Triazacyclohexane chromium triflate complexes as precursors for the catalytic selective olefin trimerisation and its investigation by mass spectrometry

Triazacyclohexane Chromium Triflate Complexes as Precursors for the Catalytic

Selective Olefin Trimerisation and its Investigation by Mass Spectrometry

a

b

Randolf D. Köhn* , Alexander G. N. Coxon , Samaphon Chunawat , Callum Heron , Shahram

c

d

Mihan , Catherine L. Lyall , Shaun Reeksting , Gabriele Kociok-Köhn

^aDepartment of Chemistry, University of Bath, Claverton Down, Bath, BA2 7AY, UK

^bCentre for Sustainable Chemical Technologies, University of Bath, Claverton Down, Bath, BA2

7

AY, UK

^cLyondellBasell, Basell Polyolefine GmbH, Industriepark Höchst, 65929, Frankfurt am Main,

GERMANY

d

2

Materials and Chemical Characterisation Facility (MC ), University of Bath, Claverton Down, Bath,

BA2 7AY, UK. https://doi.org/10.15125/mx6j-3r54

th

Dedicated to Prof. John Bercaw on the occasion of his 75 birthday.

Corresponding author. Tel.: +44 1225 383305

*

E-mail address: r.d.kohn@bath.ac.uk

ABSTRACT_____________________________________________________________

A novel set of linear α-olefin trimerisation pre-catalysts is presented. The R TACCr(OTf) (R TAC =

3

1

,3,5-trialkyl-1,3,5-triazacyclohexane) complexes produce an active system after facile activation with

less than 10 equivalents of trialkylaluminium. Isomerisation was observed in many cases, which is

proposed to occur via the π-allyl complex mechanism, though this was minimised by optimisation of

the reaction conditions. The pre-catalysts can be synthesised from R TACCrCl in neat TfOH or by

3

addition of R TAC to Cr(OTf) or better Cr(OTf) (THF) , the synthesis and structure of which is

3

presented here. The use of this highly defined system allowed the identification of 2-methyl-1-hexene

as a side product of activation with AlMe , in agreement with the proposed metallacyclic mechanism.

3

Isomer distribution of the trimer product is similar to R TACCrCl /MAO activated systems and depends

3

mostly on the ligand substituent R. ESI mass spectrometry of an ortho-difluorobenzene solution of the

activated catalyst was analysed at different stages of reaction. A series of signals was observed that

matched those expected for cationic chromium species predicted by the metallacyclic mechanism. In

+

particular, [R TACCrMe ] was observed to form immediately after alkylation with AlMe , while

3

2

3

+

[

R TACCr(olefin) ] (n = 0,1,2,3,4 and olefin= 1-hexene or 1-octene) formed after addition of 1-hexene.

Absence of any detected tetramer and observation of [R TACCr(olefin) ] leads to the conclusion that

3 n

+

3

4

a metallacycloheptane-olefin complex may lead to -H abstraction rather than insertion and may even

+

be required to avoid formation of [R TACCr(olefin) ] which is observed in only very small amounts

3

1

+

indicating a low stability. Chromium(I) complexes [R TACCr(arene)] with the arene solvent can also

3

be observed and their signal intensity indicate a relative stability of o-C H F < cymene < benzene <

6

4 2

PhF, toluene < p-xylene.

_

______________________________________________________________________

1

. Introduction

1

The selective oligomerisation of ethylene, especially tri- and tetramerisation, has become an essential

synthetic step in the production of widely used co-polymers such as Linear Low Density Polyethylene

(LLDPE). The adoption of these processes has been driven by strong growth in demand for 1-hexene

and 1-octene for use as co-monomers. These selective reactions are more expensive than traditional

non-selective oligomerisation routes, such as the ‘Aufbau’ reaction, due to the use of more complex

catalyst systems. Since commercialisation of the poorly defined Phillips Trimerisation Catalyst

considerable research has focused on the preparation of well-defined pre-catalysts capable of similar

efficiencies. [1,2]

Significant progress in recent years has led to the discovery of several systems that are competitive

with the commercialised catalyst in terms of selectivity and activity. We have introduced very active

triazacyclohexane based chromium complexes as pre-catalysts for the polymerisation and selective

trimerisation of ethylene.[3,4] However, these systems are reliant on highly expensive

trimethylaluminium derived activating agents such as various forms of methylaluminoxane (MAO or

dried modified MAO, DMMAO), which severely reduces their economic viability. The activating

agent typically accounts for a significant proportion of overall production costs due to the large excess

required, typically 300-10,000 equivalents relative to the pre-catalyst. In order to find a true

competitor to the Phillips system, which is activated with Et AlCl, a novel activation process is

2

desirable in which the costly use of MAO variants is avoided.

In an attempt to overcome this barrier to commercialisation, increasing research is being performed

into alternative activation methods. Most notably, Bercaw et al. recently described a two-step

activation procedure for the highly active (FI)TiCl ethylene trimerisation catalyst that avoids the need

3

for MAO.[5] Wass et al. explored a different two-step approach, in which the pre-catalyst was

oxidised in order to form the cation prior to alkylation, again avoiding the use of MAO.[6] We have

i

been able to activate R TACCrCl also with [PhNMe H] [B(C F ) ]/Al Bu .[7] In all cases two

3

2

6

5 4

3

distinct activation steps were required in order to achieve activation, which would in turn result in

considerable increases in production costs. To be truly viable, modification of the activation

procedure should not result in additional synthetic steps and should require only a minimal excess of

activating agent while avoiding MAO based species.

While there is extensive literature on selective ethylene trimerisation [1,2] only two systems capable

of selective trimerisation of -olefins have been described, a titanium based system by Bercaw et al.

[5,8] and our triazacyclohexane chromium systems [9,10].

When undertaking a mechanistic investigation into catalytic systems a key challenge is the

characterisation of intermediates within the catalytic cycle. This has proved to be the case for selective

α-olefin trimerisation in just the same way as it has for numerous other systems. The metallacyclic

mechanism originally proposed by Manyik [11] and Briggs [12] for this process has been supported by

numerous experiments that have looked at different aspects of the reaction. For example, the strongest

evidence to date was presented by Bercaw et al. who ruled out a chain growth mechanism with the use

of a deuterium scrambling experiment.[13]

In contrast, direct observation of the proposed intermediates has seldom been reported due to the

inherent difficulties associated with characterisation of such short-lived species. In 2009, the first

NMR derived evidence for the existence of the proposed intermediates was presented for a tantalum

based system.[14] However, tantalum systems demonstrate very low trimerisation activities in

comparison to chromium based systems, which are therefore of more interest but cannot be

investigated in the same manner due to their paramagnetic nature. For this reason, only limited

spectroscopic analysis of chromium-based intermediates has been possible to date. In 1997 a

chromacyclopentane species was synthesised directly by Jolly et al. [15] and successfully crystallised,

confirming the proposed intermediates are viable but under very different conditions to α-olefin

trimerisation. More recently, Bercaw et al. conducted an extensive UV-Vis and EPR spectroscopy

investigation into ethylene trimerisation.[16] The results strongly support the proposed alkylation of

the pre-catalyst during activation and provide considerable information on the oxidation state of the

chromium species present.

Despite these advances, direct observation of chromacyclic and other proposed intermediates has

remained elusive. One method for the identification of such species that has, until now, remained

2

unexplored is the use of electrospray mass spectrometry (ESIMS) although it has been successfully

applied for other catalytic systems [17] and is even used to analyse MAO solutions[18]. This

technique is a prime candidate for investigation of α-olefin trimerisation because the active

intermediates are known to be cationic and are therefore readily detectible under the right conditions.

We describe herein an investigation into the substitution of chloride ligands, almost universally used

in defined trimerisation pre-catalysts, with triflate anions to produce R TACCr(OTf) species. The

3

activation pathways available to these pre-catalysts were then explored in relation to the trimerisation

of linear α-olefins. Many of the proposed cationic catalytic intermediates were detected by the use of

in situ mass spectrometry.

2

. Experimental

2

.1. Materials

All manipulations of air/moisture sensitive compounds were carried out under an atmosphere of argon

or nitrogen using standard Schlenk techniques or in an argon atmosphere in a Saffron glove box. All

reagents were obtained from major suppliers. Synthesis of complexes 1a,b has been described

previously[19] and crystals could be grown by evaporation of dichloromethane/acetonitrile solutions.

CrCl (THF) [3] and complexes 1c,d[10] were prepared in analogy to methods described previously.

3

1

,2-difluorobenzene was dried with molecular sieves, degassed, vacuum transferred and stored over

molecular sieves in the glove box. Other dry solvents were obtained from the Innovative Technology

Solvent Purification System (SPS).

2

.2. Instrumentation and Characterisation Procedures

1

NMR spectra were obtained on a Bruker AVII+ 500 NMR spectrometer [500MHz ( H), 125MHz

(¹³

C), 470MHz ( F)] at 298K. All C spectra are H decoupled unless otherwise stated. Shift values

19

13

are quoted in ppm relative to TMS or set internal solvent signals. Coupling constants and line widths

at half height (W) are quoted in Hz. NMR spectra are often taken without the use of deuterated

solvent and referenced to the solvent signal externally referenced to TMS. The shift of 1,2-

1

13

19

difluorobenzene (PhF ) was set to 6.976 ( H), 117.98 ( C) and -139.4 ( F) ppm determined for neat

2

solvent relative an added trace of TMS. Effective magnetic moments were measured using the Evans

method and corrected for the diamagnetic contribution.[20,21]

Routine mass spectra were obtained using a micrOTOF electrospray time-of-flight (ESI-TOF) mass

spectrometer (Bruker Daltonik) using acetonitrile as a solvent collecting data in positive mode. The

peaks quoted correspond to the calculated exact mass, the correct isotope patterns are present where

.

the peak is quoted. For metal complexes a small amount of Me N HCl was added to obtain a

3

+

predictable ion formation (M+C H N ) unless otherwise stated.

3

10

Elemental analysis was performed on an Exeter Analytical CE440 Elemental Analyzer by Mr. A. K.

Carver in the University of Bath, Department of Chemistry, or externally by London Metropolitan

University Elemental Analysis Service, UK.

GCMS analysis was performed with an Agilent 7890B with Agilent 5977A MSD and FID detectors.

A DB-FFAP column 30 m in length, with a diameter of 0.250 mm and a 0.25 µm film thickness was

used in all cases. A ramp rate of 3°C per minute was used from 40°C to 350°C.

