On the Origin of the Promoting Effect Exerted by Magnesium in the ZnCl2-Catalyzed Synthesis of N,N-Diisopropylethylamine.

The reaction of magnesium or zinc amides with alkyl or benzyl halides is an attractive approach to make C-N bonds, especially for electron-poor organic halides. The magnesium-promoted preparation of hindered non-nucleophilic amine (N,N-diisopropylethylamine) from ethyl chloride and zinc diisopropylamide has been studied. In this paper, instead of the application scope of this method, we focused on the mechanisms of the catalytic processes and the associated electronic origins. According to the calculations, the C-N coupling process in all selected systems proceed preferably in an ethylium-transfer mode. Further, rather than undergoing the Grignard reaction route, the more pronounced electronic interactions within the transition structure as induced by the "innocent" magnesium atom should be responsible for the observed high catalytic activity of the Mg/ZnCl2 combination.

Introduction

Hindered non-nucleophilic amines are widely used as base catalysts in alkylations,[1,2] eliminations,[3,4] and coupling reactions,[5,6] and they constitute important function blocks in various biologically active compounds.[7,8] To obtain such an amine, like N,N-diisopropylethylamine (DIPEA), considerable efforts have been made on synthetic strategy, for example, reacting aldehydes or ketones with primary or secondary amines to produce secondary amines or tertiary amines.[9] However, in these processes, usually catalysts with noble metals are required,[10,11] and the relatively expensive sodium borohydride[12] or sodium cyanoborohydride[13] serves as the reducing agent. In addition, it is difficult to convert secondary amines to tertiary amines because the enamine cation, which serves as the intermediate, is hard to form.[14] Other methods such as the reaction of N-chloro dialkylamines with alkyl Grignard reagents, however, afford relatively low yields of tertiary amines even without steric hindrance.[15] Alternatively, the aminolysis of haloalkanes is an attractive approach to make C–N bonds, using catalysts involving metals such as Pd[16,17] and Cu.[18,19] Although Lewis acids such as ZnCl2 can mediate a SN2 nucleophilic substitution like the aminolysis of haloalkane, unsatisfactory yields are expected.[20,21] Since Hauser and Walker[22] first reported magnesium amide bases with the general formula R2NMgX and (R2N)2Mg (Hauser bases) in 1947, the reactions of magnesium or zinc amides with alkyl or benzyl halides have become an efficient method for constructing the C–N bonds.[23,24] Owing to the fact that magnesium or zinc amides are of stronger nucleophilicity than their precursor amines, they are more reactive as nucleophilic reagents. This reaction, as far as we know, is so far mostly achieved using palladium or copper complex catalysis,[25] and the amides are prepared by metalation of the corresponding amines by Grignard reagents prior to the coupling reaction.[26]

In the mechanism proposed in this paper, according to the experimental and computational results, we depicted the preceding elimination of an intact HCl molecule from the intermediate iPr2HN → ZnCl2. The latter is a kind of Lewis acid/base pair that provides efficient access to iPr2NZnCl in situ at the alkaline environment. Subsequently, the C–N coupling takes place to afford DIPEA. The analogous deprotonation of diisopropylamine can also be found in the literature reported previously.[27] In our case, moderate yield was obtained when zinc amide reacted with chloroethane via nucleophilic substitution to produce DIPEA. However, owing to the high cost of palladium, it is still desirable to develop a non-noble, cheap catalyst-mediated system.

Regardless of using a noble catalyst,[28] iron-catalyzed cross-coupling methodologies between alkyl Grignard reagents and organic n class="Chemical">halides are promising.[29,30] Gerard Cahiez[31] demonstrated that the major bisarene-ligated Fe0 complex generated in situ could be an efficient precursor to promote the activation of sp2 electron-poor organic halides in the absence of a stabilizing ligand. Inspired by this result, we envisioned if constructing a coordination environment for single metal was able to accelerate the reaction. So, first we put Fe power into our system, and no obvious promotion effect was found. Then, we put “innocent” magnesium to our system. Interestingly, the yield can be dramatically improved when magnesium is introduced into the above reaction system. Further, magnesium was also found to be an effective additive to accelerate some chemical processes.[32−35] However, the origins for the behavior of magnesium in any of the above reactions remain unclear.

In our context, the electronic origins for the auxiliary performance of Mg are revealed via a combination of experiments and theoretical calculations to get insight on this interesting phenomenon.

