private Communication.

34. Becker, H.D.., J. Org. Chem., 29, 3070 (1964)

35. Weil, von Th., Prijs, B., and F.rlenmeyer,

Helv. Chem. Acta, 36, 1314 (1953^

36. Kosolapoff, G. M. "Organophosphorus Compounds". P. 362

(John Wiley: Mev; York 1950).

37. Grayson, M. J. Am. chem. Sop-, pm, 79 (1963)

38. Frank, A.W., Chem. Rev., 61. 389 (1961)

39. Michaelis, A., and Kachne, R., Chem. Be^., 33 3C43 (1898)

40. Arbusov, A.E., J. Russ. Phvs. Chem., 38, 637 (1906).

41. Harvey, R.G., and De Sombre, E.R. "Topics in Phosnhnrus 0 c;rn strv"

Vol. l.;'P. 57 (Interscience:

Mew Yoru. 3965'. 93.

42. Aksnes, G., and Aksnes, D., Acta* Chem. .scand., 18 38 (1964).

43. Henderson, W.A., and Buckler, S.A., J. Am. chem. Soc., 82, 5794

(I960).

44. Horner, L., Pure appl. Chem., 9 (1964).

45. Borowitz, I.J., and Vinkhaus, R., J. Am. chem. Soc., 85, 2183 (1963)

! 46. Hudson, R.F., "Organic Reaction Mechanisms", Chem. Soc,

Spec. Publ. No. 19, London (1965).

47. Hudson, R.F., and Searle, R.J.G., Chi mia (Switz.) 20, 117 (1966).

48. Pearson, R.G., J. Am. chem. Soc., 85, 3533 (1963)

49. Kosolapoff, G.M., J. Am. chem. Soc., 69, 1002 (1947)

50. Burn, A.J., and Cadogan, J.I.G. J. chem. Soc., 57.Q3 (1963).

51. Mackenzie, J.E., J. chem. Soc., 121, 1695 (1922).

52. Griffen, C.E.. and Gordon, M., J. Organometalllc Chem., 3

414 (1965).

53. Seyferth, D., Grim., S.O. and Read, T.O.,

J. Am. chem. Soc., 82, 953 (i960).

54. Oda, R., Ito, Y. and Okana, M. Tetrahedron Lett., 37 fl964>'.

55. Derskowitz, S., and Proskauer, S., J. Org. Che". , 26. 3595 (196.1 ' •

56. Partos, R.D., and Speziale, A.J. J. Am. chem. Son.. B’7. 50*-

57. Schmidlin, J. , and von Escher, R., Ber. dt. Chem. Gc-. , 4 . -'-'/i i

58. Finklestein, Pi., Ber. dt. chem. Ges. , 43, 1533

59. Norris, J.F., Thomas, R., and Brown B.M. Ber. dt. chc^. f>s., 43. 94.

60. Mark, V. , J. Am. chem. Soc., 85, 1884 (1963).

61. Jones, E.R.H., and Mann, F.G., J. chem. Soc., 294 (1958).

62. Gallagher, M.G., Kirby, E.C., and Mann, F.G.

J. chem. Soc., 4846 (1963).

63. Markl, G., Annew. Chem. inst. Edn., 4, 1023 (1965). 5 64. Levy, J.B., Doak, G.O., and Freedman, L.D.

J. 2J!£* Chem., 30, 660 (1965).

65. Levy, J.B., Doak, G.O., and Freedman, L.D.

J. orq. Chem., 29, 2382 (1964).

66. Davis, M.., and Mann, F.G. Chem. and Ind.? 1539 (1963^.

67. Haller, H.L.,' Bartlett, P.O., Drake, M.L. Newman, V.S.

Cristol, S.J., Eaker, C.M., Hayes, R.A., Kilmer, G.V7.,

Magerlein, B., Mueller, G.P., Schneider, A., and Wheatley, W.

2* Soc., 67, 1601 (1945).

68. Albright, J.D., and Goldman, L.

J. Am. chem. Soc., 87. 421^ (1965).

69. Hellmann, H., and Bader, J. Tetrahedron I^tt^rs, 724 (1961'.

70. Eyles, C.T., and Trippett, S. J. chem. Soc., (c\ 67 C1966 .

71. Hands, A.R., and Mercer, A.J.H. J. chem. Soc., (n), 109° '■ 196'' •.

72. Henderson, W.A., Epstein, M., and Seichter, F.S.,

J. Am. chem. Soc., 85, 2462 (1963)

73. Dannley, R.L., and Kabre, K.R. J. Am. chem. Soc., 37. 4 805 (1969'-) 95.

74. Fritsche, H., Hasserodt, U., and Korte, F.,

Chen. Rer. 98, 1681 (1965).

75. Peterson, D.J., J. org. Chen., 32, 1717 (1967).

76. Hatt, H.H., J. cherrt., Soc., 2412 (1929).

'I 77. Baum, G., Lloyd H., and Tamborski, C.,

J. ora. Chem., 29, 3410 (1964).

78. Burgada, R., Ann. Chin. (Paris) 8, (5-6), 347 (3963).

79. Peterson, D.J. and Collins, J.H., J. org. Chen., 31, 23"73 (1966).

80. Corey, E.J., and Jautelat, M. J. Am. chem. Soc., 89, 3912 (1967).

81. Davis, M., and Iv'ann, F.G., J. Chem. Soc., 3786 (1964). 96.

ACKNOWLEDGEMENTS

The author is greatly indebted to his Supervisor,

Dr. M.G. Gallagher, for his continuing interest in

this project and numerous helpful discussions - a

great many taking place in the Supervisor's leisure

time.

Dr. Challen.is thanked for many of the micro-analytical

figures quoted in this thesis.

The generous supply of samples of tris(dimethylamino)

phosphine and phosphorobenzene from r'r. I.D. Jenkins

is acknowledged.

Lastly, the author would like to thank his wife, "’ab'n^tte, whose patience and even temper have made this project possible. Kepnnted from TETRAHEDRON LETTERS

The International Organ for the Rapid Publication of Preliminary Communications in Organic Chemntrr

PERGAMON PRESS OXFORD • LONDON • NEW YORK • PARIS Tetrahedron Letters No.l, pp. 121-127, 1966, Pergamon Press Ltd, Printed in Great Britain.

THE ALUMINIUM CHLORIDE-CATALYSED ADDITION OF P(III) HALIDES TO

CARBONYL COMPOUNDS ly

K. L. Freeman and M. J. Gallagher

School of Chemi atry, University of New South Vales

P.0. Box 1, Kensington, Sydney, Australia

(Received 18 November 1965)

Michaelis reported1 that benxophsnone and phosphorus trichloride

do not react in the presence of aluminium chloride and Conant and his

co-workers in their extended study^ of the uncatalysed reaction of

phosphorus(IIl) halides with carbonyl compounds required forcing

conditions^ to bring about reaction. In the course of unrelated work

we had occasion to Investigate the reaction of phenylphosphonous

dichloride and aluminium chloride with benzophenone. The reaction

was carried out in exoess halide as solvent and using two moles of

partially hydrated aluminium chloride. The reaction mixture was

heated at 100°C until hydrogen chloride evolution ceased (3 hrs.),

the product poured into aqueous sodium hydroxide and extracted with benzene to give a 65^ yield of diphenylchloromethylphenylpho sphlnic chloride (i), m.p. 103-104°C.

121 122 No, 1

2ilCl3 Cl Cl PhPCl2 + Pl^C » 0-----

(I)

If “anhydrous* aluminium chloride la used the yield la much

lover* The structure (I) was confirmed by its independent synthesis

as follows :

Cl 0

PbP(CHe)2 ♦ PhgCCl, -> Ph2C - PPh +

(II) ^

h2o/h+

Cl o

(!)«■ PhgC—PPh

(III)

The acid chloride (I) is remarkably inert aa indicated by its mode of Isolation and the fact that it was unaffected by boiling in methanol. However, sodium nethoad.de in boiling methanol converted it to the eater (II; m*p. 100-101°; 73%), prolonged (14 hr.) aqueous alkaline hydrolysis afforded the hydroxy acid (IV; imp. 191-192°), and boiling ethanolic hydrochloric acid afforded the chloro-aeid (III). OH 0

(IV) No. 1 123

This sluggish reactivity is interpreted as being due to steric hindrance to Sjj2 attack, Horner has observed similar tut more pronounced effects for a-chlorobenzyldiphenylpho Ephine oxide and attributed it to the

"positive" character of the halogen atom.