3

2

.3. X-ray Crystallographic Analysis

Intensity data for all structures were collected at 180(2) K on a STOE IPDS (1b), and at 150K on a

Nonius KappaCCD diffractometer (3 and 4b) or a RIGAKU SuperNova (1a, 4a, CrCl (THF) )

3

equipped with an Oxford Cryostream. Data were processed using the STOE- [22], Nonius-[23] and

RIGAKU Software [24]. For all structures a symmetry-related (multi-scan) absorption correction had

been applied. Crystal parameters and details on data collection, solution and refinement for the

.

complexes are provided in Table 2 and for [Cr(H O) ](OTf) 3H O in Table S2. Structure solution,

2

6

3

2

followed by full-matrix least squares refinement was performed using the WINGX-1.80 suite of

programs throughout.[25]

2

.4. Preparation

.4.1. Synthesis of (Pr CHCH ) TACCrCl (1c)

2

2 3

3

.4.1.1. Synthesis of methyl 2-cyano-2-propylpentanoate (MeOC(=O)C(Pr) CN)

2

Sodium methoxide (12.00 g, 222 mmol) was dissolved into 100 mL of dry methanol under nitrogen,

with the evolution of a considerable amount of heat. Half (50 mL) of the base solution was added

drop-wise to ethyl cyanoacetate (11.7 mL,110 mmol, 12.44 g) and the solution left to stir for 15

minutes, leading to a yellowing of the mixture. 1-Bromopropane (10 mL, 110 mmol, 13.54 g) was

then added. The system was kept under nitrogen for three hours with vigorous stirring, leading to a

cloudy orange solution. The second half of base solution was then added followed by another 10 mL

of 1-bromopropane and the mixture left stirring overnight. The vivid red suspension was allowed to

cool to room temperature before being decanted from the white precipitate. The salts were extracted

with petroleum ether (3 x 100 mL) and the solvent of the combined extractions removed under

-2

reduced pressure. The resulting liquid was vacuum transferred at 10 mbar and 300 ˚C using a water

1

bath to cool the collection vessel. 16.8 g (92 mmol, 84%) of a clear liquid was collected. H NMR

(

1

CDCl ): δ = 3.79 (3H, s, -OCH ), 1.87 (2H, t of d, J = 13.1/4.8, -CH Et), 1.77 (2H, t of d, J =

2.8/4.0, -CH Et), 1.58 (2H, m, -CH Me), 1.33 (2H, m, -CH Me), 0.96 (6H, t, J = 7.3, -CH ). C

2 2 2 3

3

2

1

3

NMR (CDCl ): δ = 169.09 (-C=O), 118.56 (-CN), 52.49 (-OCH ), 49.35 (-C(Pr) CN), 38.99 (-

3

2

CH Et), 18.32 (-CH Me), 13.06 (-CH ).

2

3

2

.4.1.2. Synthesis of 2-propylpentanenitrile (Pr CHCN)

2

De-ionised water (2.5 mL, 2.48 g, 138 mmol) was dissolved into 100 mL of DMSO. Lithium chloride

11.73 g, 277 mmol) was then dissolved into the mixture (very exothermic!) and the system was put

(

under nitrogen. Once dissolved, MeOC(=O)C(Pr) CN (16.81 g, 92 mmol) was added and the mixture

2

heated to 150 ˚C whilst stirring overnight. The resulting brown solution was decanted off the yellow

solids and 100 mL de-ionised water was added. The aqueous solution was then extracted with pentane

(100 mL), which was washed with further de-ionised water (3 x 100 mL). The water washings were

extracted with pentane (25 mL) which was then combined with the original pentane solution. The

solvent was removed under reduced pressure to give a yellow oil. This was vacuum transferred (~150

4

-2

1

˚

C, 10 mbar, liquid nitrogen bath) to give 8.90 g (71 mmol, 77%) of a colourless oil. H NMR

1

3

(

CDCl ): δ = 2.54 (1H, m, -CHCN), 1.42-1.62 (8H, m, -CH CH Me), 0.96 (6H, t, J = 7.5, -CH ). C

3

2

3

NMR (CDCl ): δ = 121.70 (-CN), 34.04 (-CH Et), 30.80 (-CHCN), 20.05 (-CH Me), 13.11 (-CH ).

3

2

3

2

.4.1.3. Synthesis of 2-propylpentan-1-amine (Pr CHCH NH )

2 2 2

AlCl (10.69 g, 80.1 mmol) was carefully dissolved into 400 mL of ice-cooled dry Et O under a

3

2

nitrogen atmosphere. LiAlH (9.07 g, 240.0 mmol) was then slowly added to avoid excessive boiling

4

of the solvent during the highly exothermic reaction. The grey suspension was stirred for an hour at

ambient temperature. A solution of Pr CHCN (8.90 g, 71.0 mmol) in dry Et O (50 mL) was added

2

drop-wise over 30 minutes and left to stir for half an hour. Significant bubbling was observed

on addition of each drop due to the highly exothermic nature of the reaction. After complete addition

the mixture was cooled in an ice bath and hydrolysed by the careful addition of 50 mL of water, 50

mL of 20% NaOH in water, 100 mL water and another 100 mL of 20% NaOH solution, in that order.

The mixture was stirred vigorously for an hour to ensure complete hydrolysis. The suspension became

white over this period. The suspension was left to settle before decanting off the clear Et O solution,

2

followed by extractions with Et O (3 x 150 mL). The solvent was removed from the combined

2

-2

extracts under reduced pressure. The product was vacuum transferred (2 x 10 mbar, ~150 ˚C, ice

1

bath) to give 7.05 g (54.6 mmol, 77%) of a colourless oil. H NMR (CDCl ): δ = 2.60 (2H, d, J = 5.3,

3

-

CH NH ), 1.18-1.38 (9H, m, -CHCH CH Me), 1.01 (2H, broad s, -NH ), 0.90 (6H, t, J = 7.6, -CH ).

2 2 2 2 2 3

1

3

C NMR (CDCl ): δ = 44.70 (-CH NH ), 39.99 (-CHCH NH ), 33.43 (-CH Et), 19.40 (-CH Me),

3

2

1

3.89 (-CH₃).

2

.4.1.4. Synthesis of 1,3,5-tris(2-propylpentyl)-1,3,5-triazinane ((Pr CHCH ) TAC)

2

2 3

Paraformaldehyde (1.65 g, 55.0 mmol) was added to Pr CHCH NH (7.05 g, 54.6 mmol) in 100 mL

2

toluene. The suspension was stirred for 3 days. During this time the particles were consumed and

several droplets of water formed, indicating a complete reaction. The water was removed via

azeotropic distillation of toluene (3 x 50 mL) and the remaining solvent removed under reduced

-2

pressure. The mixture was further dried at 10 mbar and 150 ˚C for 16 hours. The yield after drying

1

was 6.44 g (15.2 mmol, 90%) of a colourless viscous oil. H NMR (CDCl ): δ = 3.26 (6H, broad s, -

3

NCH N-), 2.26 (6H, d, J = 7.2, -CH TAC), 1.43 (3H, m, -CHCH TAC), 1.18-1.38 (24H, m, -

2

1

3

CH CH Me), 0.88 (18H, t, J = 7.2, -CH ). C NMR (CDCl ): δ = 75.48 (-NCH N-), 57.38 (-

2

3

2

CH TAC), 35.77 (-CHCH TAC), 34.97 (-CH Et), 19.98 (-CH Me), 14.66 (-CH ). Anal. Calc. for

2

3

C H N (%): C, 55.71; H, 9.87; N, 7.22. Found: C, 55.59; H, 9.99; N, 7.33.

2

7

57

3

2

.4.1.5. Synthesis of (Pr CHCH ) TACCrCl (1c)

2 2 3 3

CrCl (THF) (3.71 g, 10.0 mmol) was added to (Pr CHCH ) TAC (4.28 g, 10.1 mmol) under an argon

3

2

2 3

atmosphere. The reagents were then dissolved into 20 mL dry DCM and the solution stirred for 2

days. The resulting purple solution was separated by column chromatography, eluting with DCM. The

solvent of the purple fractions was removed under reduced pressure and the resulting solid dried for

1

3

2

4 hours under high vacuum, yielding 4.97 g (8.5 mmol, 85%) of a purple solid 1c. C NMR (2w% in

5

o-C H F ) : δ = 51.3 (W = 130, -CH Et), 23.9 (W = 136, -CH Me), 16.8 (W = 127, -CH ), -39.2 (W =

6

4

2

3

+

8

66, -CHPr ). ESI-MS (m/z) [C H Cl CrN .C H N] : Calculated exact mass: 640.3836; found:

2 27 57 3 3 3 10

6

40.3875. Anal. Calc. for C H Cl CrN (%): C, 39.04; H, 6.23; N, 4.55. Found: C, 38.79; H, 6.40;

27 57 3 3

N, 4.36.

i

2

.4.2. Synthesis of ( Bu CH) TACCrCl (1d)

2 3 3

i

2

.4.2.1. Synthesis of 2,6-dimethylheptan-4-amine ( Bu CHNH )

2 2

Ammonium acetate (35.02 g, 454 mmol) was dissolved into MeOH (100 mL) under nitrogen before

addition of 10 mL (8.08 g, 56.8 mmol) of 2,6-dimethylheptan-4-one. This was stirred for 90 minutes

at room temperature. Sodium cyanoborohydride (2.58 g, 41.0 mmol) was dissolved into MeOH (20

mL) and added to the reaction mixture which was left stirring overnight at 30 °C. The reaction vessel

was placed in a cold water bath and a liquid nitrogen cooled trap was added. 10 mL of 32% aqueous

HCl was added dropwise over the course of 30 minutes, leading to a highly exothermic reaction and

considerable white precipitate. CARE: Hydrolysis of the cyanoborohydride leads to the evolution of

hydrogen cyanide which should be caught in the trap and disposed of properly. The MeOH was

removed under reduced pressure to give a white slurry. This was extracted with pentane (3 x 50 mL)

and Et O (3 x 50 mL) and the solvents again removed to give a clear oil. This was dissolved in Et O

2

(50 mL) before addition of 5.32 g (133 mmol) of NaOH. The mixture was stirred for 2 hours before

addition of 5 mL of water to form two phases and dissolve the solids. The organic phase was removed

and collected and the aqueous phase extracted with Et O (3 x 50 mL) and pentane (2 x 50 mL). The

2

solvent was then removed at reduced pressure. The resulting oil was distilled at 160 °C and 300 mbar

1

to give 5.44 g (38 mmol, 67%) of a colourless non-viscous oil. H NMR (CDCl ): δ = 2.83 (1H,

3

quintet, J = 6.6, -CHNH ), 1.73 (2H, nonet, J = 6.7, -CHMe ), 1.17 (4H, m, -CH CHMe ), 0.90 (12H,

2

1

3

t, J = 6.4, -CH ). C NMR (CDCl ): δ = 47.99 (-CH CHMe ), 46.24 (-CHCH NH ), 24.44 (-CHMe ),

3

2

3.24 (-CH ), 21.79 (-CH ).

3

i

2

.4.2.2. Synthesis of 1,3,5-tris(2,6-dimethylheptan-4-yl)-1,3,5-triazinane (( Bu CH) TAC)

2

3

i

Paraformaldehyde (1.03 g, 34.4 mmol) was added to Bu CHNH (4.93 g, 34.4 mmol) in 100 mL

2

toluene. The suspension was stirred for 2 days. During this time the particles were consumed and

several droplets of water formed, indicating a complete reaction. The water was removed via

azeotropic distillation of toluene (3 x 50 mL) and the remaining solvent removed under reduced

pressure. The mixture was further dried at 50 mbar and 50 ˚C for 6 hours. The yield after drying was

.00 g (25.7 mmol, 75%) of a colourless viscous oil. NMR spectroscopy indicated a mixture of imine

4

and triazacyclohexane was formed at a ratio of 3:2.

