THE PREPARATION AND
CHARACTERIZATION. OF SOME NEW
ORGANOSCANDlU-M WOUNDS

Thesis for the Degree of M. S.
MICHEGAN STATE UNIVERSITY
JEFFREY A. WM
1972

'J 1 IBRARY

Michigan Ste»:
University

' amnms 99 1 ‘s

HUAG & SUN NS'

' aunx amum mus

LIBRARY BINDERS
SPIIIBPORT ulcmm .

 

ABSTRACT

THE PREPARATION AND CHARACTERIZATION OF
SOME NEW ORGANOSCANDIUM COMPOUNDS

By
Jeffrey A. Witt

Thermally stable organoscandium compounds of the
general formula RSSC have been prepared containing triphenyl—
methyl, trimethylsilylmethyl, pentafluorOphenyl, tolyl and
benzyl groups. Compounds of uncertain composition containing
scandium and naphthalyl and t-butyl groups have also been
prepared. Infrared, mass spectral, elemental and 1H nmr
techniques were used where possible to characterize these
compounds.

Polymeric structures are proposed for these compounds
with the scandium (III) ion in a 4 or 5 coordinate
environment. Normal 2 electron 2 center bonds and 2 electron
3 center bonds are proposed for the terminal and bridging
ligands respectively. The ability of the trimethylsilylmethyl
and pentafluorOphenyl compounds to form dioxane complexes
lends further support to the proposed coordinatively

unsaturated polymeric structures.

 

 

 

 

"A‘llllllll‘ll
l

Jeffrey A. Witt

All of the compounds isolated as solids are decomposed
by the addition of dilute nitric acid but are not perphoric
in air.

Reactions between scandium metal and the alkylhalides
triphenylchloromethane, benzylbromide and t-butylbromide
have produced products of uncertain composition and unusual
properties. Further investigations concerning the [—
composition and structure of these products are necessary
before any conclusions can be reached.

The triphenylmethyl and benzyl compounds of scandium
act as catalysts for the polymerization of styrene in the
absence of other metals, halides or other organometallic
compounds. Further studies to determine whether these and
other organoscandium compounds may act as catalysts in

hydrogenation reactions are indicated.

 

 

 

lil'll’l'll‘l‘f‘fll‘llllfll

THE PREPARATION AND CHARACTERIZATION
OF SOME NEW ORGANOSCANDIUM COMPOUNDS

By

Jeffrey AE Witt

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Chemistry

1972

 

0

ii

 

 

 

ACKNOWLEDGMENTS

The author is deeply indebted to Dr. Gordon A. Melson
for his friendship, inspiration and invaluable suggestions

throughout the course of this research.

The author wishes to express his gratitude to the
Department of Chemistry, Michigan State University, for

the financial aid while attending this University.

iii

 

 

 

TABLE OF CONTENTS

Page
INTRODUCTION 0 O O O O O O O O O O O O O O O O O O O O 1
EXPERIMENTAL . . . . . . . . . . . . . . . . . . . . . 5
A. REAGENTS AND SOLVENTS . . . . . . . S
B GENERAL TECHNIQUES. 5
C. PHYSICAL MEASUREMENTS . . . 6
D SYNTHESES 6
l. Tris(triphenylmethyl)scandium (III). 6
2. Tris(trimethylsilylmethyl)scandium
(III). . . . . . . . . . . . . 7
a. Tris(trimethylsilylmethyl)-
scandium (III) (1,4 dioxane) . . 8
3. Tris(pentafluorOphenyl)scandium (III). 8
a. Tris(pentafluor0phenyl)scandium
(III) (1,4-dioxane) 9
4. Tris(tolyl)scandium (III). 10
S. Tris(benzyl)scandium (III) 11
6. Preliminary Results. . . . . . 11
a. Reaction between Sc3+ and the
naphthalenide ion . . 11
b. Reactions between Sc3+ and the
t-butyl ion, and scandium metal
and t-butylbromide. . 12
c. Reaction between scandium metal
and benzylbromide . . . . . . . 13

iv

5;)“; “a

TABLE OF CONTENTS (Continued)

RESULTS AND DISCUSSION .

CONCLUSIONS AND SUGGESTIONS FOR FURTHER WORK
BIBLIOGRAPHY

Page
14
33
36

n .‘n 1 \n'fii‘nbl‘

 

 

TABLE

LIST OF TABLES

Infrared Absorptions in the Metal-Carbon
Region . . . . . . .

Mass Spectral Data

vi

Page

26

29

'I

r .n— .1...‘

 

LIST OF FIGURES

FIGURE Page
1 Modified sublimation apparatus . . . . . . . 9
2 Products of pressure tube reaction . . . . . l6
3 Possible structures of polymeric
organoscandium compounds . . . . . . . . . . 27

vii

 

 

ill-!l1|l|"lllllll(lj1( (l1ll|[{‘ll[(

 

INTRODUCTION

In 1938 Plets1 reported the isolation of
(CZHS)SSCO(C2H5)2' However, recent attempts to repeat

2’3 and it is doubtful

the synthesis were unsuccessful
that the compound was actually prepared.

The first authentic report of the isolation of an
organoscandium compound was made in 1956 by Birmingham
and Wilkinson.4 They obtained tricyclopentadienylscandium,
(CSH5)3Sc, by sublimation from the product of the reaction
between anhydrous scandium chloride and sodium cyclo-
pentadienide. It is a yellow solid and the authors pr0posed
that the bonding in the compound was ionic because of the
instantaneous liberation of cyclopentadiene upon the
addition of water.

Hart and coworkers3 reported the synthesis of triphenyl-
scandium, (C6H5)38c, and tris(phenylethynyl)scandium,
(C6H5-CEC-)38c in 1968. The compounds were prepared by the
reaction between anhydrous scandium chloride and the
corresponding organolithium compound and were brown, non-
sublimable and perphoric in air. Because of the insolubility

and involatility of these compounds, the authors concluded

that they were not monomeric.