Varying the proportions of the reagents resulted in lowered yields, Extension of the reaction time to 6 bre, also lowered the yield and led to the isolation of substantial amounts (4C%) of a dibasic acid (m,p« 275-278°;) isolated by acidification of the alkaline hydrolysate. This apparently has one of the structures (V) and is currently under investigation

Substitution of aralkyl or dialkyl ketones for benzophenone led to complex mixtures of products. Aluminium chloride readily induces aldol type condensation^ and the unsaturated carbonyl compounds resulting would be expected to add P(III) halides readily. An example of this is the reaction of acetone with phosphorus trichloride and aluminium chloride.'' Nevertheless, cyclohexanone afforded the acid (VI; m.p, 149-150°; 42%) f and o-chlorobenzaldehyde the hydroxy acid (VII; m.p. 283-285°; 66%). 124 No. 1

In view of those resalts vs rsinrestigated the reaction of pho spheres trichloride with bensophenone bat ve could induce no reaction under a vide variety of conditions. On the other hand diphenylpho ^shinous chloride reacted readily with bensophenone under the sane conditions as were used for phenyl pho sphonous dichloride, but the product was benshydryl- diphenylpho ephine oxide (VIII; n.p. 304-305°, lit.^ 303° ! 35%)• This substance has prerioualy been prepared in unstated yield by the action of bensbydiylohlorlde on diphenylphosphinous chloride.^ Its structure was confirmed as follows t

Li * PhjPCHPh,

(vin)

This unexpected result presumably arises by attack of diphenylphosphinous chloride on initially forned diphenylchloronetfayldiphenylpho ephine oad.de to give the ion pair (IX) which on hydrolysis would give (VIII) and diphenyl phosphide acid.

Cl 0 0 I A -II Ph2C_PPh2 ♦ PhjjPCl

PhjPofcH) ♦ (vm) No. 1 125

Attack of tervalent phosphorus compounds on "positive1* halogen has 7 received much attention recently and the reduction of a related compound g In similar fashion has been observed. In part confirmation we find

that the acid chloride (I) is reduced to benzhydrylphenylpho sphinic acid

(Zj m.p. 243-245°) by diphenylphasphinouschloride in the presence of

aluminium chloride

Cl 0 1. A1C1 0 I l 3 II Ph2C —PPh + PhgPCl ------> Ph^CHPPh Cl 2. H20 OH

(I) (X)

In the absence of aluminium chloride no reaction occurs and the catalyst presumably functions by co-ordinating with the P * 0 group of (i) and

thereby facilitating "positive" halogen removal. Phenylphosphonous

dichloride is presumably insufficiently nucleophilic to effect this reduction since (X) could not be deteoted in the reaction mixtures from

the preparation of (l)«

The following alternative mechanisms are suggested to explain

the ready reaction between benzophenone and phenylphosphonous dichloride : Cl Cf l+ I Ph PCI2 + PkjC = 0 ------> Ph P----CPh2 Cl

(XI) 126 No, 1

2 The initial addition step is supported by Conant's detailed study of the uncatalysed reaction. No evidence is available to clearly distinguish between the two subsequent paths but it is hoped to carry out the experiment with Ph^C =0 18 . *

References

1. A. Michaelis, Annalen. 223, 193 (1896); ibid., 23A, 1, (1897).

2. J. B. Conant and V. H. Wallingford, £. Amer. Chem. Soc. ££, 192, 1924)

and previous papers in this series. 3. J. B. Conant, A. D. MacDonald and A. McB. Kinney, £. Amer. Chem. SaSi. 1928, (1921).

4. C. A. Thomas, Anhydrous Aluminium Chloride 1n Organic Chernlstrv.

Reinhold, New York, 1941» p. 634. 5. L. R. Drake and C. S. Marvel, J. Ore. Chem. 2, 387, (1937). No. 1 127

6. L. Rorner, H, Hoffman, G. KLahre, V. G. Toscano and H. Ertal,

Chan. Sgr. 1987, (1961)

7• For a review see B. Miller, Topics In Phosphorus Chemistry. Wiley,

New York, 1965, p. 133.

3. L. Homer, H. Hoffmann, H. Ertel and G. Klahre, Tetrahedron Letters.

2, (1961).

* All compounds gave satisfactory analytical, Infra-red and

p.m.r. figures. All reactions involving P(III) compounds

were carried out tinder nitrogen. Reprinted from the AUSTRALIAN JOURNAL OF CHEMISTRY

THE ALUMINIUM CHLORIDE CATALYSED REACTION OF PHOSPHORUS(m) HALIDES AND CARBONYL COMPOUNDS

I. REACTION WITH BENZOPHENONE*

By K. L. Freeman! and M. J. Gallagher!

[Manuscript received May 25, 1966]

Summary In the presence of partially hydrated aluminium chloride, phenylphosphonous dichloride and diphenylphosphinous chloride add smoothly to benzophenone to give chlorodiphenylmethylphenylphosphinic chloride and benzhydryldiphenylphosphine oxide respectively. Little reaction occurs with anhydrous aluminium chloride, and phosphorus trichloride does not react under any conditions. The factors affecting these reactions are described and possible mechanisms discussed. It is suggested that the intermediacy of species of the type [PhPCl3]+[PhPCl,AlCl3p best explains the observed results.

Introduction This study arose from an investigation concerned with the synthesis of the ring system (I; X = C=0). Methods for the synthesis of these systems have been well worked out1 but we were attracted by the possibility of a one-step approach such as has been successfully employed1 for the nitrogen analogue (I; X = NH; R = Cl) from phosphorus trichloride and diphenylamine. Such a reaction seemed unlikely but was felt to be (I) worthwhile investigating since even a low yield would be attractive in terms of time and materials. Substantial side reactions were not anticipated since it has been reported2 that benzophenone and phosphorus trichloride do not react in the presence of aluminium chloride.

Results Benzophenone and aluminium chloride (partly hydrated; 2 mole) were heated together in an excess of phenylphosphonous dichloride (5 mole) as solvent for 3 hr at 100°. Two moles of catalyst were used since benzophenone is known3 to form a

* A preliminary note has appeared: Freeman, K. L., and Gallagher, M. J., Tetrahedron Lett., 1966, 121. f Department of Organic Chemistry, University of New South Wales, Sydney. 1 Markl, G., Angew. Ghem. int. Edn, 1965, 4, 1023. 2 Thomas, C. A., “Anhydrous Aluminium Chloride in Organic Chemistry.” p. 168. (Reinhold: New York 1941.) 3 Menshutkin, B., Zh. russk. fiz.-khim. Obshch., 1910, 42, 1298 (Chem. Abstr., 1911, 5, 1434).

Aust. J. Chem., 1966, 19, 2025-33 2026 K. L. FREEMAN AND M. J. GALLAGHER stable adduct with aluminium chloride. This fortuitous combination proved to be optimum. The reaction mixture was worked up by pouring into aqueous alkali and extracting with benzene. The crystalline product, m.p. 102-103°, gave analyses corresponding to a 1 : 1 adduct of phenylphosphonous dichloride and benzophenone. Its infrared spectrum showed a strong P=0 band and its proton magnetic resonance spectrum only aromatic protons. It reacts sluggishly with silver nitrate in ethanol, and more rapidly with sodium methoxide in boiling methanol with loss of one halogen atom and formation of a methyl ester. Prolonged aqueous alkaline hydrolysis removes both halogen atoms and affords a monobasic hydroxy acid. These reactions indicate the structure (II) and this was confirmed by the synthesis shown.

Ph2CCl2+PhP(OMe)2 -> Ph2C(Cl)P(0)0Me+Ph2CHP(0)0Me (IV) (V) MeONa/MeOH t HO PC15 Ph2C(Cl)P(0)Cl Ph2CP(0)0H + Ph2CHP(0)0H OH-/H2O Ph Ph Ph (II) (HI) (VI) moist yf A1C13 PhPCl2 + Ph2CO

Methyl benzhydrylphenylphosphinate (V) was presumably the major product of the Arbusov reaction between dichlorodiphenylmethane and dimethyl phenylphosphonite and the mixture could only be successfully separated after hydrolysis, which unfortunately converted the a-chloro ester (IV) into the hydroxy acid (III). Attempts to chlorinate the acid (VI) in the apparently favourable benzylic position were unsuccessful. Hydrolysis of the acid chloride (II) with ethanolic hydrochloric acid also affords the hydroxy acid (III) which gives a positive Beilstein test (due probably to the presence of traces of (II)) for halogen which initially led us to believe* it was the a-chloro acid. If the heating of the mixture of benzophenone and phenylphosphonous dichloride is prolonged then the yield of (II) decreases and a further product (VII) accumulates Avhich is isolated, by acidification of the aqueous layer after work-up, as an intractable solid, m.p. 273°, difficultly soluble in most organic solvents. Titration indicated that it was dibasic and this was confirmed by the p.m.r. spectrum of its dimethyl ester. This spectrum also showed a benzylic proton split (18 c/s) by phosphorus. Alkaline fusion4 gave (p-benzylphenyl)phenylphosphinic acid (VIII) which was oxidized to the keto acid (IX) whose structure was established by the synthesis shown below