1

Imine: H NMR (CDCl ): δ = 7.35 (1H, d, J = 17.4, =CH-cis), 7.14 (1H, d, J = 17.4, =CH-trans), 3.06

3

(

1H, m, -CHN=CH ), 1.24 (2H, m, -CHMe ), 1.05 (4H, d of t, J = 13.7/7.0, -CH CHMe ), 0.86/0.83

2 2 2 2

1

3

6H/6H, d/d, J = 6.5/6.5 -CH ). C NMR (CDCl ): δ = 250.91 (-N=CH ), 68.63 (-CHN=CH ), 40.51

3

2

-CH CHMe ).

2

1

Triazacyclohexane: H NMR (CDCl ): δ = 3.48 (6H, s, -NCH N-), 2.62 (3H, quintet, J = 6.9, -CHN),

1

3

2

1

3

.69 (2H, m, -CHMe ), 1.46 (4H, m, -CH CHMe ), 0.88/0.87 (6H/6H, d/d, J = 6.8/6.8, -CH ). C

2 2 2 3

6

1

3

NMR, CDCl : δ = 67.39 (-NCH N-), 56.46 (-CHN), 45.36 (-CH CHMe ). Six further C peaks were

3

2

observed which represent (-CHMe ) and the pairs of (-CH ) for each molecule. These were not

2

3

assigned but were recorded at 24.98, 24.24, 23.59, 23.13, 22.88 and 21.54. Anal. Calc. for C H N

3

0

63

(%): C, 57.73; H, 10.17; N, 6.73. Found: C, 57.66; H, 10.28; N, 6.86.

i

2

.4.2.3. Synthesis of ( Bu CH) TACCrCl (1d)

2

3

i

CrCl (THF) (3.23 g, 8.7 mmol) was added to the ( Bu CH) TAC / Bu CHNCH mix (4.00 g) under

3

2

3

2

an argon atmosphere. The reagents were then dissolved into 20 mL dry DCM and the solution stirred

for 2 days. The resulting purple solution was separated by column chromatography, eluting with

DCM. The solvent of the purple fractions was removed under reduced pressure and the resulting solid

1

3

dried for 24 hours at 50 °C under high vacuum, yielding 3.81 g (6.1 mmol, 71%) of a purple solid. C

NMR (CDCl ): δ = 37.64 (W = 111, -CHMe ), 28.55 (W = 91, -CH ), 20.16 (W = 32, -CH ), -36.4

3

2

3

i

+

(

W = 1490, -CH Pr). ESI-MS (m/z) of [C H Cl CrN .C H N] : Calculated exact mass: 684.4276;

2 30 63 3 3 3 10

found: 684.4388. Anal. Calc. for C H Cl CrN (%): C, 41.07; H, 6.58; N, 4.35. Found: C, 40.86; H,

3

0

63

3

6

.72; N, 4.33.

2

.4.3. Synthesis of Cr(OTf) 2

3

Trifluoromethanesulphonic acid (32.64 g, 19.2 mL, 218 mmol) was dissolved into de-ionised water

75 mL) by drop-wise addition due to the highly exothermic reaction. Considerable fuming was

(

observed on addition of each drop. Granulated chromium (4.16 g, 80 mmol) was then added and

stirred vigorously with no initial reaction observed. The solution was heated to 90 °C in the presence

3

+

of oxygen and over the course of 48 hours became blue/violet in colour, indicative of [Cr(H O) ]

2

6

formation. The solution was filtered through sintered glass to remove the excess chromium before the

solvent was removed under reduced pressure to give a viscous blue oil. Further drying of some of the

viscous oil in a desiccator with P O for a week resulted in highly hygroscopic blue crystals of

4

10

.

[

Cr(H O) ](OTf) 3H O suitable for X-ray crystallography. The oil was heated to 100 °C under

2 6 3 2

reduced pressure for one hour, leading to a pale green solid. Further heating to 200-300 °C in a

-2

regeneration oven under high vacuum (10 mbar), led to a dark green solid and a considerable

quantity of water being collected over the course of an hour. The product demonstrated strong

hydrophilicity, reverting back to a blue/green oil within seconds of exposure to air. The solid was

ground by pestle and mortar under an argon atmosphere to give an extremely fine powder, which was

further dried, under high vacuum. After drying, 32.2 g (64.4 mmol, 89%) of a green powder 2 was

collected. Anal. Calc. for C CrF O S (%): C, 7.22; H, 0.00. Found: C, 7.12; H, 1.17. This

3

9

9 3

experimental value would suggest 0.4 H O (C%) or 2.9 H O (H%) per chromium and may be due

2

some residual water and/or uncertainty in the analytical method including sample preparation of the

highly hygroscopic product.

Alternative synthesis from CrO : 10g water is cooled in an ice bath and 17.3g triflic acid (M=150,

3

0

.115mmol) added slowly (warm!). Then 3.76g CrO (M=100, 0.0376mmol) are added, residues

3

i

washed in with a few more ml water, followed by ca 20ml PrOH (slow, gets hot!). After one hour, the

deep blue solution is concentrated on the rotary evaporator (50mbar, 50 ºC) for 2 hours and then at

high vacuum with bridge and LN2 cooled Schlenk flask. When the solution solidifies the temperature

is slowly raised as above while the solids turn light green. It was found that (probably due to organic

residues) this solid tends to give dark products on heating above 200 ºC.

7

2

.4.4. Synthesis of Cr(OTf) (THF) 3

3 3

A Schlenk flask with 555mg of Cr(OTf) (2) (M=499, 1.11mmol) is evacuated and THF (from NaK,

3

ca 10ml) is condensed in. The dark solution is stirred for 4hrs (occasional warming in hot water bath

as it becomes more viscous) and then all volatiles are pumped off (high vacuum, 40 ºC). The sticky

solids are dissolved in ca 10 ml of o-difluorobenzene (slow to dissolve and very viscous at first). After

2

hrs of stirring, the now more fluid green solution is filtered and then left in a -20 C freezer for two

days. Some 3 crystallised out. The solution is decanted and the residue washed with two 5ml portions

of hexane. The combined solution, washings and further 20ml of hexane are left in the freezer to

obtain additional 160mg product. The combined solids are dried in vacuo to give 360mg of a light

green solid 3 (for M=715.5: 0.50mmol, 45%). Anal. calcd for C H CrF O S (%): C, 25.18; H 3.38.

1

5

24

9

12 3

Found: C, 25.09; H, 2.85.

NMR: 34.6mg are dissolved in 1438mg PhF₂:

1

H: -0.51 (-CH of THF, W=655Hz), -CH too broad to be observed;  =3.74 (Evans method)

2

eff

B

1

3

C: -84.5 (-CH of THF, W=3300Hz), -CH too broad to be observed

2

¹⁹F: -74.4 (3F, W=510Hz), -77.7 (6F, W=550Hz)

2

.4.5. Synthesis of Cy TACCr(OTf) 4a

3 3

Route A: Cy TACCrCl (1a) (1.982 g, 4.03 mmol) is weighed out into a Schlenk tube, and then

3

removed from the glove box and connected to a Schlenk flask containing triflic acid (TfOH). The

Schlenk tube is then placed in liquid nitrogen and the TfOH condensed over. After condensing over

all the acid, it is condensed back over to yield a sticky dark blue solid which is then washed with

diethyl ether 10 times, by vacuum transferring it over then decanting the washings back across, until

the ether appeared colourless. The dark blue solid is then dried on a vacuum line to yield 3.083 g of

4

3

a (Yield = 92%). Anal. Calc. for C H CrF N O S (%): C, 34.62; H, 4.72; N, 5.05. Found: C,

24 39 9 3 9 3

4.53; H, 4.89; N, 4.92. UV/Vis(KBr, cm ): 15120, 21500.

-1

Route C: 34.6mg of Cr(OTf) (THF) (0.0484mmol) are dissolved in 1438mg of o-difluorobenzene in

3

an NMR tube and Cy TAC (M=333.56, 16mg, 0.048mmol) is added – no visible change at first! After

3

1

hr turquoise crystals cover the walls of the NMR tube. Shaken down the much lighter solution is

taken to NMR: nearly 2.5eq THF are shown with little (<0.002mmol, 0.04eq) Cy TAC left.

3

2

.4.6. Synthesis of Bz TACCr(OTf) 4b

3 3

Route B: 20mg Bz TAC (0.056mmol) and 28mg Cr(OTf) (0.056mmol) are stirred with 945mg of o-

3

difluorobenzene. After 4hrs, the largely dissolved mixture is filtered and the filtrate placed into vial

with hexane. The filter is extracted with further two 1ml portions of o-difluorobenzene and placed in

another vial with hexane and a final o-difluorobenzene extract is placed into a freezer to give crystals

good enough for X-ray crystallography. Further product is obtained from the hexane vials.

Route A: 238mg of Bz TACCrCl (1b) (0.46mmol) are dissolved in 3.5g TfOH which is subsequently

3

distilled off. The residue is washed with Et O (5ml) to give 356mg (91%) of 4b, m.p. 207-208C. Anal.

2

8

Calc. for C H CrF N O S (%): C, 37.85; H, 3.18; N, 4.90; S, 11.23; Cr, 6.07. Found: C, 36.48; H,

2

7

27

9

3

9 3

-1

3

.13; N, 4.63; S, 11.04; Cr, 6.65. UV/Vis(KBr, cm ): 16660, 23370.

