Accelerating the insertion reactions of (NHC)Cu–H via remote ligand functionalization

Most ligand designs for reactions catalyzed by (NHC)Cu–H (NHC = N-heterocyclic carbene ligand) have focused on introducing steric bulk near the Cu center. Here, we evaluate the effect of remote ligand modification in a series of [(NHC)CuH]2 in which the para substituent (R) on the N-aryl groups of the NHC is Me, Et, tBu, OMe or Cl. Although the R group is distant (6 bonds away) from the reactive Cu center, the complexes have different spectroscopic signatures. Kinetics studies of the insertion of ketone, aldimine, alkyne, and unactivated α-olefin substrates reveal that Cu–H complexes with bulky or electron-rich R groups undergo faster substrate insertion. The predominant cause of this phenomenon is destabilization of the [(NHC)CuH]2 dimer relative to the (NHC)Cu–H monomer, resulting in faster formation of Cu–H monomer. These findings indicate that remote functionalization of NHCs is a compelling strategy for accelerating the rate of substrate insertion with Cu–H species.


General Considerations
All reactions were conducted under a nitrogen atmosphere in a Vacuum Atmospheres glovebox or using standard Schlenk techniques unless otherwise noted. Reagents were purchased from Sigma-Aldrich, Alfa Aesar, TCI America, Combi-Blocks, or Oakwood Chemicals, and used as received unless otherwise noted. Toluene, THF, diethyl ether, pentane, and hexanes were sparged with N2, purified by passage through neutral alumina using an Innovative Technology, Inc., PureSolv solvent purification system, and stored over activated 4 Å molecular sieves. 1-hexene and 1,5-hexadiene were purified by passage through a plug of silica or alumina, freeze-pump-thaw degassed, and stored over activated 4 Å molecular sieves. 3-hexyne and C6D6 were freeze-pump-thaw degassed and dried over activated 4 Å molecular sieves.
NMR spectra were acquired on Varian NMR spectrometers at 25 °C unless otherwise noted. The chemical shifts are referenced to tetramethylsilane (0.0 ppm) using internal CDCl3 (

Synthesis of IPr*Cl•HCl
To a 500 mL round-bottom flask was added 2,6-dibenzhydryl-4-chloroaniline (30.6 g, 66.5 mmol), 40% glyoxal (10.5 g, 73.5 mmol), solid paraformaldehyde (4.20 g, 73.5 mmol), 200 mL CHCl3, and a large stir bar. The reaction mixture was heated to 70 ºC, and HCl (6.30 mL, 73.5 mmol) was slowly added. The reaction mixture was stirred at 70 ºC for 24 h, and then all volatiles were removed to give a brown foam. To the crude mixture was added EtOAc (20 mL) and hexanes (150 mL). The mother liquor was decanted, leaving a sticky brown residue. EtOAc (20 mL) was added to the flask. The flask was fitted with a reflux condenser and heated at 70 ºC with vigorous stirring for 1 hr producing a white precipitate in a brown solution. The hot mixture was filtered on a medium porosity glass frit and washed with room temperature EtOAc (20 mL x 2) and hexanes (10 mL) to give a white solid that was dried under vacuum. Yield = 11% (3.50 g, 3.53 mmol).     Toluene H2O S16