* See the preliminary note. 4 Freeman, K. L., and Gallagher, M. J., Aust. J. Chem., 1966, 19, 2159. PHOSPHORUS(iii) HALIDES AND CARBONYL COMPOUNDS 2027

(-C6H4- refers to para-phenylene). These data establish the structure of the dibasic acid as p-hydroxy(phenyl)phosphinyldiphenylmethylphenylphosphinic acid (VII). OOO 0 NaOH 11 KM11O4 PhP-C6H4-CH-PPh------> PhP-C6H4-CH2Ph------> PhP-C6H4-COPh | | | 300° | Me2CO OH Ph OH OH OH (VII) (VIII) (IX)

(1) PhPCl2 (2) H20

(1) hno2 PhCO-C6H4-NH2------> PhCO-C6H4-N2BF4 (2) NaBF4 Diphenylphosphinous chloride, aluminium chloride, and benzophenone react under the same conditions as the dihalide to give a neutral, halogen-free product,

Table 1

ALUMINIUM CHLORIDE CATALYSED REACTIONS OF PHOSPHORUS(lIl) HALIDES “Moist” (exposed 24 hr) aluminium chloride, excess halide, one mole of substrate, and heat at 100° for 3 hr, were used except where indicated. S indicates substrate recovered in high yield

Yield (%) Catalyst Halide Substrate (mole) (II) (VII) (X)

PhPCl2 Ph2COa — S PhPCl2 Ph2CO 1 37 — — PhPCl2 Ph2CO 2 66 — — PhPCla------—Bh^GO----- 3 $ '-ri > PhPCl2 Ph2CO 4 S PhPCl2 Ph2CO 2b 7 — — PhPCl2 Ph2COc 2 19 41 — PhPCl2d Ph2CO 2 S PhPCl2 Ph2CO 2e.f S PhP(0)Cl2 Ph2CO 2 s PC13& Ph2CO 2 s Ph2PCl Ph2CO 2 — — 35 PhPCl2 Ph2CClPPh(0)Cl 2 87 12-5 — PhPCl2 Ph2CClPPh(0)Cl 2b 28 58 — Ph2PCl Ph2CClPPh(0)Cl 2 11% of (VI) PhPCl2 Ph2CO 2h 72 — —

a Reaction time 1 hr. b Anhydrous AICI3 was used. c Reaction time 6 hr. d One mole only was used. e Boron trifluoride etherate was used in place of AICI3. f Addition of water did not affect the result. s Same result in chlorobenzene as solvent. h “Moist” AICI3 (exposed 72 hr) was used. m.p. 304-305°. It shows a P=0 band in the infrared, and the p.m.r. spectrum shows 20 aromatic and one benzylic protons, the latter split (12 c/s) by phosphorus. The 2028 K. L. FREEMAN AND M. J. GALLAGHER only reasonable structure is benzhydryldiphenylphosphine oxide (X), and analysis and an unambiguous synthesis confirmed this.

moist Ph2PCl+Ph2CO------> Ph2CHP(0)Ph2 A1C13 (X) A

h2o2 Li Ph2CHCl Ph2PCl---- > Ph2P-Li+------> Ph2PCHPh2 (XI)

The phosphine (XI) was not isolated. The factors influencing these reactions have been extensively investigated and the results are summarized in Table 1.

Discussion Conant and his collaborators studied the uncatalysed addition of phosphorus halides to carbonyl compounds nearly 50 years ago, and concluded5 that the results were best rationalized as in the following example:

0 OH PhCHO + PC13 = PhCH—PC13 PhCH—P(0)(0H),,

slow H20 A A PhCH—PC1:. - - ► PhCH—P(0)C1 “f* In our case the intermediate would be PhPCl2-C(Ph2)-0~ (XII). This simple mechanism explains the necessity for the presence of water (in excess of that normally required6 in Friedel-Crafts reactions) and the catalyst would facilitate attack by the halide if it coordinated with the carbonyl oxygen. It does not explain why two moles of aluminium chloride are required, nor why four moles should stop the reaction completely. An alternative mechanism:

(xn) — PhpA>Cph2cr —- (11) ci suffers from the same drawbacks, and further, when the reaction is carried out using only one mole of phosphorus halide, no reaction is detectable. A decision between these two possibilities could be made by the use of Ph2C180 but it seems unlikely that either mechanism provides an adequate explanation. Unfortunately, the system is complex and does not lend itself to a more detailed study. The catalyst used by us

5 Conant, J. B., and Wallingford, V. H , J. Am. chem. Soc., 1924, 46, 192. 6 Olah, G. A., “Friedel-Crafts and Related Reactions.” Vol. 1. (Interscience: New York 1964.) PHOSPHORUS(iii) HALIDES AND CARBONYL COMPOUNDS 2029

had been stored in a loosely stoppered vessel until an increase in weight of 3-4% had occurred. Further exposure increased the yield of the acid chloride (II) only slightly. Since aluminium chloride fumes on exposure to air, it appears certain that hydrolysis as well as absorption occurs and hence the actual amount of water taken up cannot be estimated accurately. Another very puzzling feature is the complete failure of phosphorus trichloride to react, especially in view of its reactivity in the uncatalysed reaction. The nature of the species formed from aluminium chloride and phosphorus (hi) halides is not known. Van Wazer has noted7 that the salt often enhances the reactivity of tervalent phosphorus compounds. This often occurs in an unexpected fashion, e.g. the reaction of alkyl halides with the phosphine-aluminium chloride adduct to give primary phosphines.8 We suggest that the following equilibrium could explain the apparently enhanced nucleophilicity of the phosphorus halides in our reactions:

PhPCl2+PhPCl2,A1C13 ^ [PhPCl3]+[PhPCl,AlCl3]- (Xlla) followed by attack of the anion of this ion pair at carbonyl carbon. Formation of the ion pair could be looked on as an assisted auto-ionization of the phosphorus halide.7-9 An excess of catalyst might be expected to coordinate with the anion and prevent further reaction. There is some indirect support for reactions of this type. The same intermediate could explain the reaction:10

AICI3 2PhPCl2------> Ph2PCl+PCl3 as an alternative mode of recombination of the ion pair (Xlla). It may be significant that aluminium triiodide only forms a stable complex with tivo moles of phosphorus triiodide.11 Ion-pair formation has been widely proposed to explain the reactions of trivalent phosphorus compounds with the so-called positive halogen compounds12 and a good analogy, though uncatalysed, is the reduction of phosphorus(m) halides with tributyl phosphine.13 Failure of phosphorus trichloride to react may reflect its low nucleophilicity in the equilibrium or unreactivity of the species [PC12,A1C13]~ if formed. We hope to carry out further experiments to test this hypothesis. The chemistry of the acid chloride (II) is of interest since both chlorine atoms are in neopentyl-type environments and should be hindered to S-$2 attack. This is reflected in its sluggish reactivity towards nucleophilic reagents and the sole formation of the a-hydroxy acid (III) on aqueous hydrolysis rather than the a-chloro compound, indicative of an $^1 rather than an 8-^2 mechanism for the loss of the benzylic

7 Van Wazer, J. R., “Phosphorus and its Compounds.” Vol. 1. (Interscience: New York 1958.) 8 Pass, F., Steininger, E., and Zorn, H., Mh. Chem., 1962, 93, 230. 9 Failkov, Y. A., and Bur’yanov, Y. B., Dokl. Akad. Nauk SSSR, 1953, 92, 585. 10 Brown, M. P., and Silver, H. B., Chemy Ind., 1961, 24. 11 Baudler, M., and Wetter, G., Z. anorg. allg. Chem., 1964, 329, 3. 12 Miller, B., “Topics in Phosphorus Chemistry.” Vol. II. (John Wiley: New York 1965.) 13 Frazier, S. E., Nielsen, R. P., and Sisler, H. H., Inorg. Chem., 1964, 3, 292. 2030 K. L. FREEMAN AND M. J. GALLAGHER chlorine. Reduction of the acid chloride (II) with zinc amalgam in tetrahydrofuran, followed by hydrolysis, gave benzhydrylphenylphosphinic acid (VI), possibly via the phosphorane (XIII): Cl Cl Zn HaO Ph2C—P(0)Ph---- > Ph2C=P(0)Ph----- > (VI) (XIII) The reaction is of interest as a novel route to this type of compound but is unlikely to provide a useful alternative to the available methods of preparation. The formation of benzhydryldiphenylphosphine oxide (X) from diphenylphosphinous chloride and benzophenone is noteworthy. Presumably, the chloro compound (XIV) is formed first, as in the reaction of phenylphosphonous dichloride, and is subsequently dechlorinated by nucleophilic attack of the phosphinous halide at halogen to give the ion pair (XV7) which with water would give the product (X). Cl I Ph2CP(0)Ph2+Ph2PCl -> [Ph2PCl2]+[Ph2C~-P(0)Ph2] (XIV) (XV) Our attempts to prepare the chloro compound (XIV) to check this point were unsuccessful. Since they are of interest in the general problem of nucleophilic attack at halogen we have extended them and will report them separately. A related reaction occurs with the acid chloride (II), and diphenylphosphinous chloride will also dechlorinate this compound, though only in the presence of aluminium chloride, which probably coordinates with the phosphoryl oxygen and facilitates halogen removal. No reduction occurs in the absence of a catalyst. The formation of the diacid (VII) can be understood on this basis. Phenylphos phonous dichloride and anhydrous aluminium chloride convert the acid chloride (II) to the ion pair (XVI) in which the phenyl groups would be strongly activated to electrophilic attack. More probably the reaction can be looked upon as an ion-pair