_"_____

The most recent paper concerning the synthesis of new

5 in 1970.

organoscandium compounds was by Coutts and Wailes
The authors reported the isolation of dicyclOpentadienyl-
scandium chloride, (CSHS)ZScCl, by sublimation of the
product of the reaction between anhydrous scandium chloride
and magnesium cyclopentadienide. On the basis of
ebulliometric molecular weight measurements, they proposed

a dimeric structure with bridging chloride ions. The
authors also prepared four new compounds by replacing the
chloride ion with acetate, acetylacetonate, allyl and
phenylethynyl ions. On the basis of ebulliometric and
infrared data, they proposed that dicyclopentadienylscandium
acetate has a dimeric structure with bridging acetate ions
and (dicyclopentadienyl)phenylethynylscandium is at least a
dimer. The 1H nmr spectrum of (allyl) dicyclopentadienyl-
scandium confirmed the symmetrical nature of the allyl group
in the prOposed monomeric structure, and dicyclopentadienyl-
scandium acetylacetonate was also prOposed to be monomeric
on the basis of its molecular weight in boiling benzene and
infrared absorptions characteristic of bidentate acetyl-
acetonate.

Shortly after the report of the synthesis of dicyclo-
pentadienylscandium chloride, Smith and Atwood6 reported its
crystal and molecular structure. Their study confirmed the
presence of bridging chloride ions in the dimeric molecule
and showed that the cyclopentadienide ions exhibited penta-

hapto-coordination.

bbmwu'fi
.
4L -.'-

 

III].

 

II III"!

 

llll' In I ‘| All. fll.lA

In 1965, Kirmse and Zwietasch7 reported that scandium
chloride would form complexes with olefins such as tetra—
methylethylene, cyclohexene, and 1,5-hexadiene. They also
found that olefins such as l,l-diphenylethylene and 1,3-
cyclohexadiene formed oligomers in the presence of scandium
chloride. 1
A patent8 granted in 1959 claimed the use of a
solution of scandium chloride and phenylmagnesium bromide
as a catalyst for ethylene polymerization. Although it was
not completely understood what made the system
catalytically active, it was proposed that a complex was
formed that was capable of coordinating ethylene or similar
olefins. The nature of this complex was not discussed
however.
It had been accepted for some time that transition
metal-carbon bonds were relatively weak, but could be

9,10

stabilized by using fl accepting ligands. A more recent

theory proposed that the function of these ligands was to

11

block coordination sites on the metal. Since stable

compounds have been prepared that do not contain stabilizing
ligands, it now appears that a judicious choice of ligand
will eliminate the need for supporting ligands.11’12
There are several criteria to be considered in choosing
a ligand. An anionic ligand that is able to delocalize its
charge, such as the benzyl ion, will be less likely to be

13

oxidized. If the 8 atom of the anionic ligand has no

 

 

readily transferable groups, there will be less chance of

11,12 An

an alkene elimination reaction taking place.
anionic ligand that is bulky could block coordination
sites on the metal and prevent possible decomposition
reactions involving attack at the metal-carbon bond.14
These criteria, as well as mechanisms for possible
decomposition reactions, have been summarized by Braterman
and Cross.15
A primary aim of investigating the area of organo-
scandium chemistry is to prepare stable organoscandium
compounds and determine whether they can act as catalysts

in reactions such as polymerization of olefins.

I I. \III. E I I | [II-1| I’ll II" I II. II." .III

EXPERIMENTAL

A. REAGENTS AND SOLVENTS

Benzene, diethylether, 1,4-dioxane, hexane and
petroleum ether were purchased from J. T. Baker Chemical
Company and dried and deoxygenated over sodium before use.
Magnesium (Fisher Scientific Company), lithium (Matheson
Coleman and Bell) and sodium (Matheson Coleman and Bell)
were used as purchased.

2-Bromo-Z-methylpr0pane (t-butylbromide), p-bromotoluene
and o-bromotoluene (benzylbromide) were used as purchased
from Matheson Coleman and Bell. Naphthalene (Mallinckrodt),
iodopentafluorobenzene (PCR, Inc.) and chloromethyltrimethyl-
silane (PCR, Inc.) were also used as purchased. Triphenyl-
chloromethane (PCR, Inc.) was purified by sublimation before
use.

Anhydrous scandium chloride was prepared from scandium
oxide (Roc/Ric, 99.9%) by the method of Melson and Stotz.16

Scandium foil was used as purchased from Alfa, Inc.

B. GENERAL TECHNIQUES
All operations were carried out where possible in a
moisture and oxygen free nitrogen atmosphere provided by 3

Vacuum Atmospheres "Dri-Lab, Dri-Train" system.

'11:“ch T: m

lilllllitl'liII'I'I‘i'l I‘lulllllrul‘l'l’ll'l‘E'l'I'l

C. PHYSICAL MEASUREMENTS

Infrared spectra in the region 4000-250 cm.1 were
recorded with a Perkin-Elmer Model 457 infrared spectro-
photometer using nujol mulls between CsI plates. Mass
spectra were determined with a Hitachi-Perkin-Elmer RMU-60
Spectrometer. 1H nmr spectra were recorded with a Varian
HA-lOO spectrometer.

Elemental analyses for scandium were determined by
hydrolysis of the sample and igniting the residue to scandium
oxide. Carbon and hydrogen analyses were performed by

Galbraith Laboratories, Knoxville, Tennessee.

D. SYNTHESES
l. Tris(triphenylmethyl)scandium (III)
An excess of freshly cut sodium metal was added to a
flask containing 50 ml. diethylether and 2.51 grams (9 mmoles)

triphenylchloromethane, (C6H5)3CC1. The solution immediately

became yellow thus indicating the presence of the
triphenylmethyl radical.17 After half an hour, the solution
became red which indicated the presence of the tri-

18 The solution was filtered and the

phenylmethyl anion.
filtrate added to a flask containing 0.46 grams (3 mmoles)
scandium chloride. The solution was very dark after two
hours stirring and the solvent was removed under reduced

pressure. The dark solid remaining was extracted with 50 m]

benzene, filtered and the filtrate taken to dryness under

amt-Tia]

reduced pressure. The brown solid obtained was placed in a
sublimator and heated under vacuum at 120° for several

hours to remove any volatile impurities. Less than 0.5

grams of brown tris(triphenylmethyl)scandium, (III), I, remained;
yield 20%. Rather than generate the triphenylmethyl anion

with sodium metal, an alternate reaction using scandium

metal was chosen. An excess of triphenylchloromethane and

 

0.1 grams (2.2 mmoles) scandium foil were placed in a I
pressure tube. The pressure tube was evacuated and heated

at 140° for four days in a vertical position. The brown

solid at the bottom of the pressure tube was extracted with

50 ml benzene, filtered and the filtrate taken to dryness.

