Combined Experimental and Computational Study of Ruthenium N-Hydroxyphthalimidoyl Carbenes in Alkene Cyclopropanation Reactions

A combined experimental–computational approach has been used to study the cyclopropanation reaction of N-hydroxyphthalimide diazoacetate (NHPI-DA) with various olefins, catalyzed by a ruthenium-phenyloxazoline (Ru-Pheox) complex. Kinetic studies show that the better selectivity of the employed redox-active NHPI diazoacetate is a result of a much slower dimerization reaction compared to aliphatic diazoacetates. Density functional theory calculations reveal that several reactions can take place with similar energy barriers, namely, dimerization of the NHPI diazoacetate, cyclopropanation (inner-sphere and outer-sphere), and a previously unrecognized migratory insertion of the carbene into the phenyloxazoline ligand. The calculations show that the migratory insertion reaction yields an unconsidered ruthenium complex that is catalytically competent for both the dimerization and cyclopropanation, and its relevance is assessed experimentally. The stereoselectivity of the reaction is argued to stem from an intricate balance between the various mechanistic scenarios.


General experimental
Materials. All reactions were carried out using oven-dried glassware under an atmosphere of argon or nitrogen. Dry solvents were obtained using a solvent purifier equipped with activated alumina columns. All solvents used in work-up procedures, silica gel column cromatography, HPLC analysis and/or purification were obtained from commercial suppliers and used without further purification. When appropriate, degassing of anhydrous solvent was achieved through three freeze-pump-thaw cycles or by bubbling Ar for 30 minutes under sonication. Reactions were performed in common pyrex round bottom flasks, microwave vials 2 -8 ml (VWR or Biotage®), or 5 -20 ml flat bottom vials (Cronus, SMI-LabHut Ltd. or VWR®) crimped on top with 20 mm Sil/PTFE Septa. Reaction temperatures were maintained using water-ice baths. Slow additions were performed using Landgraf HLL LA-30 or Harvard Apparatus Pump11Elite syringe pumps.
Kinetic measurements. All experimental kinetic measurements were performed using the Man on the Moon X103 kit according to the specifications by the manufacturer (for more specifics on the measurements, see Kinetic Profiling Details below).
Reagents were obtained from commercially available sources and used as received unless stated otherwise. NHPI-DA was prepared as previously reported by Mendoza and co-workers. 1 Catalyst (S)-Ru-Pheox(CH3CN)4PF6, was prepared as described by Iwasa. 2 Commercially available alkenes were purchased from Sigma-Aldrich, Fluorochem or Acros Organics. Liquid alkenes were quickly distilled, using a Hickman head, prior to use.

Kinetic Profiling Details
Crystalline NHPI-DA 1a was used and EDA 1b concentration was checked by 1 H-NMR (usually around 85% Wt, in CH2Cl2, SigmaAldrich) before use. Typically, a 0.15-0.21 M solution of the corresponding diazocompound in dry CH2Cl2 was prepared in a volumetric flask under argon. Alkenes were distilled using a Hickman head and a stock solution in dry CH2Cl2 was prepared under argon. A stock solution of (S)-Ru-Pheox (0.015 M) was prepared in the glovebox, and taken out just before the experiments to avoid decomposition of the catalyst (Note: we have observed that solutions of the catalyst change color yellow to green when stored for 1-2 hour under a positive pressure of argon; the resulting green solutions lead to lower and irreproducible activity).
A dried 12 mL "Man on the Moon" reactor was backfilled with argon (Note: the space between the valve and the sensor was also purged using the 3-way valve). The corresponding diazocompound (0.225 mmol) and alkene were charged into the reactor and the volume of the mixture was adjusted to 2.1 mL with dry CH2Cl2 (this will give a final concentration of diazocompound of 0.1 M, after the addition of the catalyst solution). The pressure of the reaction after the addition has been calculated to be around 0.5 bar relative to the initial pressure inside the vessel. The mixture is then placed in an ice-bath and stirred for 2 minutes. Then the valve is changed from the argon flow to the detector and the measurement is started. The system is allowed to stabilize, until a constant pressure reading is achieved (approximately 5-8 min). The solution of catalyst is then injected in the reaction mixture and the syringe is carefully removed from the setup.
A typical data set looks as shown in Figure S1: Figure S1: An example of raw pressure data set.

S3
Note: The process of the injection of the catalyst through the septum also causes a variation in the pressure, but it is usually minor compare to the variation caused by the N2 evolution. However, in some cases like in the example depicted above, this effect is more important. For a better comparison between experiments, the raw data is normalized from the first point after the injection spike (t=0) . In Figure  S1, the first data point (for both time and pressure), was established after 7.00 minutes. The raw data is then transformed into the following normalized plot ( Figure S2): Figure S2: A pressure data set after normalization.
As a control, the dimerization reaction has been used as a control experiment between reaction sets. In Figure S3, three different runs are shown. The red and blue profiles show how the robustness of the method. In contrast, the grey profile shows a slightly slower reaction and lower final pressure, most likely due to a wrong placing of the septum.

Reproducibility study
Run 1 Run 2 Run 3

Kinetic measurement -Dimerization reaction
Dimerization of ethyl diazoacetate (EDA) 2a: The measurement was performed following the general description in Section 2, employing ethyl diazoacetate 1b (

Dimerization of N-hydroxyphthalimidoyl diazoacetate (NHPI-DA) 1a:
The measurement was performed following the general description in Section 2, employing NHPI-DA 1a To prove that the inizial induction period is not caused by the dissolution of N2 in the reaction solvent, the same reaction was performed employig N2-saturated dichloromethane. A similar profile was obtained.