Calculation of %Vbur and G
To quantify steric effects, we calculated %Vbur and G (solid angles) using the XRD structures of [(IPr*R)CuCl] (R = Cl, Et, OMe, t Bu) as well as the structure of [(IPr*Me)CuCl] from the literature. 8 %Vbur was calculated with SambVca 2.1 9 (https://www.molnac.unisa.it/OMtools/sambvca2.1/index.html ) using a 5.5 Å sphere radius with the metal-NHC bond length normalized to 2 Å and H atoms omitted. The results of these calculations are shown in Table S1. The calculations indicate ~58% %Vbur for R = Cl, Me, Et, and OMe with only a 1.3% increase along the series. For context, Nolan and co-workers suggest that a range of ~1.2% can be expected for different conformations of a flexible ligand. 10 A slightly higher value (%Vbur = 61.5%) is observed for R = t Bu. However, the steric maps ( Figure S10) suggest that the changes may be largely due to minor differences in the orientation of the CHPh2 substituents. Accordingly, when the CHPh2 groups are removed and %Vbur is reevaluated, the values are identical within error, since the R-groups are not contained in the coordination sphere evaluated in the calculation. This finding also indicates that there is no significant change in the primary coordination sphere of Cu as a function of the remote steric bulk.
Solid angles (G) have been proposed as an alternative to %Vbur in situations where the steric bulk is further from the metal center. 11, 12 G values were calculated using the SolidG program (https://xray.chem.wisc.edu/solid-g/ ) with the metal-NHC bond length was normalized to 2 Å. The results of these calculations are shown in Table S1. Based on the difference in G calculated for the two crystallographically inequivalent molecules in the XRD structure of [(IPr*OMe)CuCl], changes of ~1.5% are likely insignificant. The values of G for IPr*Cl, IPr*Me, IPr*Et, and IPr*OMe span a very small range and G for IPr*Cl » IPr*Me < IPr*OMe » IPr*Et << IPr* t Bu, as expected. The same trend in G is observed for models where the CHPh2 groups are removed.
Overall, this analysis indicates that %Vbur and (to a lesser extent) G are significantly influenced by the orientations of the CHPh2 groups, which are subject to influences from crystal packing. The calculated steric parameters are roughly the same within error. These methods are therefore not suitable for analysis of structures like the IPr*R ligands reported here that have only subtle structural differences far from the metal center and contain conformationally flexible groups.  Cyclic voltammograms were recorded on ~4 mM solutions of [(IPr*R)Rh(COD)Cl] in CH2Cl2 containing 0.1 mM NBu4PF6 as supporting electrolyte. Measurements were performed on a CH Instruments 700D potentiostat using a standard three-electrode cell consisting of a glassy carbon working electrode, glassy carbon rod counter electrode, and silver wire pseudo-reference electrode. The potentials are reported versus Cp2Fe 0/+ . Table S2. Comparison of ratio of anodic to cathodic current and peak-to-peak separation of ferrocene and [(IPr*R)Rh(COD)Cl] (scan rate = 100 mV/s).

General Notes
[(IPr*Me)CuH]2 was prepared as previously described. 13 With the exception of [(IPr* t Bu)CuH]2, which was prepared from isolated [(IPr* t Bu)Cu(O t Bu)], all [(IPr*R)CuH]2 were synthesized by addition of HSi(OEt)3 to Cu alkoxide complexes generated in situ. The copper hydride complexes are extremely sensitive to air and moisture, which prevents thorough drying and recrystallization. We have therefore not attempted to obtain elemental analysis data. The purity of the copper hydrides is estimated at ~90-95% based on 1 H NMR spectroscopy.
During attempts at recrystallization of [(IPr*R)CuH]2 or upon storage of isolated material for more than a few weeks, even in a glovebox freezer at -30 °C, [(IPr*R)Cu(OH)] was observed as a major decomposition product. The Cu-OH complexes can be identified using 1 H NMR spectroscopy. Their assignment is based on comparison to [(IPr*Me)Cu(OH)], which has been reported in the literature. 14 We also describe the isolation of [(IPr*OMe)Cu(OH)] below. In the Cu-OH spectra, characteristic singlets appear in between the resonances corresponding to CHPh2 and H-C=C-H in the corresponding copper hydrides. A singlet corresponding to the Cu-OH is also present at approximately -0.5 ppm but is broadened beyond detection in some cases. We have not isolated and fully characterized all of the Cu-OH complexes, but the key resonances described above are provided for each complex to allow identification of this common impurity.

Synthesis of [(IPr*Et)CuH]2.
In a glovebox, 100 mg (0.096 mmol) of [(IPr*Et)CuCl] was dissolved in 2 mL toluene and 30 µL (0.10 mmol, 1.04 equiv) of 40% sodium tert-pentoxide in toluene was added. After 2 hours, the suspension was filtered through Celite. The filtrate was concentrated to 2 mL, and 30 µL (0.16 mmol. 1.7 equiv) triethoxysilane was added, resulting in rapid formation of an orange solution. After 5 minutes, 10 mL of pentane was added causing precipitation of a yellow solid. The solid was isolated by vacuum filtration, washed with pentane and dried under vacuum. Yield: 70 mg (72%) Yellow crystals suitable for XRD were grown by layering a concentrated THF solution of [(IPr*Et)CuH]2 in the presence of excess (~3 equiv) HSi(OEt)3 with hexanes at room temperature.