The brown solid obtained was placed in a sublimator and

heated at 180° for 20 hours under vacuum to remove unreacted
triphenylchloromethane or volatile impurities such as

hexaphenylethane; yield 0.3 grams 17%.

2. Tris(trimethylsilylmethyl)scandium (III)
An excess of chloromethyltrimethylsilane, (CH3)SSiCH2Cl,

was added to a flask containing 75 ml diethylether and
0.29 grams (12 mmoles) magnesium. The mixture was heated
periodically until all the magnesium had reacted. The
resulting solution was added to a flask containing 0.61
grams (4 mmoles) scandium chloride and stirred overnight at
room temperature. The solvent was removed under reduced

pressure and the white solid remaining was extracted twice

with 150 ml of warm hexane and filtered. The two filtrates
were combined and the hexane removed under reduced pressure
yielding 0.78 grams of tris(trimethylsilylmethyl)scandium, III,
II; yield 64%.

Anal. calc. for ((CH3)SSiCH2)SSc: Sc, 14.7%.
Found: Sc, 15.7%.

a. Tris(trimethylsilylmethyl)scandium (III)
(1,4 dioxane)

A solution, as prepared above, was stirred overnight
and the solvent removed under reduced pressure. The white
solid remaining was extracted with two 100 m1 portions of
petroleum ether and filtered. The filtrates were combined
and almost all of the petroleum ether was removed under
reduced pressure kmving a gelatinous substance. A portion
of this product was redissolved in 75 ml of petroleum ether
and 15 ml of 1,4 dioxane were added. A white precipitate
formed immediately which was filtered off and dried under
vacuum, 113.

Anal. calc. for ((CH3)SSiCH2)38c(C4H8Oz): Sc, 11.4%.
Found: Sc, 11.0%.

3. Tris(pentafluorophenyl)scandium (III)
A flask containing 75 ml diethylether, 0.145 grams
(6 mmoles) magnesium and 1.76 grams (6 mmoles) iodopentafluoro—
benzene was heated periodically until all of the magnesium
had reacted. The resulting solution was added to a flask

containing 0.31 grams (2 mmoles) scandium chloride and

[I I III I} ll] 1‘! II I Ill- : III 1" 1" I l' l I 1" l I I \III“ ’II‘ ‘III I! l ’I.‘ III l l A IF!“ I Isl. I {I 1"

stirred overnight at room temperature. The solvent was
removed under reduced pressure, leaving a dark tarry
substance which was placed in a sublimator. The sublimator
was heated at 150° under vacuum, and a colorless liquid was
observed condensing on the coldfinger and dripping down to
the bottom of the sublimator. A modified sublimator,

shown in Figure l, was used to collect the fraction of

liquid with a boiling point from 130-140°, III.

Modmgd sublimation Cold finger of sublimator

apparatus

 

 

 

 

L
[Li Bucket to collect llquld

Figure 1. Modified sublimation apparatus.

a. Tris(pentafluorophenyl)scandium (III)
(1,4-dioxane)

The solid remaining in the sublimator was extracted
with 100 m1 of a 5:1 mixture of diethylether and 1,4-dioxane.
The solution was filtered to remove precipitated magnesium
salts and the volume of the filtrate reduced. High boiling

petroleum ether was added and a dark brown solid precipitated

. I'm 12' 3-qu

.I.Illl.l ll. III-ll II II III ‘I' 1 l l ,l I II. III I III II! ‘IIIII II" II I ‘1' II" III III’ I III III I I I II- I I l I .JIII.

10

and was filtered off and dried under vacuum; yield .3 grams

of 1113.

4. Tris(tolyl)scandium (III)

An excess of p-bromotoluene and 0.145 grams (6 mmoles)
magnesium were added to a flask containing 75 ml diethylether.
The solution was heated periodically until all of the
magnesium had reacted and added to a flask containing 0.31
grams (2 mmoles) scandium chloride. Initially
the resulting solution was yellow, but after stirred over-
night at room temperature it became brown. The solvent was
removed under reduced pressure leaving a brown solid. The
brown solid was extracted with 50 ml benzene, filtered to
remove magnesium salts, and the filtrate taken to dryness
under reduced pressure. The brown solid obtained from the
filtrate was again extracted with 50 ml of benzene. During
a slow filtration of the solution, a great deal of yellowish
brown solid came out of solution and was collected on the
filter and dried under vacuum. Analysis of powder obtained
showed that the sample contained 15.6% scandium; theoretical
14.2% for (CH3C6H4)3SC. The filtrate from which the powder
was obtained was allowed to evaporate to dryness in the dry
box producing 0.1 grams of small yellowish brown crystals,

IV.

 

11

5. Tris(benzyl)scandium (III)

To a flask containing 75 m1 diethylether were added
0.145 grams (6 mmoles) magnesium and an excess of benzyl-
bromide, C6HSCH2Br. The solution was heated periodically
until all the magnesium had reacted, added to a flask
containing 0.31 grams (2 mmoles) scandium chloride and A

stirred overnight at room temperature. The solvent was

removed under reduced pressure leaving a brown solid. The

 

solid was extracted with 75 m1 benzene and the solution
filtered to remove magnesium salts. The filtrate was taken
to dryness under reduced pressure and the brown solid
obtained again extracted with 75 m1 benzene. During a slow
filtration a great deal of brown powder came out of solution
and was collected on the filter. Evaporation of the filtrate

left 0.1 grams of brown tris(benzyl)scandium (III), v; yield 18%.