Results and Discussion

Thus, we first focus on the catalyn class="Chemical">tic performances of Zn and Mg in the aminolysis process when they are serving as Lewis acids. The reaction of ethyl chloride with diisopropylamine was examined with various catalysts. As shown in Table , the aminolysis reaction takes place slowly without adding any species at 175 °C (Table, entry 1), and the yield was slightly promoted to 32.7% when only Mg power was employed (Table , entry 2), so pure Mg itself does not invoke an efficient reaction. Next, when MgCl2 and ZnCl2 were employed, the yield increased to 42.5 and 71.0%, respectively (Table , entry 3, 4). Most likely, this finding can be attributed to the higher Lewis acidity of ZnCl2 compared to MgCl2. So, we investigated this reaction catalyzed by FeCl3 which has stronger Lewis acidity. However, the yield was only 40.8%, even lower than MgCl2 (Table , entry 5). Compared to the ZnCl2 system, there is no further enhancement in the yield when Fe and ZnCl2 were employed (Table , entry 6). Further, surprisingly, the yield reached up to 82.4% when the combination Mg/ZnCl2 (Table , entry 7) was used as the catalyst; a similar effect was obtained when zinc was added to MgCl2 system that the yield changed to 67.7 from 42.5% (Table , entry 10). By contrast, when adding Mg to MgCl2 or Zn to ZnCl2, no obvious promotion effect was found (Table, entry 9, 11). However, unsatisfactory yield was obtained when lowering the reaction temperature from 175 to 150 °C on using Mg/ZnCl2 as the catalyst (Table , entry 8). At last, we optimized the molar ratio of Mg and ZnCl2.The yield was slightly reduced to 79.5% (Table , entry 7) from 82.4% when decreasing the amount of Mg (5 mol %). By contrast, the yield was significantly reduced from 79.5 to 65.8% (Table , entry 7) when employing (3 mol %) ZnCl2. This results seem to manifest that the real catalyst is ZnCl2 while Mg promotes the reaction. Finally, an optimum molar ratio of Mg/Zn was found to be 1:1.

Table 1

Results on Reaction of Chloroethane with Diisopropylamineb

entry	catalyst	temperature/°C	yieldb, %
1		175	26.6
2	Mg	175	32.7
3	MgCl2	175	42.5
4	ZnCl2	175	71.0
5	FeCl3	175	40.8
6	Fe/ZnCl2	175	69.4
7	Mg/ZnCl2	175	82.4/79.5c/65.8d
8	Mg/ZnCl2	150	41.6
9	Mg/MgCl2	175	44.9
10	Zn/MgCl2	175	67.7
11	Zn/ZnCl2	175	66.8

Diisopropylamine (91 g, 0.89 mol), chloroethane (30 g, 0.465 mol), reaction time: 8 h.

GC yield. The amount of all metals used in the experiment is 0.033 mol (7 mol %), and the amount of all Lewis acids used in the experiment is 0.018 mol (4 mol %).

0.021 mol Mg (5 mol %), 0.018 mol ZnCl2 (4 mol %).

0.021 mol Mg (5 mol %), 0.012 mol ZnCl2 (3 mol %).

Solvent effect[47] was simulated using n class="Chemical">dipropylamine as the solvent as the associated parameters for diisopropylamine are lacking; numbers in the left columns are the barriers for forward reactions, and the ones in the right columns are for reverse barriers (the same for the tables below).

To investigate the role played by the pure n class="Chemical">metals in the reaction, X-ray photoelectron spectroscopy (XPS) analysis was performed to determine the surface elemental chemical valence and electronic configurations of the catalysts. As the catalyst can be coordinated by and mixed with the substrates/products and is buried in the chloride salts after the reaction, we performed “blank” reaction with the catalysts for XPS analysis. Thus, Mg/ZnCl2, Zn/MgCl2, and Zn/ZnCl2 were dissolved by bis(2-butoxyethyl)ether at reaction conditions. All lines in the XPS spectra are corrected with carbon C 1s at 284.8 eV. As can be seen in Figure a, for Mg/ZnCl2, the Mg 1s peak is composed of the Mg metal peak [full width at half-maximum (FWHM) 1.683 eV) centered at 1303.8 eV[36] and a broader Mg2+ peak (FWHM 1.859 eV) centered at 1304.8 eV.[37] This result indicates that magnesium can be partly oxidized at the aminolysis temperature and, most likely, the charge transfer from magnesium to the reactive center matters for its promotion effect. This finding holds true for Zn/MgCl2 system (see Figure c). The chemical state of zinc was determined by Zn auger spectra since it is difficult to precisely differentiate the peaks corresponding to Zn (1021.7 eV) and Zn2+ (1022e V) in the Zn 2p3/2 region (see Figure S3, Supporting Information),[38] and no attempts at quantification of zinc metal/oxide components were found in the literature. As shown in Figure b,d,e, zinc auger spectral region at 986–988 eV[18] shows sharp auger peaks, indicating that all of them are oxidized zinc.

Figure 1

Representative (a) Mg 1s XPS spectra of Mg/ZnCl2, (b) Zn LMM XPS spectra of Mg/ZnCl2, (c) Mg 1s XPS spectra of Zn/MgCl2, (d) Zn LMM XPS spectra of Zn/MgCl2, and (e) Zn LMM XPS spectra of Zn/ZnCl2.