Synthesis of [Cu(IPr* t Bu)(O t Bu)].
In an N2-filled glovebox, a solution of 500 mg (0.46 mmol) [Cu(IPr* t Bu)(Cl)] in 10 mL THF was treated with a solution of 54 mg (0.48 mmol, 1.05 equiv) potassium tert-butoxide in 5 mL THF. After 1 hour, the solvent was removed under vacuum. The resulting residue was redissolved in toluene and filtered through Celite. The filtrate was concentrated to 5 mL, and 80 mL pentane was added. This solution was stored in a -30 °C freezer overnight, resulting in the precipitation of a white solid. The solid was collected on a fritted glass funnel, washed with 3x10 mL pentane, and dried under vacuum. Yield: 440 mg (85%).

Synthesis of [(IPr* t Bu)CuH]2.
A slurry of 202 mg (0.18 mmol) [(IPr* t Bu)Cu(O t Bu)] in 5 mL diethyl ether was treated with 50 mmol (0.27 mmol, 1.5 equiv) HSi(OEt)3 causing formation of a homogeneous orange solution. The solution was filtered through a fiberglass plug, and the volatile materials were removed under vacuum. Upon addition of 10 mL pentane, a yellow solid precipitated. The solid was isolated by vacuum filtration, washed with 3 x 3 mL pentane, and dried under vacuum to afford 160 mg of crude [(IPr* t Bu)CuH]2 (approximately 80% yield) as a yellow solid. Based on 1 H NMR spectroscopy, this material is 90-95% pure, and contains pentane as well as other minor unknown impurities. We have been unable to fully purify [(IPr* t Bu)CuH]2 due to its thermal, moisture, and light sensitivity (see Figure S29).
Orange crystals suitable for XRD were grown in a -30 °C freezer by vapor diffusion of pentane into a concentrated THF solution of [(IPr* t Bu)CuH]2 in the presence of excess (~3 equiv) HSi(OEt)3.

Note on the synthesis of [(IPr*OMe)Cu(µ-H)]2.
During our attempts to isolate [(IPr*OMe)Cu(µ-H)]2, [(IPr*OMe)Cu(OH)] was consistently present as a 5-10% impurity. We were unable to develop a reproducible procedure to remove it. We verified that the impurity arises from adventitious water by treating as-isolated [(IPr*OMe)CuH]2 with a solution of H2O in THF and isolating and characterizing the product, as described below. The 1 H NMR spectrum of the resulting material is identical to the impurity in the as-isolated [(IPr*OMe)CuH]2 ( Figure S27). Based on 1 H NMR spectroscopy, quantitative conversion (based on Cu-H) to inserted products is observed for the reaction of the as-isolated [(IPr*OMe)CuH]2 with substrates; the small amount of [(IPr*OMe)Cu(OH)] remains unreacted. Furthermore, [(IPr*OMe)Cu(OH)] has no detectable absorbance in the visible region ( Figure S28). Based on these control experiments, we are confident that the presence of the minor Cu-OH impurity does not impact the results of our reactivity studies with [(IPr*OMe)CuH]2.

Synthesis of [(IPr*OMe)CuH]2.
A slurry of 251 mg (0.24 mmol) [(IPr*OMe)CuCl] in 3 mL of toluene was treated with 75 µL (0.25 mmol, 1.04 equiv) of 40% sodium tert-pentoxide in toluene. A white solid precipitated over the course of 10 minutes. A 75 µL (0.41 mmol, 1.7 equiv) portion of HSi(OEt)3 was added, causing formation of an orange solution. The resulting suspension was filtered through a Celite pad, and the pad was washed with toluene until the washes became pale yellow (approximately 15 mL). The filtrate was concentrated to 5 mL, and 25 µL HSi(OEt)3 and 15 mL pentane were added, causing precipitation of an orange solid. The solid was isolated on a fritted glass filter funnel, washed with 10 mL pentane, and dried under vacuum to afford 221 mg of a bright orange solid. Based on 1 H NMR spectroscopy, this material contains residual pentane and toluene, as well as 0.