6. Preliminary Results

a. Reaction between Sc3+ and the naphthalenide ion

An excess of lithium ribbon and 0.76 grams (6 mmoles)
naphthalene were added to a flask containing 75 ml diethyl-
ether. After half an hour of stirring, the solution was green,
thus indicating the presence of the naphthalenide anion.5
The solution was added to a flask containing 0.31 grams
(2 mmoles) scandium chloride and stirred at room temperature
for 3-5 hours. The solution was brown and the

solvent was removed under reduced pressure. The brown solid

remaining was extracted with 100 m1 benzene and the solution

 

12

was filtered. Hexane was added to the filtrate and a light
brown precipitate formed which was collected by filtration
and dried under vacuum. The amount of solid, VIa, obtained

was 0.15 grams.

b. Reactions between Sc3+ and the t-butyl ion, and

scandium metal with t-butylbromide

To a flask containing 75 m1 diethylether were added
0.85 grams (6 mmoles) Z-bromo-Z-methylpropane, (CH3)3CBr,
and 0.38 grams (6 mmoles) n-butyllithium. The solution was
stirred for an hour at room temperature and added to a flask
containing 0.31 grams (2 mmoles) scandium chloride. The
solution was stirred for 3 hours and the solvent was removed
under reduced pressure.

50 ml of benzene was added to the solid and the solution
was filtered. Hexane was added to the filtrate and a tan
powder precipitated which was filtered from the solution and
dried under vacuum. The product, VIb, .09 grams, was placed
in a sublimator and gradually, under vacuum, heated to 150°.
Examination of the cold finger revealed that nothing had
sublimed, which indicated that no t-butyllithium had been
present in the sample.19

An alternate method of preparing tris(t-butyl)scandium
(III) was also attempted. 50 m1 of Z-bromo-Z-methylpropane,
50 ml of high boiling petroleum ether and 0.09 grams (2
mmoles) scandium metal were placed in a flask and heated at

75° for 3-5 days. The solution was taken to dryness under

 

13

reduced pressure and 50 ml benzene was added to the solid
that remained. The solution was filtered and high boiling
petroleum ether was added to the filtrate. The solid
precipitate that formed was filtered from the solution and
dried under vacuum. The amount of solid, VIc, obtained was

.05 grams.

“1

c. Reaction between scandium metal and benzylbromide

\l

50 ml of colorless benzylbromide and 0.09 grams 1mm
(2 mmoles) scandium metal were placed in a flask, and after
heating at 75° for 3-5 days the solution was brown in color.
The solution was filtered to remove unreacted scandium metal
and scandium bromide. High boiling petroleum ether was added
to the filtrate and the dark tar which precipitated was
filtered and dried under vacuum. The amount of blackish

solid obtained, VId, was .15 grams.

RESULTS AND DISCUSSION

Reactions between scandium (III) and a series of

organic anions have produced thermally stable, moisture

«a
L

sensitive organoscandium compounds. The anions were chosen

because they fulfill the criteria of being bulky enough to

at?!

shield the metal, charge delocalization is possible, and no
readily transferable groups on the 8 atom, or because other
organometallic compounds have been prepared using the anion.
The reactions were undertaken with the aim of preparing
compounds containing scandium-carbon sigma bonds.

None of the compounds reported herein are perphoric
in air, although all those isolated as solids reacted
immediately with dilute nitric acid except tris(triphenyl-
methyl)scandium (III) which reacted slowly. The reaction
presumed to be responsible for this behavior is that of the
anionic ligands with H+ to form the corresponding hydro—
carbons. The phenyl and phenylethynyl compounds of scandium
are perphoric in air, but no explanation of the differing
behavior is possible at this time.

Carbon analyses for all of the compounds isolated as
solids, I-VIb excluding III, were low. In
addition, the carbon to hydrogen ratio was low for all of

the analyses except for VIb. Carbide formation is thought

14

15

to be one of the factors involved, as was reported for the
phenyl and phenylethynyl compounds, but it is apparently
not the only factor to be considered.

The triphenylmethyl group satisfies the three criteria
listed earlier, and it was, perhaps, for these reasons
that the triphenylmethyl group was used to prepare compounds

18 and nickel.20

of calcium Hexaphenylethane was added to
either calcium amalgam or a nickel (0) complex to prepare
the compounds. The first reaction utilized to prepare tris
(triphenylmethyl)scandium (111) did not make use of scandium
metal, but involved a reaction between the triphenylmethyl
anion and scandium (III). A large excess of sodium was used
to prepare the anion, because a stoichrometric amount merely
produces hexaphenylethane by means of a Wurtz reaction, which
readily dissociates to form triphenylmethyl radicals.17

The reactions between scandium chloride and triphenyl-
methylsodium and scandium metal and triphenylchloromethane
both produced a brown, nonsublimable solid, I. The phenyl
and phenylethynyl compounds are also brown, nonsublimable
solids but are insoluble, whereas I is soluble in benzene and
produces a brown solution. The solution is unstable,
however, and decomposes on standing to give a yellow solution
of triphenylmethyl radicals.

The infrared spectrum of tris(triphenylmethyl)scandium
(III) contained bands of weak intensity at 522 and 426 cm-1
in addition to the expected absorptions of the triphenylmethyl

group.

16

The mass spectrum of I did not contain a parent ion
peak. However, the temperature needed to produce a spectrum
was sufficiently high, SOC-395°, that extensive decomposition
had taken place, as evidenced by large peaks corresponding
to hexaphenylethane, m/e = 486, and its fragments. Peaks
were observed corresponding to ScLz+ and ScL+, m/e = 531
and 288 respectively.

The reaction between scandium metal and triphenylchloro-
methane was more successful than expected in that tris(tri-
phenylmethyl)scandium (III) was not the only product obtained.

As shown in Figure 2, three distinct products were observed.

Products of pressure
tube reaction between
St:0 and (C6 I‘l5)3C Cl

Colorless llquld

White Solid

 

 

Brown Sol ld

U

Figure 2. Products of pressure tube reaction.

17

The brown solid was a mixture of scandium chloride,
tris(triphenylmethyl)scandium (III), and unreacted scandium
metal and triphenylchloromethane. The white solid has been
identified as hexaphenylethane on the basis of infrared
and mass spectral data. The mass spectrum of the clear
liquid had peaks corresponding to (C6H5)3CH+ and (C6HS)3C+’
m/e = 244 and 243 respectively, but nothing heavier was
observed. The infrared spectrum of the liquid had a few
absorptions similar to those of I, but was significantly
different in that additional absorptions at 1662 vs, 1620 sh,
1319 s, 942 s, 720 m, and 375 w (br) cm.1 were observed
along with much stronger absorptions at 920 s, 640 s, and
605 s cm-l. The spectrum is no closer to resembling that
of I than it is those of hexaphenylethane or triphenylmethyl-
sodium.