Cl Cl I I PhPCL + Ph2C—PPh Ph)CPPh

0—A1C13 0—A1C1

Ph Cl (VII) C P Ph

recombination of (XVI) which, if very rapid, could explain the absence of benzhydryl phenylphosphinic acid from the products of aqueous work-up. This reaction proceeds to a negligible extent in the presence of hydrated aluminium chloride as shown by the PHOSPHORUS(iii) HALIDES AND CARBONYL COMPOUNDS 2031 absence of the diacid from the preparations of the acid chloride (II) (see Table 1). Extended reactions, however, lead to substantial yields of (VII) presumably corre sponding to eventual elimination of the partially hydrated catalyst.

Experimental

All reactions involving tervalent phosphorus compounds were carried out under dry nitrogen. All compounds are colourless unless otherwise stated. Yields quoted are for pure compounds.

General Procedure Benzophenone (1 mole) and “moist” aluminium chloride (2 mole) (i.e. commercial anhydrous material exposed to the air in a loosely stoppered vessel till an increase in weight of 3-4% occurred) were added to phenylphosphonous dichloride (< 5 mole) and the mixture heated in an oil-bath to 100° and kept there for 3 hr. The product was poured into a large excess of cold, 2n sodium hydroxide solution, stirred and extracted with benzene. The benzene layer was washed with water till the washings were neutral, dried, and the solvent removed; the aqueous layer was acidified and any precipitate collected, dissolved in hot, saturated aqueous NaHCC>3, and reprecipitated with acid. The results are summarized in Table 1.

Chlorodiphenylmethylphenylphosphinic Chloride (II) Benzophenone (3-6 g; m.p. 42—44°) gave, in the neutral fraction, the acid chloride (II), 4 • 7 g (66%), m.p. 103—104° from ethanol (Found: C, 63-5; H, 4-4; Cl, 20-0; P, 8 • 7. C19H15CI2OP requires C, 63• 2; H, 4-2; Cl, 19-6; P, 8-6%); v(P=0) 1223 cm-1. Benzophenone was the only other detectable constituent of the neutral fraction. Acidification of the aqueous layer afforded diacid (VII), 70 mg (see below). In the reaction utilizing 3 moles of moist catalyst there was obtained a low yield (c. 10%) of an unidentified neutral material, m.p. 202—203°, spectrally very similar to but not identical with the acid chloride. The bulk of the neutral material was benzo phenone (80% recovery).

Methyl Chlorodiphenylmethylphenylphosphinate (IV) The acid chloride (II, 0 • 6 g) was unaffected by prolonged boiling in methanol. After boiling for 1 hr with an excess of sodium methoxide in methanol, 0 • 5 mole of chloride ion was liberated (Found as AgCl: 9-9; Calc.: 9-8%) and 0-54 mole of methoxide was consumed (by titration). The reaction mixture was poured into water and the precipitated ester recrystallized from light petroleum (60-80°), 0• 5 g (76%),m.p. 100-101° (Found: C, 67-3; H, 5-3. C2oHi8C102P requires C, 67-3; H, 5-1%); v(P=0) 1230 cm-1; 8(CDC13) 7-15-7-73 (20H, ArH); 3-72 (J(POCH) 11 c/s; 3H, OCH3).

Hydroxydiphenylmethylphenylphosphinic Acid (III) (i) By hydrolysis of the acid chloride (II).—A sample of the acid chloride (0-29g) was refluxed with a large excess of aqueous NaOH for 14 hr, cooled, and the precipitated solid recrystallized from aqueous ethanol to afford the hydroxy acid, 0-26 g (98%), m.p. 191-192° (Found: C, 70-45; H, 5-3; equiv. wt., 325. C19H17O3P requires C, 70-4; H, 5-3%; equiv. wt., 324-3); v(OH) 3250 cm-1. The S-benzylisothiouronium salt had m.p. 170-171° from aqueous ethanol (Found: C, 65-95; H, 5-55; N, 5-5. C27H27N2O3PS requires C, 66-1; H, 5-55; N, 5-7%). Hydrolysis of the acid chloride (II) with aqueous ethanolic hydrochloric acid for 18 hr at reflux afforded the same acid. The acid chloride (51%) was recovered from the mother liquors. (ii) Synthesis.—Dimethyl phenylphosphonite (5g; 3 mole) was added dropwise with stirring to hot (100°) dichlorodiphenylmethane (12 g; 5 mole) during 0-5 hr and the deep red mixture maintained at 100° for a further 0-5 hr. The cooled product was boiled under reflux for 8 hr with ethanolic hydrochloric acid and the product worked up for acids. Repeated recrys tallization afforded a crude sample (0 • 48 g) of the a-hydroxy acid (III), m.p. 184—186°, undepressed by admixture with an authentic sample. Work-up of the first NaHC03 wash and the mother 2032 K. L. FREEMAN AND M. J. GALLAGHER liquors from the hydroxy acid furnished benzhydrylphenylphosphinic acid, 0-75g (8%), m.p. 244-246°, undepressed by admixture with a synthetic4 sample.

Conversion of the oc-Hydroxy Acid (III) to the a-Chloro Acid Chloride (II) Synthetic hydroxy acid (0-15 g) and PCI5 (0-22 g; 2-2 mole) were heated together (155°) till evolution of POCI3 ceased. The residue was extracted with an excess of 2n NaOH and benzene. Concentration of the benzene layer afforded the acid chloride (II), 32 mg (20%), m.p. 102-103°, undepressed by admixture with authentic material. Thionyl chloride did not effect the conversion even on prolonged boiling.

The Diacid (VII) This was prepared from benzophenone (l-8g), phenylphosphonous dichloride (5 ml), and partially hydrated aluminium chloride (2-6 g) by the general method, but the reaction time was extended to 6 hr. Work-up afforded the acid chloride (II), 0-67 g (19%), and acidification of the aqueous layer gave the diacid (VII) which was purified by extraction with NaHCC>3 solution, filtration through diatomaceous earth, and acidification. Recrystallization from methanol gave a product, m.p. 278-282°, which would not redisshlve in methanol (1 • 83 g; 40%). It subsequently dissolved only in hot dimethylformamide or hot dimethyl sulphoxide and could not be conveniently recovered from these solvents (Found: C, 64-35; H, 5-5; P, 12-8; equiv. wt., 245, 246. C25H22O4P2 requires C, 66-95; H, 4-9; P, 13-8%; equiv. wt., 224). The persistent contaminant was probably a small amount of aluminium salts which are always difficult to remove and in a dibasic acid of this type could survive even NaHC03 treatment. The acid as precipitated from the reaction mixture was always badly contaminated with these salts and dissolved very slowly in hot, saturated, aqueous NaHC03 in marked contrast to the purified material. Acceptable analytical figures and structural information were obtained by conversion to the dimethyl ester. The purified acid (1 g) and PCI5 (0- 71 g; 2-2 mole) were heated together (155°) till evolution of POCI3 ceased. The residue was taken up in methanol, an excess of sodium methoxide in methanol added, and the solution boiled under reflux for 1 hr. The mixture was poured into water and the diester isolated by ether extraction and recrystallization from ethyl acetate, m.p. 190-195° (Found: C,67-8; H,5-8; P,12-8. C27H25O4P2 requires C, 68• 1; H,5-5; P, 13-0%); v(P=Q) 1230 cm-1; S (CDCI3) 7-05-7-78 (ArH, 19H); 4-5 (J(PCH) 18 c/s; 1H); 3-65 triplet (JXPOCH3) 14 c/s, 6H). The broad m.p. and low yield (0-10g) are not surprising since four diastereoisomers are possible.

Alkaline Degradation of the Diacid (VII) This was carried out as described elsewhere4 and the (p-benzylphenyl)phenylphosphinic acid thus obtained (0-1 g) was converted to p-(benzoylphenyl)phenylphosphinic acid, m.p. 183- 185°, by oxidation with an excess of potassium permanganate in boiling acetone (2-5 hr).