Synthesis of [(IPr*OMe)Cu(OH)].
A 91 mg portion of [(IPr*OMe)CuH]2 containing 15% [(IPr*OMe)Cu(OH)] (0.089 total mmol Cu) was dissolved in 5 mL of THF and treated with 1 mL of 0.11 M H2O in THF (0.11 mmol, 1.2 equiv). After 3 hours, the solution became colorless. The solvent was removed under vacuum, and the resulting material was redissolved in 2 mL THF and filtered through a fiberglass plug. The filtrate was layered with 15 mL hexanes and stored in a -35 °C freezer overnight resulting in precipitation of a white solid. The solid was collected on a fritted glass filter funnel, washed with pentane, and dried under vacuum. Yield: 53.6 mg (59%)

Synthesis of [(IPr*Cl)CuH]2.
In an N2-filled glovebox, a suspension of 99 mg (0.094 mmol) [(IPr*Cl)CuCl] in 2 mL toluene was treated with 30 µL (0.10 mmol, 1.05 equiv) of 40% sodium tert-pentoxide in toluene . After 15 minutes, 25 µL (0.14 mmol. 1.4 equiv) of triethoxysilane was added, resulting in formation of an orange solution. The reaction mixture was filtered through Celite and the filtrate was concentrated to 5 mL under vacuum. Pentane (15 mL) was added, causing precipitation of a yellow solid. The solution was left in a -30 °C freezer for 1 hour. The solid was then collected on a glass frit, washed with 3 x 5 mL pentane, and dried under vacuum. Yield: 80 mg (83%).
Orange crystals suitable for XRD were grown by vapor diffusion of pentane in to a concentrated toluene solution of [(IPr*Cl)CuH]2 in the presence of excess (~3 equiv) HSi(OEt)3 at room temperature.

Synthesis of [(IPr*Me)CuD]2.
Inside a nitrogen-filled glovebox, to a 20 mL scintillation vial was added [(IPr*Me)CuCl] (500 mg, 0.490 mmol), a small stirring bar, and toluene (10 mL). To the slightly soluble mixture was added 150 µL of sodium pentoxide (33% in toluene). The reaction was stirred at ambient temperature for 30 min to give a yellow solution with no remaining insoluble materials. The reaction mixture was filtered through a medium porous glass frit containing Celite and washed with 3 mL of toluene. The filtrate was transferred to a 100 mL round-bottom flask and 15 mL of THF was added. To this mixture was added LiAlD(O t Bu)3 (138 mg, 0.540 mmol) in 5 mL of THF. The reaction mixture was stirred at ambient temperature during which time it turned intensely yellow, then orange. After 3 h, the volume of the reaction was reduced by 1/3 and the suspension was stirred for another 1 h. The resulting bright yellow precipitate was collected on a medium porous glass frit, washed with 10 mL of THF, washed with 10 mL of pentane, and dried under vacuum to give a yellow-orange solid. Yield = 35% (168 mg, 0.350 mmol). 1 H NMR data of [(IPr*Me)CuD]2 matches that of [(IPr*Me)CuH]2 except that the hydride resonance is not observed.    Single crystals for XRD measurement were obtained from layering pentane over a saturated THF solution containing the complex at 25 ºC.

Synthesis of [(IPr*Me)Cu-C(Ph)=CH(Ph)] (2)
Inside a nitrogen-filled glovebox, to a 20 mL oven-dried scintillation vial was added [(IPr*Me)CuH]2 (100 mg, 0.0510 mmol), a small stirring bar, and THF (5 mL). To this yellow reaction mixture was added diphenylacetylene (19.0 mg, 0.107 mmol) leading to formation of a colorless mixture. After 12 h, the reaction mixture was filtered through as glass pipette into a new 20 mL scintillation vial, and was layered with pentane (15 mL). An off-white crystalline solid precipitated and was isolated by collected by decanting the mother liquor, rinsing with pentane (2 mL), and drying under vacuum. Yield = 65%. (169 mg, 0.150 mmol).  Single crystals for XRD measurement were obtained by the slow diffusion of pentane into a saturated THF solution containing the complex at 25 ºC.

Synthesis of [(IPr*Me)Cu-C(Et)=CH(Et)] (3)
Inside a nitrogen-filled glovebox, a 20 mL scintillation vial was charged with [(IPr*Me)CuH]2 (52 mg, 0.026 mmol), a small stirring bar, and toluene (2 mL). Neat 3-hexyne (150 µL, 1.32 mmol, 50 equiv) was added to the resulting orange suspension. The reaction mixture was stirred overnight at ambient temperature to give a colorless solution which was filtered through a fiberglass plug, layered with 15 mL pentane, and placed in a -30 °C freezer. After 5 days, colorless crystals formed. The mother liquor was decanted off, and the crystals were washed with pentane and dried under vacuum. Yield: 40 mg (70%).