Because of the rather unusual characteristics of the
clear liquid, an attempt was made to prepare more of it for
further study. A few crystals of colorless triphenylchloro—

methane were near the neck of the pressure tube, and as the

reaction progressed, some of the liquid condensed on the

crystals of triphenylchloromethane and produced a brown solid.

Since tris(triphenylmethyl)scandium (III) is a brown solid,
it was concluded that the liquid contained scandium. As of
this date, no conclusions have been reached concerning the

composition of the liquid or the oxidation state of the

scandium present in it.

18

In recent years, the trimethylsilylmethyl group has
been found to be useful in preparing thermally stable

11,12,21-23 This

organometallic compounds and complexes.
group fulfills the criteria of being bulky enough to shield

the metal and having no readily transferable groups on the

8 atom. The reaction used to synthesize tris(trimethyl- f

silylmethyl)scandium (III), II is similar to that used for

other transition metals, but the product could not be induced

(I... _
. -‘ .z‘. .
..

to sublime.
The infrared spectrum of 11 contained the expected
absorptions of the ligand along with absorptions at 3360 w,

1

1620 w, and 430 sh cm' which indicated that some

decomposition had taken place, and peaks at 530 sh and 505 m
cm-l. On the basis of the infrared spectrum, it was not
surprising that the amount of scandium found was higher than
that calculated on the basis of ((CH3)SSiCH2)38c. Water
molecules could attack the Sc-C bonds and produce tetra-
methylsilane and scandium hydroxide.

As was mentioned earlier, II could not be induced to
sublime, thus it was not surprising that the mass spectrum
of 11 did not contain a parent ion peak. The heaviest peak
observed corresponded to ((CH3)3SiCH2)2+ or L2+, m/e = 174-
In addition to decomposition fragments of L2+, peaks were
observed for ScL+ and Sc+, m/e = 132 and 45 respectively.
Due to the insolubility of tris(trimethylsilylmethyl)
scandium (III) once it has been isolated as a solid, it was

impossible to obtain an 1H nmr spectrum of it.

II II I I II. I. ll 1 .l.l l 1" I I ll. 1' I III II "I‘ lllll

19

The dioxane complex of II, IIa, is a 1:1 complex on the
basis of the scandium analysis. The infrared spectrum of
11a had all of the absorptions found for II along with

bands for dioxane, including a strong band at 340 cm-1

assigned to v (Sc-O).24
The mass spectrum of 11a was similar to that of II and
I in that no parent ion peak was observed and peaks

corresponding to L2+ and ScL+, m/e = 174 and 132 respectively,

Iii-f1 filmn fl
' .

were observed. Two other peaks at m/e = 219 and 220 were
observed that corresponded to ScLz+ and ScL (dioxane)+
respectively. Peaks heavier than m/e = 220 were observed
but could not be assigned.

The pentafluorOphenyl group has been widely used for
preparing compounds and complexes of transition metals and

25

main group metals. More recently, tris(pentafluorophenyl)-

indium (III) and several complexes of this compound have
been prepared.26’27
When tris(pentafluorophenyl)indium (III) is sublimed at
140-150° onto a room temperature probe it condenses as a
liquid. Tris(pentafluorophenyl)scandium (III), III, is a
liquid that condenses at l30-l40° onto a room temperature
probe. The mass spectrum of 111 had a fairly strong parent
ion peak, m/e = 546, as well as peaks corresponding to
ScL4+ and ScL2+, m/e = 713 and 379 respectively. There was
no large peak corresponding to L2+, (C6F5)2+ m/e = 334, as
there was in the mass spectrum of a lower boiling fraction,

60-125°.

IIIIIIIIIII

I I I. I! I Ill I l I I l‘
I .

20

The infrared spectrum of the liquid remaining on the
cold finger in the first sublimation attempt was very complex,
indicating that several compounds were present. The mass
spectrum of this liquid indicated the presence of LZScF+,
LZScI+, L38c+, L2+ and LH+. The infrared spectrum did have
two absorptions of weak to medium intensity at 560 and 580
cm.1 that could not be assigned as absorptions of the penta-
fluorophenyl group.

The infrared spectrum of the brown dioxane complex, IIIa,
was very similar to that of the indium complex which was
prepared in the same manner. In addition to the expected
absorptions of the ligands present, a very broad peak of
medium intensity was observed at 520 cm'l. The mass spectrum
of 1113 also did not contain a parent ion peak, but peaks
corresponding to ScL2+, LH+, dioxane+, and Sc+ were observed,
m/e = 379, 168, 88 and 45 respectively.

The tolyl group was chosen to prepare an organoscandium
compound because the phenyl compound has been prepared and
the two compounds should be very similar. The phenyl
compound is a yellowish brown, insoluble and involatile solid,
and tris(tolyl)scandium (III), IV, is a yellowish brown,
insoluble and involatile solid.

The infrared spectrum of IV contained the expected
absorptions for the ligand in addition to a broad peak of

medium intensity at 542 cm-1, a broad, weak peak at 3300 cm-1,

1

v (O-H); and a shoulder at 400 cm' , v (Sc-O). The presence

21

of the last two bands mentioned indicated that partial
decomposition had taken place and could account for the
high value for scandium in the analysis.

The lowest temperature at which a mass spectrum could
be obtained was 220°, and, as expected, no parent ion peak
was observed. The major peaks in the spectrum were assigned
to (CH3C6H4)2+ and its fragments. Other peaks observed,

m/e = 136, 91 and 45, were assigned to ScL+, L+ and Sc+
respectively.

The benzyl ion has been used to prepare thermally stable
organometallic compounds such as tetrabenzyltitanium.13
The benzyl ion fulfills the criteria in that it is bulky
enough to shield the metal to some extent, there are no
readily transferable groups on the 8 atom, and charge
delocalization is possible. The methyl group only fulfills
the second criterion and tetramethyltitanium decomposes
above -78°.