2

.4.7. Synthesis of (Pr CHCH ) TACCr(OTf) (4c)

2

2 3

3

(

Pr CHCH ) TACCrCl (1c) (1.97 g, 3.4 mmol) was evacuated and cooled with liquid nitrogen and

2 2 3 3

approximately 10 mL of triflic acid was condensed onto the purple solids across all-glass linkages. The

acid was then allowed to melt and the system opened up to a trap cooled with liquid nitrogen to catch

any hydrogen chloride given off. Once the acid melts a fast reaction proceeds with considerable

-2

effervescence. Once the reaction was complete the acid was distilled off at 40 °C and 10 mbar over

the course of a day. The dark blue solids were then washed twice with 25 mL of dry Et O under a

2

nitrogen atmosphere to remove any further acid. Subsequent drying at reduced pressure led to 2.73 g

1

3

(

1

3.0 mmol, 88%) of a pale blue solid 4c. C NMR (1w% in o-C H F ): δ = 42.32 (W = 83, -CH Et),

6 4 2 2

1

9

9.21 (W = 41, -CH Me), 13.57 (W = 37, -CH ), -43.9 (W = 5100, -CHPr ). F NMR (o-C H F ): δ =

2 3 2 6 4 2

-

45.56 (W = 3300). Anal. Calc. for C H CrF N O S (%): C, 39.04; H, 6.23; N, 4.55. Found: C, 38.81;

30 57 9 3 9 3

H, 6.38; N, 4.32.

i

2

.4.8. Synthesis of ( Bu CH) TACCr(OTf) (4d)

2 3 3

(ⁱBu CH) TACCrCl (1d) (1.06 g, 1.7 mmol) was cooled with liquid nitrogen and approximately 10 mL

2

3

of triflic acid was condensed onto the purple solids across all- glass linkages. The acid was then allowed

to melt and the system opened up to a trap cooled with liquid nitrogen to catch any hydrogen chloride

given off. Once the acid melts a violent reaction proceeds with considerable effervescence. Once the

-2

reaction was complete the acid was distilled off at 40 °C and 10 mbar over the course of a day. The

dark blue solids were then washed four times with 25 mL of dry Et O under a nitrogen atmosphere to

2

remove any further acid. Subsequent drying at reduced pressure led to 1.43 g (1.5 mmol, 87%) of a pale

1

3

blue solid 4d. C NMR (o-C H F ): δ = 37.3 (W = 320, -CHMe ), 20.3 (W = 120, -CH ), 19.6 (W =

1

6

4

2

3

i

19

50, -CH ), -30.1 (W = 2740, -CH Pr). F NMR (o-C H F ): δ = -50.3 (W = 2400). Anal. Calc. for

3 2 6 4 2

C H CrF N O S (%): C, 41.07; H, 6.58; N, 4.35. Found: C, 40.90; H, 6.75; N, 4.34.

3

63

9

3

9 3

2

.5. Typical catalysis of 1-hexene trimerisation

.5.1. NMR observation:

1

Trimerisation was observed by H NMR and quantified by integration (relaxation delay of at least 20s

was used to ensure full relaxation) of four regions: A is the integral of the =CH- signal of 1-hexene at

about 5.8 ppm, B is the integral of the remaining olefinic signals at 4.5-5.5 ppm, C is the integral of the

whole hexene region from 6.0-0.5 ppm and D is the Al-CH region at around 0.0-0.5 ppm (not including

2

i

Al-Me). Integral C is set to 1200+x with x as 1.5D for AlEt or 3.5D for Al Bu (taking into account

3

the remaining signals of AlR in the hexene region). As all olefins have 2H per molecule in the region

3

B, it can be shown that the % 1-hexene remaining is A, the % trimer formed is 0.75(200-B) and the %

isomers formed is 100-A-0.75(200-B).

Cy TACCr(OTf) (4a) (5.1mg, 0.0061mmol) is suspended in 315.7mg of 1-hexene (3.76mmol, 614eq)

3

and 394 mg of o-difluorobenzene (3.45mmol, 564eq). Then Me Al (9 mg weighed into a pipette

3

(

0.125mmol, 20eq) is added. All dissolves first to a blue/turquoise and then within minutes to a light

1

9

green solution and is transferred into an NMR tube. F NMR after 25min shows 2.7eq OTf at -76ppm.

9

NMR after 3hrs shows 91% hexene left, 9% trimer (55 turnover per Cr) and no isomer (0.17eq CH at

4

0

.15ppm, 0.062mmol AlMe or about half of added AlMe at -0.22ppm, additional Al-Me at 0.2

3 3

(0.018mmol Me or 3.0eq per Cr) and 0.0ppm (0.011mmol or 1.9eq per Cr)), . Thus, very slow progress.

The colour had changed to more blueish green after 3 hrs.

After 24hrs: 30.4% hexene, 67.6% trimer and 2.0% isomer (5% conversion in last hour!)

After 48hrs: 3.8% hexene left with 89.2% trimer and 7% isomer.

i

2

.5.2. Comparison of activation with AlMe , AlEt and Al Bu :

3

(

Pr CHCH ) TACCr(OTf) (4c) (16.3mg, 0.0177mmol) is dissolved in 3.10g of o-difluorobenzene.

2 2 3 3

1

.04g each (0.0059mmol Cr) of this solution is added to the following:

A: AlMe (46mg, 0.64mmol, 108eq) and 1-hexene (620mg, 7.4mmol, 1250eq); NMR next day: 91.9%

3

1

-hexene, 6.1% trimer, 2.0% isomers;after 5d: 84.5%, 7.4%, 7.2% - 0.0016mmol or 27% of 3d at 4.23d

1

3

(3H, ring), 3.66d (3H, ring), 2.69d (6H, -CH ) – C by HSQC and HMBC:74 (ring), 57 (-CH ) and

2

3

5ppm (-CH) as visible signals of the triazacyclohexane transferred to aluminium.

1

13

Additional observable signals in H: 0.15ppm (CH , about 2eq per Cr; HSQC C: -4.5ppm); apart

4

from a larger signal at -0.30ppm for Al-Me, an additional set of signals at 0.18dd (5.1 and 14.0 Hz)

1

3

and -0.01dd (8.0 and 14.0 Hz) is detected for -CH (HSQC C: 21.7ppm (C )) of about 6eq per Cr

2

1

3

of Al-CH CHMeBu; further C signals from long range HMBC: 23.7 (Me-C ), 30.3 (C ) and 41.8

2

7

2

(

C ) ppm [26]

3

B: AlEt (83mg, 0.73mmol, 124eq) and 620mg 1-hexene (620mg, 7.4mmol, 1250eq); NMR next day:

3

5

4.1% 1-hexene, 38.1% trimer, 7.8% isomers; after 4d: 4.9%, 59.3%, 35.9% - 0.0015mmol or 25% of

1

3

d at 4.28d (3H, ring), 3.57d (3H, ring), 2.69d (6H, -CH ) – C by HSQC and HMBC:75 (ring), 58

2

(

-CH ) and 35ppm (-CH) as visible signals of the triazacyclohexane transferred to aluminium.

2

1

13

1

Additional observable signals: ethane can be detected as H- C-HSQC cross peak at 0.81ppm ( H) and

1

3

13

7

1

.18ppm ( C), apart from a large set of signals for AlEt at 1.10t and 0.31q ( C-HSQC at 9.1 and

3

.3ppm, respectively),

i

C: Al Bu (120mg, 0.61mmol, 103eq) and 1-hexene (578mg, 6.9mmol, 1160eq); NMR after 4 hrs:

3

5

.5% 1-hexene, 29.5% trimer, 65.0% isomers; after 4d: 0.7%, 29.3%, 70.0% - 0.0037mmol or 63% of

1

3

d at 4.31d (3H, ring), 3.55d (3H, ring), 2.71d (6H, -CH ) – C by HSQC and HMBC:75 (ring), 58

2

(

-CH ) and 35ppm (-CH) as visible signals of the triazacyclohexane transferred to aluminium.

2

1

i

13

Additional observable signals in H: 0.31d for -CH of Bu-Al (HSQC C: 25.7ppm (C ) – HMBC:

2

1

3

13

2

8.0 (C and C )) and an additional signal at 0.45t (HSQC C: 13.4 (CH ) with further C signals

2 3 2

i

from long range HMBC: 26.1 (C ) and 36.5 (C ) ppm matching those observed for Bu Al-hexyl [27].

2

3

2

Tests for mixed AlR₃:

Et AlMe: 20mg of AlMe (0.277mmol), 63mg of AlEt (0.552mmol), 1786mg of o-difluorobenzene

2

3

1

(

1

15.65mmol) and 1560mg of 1-hexene (18.5mmol) are mixed and checked by H NMR: based on

5.65mmol o-difluorobenzene it contained 18.0mmol of hexene (17.6mmol 1-hexene), 0.49mmol of

AlEt at 1.03t and 0.14q ppm, 0.26mmol of AlMe at -0.004 (0.70mmol bridge Me), -0.13 (0.07mmol)

3

and -0.8 (0.01mmol). 590mg of the mixture (0.048mmol of AlMe , 0.095mmol of AlEt , 2.692mmol

3

of o-difluorobenzene, 3.188mmol of hexene) is added to 7.5mg of 4c (0.0081mmol). NMR after 10hrs:

1

0

3

0

4

7% hex, 58% trim, 5% isom, 0.0010mmol/17% TACAl; after 3d: 11% hex, 71% trim, 18% isom,

.0015mmol/25% TACAl; after 9d: 0.3% hex, 74.6% trim, 25.1% isom, 0.0028mmol/47% TACAl at

.23, 3.63 and 2.69ppm. After 18d, the contents of the NMR tube is distilled in air to isolate the volatile

solvent and hexenes: contains 15w% hexenes, 0.7w% hexane and hexene isomer: 67% trans-2-hexene,

1

7% cis-2-hexene, 14% trans-3-hexene, 2% cis-3-hexene.

i

Bu AlMe: Preparation of a 1-hexene/o-difluorobenzene solution of Bu AlMe: 57mg of Al Bu

3

2

(

0.287mmol), 10.5mg of AlMe (0.1457mmol), 3496mg of 1-hexene (41.54mmol) and 2096mg of o-

3

difluorobenzene (18.37mmol) are mixed and checked by NMR: contains relative to 18.37mmol of o-

i

difluorobenzene: 40.8mmolof hexene (39.8mmol 1-hexene, rest isomers), 0.254mmol of Bu Al and

3

1

i

0

.139mmol AlMe ; H NMR of Al Bu at 0.143d, AlMe at 0.022br in 4:3 ratio.

3

i

About 4mg of 4c are dissolved in 650mg of this solvent mix (0.033mmol/9 eq Al Bu ; 0.017mmol/5 eq

3

AlMe ; 4.77mmol/1300 eq hexene) and observed by NMR. Proceeds well – after 4d: 13.4% hexene,

3

1

9

6

2

7

9.2% trimer and 17.5% isomer – F NMR after 5d shows 0.0407mmol OTf at -77.7ppm relative to

.11mmol of o-difluorobenzene, thus, 0.0036mmol Cr – thus, [Cr] = 3.7mM. After 23d: 11% hexene,

3% trimer and 16% isomer, isobutene (4.63 and 1.64ppm) quickly rises to about 0.002 mmol and then

more slowly to 0.008mmol at the end (0.5 to 2eq per Cr) – thus nearly 10% of AlR is hexyl. Very little

TACAl, about 0.0002mmol or 6% of Cr.