Synthesis of [(IPr*Me)Cu(n-hexyl)] (4)
Inside a nitrogen-filled glovebox, a 20 mL scintillation vial was charged with [(IPr*Me)CuH]2 (52 mg, 0.026 mmol), a small stirring bar, and toluene (2 mL). Neat 1-hexene (350 µL, 2.8 mmol, 100 equiv) was added to the resulting orange suspension. The reaction mixture was stirred for 5 hours at ambient temperature to give a colorless solution which was filtered through a fiberglass plug, layered with 15 mL pentane, and placed in a -30 °C freezer. Colorless crystals formed overnight. The mother liquor was decanted off, and the crystals were washed with pentane and dried under vacuum. Yield: 45 mg (79%).  Colorless single crystals suitable for XRD were grown by layering a concentrated THF solution of 4 with pentane and storing overnight in a -30 °C freezer.
NaOCH2C3H6 was prepared from the reaction of HOCH2C3H6 (750 mg, 10.4 mmol) and NaHMDS (2.10 g, 11.4 mmol) in toluene (20 mL) at ambient temperature. After 3 h, the reaction mixture was concentrated to dryness to give a white solid that was triturated with hexanes, collected on a medium porosity glass frit, washed with hexanes (10 mL

UV-Visible Kinetics Experiments
UV-Visible kinetics studies were performed at 25 °C on a Cary 60 UV-Visible spectrophotometer equipped with a Peltier temperature-controlled cuvette holder. Reactions were monitored using the scanning kinetics function (280 nm -650 nm) until at least 95% of the initial UV-Visible signal had decayed. For reactions with 3-hexyne, N-benzylideneaniline, and benzophenone, kobs was determined from linear fits of ln([Cu2H2]) vs time (kobs = -1 x slope). For reactions with 1-hexene and 1,5-hexadiene, kobs was determined from linear fits of [Cu2H2] 0.5 vs time at 3 -4 substrate concentrations (kobs = -2 x slope), and k3/2 was calculated from a linear fit of kobs vs substrate concentration (k3/2 = slope).
For reactions of 0.1 mM [(IPr*R)CuH]2 (R = Me, Et, OMe, t Bu) with 3-hexyne, N-benzylideneaniline, and benzophenone, experiments were performed by injecting substrate solutions in to 1 cm cuvettes sealed with a rubber septum. In each case, control reactions (injection of neat toluene) confirm that negligible background decomposition occurs in this setup on the timescale of the insertion. For R = Cl, the longer reaction time resulted in more significant background decay, so a modified cuvette sealed with a Kontes valve was used. The very similar rates observed for different substrate identities and concentrations demonstrate the high reproducibility of this methodology (Table S5).
Reactions of [(IPr*R)CuH]2 with 1-hexene and 1,5-hexadiene were performed at slightly higher concentration of copper hydride (0.2 mM) to minimize the impact of background decomposition during the longer reaction times needed for these substrates. Reactions were performed in 5 mm cuvettes sealed with Kontes valves. Substrate was injected through a sidearm sealed with a septum that is isolated from the cuvette when the Kontes valve is sealed. In some cases, the substrate solution was injected prior to removing the cuvette from the glovebox. Negligible differences in rate were observed for these two methods. Control reactions (injection of neat toluene) were performed to confirm that negligible background decay occurs in this setup on the timescale of the insertion reactions. In addition, the absorbances of the products were similar across all substrate concentrations; since the reactions are slower at lower substrate concentrations, this suggests that reaction with substrate is much faster than background decomposition of the copper hydride. Furthermore, the plots of kobs vs substrate concentration are highly linear. Taken together, this demonstrates the high reproducibility that can be achieved with this setup.