Like I, IV, and the phenyl and phenylethynyl compounds
of scandium, tris(benzyl)scandium (III), V, is a brown,
involatile solid. It is slightly soluble in deuterochloro-
form, and an 1H nmr spectrum was obtained as shown on the
following page. It contained a complex multiplet 7.15 ppm
from TMS, (CH3)4Si, corresponding to the aromatic proton
absorption, and a sharp singlet 2.9 ppm from TMS for the
methylene proton absorption. The 1H nmr spectrum of tetra-
benzyltitanium had two multiplets at 6.42 and 6.81 ppm from

TMS and a sharp singlet at 2.57 ppm respectively.

 

22

2....» ad 2.3m SN

32...:

 

 

 

 

 

 

omeIOnIoOH do £330QO CE: I.

 

23

The infrared spectrum of V contained the expected
absorptions of the benzyl group in addition to a broad peak
of medium intensity at 512 cm-1. The mass spectrum of this
involatile solid had peaks corresponding to L+ and Sc+,

m/e = 91 and 45 respectively.

The infrared spectrum of the product obtained from
the reaction between scandium chloride and lithium
naphthalenide, VIa, had a broad peak of medium intensity at
525 cm.1 in addition to the expected absorptions of the
ligand. The mass spectrum of VIa had peaks corresponding
to C10H8+ and Sc+, m/e = 128 and 45 respectively.

The infrared spectrum of VIb, the product of the
reaction between scandium chloride and t-butyllithium, had
a broad absorption of medium intensity at 540 cm'1 in
addition to the expected bands of the ligand. The mass
spectrum of VIb had peaks corresponding to ScL+ and L+,

m/e = 102 and 57 respectively.

Recent reports have shown that gallium and indium metals

will undergo some type of reaction in the presence of alkyl—

28,29

halides. The results indicate that mixed compounds are

formed, RMX and RZMX (M = Ga, In; X = Br, I; R = CH

2 3.

CH CH 3CH2CH2)' In view of these reports, and the

3 2’
successful reaction between scandium metal and triphenyl-

CH

chloromethane, reactions between scandium metal and benzyl-
bromide and t-butylbromide were attempted. In both cases,
the surface of the metal immediately became discolored and

indicated some sort of reaction.

24

The infrared spectrum of the benzyl product, VId,

had absorptions at 280 m (br), v (Sc-Br):24

1

497 m (br)

and S45 sh cm- in addition to the expected absorptions of

the ligand. The infrared spectrum of the t-butyl product,
VIc, had a broad, strong absorption at 1575 cm'1 and no

absorptions were observed below 600 cm.1 except for the

beginning of a broad band of medium intensity estimated to 7“”
be at 220 cm’l.

 

The mass spectrum of VId at 115° had peaks corresponding

+
to C6HSCH2

and 45 respectively. The mass spectrum of VIc from 210-300°

Br+, ScL+, L and SC+, m/e = 172 and 170, 136, 91
had a small peak at m/e = 57, (CH3)3C+, which was one of a
series of peaks of decreasing intensity from m/e = 55-59.
Another peak of equally low intensity was observed at m/e =
4s, Sc+.

On the basis of infrared and mass spectra, VId
is similar to tris(benzyl)scandium (III) but has both
scandium-carbon and scandium-bromine interactions.

The infrared spectrum was very similar to those reported
for the mixed products of gallium and indium, and it is
concluded that VId is a mixture of RScBr2 and RzScBr (R =
C6H5CH2) with benzylbromide impurities.

The infrared and mass spectra do not give
conclusive evidence on the composition of VIc. The
lack of any absorptions in the expected metal—carbon
absorption region indicates the absence of the t-butyl group

in the product. This is not an unreasonable result, as the

25

possibility of an elimination reaction to form Z-methyl-
propene, (CH3)2C=CHZ, is much higher than in the reaction
between scandium chloride and t-butyllithium. If the
elimination reaction occurs, then the product obtained could
be a mixed one of hydrogen and bromine with scandium. The

1

broad band, circa 220 cm' , can then be assigned to v

1

(Sc-Br) and the band at 1575 cm- can be assigned to a

scandium—hydrogen interaction. A titanium compound has been

30

prepared that contains bridging hydride ions, di-u-hydrido-

bis(cyclopentadienyl)titanium (III), and the authors
assigned a broad, strong absorption observed at 1450 cm-1
to the antisymetric stretch vibration.

Table 1 lists the absorptions of the compounds prepared
thus far in the expected metal-carbon absorption region.

It was concluded, on the basis of their insolubility,
involatility and infrared spectra, that the phenyl and
phenylethynyl compounds of scandium were not monomeric.3
The absorptions reported for these compounds in the metal-
carbon region, in a preliminary communication in the case
of the phenyl compound,31 are very similar to the absorptions
of the compounds whose preparations are reported herein.
However, no assignments were made for the phenyl and phenyl-
ethynyl compounds.

For polymeric or dimeric structures, two scandium-carbon
stretching frequencies or a broad band shOuld be observed

due to the difference in strength of the two types of bonds

 

26

Table 1. Infrared Absorptions in the Metal-Carbon Region

 

Absorptions in

 

compound Metal—Carbon Region a
[(C6H5)3C]38c 522 w-m, 426 w-m [
[(CH3)3SiCH2]3Sc 530 sh, 505 m (br) E
[(CH3)381CH2]SSC(C4H802] 530 sh, 505 m (br)

(C6F5)38c 580 w, 560 w
(C6F5)3SC(C4H802) 520 m (br)

(CH3C6H4)3SC 542 m (br)

(C6H5CH2)SSC 512 m (br)

Sc(C10H7)n 525 m (br)

Sc[(CH3)3C]n 540 m (br)
(C6H5CH2)S°Br2 s45 sh, 497 m (br)
(C6H5CH2)zScBr

(CSH5)3SC 370 s, 325 m, 296 m
(C6H5CEC)SSC 540 sh, 516 m, (ref. 3)
(C6H5)3Sc 490 (ref. 31)

 

27

as shown in Figure 3. The terminal ligands have a normal

 

L L L L

\ l/L\s|c/L\ \3.|:/L\L¢/L
I I / \./ \.
L L

Figure 3. Possible structures of polymeric organoscandium
compounds.

two election two center bond and the bridging ligands will
take part in two electron three center bonds. For all of
the compounds reported herein except for VIc, either two
absorptions or a broad absorption were observed in the
metal-carbon region which are assigned to v (Sc-C) and are
indicative of dimeric or polymeric structures. In addition,
the fact that the mass spectrum of tris(pentafluorophenyl)-
scandium (III) had peaks corresponding to ScL4+, ScL3+ and
ScL

+ . . .
2 15 further support for the at least dimeric structure

proposed for these compounds.