2

1

.5.3. Bulk synthesis:

-hexene (30 g, 0.357 mol) was weighed out in a 500 mL round bottomed flask, triethyl aluminium

(

AlEt ) (0.163 g, 1.43 mmol) was then added to the flask and the solution was allowed to stir for 1

3

hour. In a separate flask Cy TACCr(OTf) (4a) (0.1981 g, 0.238 mmol) was weighed out along with

3

AlMe (2.45 wt. % in toluene, 2.099 g, 0.714 mmol) and o-difluorobenzene was added as a solvent for

3

the catalyst dissolution (11.568 g, 0.101 mol) and the solution left to stir for 1 hour. After leaving both

solutions to stir, they were combined in a 500 mL round bottomed flask and left to stir overnight. By

NMR, the conversion to trimer was found to be 26.2 % from the original 1-hexene added, with no

isomerisation seen. The remaining 1-hexene was removed by vacuum transfer and reused in a second

reaction, with the 1-hexene recovered (16.381 g, 0.195 mol), triethyl aluminium (0.111 g, 0.974

mmol) was added to this 1-hexene. In a separate flask: Cy TACCr(OTf) (0.081 g, 97.4 μmol), AlMe

3

(2.45 wt. % in toluene, 0.859 g, 0.29 mmol) and o-difluorobenzene (16.2 g, 0.142 mol) were added

together and both solutions left to stir for 1 hour, before then combining the two and leaving to stir

together overnight. The conversion of 1-hexene to trimer was 28.3 %. Both trimer mixtures were then

added together and extracted by phase separation with dilute HCl and pentane. The organic phase was

washed several times with dilute HCl and the aqueous phase washed with pentane several times. The

organic phases were then combined and dried with MgSO . The solvent was removed by rotary

4

-3

evaporation, and the remaining liquid high vacuum transferred (100-150 ºC at 10 mbar) to yield

1

2.495 g of trimer product (42% isolated yield).

2

.6. Air Sensitive Mass Spectrometry Procedure

Direct infusion mass spectrometry analyses were performed using an electrospray time-of-flight

MicroTOFQ) mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany), analyses were

(

performed in ESI positive and negative mode. For positive mode the capillary voltage was set to 4500

V, nebulizing gas at 0.5 bar, drying gas at 3.0 L/min at 110°C. For negative mode the capillary

1

voltage was set to 4500 V, nebulizing gas at 0.5 bar, drying gas at 4.0 L/min at 200°C.The TOF scan

range was from 50 - 2300 mass-to-charge ratio (m/z).

This was coupled to an MBraun glove box containing an argon atmosphere. An air-tight connector in

the side of the glove box allowed for PEEK tubing to pass from the glove box to the mass

spectrometer without exposure to air or moisture. The sample was injected directly into the tubing via

syringe and the use of an adaptor. The instrument was calibrated post-acquisition using a range of

sodium formate clusters. The observed mass and isotope pattern were considered to match the

corresponding theoretical values calculated from the empirical formula when an error of less than 20

ppm was observed. The background spectrum was established before each run with the use of samples

of each reagent.

The trimerisation catalysis analysed using the positive ion detection mode was carried out with the use

of 4d (2.5 mg, 2.6 µmol), AlMe (2.5 w% in toluene, 4.9 mg, 68 µmol, 26 equivalents), 1-hexene

3

(

1.226 g, 14.6 mmol, 5600 equivalents) and o-C H F (2.07 g, 18.2 mmol, 7000 equivalents). The

6 4 2

reaction analysed using the negative ion detection mode was carried out using very similar conditions,

though fewer equivalents of activator were used. A second MS experiment used analogous conditions

with only 450 equivalents of 1-hexene.

Analysis of the sample was carried out in two stages, firstly within one minute of activator addition and

secondly within one minute of 1-hexene addition. At least three aliquots of o-C H F were run through

6

4 2

the system before each sample was analysed to reduce the risk of contamination.

3

. Results and Discussion

3

.1. Synthesis and Characterisation

The substitution of chloride ligands bound to chromium(III) has been described previously with the use

of neat trifluoromethanesulphonic acid (TfOH) and in the presence of similar amine based ancillary

ligands, ammonia [28] and 1,4,7-trimethyl-1,4,7-triazacyclononane (Me TACN) [29]. The methods

3

described were modified and applied to the synthesis of four R TACCr(OTf) 4 pre-catalysts from their

3

chromium trichloride analogues 1, Scheme 1. Despite the weaker binding of R TAC ligands compared

3

to those previously investigated, the reaction proceeded as expected and near quantitative yields were

recorded for a range of differently substituted triazacyclohexanes (route A).

1

2

air/water

Cr + 3 TfOH

THF

heat/vacuum

[

Cr(THF) (OTf) ]

[

Cr(H O) ](OTf)

Cr(OTf)₃

3 3

2

6

3

2

CrO + 3 TfOH

3

+

R TAC

i

water

3

route C

+

3 PrOH

route B

R

N

R

neat TfOH

3 HCl

R

N

R

N

Cr

-

Cl

TfO

OTf

Cl

route A

OTf

4

1

n

i

R = cyclohexyl (a), benzyl (b), Pr CHCH2 (c), Bu CH (d)

2

Scheme 1. Synthesis of triazacyclohexane triflate complexes 4 and their Cr(OTf) precursors.

3

As anhydrous triflate complexes can often be obtained from aqueous triflates by simple heating under

vacuum (e.g. [30]) we also explored a more direct route to the same complexes. Aqueous blue solutions

of [Cr(H O) ](OTf) were obtained in modification to literature procedures either from aerobic

2

6

3

dissolution of chromium metal in aqueous triflic acid or reduction of an aqueous solution of CrO and

3

triflic acid with isopropanol.[31-33] Concentration of the blue solution at high vacuum resulted in a

blue viscous oil. Further drying in a desiccator filled with P O resulted in very hygroscopic blue

4

10

.

crystals identified as [Cr(H O) ](OTf) 3H O by X-ray crystallography (see supplementary material).

2

6

3

2

Further drying under vacuum with slow heating to 100 °C resulted in a light green Cr(OTf) (H O) .

3

2

x

Further drying at temperatures slowly increasing to 200 °C resulted in a sudden release of gases at

around 160 °C. TGA/MS studies (ESI) on green solids obtained at 130 °C showed a release of water

between 160 and 240 °C. Release of a small amount of SO and TfOH is also seen at this temperature

2

range and much more pronounced at 300 to 400 °C indicating decomposition of the triflate anion. Thus,

the removal of all water must be done with care to avoid anion decomposition. We found that slow

heating to 200 °C followed by suspension of the green product in neat triflic acid with subsequent high

vacuum removal at 200 °C resulted in sufficiently pure anhydrous Cr(OTf) . The product was insoluble

3

in non-coordinating solvents. A good sample could be reacted with Bz TAC in o-difluorobenzene to

3

give a turquoise-blue solution of 4b which gave crystalline product on cooling in a freezer or vapour

diffusion of hexane into the filtered solution (route B). In order to avoid problems with contaminants in

impure 2, a well-defined THF complex 3 was obtained by dissolving 2 in dry THF and crystallisation

by hexane vapour diffusion (products still containing water or TfOH result in THF polymerisation). X-

ray crystallography showed that 3 is the triflate analogue of meridional CrCl (THF) . Figure 1 shows

3

the molecular structure of 3 and some selected bond distances and angles along with the corresponding

parameters for CrCl (THF) . While the crystal structure of the chloride analogue CrCl (THF) has

3

previously been published at 100 K [34] and 278 K [35], we obtained a structure at the same temperature

as used for 3 (150 K) at a better R value (4.2% vs 7.4% and 5.9%, respectively) used for this comparison

to 3. The Cr-O(THF) bond distances are significantly shorter in 3 compared to the chloride complex

and the THF ligands are more compressed in CrCl (THF) than in 3 based on the THF-Cr-THF angles.

3

Thus, as expected OTf is much weaker trans-labilising and exerts a small steric effect on the other

ligands than chloride. Quantitative NMR spectra in PhF show a single paramagnetically broadened and

2

1

13

shifted NMR signal for the -CH groups in THF in both H and C as well as two resolved 500 Hz

2

1

3

1

9

wide F NMR signals (1:2 ratio) for the two inequivalent triflate groups in nearly the correct amount

relative the solvent signals.

Figure 1. ORTEP representation of the molecular structures of 3 and CrCl (THF) with hydrogen atoms

3

omitted for clarity and atoms are drawn at 30% probability. Selected bond distances for 3: Cr-OTf (trans

THF) 1.953(3), Cr-OTf (trans OTf) 1.959(2) and 1.951(2), Cr-O_T_H_F(trans OTf) 2.000(2), Cr-O_T_H_F(trans

THF) 1.982(2) and 1.990(2). Selected bond angles: O_T_H_F-Cr-O_T_H_F: 178.45(11), 89.67(10), 89.03(10);

OTf-Cr-OTf: 175.54(11), 92.49(11), 91.18(11). For comparison, CrCl (THF) : Cr-O(trans THF)

3

2

.0041(14) and 2.0279(14), Cr-O (trans Cl) 2.0747(15), Cr-Cl (trans Cl) 2.3130(6) and 2.3205(6), Cr-

Cl (trans THF) 2.2931(6). Selected bond angles: O-Cr-O: 173.42(6), 87.98(6), 85.56(6); Cl-Cr-Cl

77.03(2), 90.81(2), 91.74(2) (disorder in one THF not shown)

1

Complex 3 is a suitable precursor for the near quantitative synthesis of complexes 4 by addition of one

equivalent of triazacyclohexane in o-difluorobenzene (route C). In the case of poorly soluble 4a,

crystals suitable for X-ray diffraction grew within one hour of addition leaving a solution containing

nearly three equivalents of THF and negligible amounts of remaining ligand as determined by NMR.

The more soluble compounds 4c,d were characterised using paramagnetic ¹³C and ¹⁹F NMR

spectroscopy. The broadened ¹³C NMR spectrum was little changed on substitution of the chloride

ligands and follows the trends previously observed for differently substituted complexes 3 [3]. A broad

1

9

peak could be observed for 4 by F NMR spectroscopy indicative of inner sphere co-ordination of three

triflate ions. In this manner both the chloride substitution and the R TAC retention could be confirmed.

3

.1.1. Crystal structures of 4a,b and 1a,b for comparison

Complexes 4a and 4b have been characterised by X ray crystallography. Crystals of the known [19]

analogous chloride complex 1a and 1b have been obtained by crystallisation from

acetonitrile/dichloromethane or methylene chloride, respectively, and comparison allows a direct

assessment of the effect of the different anionic ligand. Complexes 4a and 4b crystallise with half and

a full molecule of PhF , respectively, and disorders in the triflate groups in 4a while 1a crystallises

2

without solvent and 1b with 2.5 molecules of CH Cl per complex in the crystal lattice and a disorder

2

in one of the benzyl groups. The molecular structures with the solvent molecules and disorders removed

are shown in Fig 2 and selected average bond distances and angles in Table 1. The Cr-N bond distance

are significantly smaller in the triflate complexes compared to the chlorides indicating a significantly

1

4

weaker bond to the triflate as found for the THF complexes above and as expected for more weakly

coordinating triflate anion.

Figure 2. ORTEP representation of the molecular structures of Molecular structures of 4a (top left), 1a

(top right), 4b (bottom left) and 1b (bottom right) with hydrogen atoms and solvent molecules omitted

for clarity and atoms are drawn at 30% probability.