Representative procedure for reactions of [(IPr*R)CuH]2 with 3-hexyne, N-benzylideneaniline, and benzophenone
A 0.5 mM solution of [(IPr*R)CuH]2 in toluene was prepared in a 10 mL volumetric flask. A 2.3 mL aliquot of this solution was diluted to 10 mL with toluene to give a 0.12 mM solution. A 2.6 mL aliquot of this solution was transferred to a 1 cm cuvette containing a stir bar and the cuvette was sealed with a rubber septum. A 0.15 M solution of 3-hexyne was prepared by diluting 85 µL of 3-hexyne to 5 mL in a volumetric flask. A 0.4 mL portion (200 equiv 3-hexyne per dimer) of this solution was placed in a 1 mL syringe. The cuvette and syringe were brought out of the glovebox and transferred to a cuvette holder at 25 °C in the UV-Visible spectrophotometer. An initial measurement was taken to the verify the concentration of [(IPr*R)CuH]2. While stirring the reaction mixture, the 3-hexyne stock solution was syringed into the cuvette (final [(IPr*R)CuH]2 = 0.1 mM, final [3-hexyne] = 20 mM), and the kinetics measurements were started.                      Figure S79. Attempted analysis of kinetics data for the reaction of 0.2 mM [(IPr*Cl)CuH]2 with 1-hexene followed for five half-lives at l = 441 nm. Because significant deviation from linearity is observed, particularly at the end of the reaction, the data were re-analyzed at three half-lives (see Figure S80).

Computational methods
The geometries were optimized in the gas phase at the density functional theory (DFT) 16 level with the hybrid B3LYP exchange-correlation functional. 17,18 The DFT-optimized DZVP2 basis sets 19 were used for H, C, N, O and Cl atoms and aug-cc-pVDZ-PP basis set was used for Cu. 20 Vibrational frequencies were calculated to show that the structures were minima. Single point calculations using the B3LYP optimized geometries were performed at the M06 21 and wB97X-D levels. 22,23 These functionals were chosen as they have been used to predict thermodynamic properties of similar systems. 24 The cartesian coordinates for the optimized geometries have been provided as a separate file. The calculations were performed using the Gaussian16 program system. 25 The quasiharmonic approximation from Truhlar and co-workers was used to correct the entropy associated with low-frequency vibrational modes. In this case, all harmonic frequencies below 100 cm -1 were raised to 100 cm -1 before evaluation of the vibrational component to the entropy. 26 Using the gas phase optimized geometries, the solvation free energies in benzene at 298 K were calculated using the self-consistent reaction field (SCRF) approach 27 with the COSMO (B3LYP//DZVP2/aug-cc-pVDZ-PP) parameters 28,29 as implemented in Gaussian 16. The Gibbs free energy in benzene solution, DGsol, was calculated from: where DGg,298K is the gas phase free energy and DDGC6H6 is benzene solvation free energy. A dielectric constant of 2.2706 corresponding to that of bulk benzene was used in the COSMO calculations.
KIEs were calculated from thermodynamic transition state theory: k = (kBT/h)*exp(-DG ‡ /RT) with -DG ‡ calculated at the wB97XD//B3LYP level. Note that the solvent contributions calculate in the SCRF approximation used in the calculations, so this is determined by frequencies in the gas phase.
The B3LYP optimized geometries were used for the time dependent-DFT (TD-DFT) calculations 34,35 , performed to analyze the UV-Visible spectra in benzene.
The calculations were performed on our local UA Opteron-and Xeon-based Linux clusters.       S85 Figure S87. DFT-optimized structures of CuH monomer, CuH-acetaldehyde complex, transition state for hydride insertion, and Cu-alkoxide product. The ligand aryl groups are shown in wireframe representation.

NBO Analysis
NBO analysis was performed on the (IPr*R)Cu-H monomer and dimer, with R = H, Cl, Me, and OMe, as well as the transition states for insertion of acetaldehyde and propene, with R = H, Cl, and Me. Key findings for the IPr*H system are presented below. There is no significant difference in the results for the different R-groups (see Table S13 and the next section for results with other ligands).  There is now a C-H bond involving the hydride which has 1.76 e (1.16 e on the H and 0.60 e on the C). There is 0.42 e in the Cu 4s in a 'non-Lewis' valence nonbonding orbital and there is 0.43 e in a C-H σ* with opposite polarization to the bonding orbital. Thus, the NBO analysis suggests that the hydride has mostly transferred to the C in the transition state. Note also that there is a loss of Cu 3d character upon formation of the complex that is partially regained in the transition state.