28

The slightly higher absorptions assigned to v (Sc-C)
for (C6FS)3Sc are reasonable because of the increased
stability of the compound due to the high electronegativity
of the pentafluorophenyl group. Presumably, the addition
of a dioxane molecule would weaken these strong bonds more
noticeably than it would for a hydrocarbon ligand.

The increased stability of a metal-carbon bond caused

by using a very electronegative ligand such as the penta-

Remain-marina“
a

fluorophenyl group is not enough to account for the

extremely high values for v (In-C) assigned by Deacon and

-1 26

Parrott, 792 and 788 cm Other workers have prepared

organoindium compounds using hydrocarbon ligands such as

29,32 with v (In-C)

1

methyl, ethyl and n-prOpyl anions,
assigned to absorptions in the range of 455-575 cm- , and
it is concluded that the assignments made by Deacon and
Parrott are incorrect.

The infrared spectrum of sodium cyc10pentadienide has
no absorptions of even medium intensity below 600 cm-1.
The infrared spectrum of (C6H5)3Sc has three bands of at
least medium intensity, which would indicate, on comparison
with other organoscandium cpmpounds, that it too is polymeric
with bridging cyclopentadienyl groups. A recent study of
(C5H5)3Sc confirmed that its structure is polymeric in the
solid state with bridging cyclopentadienyl groups.33

Table 2 shows the mass spectral data for the organo-

scandium compounds reported herein and three lanthanide

cyc10pentadienides.

29

Table 2. Mass Spectral Data

 

 

Compound Peaks Observed
+ + +
[(C6HS)3C]38C ScL2 L2 ScL L Sc
. + +
[(CH3)3SICH2]3SC L2 SCL L Sc
[(CH ) SiCH ] Sc diox. ScL +
3 3 2 3 2 + + + +
ScL(diox) L2 SCL L Sc
+ +
(C6F5)3Sc SCL2 Sc
- + diox
(C6F5)38c d10X. SCL2 C6F5 Sc
+ +
(CH3C6H4)38c L2 SCL L Sc
.4.
(C6H5CH2)SSC L SC
4.
Sc(C10H7)n L Sc
Sc[(CH3)3C]n ScL+ L+ Sc
RScBr + R ScBr
2 2 + +
R = C6H5CH2 SCL L SC
(CSHS)3SC ScLz+ ScL+ Sc
+ + +
(C5H5)3M M12 ML M

M = Nd, Yb, Sm (ref. 35)

 

 

30

The mass spectra are consistent with the polymeric
structure proposed on the basis of infrared data, insolubility
and involatility for the compounds reported here. Either
of the structures shown in Figure 3 correlates with the
mass spectral data, the observed fragments arising from the
initial cleavage at the weaker bridging bonds with further
fragmentation dependent on the strength of the bonds between

the scandium and the remaining ligands.

 

The fragmentation patterns observed for the compounds
at temperatures under 200° suggests that decomposition
is a result of electron impact rather than thermal
decomposition, since extensive fragmentation has also been
observed for peralkylchromium (IV) compounds when the
electron impact technique is used.34
The mass spectra for the lanthanide cyc10penta-

35 exhibit similar intensities for the peaks listed

dienides
compared to those of (C5H5)38c. The samarium compound has
been shown to be polymeric in the solid state, with bridging

36 In the absence of further evidence

cyc10pentadienides.
this would imply that the scandium, neodymium and ytterbium
compounds are also polymeric. The infrared spectrum and
the recent crystal structure study have shown that the

scandium compound is indeed polymeric with bridging cyclo-

pentadienide groups.

31

One of the primary reasons for attempting to prepare
organoscandium compounds is to determine whether they will
act as catalysts in olefin polymerization reactions. The
most important property of a homogeneous catalyst is a

36 Other criteria that must be

vacant coordination site.
met for a rapid rate of polymerization are: it must

be a kinetically labile system for rapid attachment of
substrate and removal of product, and the catalyst should

be soluble in nonpolar solvents.37

Organoscandium compounds are coordinatively unsaturated
with respect to the scandium and they are initially soluble
in nonpolar solvents. Since Sc(III) has a 3d° configuration,
organoscandium compounds are relatively labile and can act as
catalysts in olefin polymerization reactions. It has been
reported that a solution containing scandium chloride and
phenylmagnesiumbromide does act as a catalyst for
polymerizing ethylene,8 but it has also been reported that
scandium chloride alone could cause olefins such as 1,3-
cyclohexadiene to oligomerize.7 It is therefore necessary
to determine whether a nonhalide containing organoscandium
compound will act as a catalyst.

A benzene solution of styrene and tris(triphenylmethyl)
scandium (III), 100:1, was allowed to stand at room
temperature for 3-5 days and taken to dryness under reduced
pressure. The mass spectrum of the solid remaining had

peaks corresponding to (C6HSCH=CH2)n+’ n = l and 4.

 

32

A solution of 9.1 grams styrene in benzene was added
to 5 m1 of a saturated benzene solution of tris(benzyl)
scandium (III) and allowed to stand at 50° for 3—5 days.

The solvent was removed under reduced pressure and the
solid obtained was identified by its infrared spectrum as
polystyrene.

A similar experiment was attempted using tris(trimethyl-
silylmethyl)scandium (III) that had been isolated as a solid
and was consequently no longer soluble. The infrared
spectrum of the solid obtained indicated that no polystyrene
was present. Presumably tris(trimethylsilylmethyl)scandium
(111) would also act as a catalyst if the extraction solution
used in its isolation is used directly.

The infrared spectra of tris(triphenylmethyl)scandium
(III) and tris(benzyl)scandium (III) did not have any
absorptions attributable to the presence of halides. Hence,
these are the first reactions where a scandium compound with

only organic ligands present acts as a catalyst.

 

CONCLUSIONS AND SUGGESTIONS FOR FURTHER WORK

On the basis of the data presented in this thesis,

the following can be concluded:

1.

