Origins of high catalyst loading in copper(i)-catalysed Ullmann–Goldberg C–N coupling reactions

A mechanistic investigation of Ullmann–Goldberg reactions using common bases led to the identification of pathways for catalyst deactivation. The solid form of the inorganic phase was found to have critical influence on the mechanism of the reaction.


Reagents and materials
Unless otherwise stated, all reactions were performed under an atmosphere of nitrogen, using flame-dried glassware. Reactions were set up using air-sensitive techniques on a Schlenk manifold or in a nitrogen-filled glovebox. Anhydrous solvents were dried by passing the solvent over activated alumina via the Dow-Grubbs solvent system (Pure Solv TM ) unless otherwise specified. Anhydrous N,N-dimethylformamide and acetonitrile (99.8%) were purchased from Sigma Aldrich. Reaction solvents were degassed by bubbling of nitrogen through for a minimum of 30 minutes. Anhydrous NMR solvents for air sensitive products were degassed using three freeze-thaw cycles. All other reagents were obtained from commercial suppliers and used without further purification.

General analytical data
1 H and 13 C NMR data were performed using deuterated chloroform (CDCl 3 ), unless otherwise stated and recorded on either an Avance 300 (Brüker Biospin GmbH), or Avance 500 (Brüker Biospin GmbH). All kinetic data was obtained on the Avance 500 and 133 Cs NMR was run on an Avance 400 or 500 MHz spectrometer. Microanalysis data was obtained by Tanya Marinko-Covell of the University of Leeds. Hydrogenation of amides was performed using a 300 mL Parr Pressure Reactor. High resolution mass spectra were collected on a Brüker Daltonics (microTOF) instrument operating in the electrospray mode. GC/MS data was obtained using Agilent HP2890 series GC system, with an Agilent HP5973 mass selective detector on EI mode. Column chromatography was performed using Geduran© Si 60 silica gel with the stated solvents. FT-IR spectroscopy measurements were taken on a Brüker Alpha Platinum-ATR. HPLC data was collected from an Agilent 1200 HPLC system using a 30 x 4.6 mm Waters Acquity BEH C18 1.7 µm column and UV/Vis detector set at 285 nm. GC data was obtained from an Agilent Technologies 7890B gas chromatograph using an Agilent J&W HP-5 GC Column, 30 m, 0.32 mm, 0.25 µm.

Base suppliers, purity, and drying protocols
Cesium carbonate was obtained from three individual sources; Sigma Aldrich (99 %, Lot # BCBP3311V), Acros Organics (99.5 %, Lot # A0359643) and Chemetall (Milled, D 50 = 20 µm, D 90 = 50 µm, Bx # 27091B032) and oven dried at 80 C before being used. Prior to SEM analysis, all bases were dried at 75 C for a period of at least 24 hours, ensuring consistency between samples.  NaH (60 % dispersion in mineral oil, 1.048 g, 26.2 mmol) was weighed out into a flame-dried round bottom flask kept under nitrogen. The NaH was washed with three 10 ml batches of degassed pentane and the pentane removed via syringe. 20 ml of anhydrous THF was added at the NaH stirred as a suspension before being cooled down to 0 °C. 2-Pyrrolidinone (2.0 ml, 26.2 mmol) was added dropwise and the reaction stirred at 0 °C until no further bubbles were observed to evolve. The colourless solution was allowed to warm up to room temperature and stirred for 30 minutes further before the solvent was removed in vacuo.

N,N-Dimethylglycine
2.35 g (84 % yield) of an insoluble white solid was obtained and stored in an N 2 filled glovebox. 1 H NMR spectroscopy in d 3 -MeCN showed no starting material peaks, indicating all 2-pyrrolidinone had reacted.

Procedure for obtaining kinetic data
The NMR sample is prepared as outlined above and is then placed into the NMR machine. A scan is taken at 25 ºC to measure t 0 for the reaction, before the temperature probe is heated to the desired temperature (70 ºC). After this temperature is reached, an NMR spectrum is collected every 10 minutes.
The resulting spectra are processed using the TopSpin TM package and the integrals of all aromatic species (6 -8 ppm) are normalised to 100. The conversions are noted for each known compound and the concentration of each species calculated based on the starting concentration of aryl iodide. Rate data were generated using differential function in Origin Pro 9.0.