Table 1. Selected average bond distances (Å) and angles (°) in 4a, 1a, 4b and 1b.

complex

Cr-N

Cr-O or

Cr-Cl

N-Cr-N

Cr-N-C(R)

4a (PhF )

2.0777(19)

1.9498(18)

1a

4b (PhF )

2.078(8)

1.956(7)

1b (CH Cl )

* 2 2 2.5

2.110(4)

*

2 0.5

*

2

2.1215(17)

2.2921(5)

65.95(6)

132.26(13)

2.2776(14)

65.72(15)

128.0(2)

67.56(7)

132.33(14)

67.2(3)

130.8(7)

1

5

O-Cr-O or

Cl-Cr-Cl

95.59(9)

94.2(3)

99.18(2)

99.06(6)

Table 2. Summary of X-ray Crystallography for 3, CrCl (THF) , 1a, 1b, 4a, 4b with selected average

3

bond distances and angles (*indicates range of values or SD, if greater, as a measure of confidence).

33 31 9 11 3

C H O F N

C

15

H

24

O

12

F

9

CrS

3

C

CrCl

12

H

24Cl

3

CrO

3

C

21

H

39Cl

1a

3

CrN

3

C

1b

26.5

H

(CH

32Cl

8

CrN

3

27 41 9 10 3 3

C H O F N CrS

Empirical Formula

CrS

(PhF

970.79

Orthorhombic

P 2

3

(THF)

3

*

2

Cl

2

)

2.5

4a

*

(PhF

2 0.5

)

4

b

*

2

)

Formula Weight

Crystal System

Space Group

Unit Cell

715.52

Monoclinic

P 2

374.66

491.90

Triclinic

P -1

2912.60

Monoclinic

C 2/c

889.81

Monoclinic

I 2/a

Monoclinic

1

P 2

1

/c

1

2

1

2

1

Dimensions

a (Å)

b (Å)

9.0638(2)

14.4268(3)

10.3798(3)

90

8.0382(3)

12.5123(5)

16.3957(5)

90

92.652(3)

90

1647.25(10)

4

(Mo Kα) 1.181

6.9098(2)

10.6220(3)

16.4715(4)

86.641(2)

87.888(2)

83.072(2)

1197.51(6)

2

20.727(3)

11.957(2)

27.509(4)

90

92.022(14)

90

6813.4(18)

8

(Mo Kα) 0.985

23.5157(3)

17.3075(2)

18.4883(2)

90

100.6390(10)

90

7395.35(15)

8

(Cu Kα) 5.073

41721

15.3849(6)

15.3679(6)

17.2012(6)

90

c (Å)







(°)

91.2011(10)

90

1356.98(6)

2

3

V (Å )

Z

4066.9(3)

4

(Mo Kα) 0.540

54270

-

1



(mm )

(Mo Kα) 0.769

(Cu Kα) 7.090

Total Reflections

Independent

17202

15267

19245

28777

5

744 (0.0369)

0.0331, 0.0831

.043

4437 (0.0495)

0.0417, 0.0967

1.061

4741 (0.0492)

0.0311, 0.0725

1.058

6752 (0.0933)

0.0771, 0.1917

1.052

7317 (0.0330)

0.0452, 0.1130

1.044

9241 (0.1286)

0.1078, 0.2798

1.250

Reflections (Rint

, wR [I > 2(I)]

Goodness-of-fit

)

R

1

2

1

2

(GOF) on F

Largest Difference

in Peak and

Hole (e/Å )

0

.374,

0.530,

-0.666

0.297,

-0.339

0.835,

-0.546

0.764,

-0.521

3.373,

-1.172

-0.506

3

={Σ[w(Fo - Fc ) ] /Σ[w(Fo ) ]}¹^/². GOF = S = {Σ[w(Fo - Fc ) ]] /(n-p)}¹^/².

2

2 2

R

1

=Σ ||Fo|-|Fc||/Σ|Fo|. wR

2

3

.2. Catalysis of -Olefin Trimerisation

This newly developed procedure enabled the facile two-step synthesis of pre-catalysts 4a-d, which were

then investigated in relation to their activation and catalysis. It was hoped that the weakly co-ordinating

nature of the triflate ligands would facilitate alkylation and abstraction during activation. Meanwhile,

the disperse charge of these anions was expected to support the formation of a weakly coordinating

anion assumed to be essential for high trimerisation activity. To test this hypothesis a range of activating

agents were screened in order to determine whether the triflate anions enabled activation by agents less

potent than MAO. The activation success was measured by the ability of the resulting species to

trimerise 1-hexene, reactivity that has been established previously for R TACCrCl complexes.

3

It was found that addition of 5-20 equivalents of AlR (R=Me, Et, iBu) to an o-difluorobenzene

3

suspension of the triflate complexes leads to dissolution to a blue/green solution which trimerises 1-

hexene on addition. In the case of AlMe , there seems to be a significant initiation period of several

3

hours before full activity is reached. This can be rationalised by the requirement for a -H on the Cr-

alkyl for agostic assisted -H abstraction with reduction to a Cr(I) complex to initiate the trimerisation

catalysis. This process may occur in two steps or directly in one step as shown in Scheme 2 and will

i

subsequently referred to as “-H abstraction”. While Et and Bu already have a -H, Me requires an

apparently slow insertion of 1-hexene to give a -H-containing alkyl group. This slow initiation may

be further hindered by the formation of alkyl-bridged dinuclear complexes as shown in Scheme 2, which

should be more stable for methyl bridges [9] compared to ethyl or isobutyl bridges. Similar activation

n

with 10-20 equivalents of Et Zn or Bu Mg instead of AlR do not result in any activity. Addition of

2

3

i

just one equivalent of Al Bu to the inactive Et Zn-activated solution does result in at least some

3

2

1

6

trimerisation activity. Thus, ZnR or MgR may be able to alkylate the chromium triflate complexes but

2

cannot generate the apparently required dialkyl cationic complexes for catalysis.

R

+

2+

R

N

R

AlMe3

R

N

R

Cr

OTf

Cr

Me

N

Cr

TfO

OTf

Me

Cr

Me

N

Bu

R

N

R

+

R

N

R

+

R

N

Cr

Migratory

Insertion

R

N

Me

Cr

Me

R

+

Bu

N

R

N

Cr

Me

or direct -H abstraction via



-Hydride

Elimination

H

Bu

Me

Bu

R

+

Reductive

Elimination

R

N

R

Bu

N

2

N

Cr

Bu

Me

H

Bu

CH4

Scheme 2. Proposed activation mechanism

In order to assess the differences between aluminium alkyl activators, equal 1g portions of 0.5w%

solutions of the most soluble complex 4c in o-difluorobenzene are added to mixtures of about 110 eq

i

AlR (R= Me, Et, Bu) in about 600 mg 1-hexene and followed by NMR. The larger than required excess

3

of AlR was used to ensure that there is no significant change in aluminium alkyl species by alkyl

3

exchange reactions during activation (see later). There is a striking difference in the behaviours of the

three activators. Figures 3 show the drop in 1-hexene and formation of trimers as well as hexene

isomers.

1

7

i

Figure 3. Comparing the reaction profile using AlMe , AlEt and Al Bu with 4c. Loss of 1-hexene

3

(top left), formation of hexene isomers (top right) and formation of trimers (bottom left) as % of

starting hexene or as % of hexene not isomerised (bottom right).

In each case the selectivity of these catalysts is inferior to their R TACCrCl /DMMAO analogues.

3

While the oligomerisation selectivity (the proportion of trimer formed relative to other oligomers)

remains extremely high at >99 mol%, as measured by GC-MS/FID, a far higher incidence of

isomerisation is observed especially at low 1-hexene concentration leading to internal hexenes. We

propose that this side reaction proceeds via a π-allyl complex mechanism, Scheme 3. This mechanism

I

III

aligns well with that of metallacyclic trimerisation because of the matching Cr /Cr redox cycle

I

proposed for both. The Cr -olefin adduct is also proposed as an accessible intermediate for both catalyst

cycles, such that inter-conversion between reaction pathways would be expected.

1

8

R

H

R

Cr^I^I^I

Cr^I

1

-C₆

2-C₆

H

Cr^I^I^I

Cr^I

R

Scheme 3. Possible mechanism for olefin isomerisation via allyl complexes.

It is hypothesised that the greater propensity of R TACCr(OTf) based catalysts to isomerise 1-hexene

3

results from the presence of a different counter-ion. Several factors (catalyst dilution, 1-hexene

equivalents, ligand bulk) significantly influence the ratio of trimerisation to isomerisation. This is

I

attributed to the availability of the Cr -olefin adduct, as it is assumed that once a second olefin has co-

ordinated to the chromium centre trimerisation becomes by far the more favoured process. Thus,

reducing the 1-hexene concentration leads to an increase of isomerisation activity. Thus, isomerisation

increases towards the end of trimerisation when the monomer concentration becomes low.

Unfortunately, this competing reactivity hinders us from studying trimerisation at the initial stages at

low olefin to Cr ratio where only isomerisation activity can be observed.

The relative isomerisation activity also depends greatly on the activating AlR . While activation with

3

AlMe gives only little isomerisation (<10%) at high enough olefin concentration (>1000 eq.),

3

i

activation with Al Bu leads to more isomerisation than trimerisation. AlEt gives fast activation with

3

mainly trimerisation but more significant isomerisation (about 3:1). We found that optimal activation

i

can be achieved by mixing some AlMe with AlEt or Al Bu leading to both fast activation and slow

3

isomerisation. The use of mainly AlEt also represents the most cost efficient reagent and therefore

3

maximises the economic feasibility of this system relative to MAO based procedures.

Once AlEt was identified as the preferred activating agent the equivalents required for complete

3

activation of the catalyst were optimised. It was found that just four equivalents relative to the pre-

catalyst were required for the formation of active catalyst. The use of smaller quantities still resulted in

a colour change from blue to green, which is associated with alkylation of the chromium, but did not

produce an active species. Therefore, the implication is that four equivalents are required to completely

abstract triflate from the chromium and incorporate them into a weakly co-ordinating anion. Based on

the demonstrated importance of a 3:4 ratio of triflate units to AlEt in forming the counter-ion, its

3

structure can be proposed, Scheme 4, which would be the simplest anion with all-four-coordinated

aluminium and all-dimetallated triflate.

R

-

CF3

S

CF3

S

CF3

S

N

R

Et

O

Et

N

Cr

Et

Al

Et

O

1

9

Scheme 4. The proposed structure of the activated catalyst and its counter-ion. Formal charges have

been included at their respective positions within the counter-ion in order to clarify the overall charge.

This counter anion will likely weakly bind to the cationic chromium complexes. Bulkier R groups

adjacent to a potential S=O binding site or alternative binding through the Al-R group similar to MAO

may explain the isomerisation dependence on the type of AlR used. Thus, the choice of the AlR will

3

i

influence the binding capability of the counter anion and Bu based anions should be less coordinating

than Me based anions. Thus, it can be hypothesised that a more weakly coordinating anion leads to

i

more isomerisation. Use of a mixture of AlMe and either AlEt or Al Bu would retain the strongest

3

i

donor site of an AlMe based anion while still showing fast activation as for AlEt or Al Bu . Indeed,

:1 mixtures of AlEt or Al Bu and AlMe give fast activation and less than 25% isomerisation as

3 3 3

3

i

2

shown in Figure 4.

i

Figure 4. Reaction profile using Et AlMe (left) and Bu AlMe (right) as activator for 4c.