(p-Benzoylphenyl)phenylphosphinic Acid (IX) p-Benzophenonediazonium fluoroborate.—p-Aminobenzophenone (1-94 g; 0-01 mole) was cooled in an ice—salt bath and to it was added a mixture of concentrated HC1 (2 ml) and a filtered solution of NaBF4 (1 -5 g; 0-02 mole) in water (6 ml). Ether was added to prevent frothing and the mixture diazotized with NaN02 (0-7 g; 0-02 mole) in water (10 ml) keeping the temperature below 10°. After stirring at 5° for 0-5 hr the precipitate (2-4 g; 80%) was collected, washed, and thoroughly dried at room temperature. The salt was used without purification for the next step. Doak-Freedman reaction.—To the salt (2-4 g) suspended in EtOAc was added CuBr (0-1 g) and then phenylphosphonous dichloride (1 • 6 g; 1-1 mole). The mixture was warmed on a water-bath till reaction began and then boiled under reflux (1 hr) and poured into aqueous NaHC'03. The aqueous layer was separated, extracted once more with ethyl acetate, then acidified (HC1) and the keto acid (IX) extracted into CHCI3, m.p. 183—185° from ethanol, 0-24 g (10%) (Found: C, 70-4; H, 5-05; P, 9-5; equiv. wt., 317. C19H15O3P requires C, 70-8; H, 4-7; P, 9-6%. equiv. wt., PHOSPHORUS(iii) HALIDES AND CARBONYL COMPOUNDS 2033

308); v (C=0) 1670. It was identical by m.p., mixed m.p., and infrared spectrum with the material obtained from the diacid (VII). The infrared spectrum of the keto acid from the degradation showed a medium intensity band at 728 cm-1 not found in the synthetic material. Apart from this feature the spectra were almost superimposable. Repeated recrystallization caused this band to diminish slowly with respect to adjacent peaks in the spectrum. Since this region is close to that observed for ortho substitution, this may indicate that some of this isomer is present though its spectrum would be expected to be considerably different.

Benzhydryldiphenylphosphine Oxide (X) Benzophenone (1 • 8 g), moist aluminium chloride (2 • 6 g; 2 mole) and diphenylphosphinous chloride (5 ml) by the general procedure yielded the oxide (X), 0-8 g (35%), m.p. 304-305° (lit.14 304-305°) (Found: C, 81-9; H, 5-9; P, 8-3. C25H21OP requires C, 81-5; H, 5-75; P, 8-4%). S (CF3COOH) 5-15 (J(PCH) 18 c/s; 1H); 7-0-7-8 (20H, ArH). It was identical in m.p., mixed m.p., and infrared spectrum with a sample synthesized as follows: A solution of lithium diphenylphosphide was prepared from diphenylphosphinous chloride (2-2 g) and lithium (0-14g; 2 mole) in tetrahydrofuran (35 ml). Benzhydryl chloride (2-03 g; 1 mole) in the same solvent (10 ml) was added dropwise till the deep red colour of the phosphide ion was discharged (80%); the mixture was boiled under reflux (0 • 5 hr). The cooled product was poured into a mixture of saturated aqueous NaHC03 (100 ml) and 100-volume hydrogen peroxide (4 ml). After the product had stood overnight the crude oxide was filtered off and recrystallized from a large volume of ethyl acetate, m.p. 304-305° (1 -4 g, 39%).

14 Horner, L., Hoffmann, H., Klahre, G., Toscano, V. G., and Ertel, H., Chem. Ber., 1961, Short Communication reprinted from the AUSTRALIAN JOURNAL OF CHEMISTRY

THE ALKALINE DEGRADATION OF PHOSPHINIC ACIDS*

By K. L. Freeman! and M. J. Gallagher!

Horner1 has observed that fusion of tertiary phosphine oxides with alkali at 200-300° affords excellent yields of phosphinic acids with loss, as a hydrocarbon, of that group which is most stable as an anion. No mention was made of subsequent decomposition of the phosphinic acids. Two instances have been reported2*3 of thermal dephosphonation (240°) of arylphosphonic acids to the corresponding hydro carbon and inorganic phosphate in reasonable yields. The corresponding degradation of phosphinic acids does not seem to have been reported. Recently, we had occasion to determine the structures of a number of phosphinic acids and we have found that heating with dry sodium hydroxide to c. 300° for 30 min results in the complete decomposition of the acid in a very clean reaction. The easily isolated products were of great assistance in determining the structures,4 and we report these results since the method may prove useful in structure determination of organophosphorus compounds. Chlorodiphenylmethylphenylphosphinic chloride (I)4 decomposes smoothly to give benzophenone, benzhydrylphenylphosphinic acid, phenylphosphinic acid, and phenylphosphonic acid. Ph2C-P(0)Ph -» Ph2C0+Ph2CHP(0)0H+PhP(0)H+PhP0(0H)2 II II Cl Cl Ph OH (I) (II) (HI) (IV) (V) Presumably this reaction proceeds via the hydroxy acid (VI), which subsequently decomposes in a manner analogous to the decomposition of a-hydroxy phosphonium salts.5 0 jcr HCT^ H-^-O-V CPh2—PPh H20 + Ph2C0 + PhPC I 'cr 0"

(VI) (VII)

Attack at phosphorus5 or carbon seems less likely on steric grounds. The anion (VII) is the dianion of phenylphosphonous acid which is a known reducing

* Manuscript received May 25, 1966. ! School of Chemistry, University of New South Wales, Kensington, N.S.W.

1 Homer, L., Hoffman, H., and Wippel, H. G., Chem. Ber., 1958, 91, 64. 2 Freedman, L. D., Doak, G. O., and Petit, E. L., J. org. Chem., 1960, 25, 140. 3 Griffin, C. E., and Brown, J. T., J. org. Chem., 1961, 26, 853. 4 Freeman, K. L., and Gallagher, M. J., Aust. J. Chem., 1966, 19, 2025. 5 Grayson, M., J. Am. chem. Soc., 1963, 85, 79.

Aust. J. Chem., 1966, 19, 2159-61 2160 SHORT COMMUNICATIONS agent in, for example, the thermally induced redox decomposition of monosub- stituted phosphinic acids :6

3RP(0)H RPH2+2RPO(OH)2 I OH This dianion presumably reduces the hydroxy acid to (III). The structure of (III) follows from its analytical and spectral data and its identity with an authentic sample obtained as follows:

PhP(OMe)2+Ph2CHBr -> Ph2CHP(0)0Me ?£ (III) Ph The crude reaction product smelled strongly of phenylphosphine, further supporting the proposed pathway. A sample of the a-hydroxy acid (VI) prepared4 by hydrolysis of the a-chloro-acid chloride (I) decomposed in the same way. A more complex example was the decomposition of the biphosphinic acid (VIII),4 but in this case the products were even more straightforward. Thus, degrada tion afforded diphenylmethane and (p-benzylphenyl)phenylphosphinic acid (IX), which has been identified4 by oxidation to the benzoyl acid and comparison of this with a synthetic sample. PhP(0Hp-C6H4)-CH-P(0)Ph -> Ph2CH2+PhCH2-(p-C6H4)-P(0)Ph OH Ph OH OH (VIII) (IX) The presence of alkali-sensitive groups leads to much more complex products. Thus, a-hydroxy-o-chlorobenzylphenylphosphinic acid gave chlorobenzene, o-chloro- toluene, o-chlorobenzyl alcohol, salicylic acid, and phenylphosphonic acid. This complexity is understandable in terms of secondary reactions arising from initially liberated o-chlorobenzaldehyde, but is of little assistance in structure determination.

Experimental General Procedure The acid or derivative (1 mole) was mixed with powdered NaOH (3 mole) and placed in a long-necked bulb, made for the purpose from soda glass tubing. The temperature was raised slowly by heating in a Wood’s metal or silicone oil-bath until fusion occurred, which usually happened about the melting point of the acid. When any reaction had subsided the temperature was further raised to 300° and held there for 30 min. The cooled melt was dissolved in water and extracted with benzene which was dried and evaporated to afford the neutral products. Acids were isolated by acidification of the aqueous layer.

Chlorodiphenylmethylphenylphosphinic Chloride (I; 3-0 g) This afforded benzophenone (0 • 86 g), identified by its infrared spectrum and the m.p. and mixed m.p. of its 2,4-dinitrophenylhydrazone. Acidification of the aqueous layer gave

6 Frank, A. W., Chem. Rev., 1961, 61, 389. SHORT COMMUNICATIONS 2161 benzhydrylphenylphosphinic acid (III), m.p. 244—246° from aqueous ethanol (0-99 g), identified by m.p. and mixed m.p. and infrared spectral comparison with an authentic sample (see below). Concentration of the aqueous mother liquors afforded phenylptosphinic acid (0-21 g). The mother liquors were taken to dryness and extracted with hot ethyl acetate to give traces of phenylphosphonic acid, m.p. 149-152°. These acids were identified by mixed m.p.