1 H NMR Assignment
An example NMR taken from the kinetic monitoring study is shown below, many of the peaks belong to the tetrabutylammonium cation due to its stoichiometry and large number of protons. The integration of the in situ NMR spectra was performed on the aromatic peaks shown zoomed in due to significant overlap between peaks in the methoxy region. Aromatic peaks were assigned from 'spiking' experiments, COSY experiments and reference spectra. The NMR spectra of the main impurity matches well with literature data. 5

4-Methoxyphenol speciation in reaction mixture (d 7 -DMF)
4-Methoxyphenol appears to exist as 4-methoxyphenolate in the in situ NMR studies. When a reference spectrum of 4-methoxyphenol was taken in d 7 -DMF, only one multiplet is seen in the aromatic region at 6.8 ppm, with the -OH peak seen at 9.1 ppm. When base is added, loss of the -OH signal is seen and splitting of the aromatic peaks is noted, giving the doublets seen in the in situ studies. Slight differences in ppm exist between reference samples and the in situ data as the in situ study is performed at 70 °C.  GC-MS analysis of the above reaction shows the presence of an impurity which appears on the chromatogram at 5.8 minutes with a mass of 151 (136 -CH 3 ), overlapping slightly with 4-methoxyphenol (m/z = 234).

General procedure for reactions with automated sampling
All experiments were performed as outlined in general procedure above. Reactions were performed in a 4 mL vial with a stirrer disk in a 24 well plate within a glovebox filled with an N 2 atmosphere. Heating, solution additions and sampling were executed by a Freeslate automation robot.
Inside a nitrogen-filled glovebox reaction vials containing magnetic stirrer disks were loaded with the corresponding amounts of CuI (0.1 eq., 0.05 mmol), ligand (L 1 H or L 2 H, 0.1 mmol) and base (Cs 2 CO 3 or K 3 PO 4 , 1.5 eq., see table below for quantities). The kinetic data above were collected using Cs 2 CO 3 (99%) supplied by Chemetall (Base1) with particle size distribution D50 = 20 m (the value of the particle diameter at 50% in the cumulative distribution), D90 = 50 m (the value of the particle diameter at 90% in the cumulative distribution). Two other batches of Cs 2 CO 3 supplied by Sigma-Aldrich (ReagentPlus® 99%, lot# BCBP3311V, Base2) and Acros Organic (99.5%, lot# A0359643, Base3) were included in the study for comparison.

Kinetic data with L 2 H/K 3 PO 4
Protocol described in section 5.1 was followed. The detailed quantities of reagents are summarised below.        A reaction protocol similar to one described in section 5.1 was employed, with L 3 replacing L 1 H. When using oven dried, milled Cs 2 CO 3 , the above reaction was extremely slow, with kinetic monitoring over the first 4 hours showing up to 8 % product. After 48 hours, GC analysis showed 45% conversion to product. In a repeat of the experiment without sampling, conversion to product reached 57 %.

Additive Conversion to product 5 (%) Side products
None 87

In situ EPR experiment
Reaction protocol from section 5.1 was scaled for 2.5 g of 4-iodoanisole 1 (10.7 mmol), with one difference: CuI was added last to the pre-equilibrated reaction mixture containing all other components to enable EPR background measurements. The reaction was performed under argon using in a jacketed glass reactor at 80 C and the solution (with in situ filtration) was circulated through the EPR cavity using an FEP tube (ID 0.8 mm) using a Milligat pump at a flow rate of 3.0 mL/min. EPR spectra were recorded on a Brücker ELEXYS spectrometer at room temperature in X-band with a microwave power of 6.9 mW, a modulation frequency of 100 kHz and modulation amplitude of up to 5 G. Figure S29. In situ EPR measurements over the course of the experiment. Immediately after addition of CuI an isotropic Cu(II) signal emerged, which vanished after about 5-10 min of heating. The Cu(II) signal was not observed afterwards

CuI/L 1 H/DMF experiment
Sample was prepared using a standard EPR tube in a glovebox with degassed DMF and measured using the same equipment as described above. The sample was cooled with liquid nitrogen during this measurement. Slight line broadening is observed in CuII signal of reaction solution (red graph) which lead to loss in intensity at high field peaks. These effects could be caused by a slower tumbling rate due to interaction with other reaction components (substrate, base etc.) which were absent in the CuI+DMG experiment (black).

Sample preparation
Bases were analysed by SEM either fresh from the bottle or following drying in a 75 ºC oven when specified. Samples were stored under argon prior to analysis.

SEM/EDX experiments
All samples were mounted in powder form, under exclusion of ambient moisture, on SEM sample stubs using adhesive carbon film. The samples were then coated with a thin layer of iridium using a Cressington 208HR sputter coater. Samples were imaged using an FEI Nova NanoSEM 450 operating typically at 3 kV. The gun voltage was increased to 10 keV to collect EDX point spectra.

Surface area measurements
Nitrogen adsorption measurements indicate that all three base powders have very small specific surface areas (<1 m 2 /g). Nitrogen adsorption measurements are not suited to accurately quantify such low specific surface areas, so a comparison between the three bases using this method is not feasible.

Characterisation of solid during and after reaction
Reactions were performed as outlined in section 5.1. Following completion of the reaction, the solution stopped stirring and was removed via syringe at 90 °C, the solid was then washed three times with hot, anhydrous Et 2 O and evaporated to dryness. The dried solid was stored under argon until analysis was performed.