2

Under optimised conditions about 1000eq of trimerised olefin per chromium can be achieved with

1

3

less than 10% isomerisation. The distribution of isomers in the trimer was determined by C NMR as

described before [10] and is found to be very similar between 1/DMMAO and 4/AlR and depends

3

much more on the N-substituent. Thus, the changing counter anion does not influence the isomer

distribution very much.

2

0

Table 3. The main regioisomers produced on trimerisation of 1-hexene with either 1/DMMAO or

4

/AlR as catalyst (average over several runs). Structure of the isomers shown to the right [10].

3

Regioisomer Abundance (%)

Bu

Cat.

A

A’

B

B’

9

C

6

9

2

4

1

D

14

18

16

17

22

D’

2

E

4

Bu

4

1

4

1

a

b

c

24 18 24

A (SS/RR), A' (SR/RS)

B (SR/RS), B' (SS/RR)

37

6

5

7

8

5

4

6

Bu

33 10

34 10

8

4

10

7

Bu

C

c

7

3

Bu

d

23

20

8

9

15 12

14

4

12

Bu

4

d

9

0

24

4

13

D (SS/RR), D' (SR/RS)

E (SS/RR), E' (SR/RS)

3

.3 Analysis of the Activation Mechanism

Having established a catalyst capable of 1-hexene trimerisation at far lower aluminium alkyl loading

than previous systems it was possible to perform the catalysis on a larger scale in order to investigate

the activation products. With the use of 1 gram of catalyst, as opposed to 5 mg typically, activated with

AlMe and just six equivalents of 1-hexene it was possible to produce a product sample highly

3

concentrated in the activation products. The proposed catalyst activation mechanism for olefin

trimerisation has been expanded and adapted in Scheme 2 based on previous proposals relating to

ethylene trimerisation.

The sample produced was analysed with the use of GC-MS/FID in an attempt to identify 2-methyl-1-

hexene as a side product and provide evidence to support the proposed mechanism. However, analysis

of the products in this manner indicated a range of side-products formed during activation. These could

be identified by comparison of the corresponding fragmentation pattern to the NIST database, Figure 5.

The key products were then confirmed with use of reference compounds and correlation of retention

time under the same conditions, Figure 6. As a result, the presence of large quantities of C and C

6

7

alcohols within the product mix was confirmed alongside 2-methyl-1-hexene.

2

1

Figure 5. The products detected by GC-MS/FID on reaction of 6 equivalents of 1-hexene with AlMe₃

activated 4a. The unlabelled peaks (2.010, 3.176, 3.823, 4.065 and 4.160) are derived from the o-

C H F /pentane solvent and impurities within it, as determined by the control.

6

4 2

Figure 6. Correlation of retention times between reference compounds and the products of activation.

2

-Hexene and the trimeric products were correlated with typical product samples within which these

species are present in high abundance.

The key observation is that 2-methyl-1-hexene is formed as an activation product, which for the first

time confirms the formation of the side product predicted by the mechanism originally proposed. The

presence of its more stable isomerisation product 2-methyl-2-hexene is also not surprising under

conditions with isomerisation activity. In contrast, the formation of several alcoholic species and 3-

methylheptane was not predicted and the alcohols at least must form during exposure of the product

2

mix to oxygen after the reaction is complete. Analysis of the GC-FID spectrum allowed quantification

of the various products and in turn the number of equivalents of 1-hexene converted to each could be

calculated, Table 4. The observation of 1.7 eq. of 2-methyl-1-hexanol indicates significant alkyl

exchange between chromium and AlMe before -H abstraction leading to more than just one insertion

3

reaction of 1-hexene into the Cr-Me bond. 1-Hexanol could be explained by -hydride elimination to

coordinated 1-hexene rather than the methyl group leading to a Cr-hexyl group which can exchange to

aluminium followed by oxidative workup to the alcohol. Small amounts of 3-methylheptane (and 2-

ethyl-1-hexanol) may be due to ethyl impurities in the AlMe or the product of C-C reductive

3

+

elimination of the [LCrMe(2-methyl-1-hexyl)] intermediate.

Table 4. Products detected by GC after conversion of 6 eq. of 1-hexene

Activation Product

1-Hexene Equivalents Converted

2

3

-Hexene

2.8

0.3

-Methyl-1-Hexene

-Methyl-2-Hexene

-Methylheptane

0.2

0.1

C Alkanes

0.1

1

2

1

2

-Hexanol

0.1

-Methyl-1-Hexanol

-Ethyl-1-Hexanol

1.7

< 0.1

0.9

C Trimer

1

8

Figure 7. Additional Al-CH₂¹H NMR signal observed during of 1-hexene trimerisation with AlMe₃

activation

2

3

The processes during activation and deactivation can also be observed by NMR during trimerisation

1

with 4c in o-difluorobenzene and different AlR described in 3.2. The region below 0.6 ppm in H NMR

3

shows signals for the -hydrogens in AlR and methane at 0.15 ppm in the case of AlMe activation.

3

Integration of the methane signal corresponds to a slightly more than one equivalent per chromium – as

some methane may be lost during initial sample preparation and to the gas phase above the solution it

is likely that more than one equivalent of methane is produced. Besides this methane signal and a large

signal for AlMe at –0.30 ppm a characteristic pair of two doublets of doublets is observed at +0.179

3

and –0.013 ppm with coupling constants of 5.1 + 14.0 and 7.9 + 14.0 Hz, respectively, as shown in

Figure 7. This can be assigned to the two diastereotopic protons of a CH group adjacent to a chiral

2

centre. It shows J coupling to C at 21.7 ppm (HSQC) and ²^/³J coupling to 23.8, 30.3 and 41.8 ppm

1

13

(

HMBC) matching well the previously reported signals found for Me Al(2-methyl-1-hexyl) in benzene

2

[36]. This signal grows rapidly in the early stages of the catalysis and increases only little later on

reaching about 6 equivalents per chromium. This signal dominates all other Al-R signals observed in

small amounts around 0.1 ppm. Opening the NMR tube to air at the end of the catalysis results in NMR

signals for Al-OR groups around 3.5 ppm. Hydrolysis and distillation of all volatiles gives an o-

difluorobenzene solution containing the remaining 1-hexene other volatile olefins and alcohols. The

other olefins correspond to about 5% hexene isomers, about 3 equivalents of 2-methyl-1-hexene per

chromium and a small amount of a more symmetrical vinylidene (possibly 2-ethyl-1-pentene) identified

1

13

by their olefinic H signals at 4.68+4.66 ppm and 4.69ppm, respectively, as well as their coupled C

signals (HSQC) at 110.3 and 106.9, respectively. The alcohols are mainly methanol at 3.35 ppm (s) and

2

-methyl-1-hexanol [37] at 3.44dd (J= 5.6 and 10.3 Hz) and 3.35dd (6.7 and 10.3 Hz) besides a small

amount of linear alcohol at 3.55t (J=6.6 Hz, likely 1-hexanol). These observations clearly show that the

catalyst is activated with AlMe by multiple 1-hexene insertions into Cr-Me bonds followed H-transfers

3

before finally reducing to chromium(I) as shown in Scheme 5.

multiple times

Me

Me2Al

+

AlMe₃

1-hexene

Bu

R

+

R

+

N

Migratory

Insertion

R

N

R

N

Cr

Me

Bu

Me2Al

+

AlMe₃

+

1-hexene

+

1-hexene

Bu

R

+

R

+

N

R

N

multiple times

N

R

N

R

Cr

N

Cr

Me

Bu

H

Bu

+

1-hexene

-

2-Me-1-hexene and CH₄

trimerisation

R

+

N

R

N

Cr

Bu

2

4

Scheme 5. Proposed mechanism of multiple activation via alkyl exchange with AlMe₃.

i

Activation with AlEt or Al Bu does not require 1-hexene insertion prior to -H abstraction as both

3

have already -hydrogen atoms. However, insertion into and -hydrogen transfer to coordinated 1-

hexene could still occur.

i

Activation with Al Bu quickly produces several equivalents of isobutene per chromium detectable at

3

4

.63s (2H) and 1.65s (6H) growing more slowly but continuously during trimerisation reaching 90

i

equivalents per chromium for a sample activated with 100 eq of Al Bu . At the same time a nearly equal

amount of the initial Al- Bu observed at doublet at 0.30ppm in 1H NMR is converted into a triplet signal

at 0.45ppm identified as Al-hexyl through its C NMR signals observed by HSQC (13 ppm) and HMBC

3

i

1

3

i

(26 and 36 ppm) matching the signals observed before for Bu Al-hexyl [27]. Thus, the initial Al Bu is

2

3

i

converted to hexyl rich AlR during trimerisation especially when smaller excess Al Bu per chromium

3

is used for activation.

1

13

Activation with 120 eq of AlEt leads to detectable ethane (0.81ppm ( H) and 7.2ppm ( C) by HSQC)

3

1

3

and some of the Al-Et signal observed as quartet at 0.36 ppm ( C at 1.3 ppm (HSQC) and 9.1 ppm

HMBC)) is converted into an unidentified triplet at 0.17 ppm ( C at 16 ppm (HSQC) and 32 and 39

1

3

(

ppm (HMBC)) as well as additional Al-R under the 0.36 ppm signal of Al-Et detectable by HSQC (7-

i

1

1 ppm)/HMBC (26, 29, 36 ppm) which could include Al-hexyl as for Al Bu activation. After

3

hydrolysis at the end of one catalysis run all room temperature volatiles at high vacuum were collected

in a liquid nitrogen trap and analysed by NMR. It shows the expected solvent, remaining 1-hexene and

isomerised hexenes but also a small amount of vinylidene at 4.689ttd (J=2.2+1.6+1.0 Hz) + 4.678ttd

(

1.7+0.6+1.0 Hz) coupling to ¹³C at 108.1 ppm (HSQC) and 152.4, 37 and 29 ppm (HMBC) which

could be assigned to 2-ethyl-1-hexene as the product of ethyl insertion into 1-hexene and subsequent

elimination. The quantity of this product would correspond to about 3 equivalents per chromium. Thus,

i

multiple insertion and alkyl transfer to aluminium seems to occur also for AlEt and Al Bu activation.

3

i

Activation with a 2:1 mixture of Al Bu or AlEt and AlMe leads to fast formation of methane and

3

isobutene in excess amounts as observed with single AlR activation. Any ethylene that must be formed

3

with AlEt activation is likely incorporated in a mixture of isomeric co-trimers with 1-hexene and could

3

not be identified.