Benzhydrylphenylphosphinic Acid {III) Dimethyl phenylphosphonite (2-3g) was slowly added to stirred benzhydryl bromide (m.p. 39-40°; 2-6 g; 0-8 mole) at 100° during 0-5 hr and the mixture further heated (2-5 hr). The crude product was hydrolysed by boiling under reflux with aqueous ethanolic 0 • 5n NaOH (12 hr). The cooled hydrolysate was extracted with benzene and the aqueous layer acidified to give benzhydrylphenylphosphinic acid, m.p. 244-246° from aqueous ethanol; l-7g (4:7%) (Found: C, 74-1; H, 5-8; P, 9-9; equiv. wt., 310. Calc, for C19H17O2P: C, 74-0; H, 5-6; P, 10-05%; equiv. wt., 308). The n.m.r. of the sodium salt in D2O showed 15 aromatic protons (8 7 -55-7-85) and one benzylic (S 4-85; Jpcn 18 c/s) proton. Concentration of the mother liquors afforded the expected by-product, methylphenylphosphinic acid, m.p. 127-128° (0-7 g; 41%), undepressed by an authentic sample.

(p-Phenylphosphonyl)diphenylmethylphenylphosphinic Acid (IX) This bis-acid4 (0-37 g) gave a neutral product identified as diphenylmethane (0-048 g; 34%) by infrared and gas chromatographic comparison with authentic samples. Purity by the latter technique was at least 95%. The acid fraction yielded (p-benzylphenyl)phenylphosphinic acid (0-15g; 59%), m.p. 155-156° from aqueous methanol (Found: C, 73-8; H, 5-7; P, 10-1; equiv. wt., 314. Calc, for C19H17O2P: C, 74-0; H, 5-6; P, 10-05%; equiv. wt., 308); S (CDCI3): 3-9 (-CH2-; 2H); 13-1 (POH; 1H); 7• 05-7• 85 (ArH; 14H). oc - H ydroxy - o - chlorohenzylphenylphosphinic A cid The neutral products (see text) were identified by comparative gas chromatography. The acids were converted to their methyl esters with diazomethane and similarly identified. Phenyl phosphonic acid (m.p. 152-154°) was isolated. A high-boiling neutral product was unidentified. Reprinted from the AUSTRALIAN JOURNAL OF CHEMISTRY

THE REACTION OF TERVALENT PHOSPHORUS COMPOUNDS WITH DICHLORODIPHENYLMETHANE

By K. L. Freeman* and M. J. Gallagher*

[.Manuscript received June 5, 1967]

Summary The reactions of dichlorodiphenylmethane with (MeO)3P, (MeO)2PPh, MeOPPh2, PPh3, Bu§P, (Me2N)3P, and Ph3As have been studied, and the origin of the products is discussed in terms of attack of phosphorus at halogen and carbon. Attack at halogen does not seem to be the initial step in these reactions.

The reaction of tervalent phosphorus compounds with organic halides is one of the principal methods of forming phosphorus-carbon bonds. Recently, great interest has been shown in the reactions of more complex halides, particularly those belonging to the class called, for want of a better name, “positive” halogen compounds,1’2 e.g. the Perkow reaction.2 Extensive studies1 on a wide variety of such compounds have led to recognition of the fact that attack by tervalent phosphorus on halogenated compounds may proceed either at carbon or halogen.

RgP+R'X ■>- r'r3p+x- (1)

R"OH [R3PX]+[R']~ *-R'H+R3P+OR" X- (2)

(I) (II)

R3PO+R"X The intermediacy of ion pairs of type (1) can usually be detected by carrying out the reaction in a pro tic solvent. The salt (II) decomposes subsequently by the normal Michaelis-Arbusov reaction pathway.3 Reaction (1) is believed to proceed by an essentially SN2

* Department of Organic Chemistry, University of New South Wales, Kensington, N.S.W. 2033. 1 Miller, B., “Topics in Phosphorus Chemistry.” Vol. 2, p. 133. (Interscience: New York 1965.) 2 Hudson, R. F., “Organic Reaction Mechanisms”, Chem. Soc. Spec. Publ. No. 19, London, 1965; see also “Structure and Mechanism in Organophosphorus Chemistry.” (Academic Press: London 1965.) 3 Harvey, R. G., and De Sombre, E. R., “Topics in Phosphorus Chemistry.” Vol. 1, p. 57. (Interscience: New York 1965.)

Aust. J. Chem., 1968, 21, 145-53 146 K. L. FREEMAN AND M. J. GALLAGHER

displacement of halide by phosphorus, and there is kinetic4 and stereochemical5 evidence in support of this where the substrates involved are primary aliphatic halides. There is no evidence for reaction (2) in any of these cases, and it becomes important only where the halide is substituted on the same carbon atom, or in the a-position, with an electron-withdrawing substituent, or where attack at carbon is severely hindered. A very wide variety of such compounds has been studied,1 but there is as yet no general guide to indicate whether attack at carbon or halogen will be operative. Hudson has applied2 the “hard” and “soft” base concept developed by Pearson6 to this problem with some success. This requires that polarizable “bases” (e.g. R3P) bind more strongly to polarizable acids (e.g. halogen atoms)

Table 1

FORMATION OF TETRAPHENYLETHYLENE (Vi) From R3P (1-0 mole) and dichlorodiphenylmethane (1 • 2 mole) at 105° for 1 hr, unless otherwise stated

r3p Reaction Conditions Yield (%)

Bu3P 49 1 mole Ph2CO added 50 1 mole BuOH present 12 2 moles BuOH present 6 1 mole xylene present 40 (Me2N)3P 77 2 moles 76 in boiling benzene 8 Ph2POMe 4 hr reaction 13 than do less polarizable bases (e.g. ethers) and vice versa. This type of reaction is relatively common, e.g. dehalogenation of vicinal dihalides with sodium iodide. Thus it is possible to explain why triphenylphosphine reacts readily with a-bromocyclohexanone but very sluggishly with a-chlorocyclohexanone.2 The distinction between attack at halogen or carbon is, however, by no means clear-cut. Since we felt that this might be due in part to lack of variation of the phosphorus reagents (which almost without exception have been of the type (RO)3P or R3P), and since we had an interest in the synthetic application of one of the reactions,7 we undertook a study of the reactions of dichlorodiphenylmethane with the series (MeO)3P, (MeO)2PPh, MeOPPh2, PPh3, and Bu3P. The reactivity of this series towards primary aliphatic halides is known to increase from phosphite to phosphine,4’8 and we hoped to correlate this with a change in reaction type. We also examined two further compounds for which supporting data were not available: Ph3As and (Me2N)3P.

4 Henderson, W. A., and Buckler, S. A., J. Am. chem. Soc., 1960, 82, 5794. 5 Horner, L., Pure appl. Chem., 1964, 9, 225. 6 Pearson, R. G., J. Am. chem. Soc., 1963, 85, 3533. 7 Freeman, K. L., and Gallagher, M. J., Aust. J. Chem., 1966, 19, 2025. 8 Aksnes, G., and Aksnes, D., Acta chem. scand., 1964, 18, 38. REACTIONS WITH DICHLORODIPHENYLMETHANE 147

All reactions were carried out under the same conditions; optimum yields were not determined. The results are summarized in the following equations and in Table 1.

P(OMe)3+Ph2CCl2 ■X (3)

PhP (OMe) 2-f-Ph2CCl2 -----► MeOP(0)PhCHPh2 (8%)+Me0P(0)PhCClPh2 (4%) (III) (4) MeOPPh2+Ph2CCl2 -----► Ph2P(0)CHPh2 (10%) (5) (IV) Ph3P+Ph2CCl2 -----► Ph3P+-CHPh2Cr (33%) +Ph2C=CPh2 (1 • 4%) (6) (V) (VI) Bu3P+Ph2CCl2 ---- ► Ph2C=CPh2 (50%) (7) (Me2N)3P+Ph2CCl2 ---- ► Ph2C=CPh2 (76%) (8) Ph3As+Ph2CCl2 -X—► (9) The yields (in parentheses) are based on the phosphorus(m) compound and on the hypothetical equation:

R3P+Ph2CCl2------► R3PCl2+i(Ph2C-CPh2) Trimethyl phosphite could probably have been induced to react under more forcing or possibly free-radical conditions. Phosphites react sluggishly with carbon tetrachloride9’10 and a steric effect may here slow the reaction still further. Benzo- phenone was recovered quantitatively from this reaction mixture after hydrolysis. Similarly, lack of reaction in the triphenylarsine case was shown by its recovery (95%). ‘ The results show an overall increase in per cent reaction with increasing nucleophilicity, with the possible exception of methyl diphenylphosphinite. The dechlorinated products (III, IV, and V) presumably arise by attack of phosphorus at the a-halogen of a phosphonium salt or phosphine oxide, e.g. ROH R3P+R3P+CClPh2------► R3P+-Cl+R3P+C-Ph2 - -► R3P+CHPh2 (10) followed by quenching of the ion pair on work-up. All the reaction mixtures possessed a deep red colour destroyed by addition of a protic solvent. The formation of tetraphenylethylene (VI) may be accounted for in a number of ways. Thus, attack at halogen by R3P could give the ion pair (VII) followed by decomposition of the anion and dimerization of the resulting carbene:

R3P +Ph2CCl2 R3P+C1 Ph2C-Cl (VII)

Ph2C=CPh2 Ph9C: (11) 9 Kosolapoff, G. M., J. Am. chem. Soc., 1947, 69, 1002. 10 Burn, A. J., and Cadogan, J. I. G., J. chem. Soc., 1963, 5788. 148 K. L. FREEMAN AND M. J. GALLAGHER

Though tervalent phosphorus compounds act as carbene traps,11 the resulting phosphoranes could react with more carbene to give olefin:12 R3P=CPh2+Ph2C: -* R3P+Ph2C=CPh2 (12) An alternative explanation is suggested by the observation by Partos and Speziale13 of the formation of tetraphenylsuccinonitrile as a by-product of the reaction of triphenylphosphine with chlorocyanodiphenylmethane. R3P+Ph2CCl2------► R3P+C1 Ph2C-Cl

Ph2CCl2 r3p T R3PC12 -f-Ph2C=CPh2 ---- Ph2CCl C ClPh2+Cl- (13) Phosphites will convert vicinal dichlorides into olefins when the substrate is suitably activated by electron-withdrawing substituents.14 In the present case, relief of steric strain in going from the dihalide to the olefin might be expected to provide a driving force. Another explanation is suggested by the dichotomy of reaction of methyl diphenylphosphinite with the dihalide. Under the standard conditions the sole product of the reaction was benzhydryldiphenylphosphine oxide (IV) but if the reaction time was extended or the temperature raised tetraphenylethylene was the principal product. This suggests the sequence

Ph2C-P(0)Ph2 +Ph2CCl2 ■>- Ph2P(0)-CPh2 ci- (VIII) ClCPh2

Ph2P(0)Cl+Ph2C=CPh2 +ci- (14) and this was supported by preparing the phosphorane (VIII) from benzhydryl diphenylphosphine oxide and phenyllithium and allowing it to react with dihalide under the same conditions. Tetraphenylethylene (37%, based on above equation) was obtained, and diphenylphosphinic acid was isolated from the mother liquors. Similarly, the phosphorane from benzhydryltriphenylphosphonium chloride and phenyllithium gives tetraphenylethylene (33%) when treated with the dichloride. Ph3P=CPh2+Ph2CCl2------► Ph2C=CPh2 (15) In order to distinguish between the above possibilities, the effect of butanol upon the products of the reaction of tributylphosphine with dichlorodiphenylmethane

11 Seyferth, D., Grim, S. O., and Read, T. O., J. Am. chem. Soc., 1960, 82, 953. 12 Oda, R., Ito, Y., and Okana, M., Tetrahedron Lett., 1964, 7. 13 Partos, R. D., and Speziale, A. J., J. Am. chem. Soc., 1965, 87, 5068. 14 Dershowitz, S., and Proskauer, S., J. org. Chem., 1961, 26, 3595. REACTIONS WITH DICHLORODIPHENYLMETHANE 149 was studied. All the above mechanisms except the last require initial attack at halogen to give the ion pair (VII), which should be decomposed in the presence of a protic solvent and the production of tetraphenylethylene thereby blocked. BuOH [R3PCl]+[Ph2CCl]------► BuCl+R3PO+Ph2CHCl (16) (VII) If however attack at carbon is the initial step, followed by dechlorination of the a-chlorophosphonium salt to give a phosphorane (as in equation (10)), this could either protonate to give a phosphonium salt or react with the excess of dihalide to give the olefin (VI). In fact butanol has no effect on the products of the reaction though the yield is severely reduced, far more than by an equal volume of an inert diluent such as xylene. This has been found to be due to a competing reaction of the butanol with dichlorodiphenylmethane BuOH+Ph2CCl2------►BuCl+Ph2CO+HCl (17) Under the conditions of the reaction dichlorodiphenylmethane is converted quantitatively into benzophenone by an excess of butanol. Nevertheless, a material balance was obtained in the phosphine-dihalide reactions carried out in the presence of butanol and the formation of more than trace amounts of benzhydryltributyl- phosphonium chloride is excluded. Hence the most likely pathway for the formation of tetraphenylethylene in these reactions would seem to be one involving initial attack at carbon; e.g. equation (18):

r3p R3P + Ph2CCl2------► R3P+CClPh2Cr —R3P=CPh2 + R3PC12

R3P+-yCPh2 Ph2C=CPh2+ R3PC12 ci^cPh, + cr (18) It could be argued that the function of the phosphorane is simply that of chloride abstractor by analogy with the action of iodide ion:10 R3P=CR2+Ph2CCl2------► R3P+CClR2+[Ph2CCl]- (19) However, such a reaction should also be blocked in a protic solvent. Further, in the experiments with tributylphosphine in the absence of solvent, gas chromatography of an ethanolic solution of the mother liquors enabled all the dihalide to be accounted for and hence no appreciable amounts of phosphonium salts were formed in these reactions. Other products observed were Bu3PO and Bu3PC12. The catalytic action of a small concentration of the phosphorane cannot be excluded, but this seems unlikely, particularly in view of the low yield of (VI) in the triphenylphosphine reaction where substantial amounts of phosphorane are formed (equation (6)). The high reactivity of the aminophosphine is worthy of comment. Quantitative data are not available for nucleophilic reactions of compounds of this type but,

18 Finklestein, H., Ber. dt. chem. Ges., 1912, 43, 1533. 150 K. L. FREEMAN AND M. J. GALLAGHER in general, they seem to display a high order of reactivity. We have qualitatively compared the reactivity of the aminophosphine and triphenylphosphine towards methyl iodide in dilute acetone solution, and found the former to react completely before any detectable reaction of the latter had occurred. This high reactivity is not exhibited in reactions not involving attack at an electrophilic centre, e.g. oxidation. A possible explanation is that the lower electron affinity of nitrogen (as compared with oxygen) allows facile electron release to phosphorus, thus stabilizing the developing charge on phosphorus and lowering the energy of the transition state. Some support for this comes from the observation of Ramirez16 that phosphorus trisdimethylamide forms a zwitterionic adduct with a-diketones, whereas other tervalent phosphorus compounds form cyclic phosphoranes. Aromatic aldehydes are converted into epoxides by phosphorus trisdimethylamide17 and the following pathway has been suggested:

RCHO + (Me2N)3P------(Me2N)3P+CHRO“

RHC—CHR + (Me2N)3PO (20)

Adducts corresponding to (X) have been isolated. Tetraphenylethylene is the usual product of the action of “soft” bases, e.g. iodide ion,15 zinc,18 and cyclohexylmagnesium bromide,19 on dichlorodiphenylmethane. “Harder” bases, e.g. alkoxides,20 usually give rise to substitution but sodium t-pentyl oxide20 gives the olefin (VI). We briefly investigated the reaction of potassium t-butoxide in t-butanol on the dihalide. A crystalline product was obtained in very low yield which, despite a sharp melting point and apparent homogeneity by thin-layer chromatography, was shown by mass spectrometry to consist of a mixture of three components whose molecular weights corresponded to the olefin (VI), a monochlorinated derivative, and the epoxide of (VI). The presence of (VI) and its epoxide were supported by gas chromatography, and microanalysis indicated an approximate 1:1:1 ratio. These products indicate that where attack at carbon is blocked, attack at halogen by relatively “hard” bases can occur. The results of this investigation suggest that attack at the chlorine of dichlorodiphenylmethane by R3P does not occur. Since the dihalide should be a moderately favourable case for such attack, it seems that before this type of reaction can occur the anion of the resulting ion pair should be relatively strongly stabilized, e.g., as in a phosphorane. In cases where such stabilization is possible, e.g. hexachloro- cyclopentadiene21 and aryldichloroacetamides,22 attack at halogen by triphenyl-

16 Ramirez, F., Patwardhan, A. V., Kugler, H. J., and Smith, C. P., Tetrahedron Lett., 1966, 3053. 17 Mark, V., J. Am. chem. Soc., 1963, 85, 1884. 18 Norris, J. F., Thomas, R., and Brown, B. M., Ber. dt. chem. Ges., 1910, 43, 294. 19 Schmidlin, J., and von Escher, R., Ber. dt. chem. Ges., 1912, 45, 894. 20 Mackenzie, J. E., J. chem. Soc., 1922, 121, 1695. 21 Mark, V., Tetrahedron Lett., 1961, 295. 22 Speziale, A. J., and Taylor, L. J., J. org. Chem., 1966, 31, 2450. REACTIONS WITH DICHLORODIPHENYLMETHANE 151

phosphine is well authenticated. Insufficient data are at present available to give a clear indication of the degree of anion stabilization necessary. There is at present no simple method of determining whether a base will be “soft” enough to attack at chlorine in the absence of strong stabilization of the resulting anion. Where attack at carbon is severely hindered sterically, attack at chlorine may also occur,23 e.g. PhCOCHClPh +PhPMeBut------► PhCOCH2Ph (23%) (21) PhCOCHClPh -f PhPBu!------► PhCOCH2Ph (90%) (22) Hudson has observed24 the sluggish reaction of hindered a-chloro ketones with trimethyl phosphite and triphenylphosphine, though the a-bromo compounds react readily and explains this in terms of the “softer” nature of bromine. It should be noted that entirely different results would be expected if diphenyl-dibromo-(or -diiodo-)methane were the substrate in the above reactions. Henning and Forner25 have studied the reactions between benzal chloride and tervalent phosphorus compounds and have interpreted their results in terms of initial quaternization followed by dehalogenation of the resulting phosphonium salt by reaction with more of the phosphorus compound. Though complex products were obtained no coupling reactions of the type reported here were observed. This emphasizes the care which must be taken in extrapolating results obtained with one complex halide to related systems.