Base Isolated Post-reaction -Chemmetal Cs 2 CO 3
In order to probe the link the morphology and surface area of the bases with their kinetic behavior, solid samples were recovered from the reaction mixtures using Base1 and Base2 at 2.5 and 5.0 hours. SEM pictures of these are shown in Figure S34. Chemical speciation was achieved by a combination of EDX data and power X-ray diffraction ( Figure S37). Comparing Figure S34a-d showed that, as the reaction progresses, bright cubic crystals of CsI (in SEM pictures, Figure S34e), assigned based on EDX characterization ( Figure  S35-S37), accumulates on the solids. More importantly, the soft-edge structure of Cs 2 CO 3 gave way to a crystalline material which was identified by EDX to contain Cs, C, and O. Powder X-ray diffraction data of the sample recovered from reaction using Base2 at 5 hours indicated that these crystals are CsHCO 3 ( Figure  S34e). No diffraction signal for Cs 2 CO 3 was detected for this sample. Thus, during the reaction, the morphology all Cs 2 CO 3 samples changes toward a common final state. The difference in the initial rates of reaction using Base1, Base2 and Base3 must therefore originated from their initial states. We hypothesized that the difference in surface area, and hence their rates of dissolution in DMF, 6 are the main reason for the observed induction period with Base1.    S63 Figure S38. EDX spectrum of the bulk material

XRD measurements
X-Ray powder diffraction data was collected on a Bruker D2Phaser Diffractometer. Powdered samples were mounted on a silicon wafer by evenly distributing the powder over the wafer. All samples were rotated during data collection to ensure more homogeneous (i.e. isotropic) diffraction patterns. Data collection was carried out at room temperature, using Cu Kα radiation (λ = 1.54184 Å). Diffraction patterns were recorded in step-scan mode with a step size of (2θ) 0.7°, from (2θ) 5° to 55° (30 secs per step) using a 0.1 or 0.6mm divergent slit. Samples showed no significant evidence of degradation within the X-ray beam. The software package Diffrac.Suite Eva14 was used to process the experimental powder X-ray diffraction data collected, and Mercury 3.3 was used to simulate the powder diffraction pattern of the single crystal structures taken from literature.

10.3.1Sampling air-sensitive reactions
To ensure no oxygen is introduced into the reaction system by standard sampling techniques, enough needles for the desired amount of samples are introduced prior to the vacuum/inert gas cycles, with each needle thoroughly flushed. The set-up as shown below has been found to be a simple method of obtaining kinetic data with these reactions.  CuI (19 mg, 0.1 mmol), 1,10-phenanthroline L 3 (36 mg, 0.2 mmol), 1,1'-biphenyl (15.4 mg, 0.1 mmol) and 4-iodoanisole 1 (234 mg, 1 mmol) were added to a flame-dried Schlenk flask under nitrogen. Three purge/flush cycles were performed prior to addition of degassed DMF (2.5 mL). In a separate flask, Napyrrolidinone 6 (160 mg, 1.5 mmol) was mixed with degassed DMF (2.5 mL) under nitrogen to give a suspension. Both flasks were heated to 90 °C and the Na-pyrrolidinone 6 suspension was added in three portions at 0, 1 and 2 hours. The reaction was stopped after 20 hours and GC analysis showed 60 % conversion to product.

10.3.3Experiment with portion-wise addition of 6
The same procedure as above was performed with three additions of added dropwise over 5 minutes which is represented by the blue line in Figure S37. Samples were quenched by exposure to air and dilution with MeCN, GC analysis was used to give conversion data and a final sample at 20 hours gave 60 % conversion to product.
S67 Figure S41. Conversion of 1 and 6 to C-N coupling product 5 following portion-wise additions of salt 6 as a suspension

11.Base Solubility by 133 Cs NMR
133 Cs NMR spectra were collected at 90  C in DMF. An insert tube was used to include 0.1 M aqueous solution of CsNO 3 as external standard, while preventing Cs exchange between the standard and the sample.
Samples were prepared by stirring saturated solutions of CsI, CsHCO 3 and Cs 2 CO 3 (Acros Organics, Lot A0359643) in anhydrous DMF at 90 C for 2 hours. After the mixture was allowed to settle, the supernatants (0.5 mL) were transferred to an NMR tube via a syringe fitted with a metal needle (pre-flushed with hot DMF). An insert tube filled with 0.1 M CsNO 3 in D 2 O was inserted into the NMR tube and the samples kept at 90 C. The samples were inserted to the NMR spectrometer at 70 C and quickly heated to 90 C before data acquisition.
The 133 Cs NMR data ( Figure S38 The saturated concentration of CsI was much higher than those of Cs 2 CO 3 and CsHCO 3 . Thus, monitoring Cs content of the reaction mixtures does not reflect the solubility of Cs 2 CO 3 in these systems.