Stable organoscandium compounds containing tri-
methylsilylmethyl, triphenylmethyl, pentafluoro-
phenyl, tolyl, benzyl, naphthalyl, and t-butyl
anions have been prepared that contain scandium-
carbon bonds. The bonds in these compounds are
not inherently weak, but are subject to attack
due to the coordinatively unsaturated environment

about the scandium.

Organoscandium compounds are polymeric with
terminal and bridging ligands. Further studies

to determine whether the scandium is 4 or 5
coordinate for various ligands may enable
predictions to be made concerning the choice of one

ligand over another for better catalytic properties.

Soluble organoscandium compounds do act as
catalysts in olefin polymerization reactions in
the absence of any other metals or organometallic

compounds. Further work is needed to determine

33

34

Optimum ratios and reaction conditions. It was
determined that (C6H5CH2)3TiCl polymerized
ethylene faster than (C6H5CH2)4T1,13 hence an
investigation concerning the catalytic properties

of mixed compounds, RnSCXS-n’ should be made.

Organoscandium compounds of uncertain composition
have been prepared. Further characterization of
the colorless liquid and black solid, from the
reactions between scandium metal and triphenyl-
chloromethane and benzylbromide respectively, is
needed to determine what type of environment the

scandium ion is in and what the compositions of these

products are.

A compound of uncertain composition has been isolated
as the product of the reaction between scandium

metal and t-butylbromide. A comparison of the
infrared and mass spectral data for [(C5H5)2TiH]2

and [(CSHS)ZSCH]2 would give some indication of the
strength of the scandium-hydrogen interaction and
would make an accurate assignment of the strong, broad
band at 1575 cm.1 in the infrared spectrum of VIc
possible. The preparation of [(CSHS)ZSCH]2 has not
been reported, but it may be possible to prepare fizfrom
a reaction between [(C5H5)2ScC1]2 and a compound such

as CaHz. If this compound can be prepared, it would

 

35

be worthwhile to determine whether it and similar
organoscandium compounds would act as catalysts

for hydrogenation reactions.

BIBLIOGRAPHY

10.

11.

12.

13.
14.

15.

16.

BIBLIOGRAPHY

V. M. Pletz, Compt. Rend. URSS, 20 (1938) 27.

 

F. A. Cotton, Chem. Rev., (1955) 554.

 

F. A. Hart, A. G. Massey and M. S. Saran, J.
Organometallic Chem., 21 (1970) 147.

 

J. M. Birmingham and G. Wilkinson, J. Amer. Chem. Soc.,
78 (1956) 42.

 

R. S. P. Coutts and P. C. Wailes, J. Organometal.
Chem., 25 (1970) 117.

 

K. D. Smith and J. L. Atwood, Chem. Commun., (1972) 593.

 

E. M. Kirmse and K. J. Zwietasch, 2. Chem., 5 (1965) 393.
U. 8. Pat., 2,907,757 (1959).
(a) J. Chatt and B. L. Shaw, J. Chem. Soc., (1959) 705;

 

(b) ibid., (1960) 718.

M. L. H. Green, "Organometallic Compounds," Methuen,
London, 1968, Vol. II, pp. 220-224.

G. Yagupsky, W. Mowat, A. Shortland and G. Wilkinson,
Chem. Commun., (1970) 1369.

 

M. R. Collier, M. F. Lappert and M. N. Truelock, J.
Organometal. Chem., 25 (1970) C36.

 

U. Giannini and U. Zucchini, Chem. Commun., (1968) 940.

 

G. E. Coates and B. R. Francis, J. Chem. Soc. (A), (1971)
1308.

 

P. S. Braterman and R. J. Cross, J. Chem. Soc. (A),
(1972) 657.

 

R. W. Stotz and G. A. Melson, Inorg. Chem., 11 (1970)
1720.

36

 

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.
32.

33.
34.
35.

37

R. T. Morrison and R. N. Boyd, "Organic Chemistry,”
Allyn and Bacon, Inc., Boston, 1967, Second Ed.,
pp. 395-397.

M. A. Coles and F. A. Hart, Chem. and Ind., (1968) 423.

 

M. Weiner, G. Vogel and R. West, Inorg. Chem., 1 (1962)
654.

 

G. Wilke and H. Schott, Angew. Chem. Internat. Edit.,
5 (1966) 583.

B. Wozniak, J. D. Ruddick and G. Wilkinson, J. Chem.
Soc. (A), (1971) 3116.

F. Hug, W. Mowat, A. Shortland, A. C. Skapski and
G. Wilkinson, Chem. Commun., (1971) 1079.

 

F. Hug, W. Mowat, A. C. Skapski and G. Wilkinson,
Chem. Commun., (1971) 1477.

 

G. A. Melson and R. W. Stotz, Coord. Chem. Rev., 7
(1971) 1330.

R. D. Chambers and T. Chivers, Organometal. Chem. Rev.,
1 (1966) 279.

 

G. B. Deacon and J. C. Parrott, Aust. J. Chem., 24
(1971) 1771.

 

G. B. Deacon and J. C. Parrott, Aust. J. Chem., 25
(1972) 1169.

 

M. J. S. Gynane and I. J. Worrall, J. Organometal. Chem.,
40 (1972) C59.

 

M. J. S. Gynane, L. G. Waterworth and I. J. Worral,
J. Organometal. Chem., 40 (1972) C9.

J. E. Bercaw and H. H. Brintzinger, J. Amer. Chem. Soc.,
91 (1969) 7301.

 

F. A. Hart and M. S. Saran, Chem. Commun., (1968) 1614.

 

T. Maeda and R. Okawara, J. Organometal. Chem., 39
(1972) 87.

 

J. L. Atwood, private communication to G. A. Nelson.
W. Kruse, J. Organometal. Chem., 42 (1972) C39.

J. L. Thomas and R. G. Hayes, J. Organometal. Chem.,
23 (1970) 487.

 

38

36. C. H. Wong, T. Y. Lee and Y. T. Lee, Acta. Cryst., B25
(1969) 2580.

 

37. J. P. Collman, Accounts Chem. Res., 1 (1968) 136.

 

38. J. Halpern, "Advances in Chemistry Series," ed. by
R. F. Gould, ACS Publication, 1968, pp. 1-24.

 

 

'Itu1111111111111113111111111”!willuIII