Experimental All reactions were performed under dry, purified nitrogen. Comparison of products with authentic samples was by infrared spectra, m.p. and mixed m.p., and where applicable proton magnetic resonance spectra (p.m.r.) measured on a Varian A60 machine. Gas chromatography was carried out on a Perkin—Elmer 800 using 6 ft by J in. columns packed with SE30 (5%) doped with Versamid 900 (0-5%) on Chromosorb G (80—100 mesh). With the injector block at 300°, the column at 275°, and a flow rate of 40 ml/min, tetraphenylethylene had a retention time of 4 min 50 sec.

Reagents Triphenyl- and tributyl-phosphine and triphenylarsine were commercial samples. The following compounds were prepared by literature methods: Ph2CCl2, PhP(OMe)2, Ph2POMe, and (Me2N)3P. All had physical constants and infrared and p.m.r. spectra in accord with their structures and in agreement with published values. Purity of liquids was further checked by gas chromatography. P(OMe)3 was a commercial sample and contained about 10% (MeO)2P(0)H by p.m.r. Tetraphenylethylene was prepared from dichlorodiphenylmethane and zinc and the product repeatedly recrystallized from AcOH, m.p. 217—219°. Traces of halogen were difficult to remove from this material. One of the samples obtained below was analysed (Found: C, 92 • 8; H, 6-05; Cl, 0-4. Calc, for C26H20: C, 93-9; H, 6-1%). Mackenzie20 also obtained poor analyses for olefin obtained by the action of sodium t-pentyl oxide on dichlorodiphenylmethane.

General Method The dichlorodiphenylmethane was added to a flask, equipped with magnetic stirring and reflux condenser, heated to 105°, and the phosphine added slowly over 30 min at this temperature.

23 Hoffmann, H., and Schellenberc'h, P., Chem. Ber., 1966, 99, 1134. 24 Hudson, R. F., and Salvadori, G., Helv. chim. Acta, 1966, 49, 96. 25 Henning, H. G., and Forner, K., Z. Chem., 1966, 86, 314. 152 K. L. FREEMAN AND M. J. GALLAGHER

After the addition of the chloride the temperature was maintained at 105° for a further 30 min. In general, the tetraphenylethylene was obtained by addition of ethanol to the cooled reaction mixture, or, in the case of low yields, was determined by gas chromatography. In the reactions of tributylphosphine in the presence of BuOH, the addition was reversed, the dihalide being added to a hot mixture of Bu3P and BuOH. The reaction of PhP(OMe)2 with Ph2CCl2 and the isolation of the products have been described.7

P(OMe)a Reaction After completion, the reaction mixture was hydrolysed and examined by gas chroma tography. A quantitative yield of benzophenone was obtained.

Ph2POMe (1 hr) Addition of EtOH to the product from Ph2POMe (3-24 g) and Ph2CCl2 (4-26 g) gave benzhydryldiphenylphosphine oxide (0-53 g, 9-6%), m.p. 290-292°, identical with authentic7 sample. No tetraphenylethylene was detected. t 4 Hr Reaction This was worked up as above and the product recrystallized from glacial AcOH to give tetraphenylethylene (0-36 g, 13%), m.p. 213—215°, identical with authentic material. The mother liquors afforded the oxide (IV) (0-2 g, 3%), m.p. 292-294°.

15 Hr Reaction at 115° Under these conditions the ester (16-6 mmole) and the dihalide (18 mmole) yielded the olefin (VI) (0-54 g, 20%) and traces of the oxide (IV) (0-05 g, c. 1%).

Phosphorus Trisdimethylamide The reaction mixture from amide (2-18 g) and dihalide (3-8 g) solidified and addition of ethanol gave tetraphenylethylene (1-7 g, 78%), m.p. 212-214°.

Tributylphosphine The phosphine (2-3 g) and halide (3-2 g) gave tetraphenylethylene (0-92 g, 49%). Hydrolysis of the mother liquors gave benzophenone. When the reactions were carried out in the presence of butanol the product was analysed by gas chromatography.

Ph3P The cooled mixture obtained by the general procedure from Ph3P (3-3 g) and Ph2CCl2 (3-6 g) gave, on treatment with acetone, benzhydryltriphenylphosphonium chloride (1-95 g; 33%), m.p. 248—250° (lit.26 240-242°), identical with a sample prepared by heating together benzhydryl chloride and triphenylphosphine. The hydrolysed mother liquors contained tetraphenylethylene (1-4%) and benzophenone (1-7 g; 62% recovery) estimated by gas chromatography.

Preparation of Carbanion from Benzhydryldiphenyl Phosphine Oxide and Reaction with Dichloro- diphenylmethane The phosphine oxide (0-2 g, 0-55 mmole) was suspended in 20 ml of dry diethyl ether in a flask equipped with stirring, reflux condenser, and nitrogen bleed. To the mixture was added ethereal PhLi solution (0-55 mmole) and the mixture refluxed for 15 min. The phosphine oxide dissolved to yield a deep red solution. The reaction mixture was quenched with D20 and the solid

26 Kosolapoff, G. M., “Organophosphorus Compounds.” p. 90. (John Wiley: New York 1950.) REACTIONS WITH DICHLORODIPHENYLMETHANE 153 that separated was filtered off and recrystallized from acetic acid. The p.m.r. of the resultant product showed the absence of the benzylic proton, indicating its replacement by deuterium. The carbanion was prepared as above using phosphine oxide (1-0 g, 2-6 mmole). After refluxing the oxide with an equimolar amount of phenyllithium in ether, dichlorodiphenylmethane was added (3-1 mmole). The ether was removed and the reaction mixture yielded tetraphenylethylene (0-32 g; 37% calculated on the phosphine oxide), after heating for 1 hr at 105°. Extraction of the mother liquors with NaHC03 followed by acidification gave diphenylphosphinic acid (m.p. 185-186°, 18 mg), identical with authentic material.

Reaction of Phosphorane from Benzhydryltriphenyl Phosphonium Chloride and Dichlorodiphenyl- methane The phosphorane was prepared by treating the phosphonium salt (V; 0-19 g, 0-4 mmole) with an equimolar quantity of phenyllithium and refluxing in dry ethyl ether, for 15 min. To the resulting deep red suspension was added an 0 • 2 molar excess of dichlorodiphenylmethane. The ether was removed and the reaction mixture heated to 110° for 1 hr. Gas chromatography of the reaction product in chloroform showed tetraphenylethylene (90 mg; 33% calculated on the phosphonium salt).

Potassium t-Butoxide Potassium t-butoxide (2-7 g; 0-02 mole) in t-butanol (30 ml) was heated to reflux, dichlorodiphenylmethane (4-74 g; 0-02 mole) was added, and the solution refluxed for 1£ hr. The solution turned brown, the solvent was removed, and the residue triturated with acetic acid, yielding 130 mg of crystals, m.p. 154-156° from acetic acid. T.l.c. showed two spots. The product was chromatographed on silica gel and eluted with benzene to give 125 mg of material showing a single spot on t.l.c. (Found: C, 86-8; H, 5-4; Cl, 6-1%). A 1 : 1 : 1 mixture of (VI), a monochlorinated derivative, and the epoxide of (VI) requires C, 88 ■ 4; H, 5 • 5; Cl, 5 • 0%. Mass spectra and gas chromatography showed that the material was a mixture of three products. From gas chromatography two products were shown to be tetraphenylethylene and tetraphenylethylene oxide; the other product would appear to be a chlorinated tetraphenylethylene from its mass spectral molecular weight (366). Parent ion peaks for (VI) and its oxide were observed in the mass spectrum.

Acknowledgments We thank Dr C. S. Barnes for the mass spectrum and Professor R. F. Hudson for the copy of a manuscript prior to its publication.