Determination of the Nucleophilicity Parameter (N) for NHPI-DA (1a)
General Experimental Details: UV-Vis measurements were performed in a quartz cuvette fitted with a septum and under Ar. All solutions were prepared under Ar using dry CH2Cl2 (distilled over CaH2 and stored over 4 Å molecular sieves in a glovebox), and were kept in the glovebox. All the glassware was dried overnight in an oven. Stock solution were prepared in volumetric flasks, and the aliquots were added using glass Hamilton syringes. The temperature in the room where the experiments were performed was measured to be 19-21 °C. Benzhydrylium cation (mfa)2CH + BF4 8 was prepared following methods described in the literature. 4 Figure S6: Structure of the benzhydrylium cation 8.

Kinetics and determination of N:
According to the method described by Mayr and co-workers, 4 the apparent kinetic constant kobs was calculated from the time profile of the absorbance data (eq. S1). Then, the bimolecular kinetic constant k2 was calculated (eq. S2). The bimolecular kinetic constants calculated from the data of each run (k2,i) were used to estimate the nucleophilicity parameter N for the run (Ni)(eq. S3). To apply this equation, we used the average of the slope parameter (s = 0.9275) obtained by Mayr and co-workers 4 for similar diazocompounds (Table S1) and the electrophilicity parameter for the benzhydrylium cation 8 (E = -3.85). Mayr and co-workers determined that the slope parameter (s) has a very similar value across a series of different diazocompounds and recommended this approximation for preliminary estimation of nucleophilicity. 4 We averaged the results of three independent runs (n=3; eq. S4) taking the data on the initial 60 s (20 data points).


To obtain the confidence interval we used the partial derivatives of these functions (eq. S5-7). We calculated the standard error of the averaged nucleophilicity parameter N (eq. S8-10) and then estimated the confidence interval at 95% confidence of the normal distribution.

Procedure for kinetic measurements using (mfa)2CH + BF4 -(8)
A quartz cuvette was charged with 2.0 mL of dry CH2Cl2 in a glovebox. After recording the blank, 8 was added from a stock solution and the UV-Vis spectrum of the blue solution was recorded. Next, partial UV-Vis spectrum (λ= 575-565 nm) was recorded every 2-2.4 s, first with 8 alone. After 1-2 minutes, the corresponding amount of diazoacetate 1 solution was added to the cuvette quickly under argon. The solution was made homogenous by shaking gently and placed in the spectrometer. The time elapsed S10 between the addition of the diazocompound and the first measurement was measured to be 6-10 s, depending on the experiment. The absorbance at t = 0 s is the average of the points obtained before the addition of the diazocompound, and it has been corrected with the corresponding dilution factor. The absorption of the diazocompound in this region of the spectrum is negligible.

Recording of the absorbance vs time for NHPI-DA (1a) and (mfa)2CH + BF4 -(8)
The procedure for kinetic measurements was followed for three different initial concentrations of 1a and 8. The variation of the absorbance at 569 nm with time (s) was plotted:

Preliminary estimation of the nucleophilicity (N) of NHPI-DA (1a)
As described in the General Experimental Details of this section, we used the kinetic data to obtain Ni for the corresponding initial concentrations in each run and calculated the statistics of the fit (Table S2). Table S2 -Calculation of N through averaging of three independent kinetic runs (see Figure S7): With these individual Ni values, we obtained an averaged nucleophilicity parameter for NHPI-DA, 1a (eq. S11): NOTE: This confidence interval should not be over-interpreted in absolute terms and only as an estimate of the consistency of our data. Further studies would be required to estimate N with high-accuracy using other benzhydrylium cations under stop-flow conditions.

Validation of the experimental set-up: nucleophilicity (N) of ethyl diazoacetate (1b)
We have reproduced in-house the nucleophilicity (N) value determined by Mayr 4 for the benchmark ethyl diazoacetate 1b to validate our experimental set-up, using the same benzhydrylium cation 8. The general procedure for kinetic measurements was followed, resulting in a faster decay of the absorbance over time (Figure S8), due to the higher nucleophilicity of 1b. Identical mathematical treatment resulted in a nucleophilicity value in agreement with the one reported by Mayr and co-workers (eq. S12-13):

Determination of reaction order in olefin
The overlay plot [5][6][7] was generated calculating the fraction of nitrogen derived from the cyclopropanation cycle. In order to do that, several reactions were analyzed by 1 H NMR to determine the concentration ratio of product 3d to dimer 7a. The reactions were monitored by N2 evolution and stopped by addition of excess methanol once the right conversion was reached. The plot below summarizes the results: Figure S11: Summary of the product:dimer (3d: 7a) ratio determination by 1 H NMR analysis.
From these experiments, it resulted that until 56% conversion, the ratio was constant with a value of 85:15 for the product 3d. This allowed us to estimate the pressure of nitrogen generated by the cyclopropanation cycle using the ideal gas approximation (eq. S14): This pressure was then used in VTNA as shown by Burés and co-workers to obtain order in olefin = 1 ( Figure S14). Additionally, the linearity of the plot informs us that the order in diazo = 0. Below are shown the VTNA plots, with order values 0 (Figure S12), 0.9 (Figure S13), 1 ( Figure S14) and 1.1 (Figure S15)

11.a HPLC traces for the stoichiometric experiment HPLC trace for compound rac-3a
HPLC traces for compound 3a (entry 2)