Careful NMR observation also allows a study of ligand transfer to aluminium as a major decomposition

pathway. During catalysis, additional signals shown in Figure 8 are observed in the otherwise empty

1

region of 2.5-4.5 ppm in H NMR. These signals include the characteristic pair of doublets previously

observed and identified as the ring hydrogens coordinated to AlR₂⁺[7]. A third doublet of double

intensity is also observed for the -CH of the N-substituent. The small but significant differences in

2

the chemical shifts between different R groups indicate that they are still coordinated to this aluminium

complex. When 4 is activated with only a minimum amount of AlR needed these signals appear as

3

complex signals of several overlapping doublets due to the formation of complexes with mixed alkyl

groups due to the alkyl exchange reaction. When trimerisation has ceased after a few days, the integral

of these signals correspond to 30-70% of the initial complex used. Thus, this is a major decomposition

route but other inactive chromium complexes must also be formed. Irreversible decomposition occurs

when all 1-hexene has been used up at low substrate loading as addition of new 1-hexene does not lead

to further catalysis. However, if insufficient amounts (< 4 eq) of AlR are added to start trimerisation,

3

catalysis can be started by later addition of a few more equivalents AlR even after several hours of

3

inactivity. Thus, decomposition does not seem to occur before trimerisation starts.

2

5

1

Figure 8. H NMR signals observed for the LAl decomposition products during 1-hexene trimerisation

with 4c. The spectrum shown is for R=Et, the chemical shifts in ppm observed for the doublets are

shown in the table.

3

.4. Electrospray MS Studies

The proposed catalytic cycle involves mainly cationic chromium species. Thus, electrospray mass

spectrometry (ESIMS) should be an ideal method to observe these intermediates. When we tried to

observe MAO-activated solutions, we found that in our hands the large excess of MAO led to fast

blocking of the thin PEEK or even steel tubing leading from a glovebox to the MS instrument,

presumable by hydrolysis, despite thorough drying in an oven and under vacuum. This problem was

much less pronounced with our new triflate/AlR catalyst system. The much lower AlR concentration

3

allowed us to observe MS spectra for about 2 hours before the tube was blocked.

The reaction was split into two key stages. Firstly, the activation of the pre-catalyst was explored

by excluding 1-hexene and simply adding the AlMe activator (20 equivalents) to the pre-catalyst in the

3

presence of the o-difluorobenzene solvent. For the MS studies, the most soluble complex

(ⁱBu CH) TACCr(OTf) 4d was used to avoid solubility problems especially after addition of large

2

3

amounts of olefin. A sample of the solution was then taken one minute after addition and injected into

the mass spectrometer while under argon. In this way, the sample could be analysed immediately and

the ionic species present at this stage of the reaction readily detected.

The mass spectrum observed for this sample was remarkably clean, with just one set of peaks

prominent. The distinctive isotope pattern (Figure 9) matched that expected for

i

+

[

(( Bu CH) TAC)CrMe ] , the alkylation product predicted by the metallacyclic mechanism. A smaller

2 3 2

i

+

signal (about 10%) of [(( Bu CH) TAC)CrMe(OTf)] can also be detected at m/z = 681.4029 (calcd

81.4177) as an intermediate of this activation process. Identification of [(( Bu CH) TAC)CrMe ] lends

2 3 2

2

3

i

+

6

considerable support to the initiation of the metallacyclic mechanism and agrees with the EPR

spectroscopy carried out by Bercaw et al..[16]

2

6

Figure 9. The observed (top) and simulated (bottom) m/z peaks corresponding to

i

+

[

(( Bu CH) TAC)CrMe ] .

2 3 2

After addition of 1-hexene (either 450 or 5000 equivalents per Cr) to a portion of the same

mixture another MS sample was taken. Now there were considerably more signals present in the

spectrum, suggesting a multitude of species are formed on addition of 1-hexene. While many of these

signals could not be assigned, the major species present and several minor species were found to

correlate with those expected for the intermediates proposed in Scheme 5. The previously dominating

i

+

signal for [(( Bu CH) TAC)CrMe ] is still present throughout the 2 hours of the MS experiments

2

3

2

supporting the slow activation with AlMe observed by NMR. Indeed, this slow activation provides

3

relatively constant conditions of trimerisation throughout the experiment making observed intensities

approximately comparable. In addition, ions corresponding mainly to species of the composition

i

+

[

(( Bu CH) TAC)Cr(C H ) ] (n=0-4) are observed. In MS, we can only detect the total mass and

2 3 6 12 n

several different species of the same composition as shown in Scheme 6 could be responsible for the

observed cations.

2

7

R

₃Bu

R

N

R

N

R

N

Cr

N

-

CH4, 2-Me-1-hexene

possibly via B1

Cr

Me

Bu

Me

A

Bu

₂Bu

Bu

B2

and isomers

R

B2'

N

Bu

R

N

Bu

Cr

X

R

C

N

R

N

R

N

R

N

Cr

H

Bu

X

Pr

Bu

C₁₈Trimer

Bu

B1

B1'

Bu

and isomers

B3

hexene isomers

R

N

R

N

Bu

N

Bu

Cr

R

N

R

Bu

N

Bu

Cr

Bu

and isomers

B4'

B3''

Bu

and isomers

B3'

R

N

R

N

Bu

Cr

Bu

and isomers

B4

Scheme 6. The proposed intermediates of selective 1-hexene trimerisation based on mass spectrometric

observation of cations of LCrMe (A), LCr(1-hexene) (B1 and B1’), LCr(1-hexene) (B2 and B2’),

2

LCr(1-hexene) (B3, B3’ and B3’’), LCr(1-hexene) (B4 and B4’), and LCr(arene) (C)

3

4

Table 5. The proposed intermediates that were found to have corresponding peaks in the mass

spectrum during reaction with 1-hexene.

rel Intensity

450 eq

rel Intensity

5000 eq

1-hexene

100

-

8

45

3

-

Observed

m/z

Calculated

m/z

Catalytic Intermediate

1

-hexene

i

+

[

(( Bu CH) TAC)Cr]

517.4338

601.5198

517.4422

601.5361

685.6300

769.7239

853.8179

615.5518

699.6457

100

20

9

3

2

3

i

+

(( Bu CH) TAC)Cr(C H )]

2

3

6

12

i

+

(( Bu CH) TAC)Cr(C H ) ] 685.6142

(( Bu CH) TAC)Cr(C H ) ] 769.7140

(( Bu CH) TAC)Cr(C H ) ] 853.7976

2

3

6

12 2

i

+

2

3

6

12 3

i

+

2

3

6

12 4

i

+

(( Bu CH) TAC)Cr(C H )]

615.5401

60

9

2

3

7

14

i

+

(( Bu CH) TAC)Cr(C H )] 699.6275

-

2

3

13 26

The relative intensities of the peaks shown in Table 5 are likely to be indicative of the relative

i

+

concentrations. Observation of [(( Bu CH) TAC)Cr] is highly significant because it confirms that

2

3

chromium(I) with coordinated ligand is present. While a naked species like this is unlikely in the

solution during catalysis, it may indicate the presence of Cr(I) complexes with weakly coordinated

2

8

neutral ligands that can be lost in the gas phase activation within the MS instrument. The observation

of significant amounts of complex cations made up from two and three olefins supports the presence of

metallacyclopentane and –heptane species as expected from the proposed catalytic cycle with increased

amounts of metallocycloheptane at higher monomer concentration. The observation of cations made up

of four olefins is surprising as no tetramerisation is detectable by GC in any of these systems. Thus, this

ion must be made up of coordinated monomer plus trimer either before (B4) or after -H abstraction

(B4’) rather than a metallacyclononane. Observation of a significant amount of cations containing only

one olefin at relatively low monomer loading while not at high monomer concentration may be due to

formation of Cr(III) allyl hydride complexes (B1’) proposed as intermediates of the olefin isomerisation

side reaction rather than a Cr(I) complex (B1). This explanation would also agree with the observation

of hardly any isomerisation during catalysis with 5000 eq. olefin, significant isomerisation at 450 eq

and almost exclusive isomerisation at fewer than 100 eq of olefins.

At low olefin concentration, two other peaks can be identified corresponding to complexes containing

C H and 1-hexene+C H . This would correspond to ions formed from inclusion of 2-methyl-hexene

7

14

7

14

(

C H ) which is formed as initiation product identified by GCMS (see above). As above, the C H

7

1

4

7

1

4

complex may be more likely present as the allyl hydride Cr(III)-complex rather than as mono-olefin-

Cr(I) complex. Either way, the observation of these as significant cations confirms 2-methyl-hexene as

initiation product.

Identification of these 1-hexene containing intermediates was confirmed by addition of 500 eq. of 1-

octene to another portion of the same AlMe activated catalyst. The analogous cationic species in

3

approximately the same proportions were detected as shown in Table 6. No cation corresponding to five

olefins was detected in either experiment. Thus, co-trimerisation of trimer product with two monomers

can be excluded within detection limits and an initially formed complex B4’ leads to olefin substitution

B1 to B2 before oxidative cyclisation can occur.

Table 6. The proposed intermediates that were found to have corresponding peaks in the mass

spectrum during reaction with 1-octene.

rel Intensity

Observed

m/z

Calculated

m/z

Catalytic Intermediate

500 eq

-octene

1

i

+

[

(( Bu CH) TAC)Cr]

517.4338

629.5528

517.4422

629.5674

741.6826

853.8179

965.9431

643.5831

755.7083

100

14

9

7

1

2

3

i

+

(( Bu CH) TAC)Cr(C H )]

2

3

8

16

i

+

(( Bu CH) TAC)Cr(C H ) ] 741.6804

2

3

8

16 2

i

+

(( Bu CH) TAC)Cr(C H ) ] 853.7997

2

3

8

16 3

i

+

(( Bu CH) TAC)Cr(C H ) ] 965.9260

2

3

8

16 4

i

+

(( Bu CH) TAC)Cr(C H )]

643.5775

170

22

2

3

9

18

i

+

(( Bu CH) TAC)Cr(C H )] 755.5165

2

3

17 34

The AlMe activated solution with 450 eq 1-hexene was also tested for incorporation of other olefins.

3

Addition of excess cyclopentene to a portion of this solution resulted in no detectable ions

i

+

[

(( Bu CH) TAC)Cr(cyclopentene)(C H ) ] . Thus, cyclopentene does not seem to bind or insert in any

2 3 6 12 n

significant amount. Addition of 1,7-octadiene to another portion does result in a clear set of signals

i

+

corresponding to [(( Bu CH) TAC)Cr(octadiene)] at m/z = 627.5366 (calcd. 627.5518) but no signal

2

3

corresponding to a cation containing both octadiene and hexene. Thus, the octadiene can bind to Cr(I)

and maybe oxidatively add to form a metallacyclopentane but seems to be incapable of inserting any 1-

hexene. In a synthetic experiment no trimerisation was observed when 1,7-octadiene was added to 1-

hexene before addition to activated catalyst and thus acts as catalyst poison.

2

9

Article Doi

Triazacyclohexane chromium triflate complexes as precursors for the catalytic selective olefin trimerisation and its investigation by mass spectrometry

DOI: 10.1016/j.poly.2020.114572

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.poly.2020.114572

Full text of DOI:10.1016/j.poly.2020.114572

Products guided by the article

R&D Labs maybe for 13310-75-3

Relevant to this article

Hot Product