Next, the reaction mechanisms for the n class="Chemical">((CH3)2CH)2NH/C2H5Cl couple (iPr2NH/C2H5Cl) as catalyzed by MgCl2, ZnCl2, Mg/ZnCl2, Zn/MgCl2, Zn/ZnCl2, and Mg/MgCl2 were investigated by density functional theory (DFT) calculations. We first considered the reaction catalyzed by solely the Lewis acids. For example, conceivably, the donor–acceptor interaction[39] between diisopropylamine and ZnCl2 plays an important role in the transformations. According to the calculation, the elimination of an intact HCl molecule from the intermediate iPr2HN → ZnCl2 is barrierless with an energy of 228 kJ mol–1. That means this reaction may not be invoked at low temperature. Subsequently, the C–N coupling process is initiated by the interaction of iPr2NZnCl with C2H5Cl (see Figure ). As shown in Figure , from an encounter complex IM1a in which C2H5Cl is bound to the Zn center via Zn–Cl interaction, the breaking of the C–Cl bond is concerted by formation of the C–N bond; the transition state TS1-2a is involved. The same route prevails as well for the reactions catalyzed by MgCl2. However, the above C–N coupling process cannot be found when FeCl3 is employed.

Scheme 1

Process of the Formation of DIPEA by Generation of Zinc Amide In Situ

Figure 3

Simplified potential energy surfaces (PESs) for the formation of DIPEA through TS1-2 as well as the deformation energies (ΔEdef) for the associated transition states.

To further confirm the process of dehydrohalogenation in our reaction system, X-ray diffraction experiments (XRD) analysis technique was performed to identify the crystal structure of diisopropylamine hydrochloride. Thus, we conducted a reaction involving diisopropylamine and ZnCl2 without adding chloroethane under the same conditions as shown in Table . The complex at the bottom of the reactor was tested by XRD after further drying. As shown in Figure , the crystal structure of diisopropylamine hydrochloride was identified by the diffraction peaks at 2θ = 12.7, 17.1, 23.3, 25.7, and 28.0° (JCPDS 09-0589). Meanwhile, the XRD pattern showed the diffraction peaks at 2θ = 31.8, 34.4, 36.2, 47.5, 56.6, 62.8, and 67.6° corresponding to (100), (002), (101), (102), (110), (103), and(112) crystal planes of ZnO with low crystallinity (JCPDS 89-7102). These results demonstrated that zinc exists in the oxidized form after the reaction, in line with the XPS results. However, there were still some diffraction peaks that could not be identified, most likely, attributing to the oxidation products of the zinc amides. As far as we know, air-stable solid aryl and heteroaryl zinc reagents have been made in the laboratory by transmetalation of a series of organomagnesium species with zinc pivalate.[40,41] The XRD patterns of catalysts after the reaction are also provided (see Figure S2, Supporting Information). Nevertheless, the available information that can be given is very limited since the related properties have been changed after filtering.

Figure 2

XRD patterns of the complex after the reaction.

Regarding the role played by the pure metal, the reaction of the active metal with chloroalkane is well known as the Grignard reaction. Although the direct insertion of transition metals into a carbon–halogen bond is usually hampered by the high lattice energy of the metal as well as difficulties associated with activation of the metal surface,[42] some Lewis acids such as ZnCl2 or FeCl2 can catalyze the insertion.[43,44] We first considered whether this process is responsible for the enhanced reaction efficiency in the reactions catalyzed by Mg/ZnCl2, Zn/MgCl2, and Zn/ZnCl2. Considering that (1) both the generation and consumption of the Grignard reagent are almost irreversible in the employed system,[45] (2) the addition of Mg/Zn to the system promotes the reaction efficiency, and (3) pure Mg/Zn does not invoke an efficient reaction, the Grignard reaction is not expected to be rate-limiting for the overall process. Thus, the interaction between the C2H5MgCl and iPr2NZnCl species was examined computationally. As shown in Figure , the incoming C2H5MgCl is bound to iPr2NZnCl via the Cl–Zn interaction (IM3c), elongating both Mg–Cl and C–Mg bonds. From IM3c, migration of the ethyl group from the Mg to N atom takes place via TS3-4c, generating IM4c. This pathway was also found for the systems employing Mg/MgCl2, Zn/MgCl2, or Zn/ZnCl2 as catalysts. In addition, a two-step C–N coupling route was also found for the reactions mediated by Mg/ZnCl2 and Mg/MgCl2: as shown in Figure , from IM3c, the ethyl group is passed from the Cl atom first to the Zn atom and then to the N atom in a step-wise manner via the sequence IM3c → TS3-5c → IM5c → TS5-6c → IM6c. However, all these pathways are energetically much less favorable as compared to the ones shown in Figure as the former are hampered by much higher barriers (See Table ).

Figure 4

PESs for the formation of DIPEA initiated via Grignard reaction as well as ΔEdef for the associated transition states.

Figure 5

PESs for the formation of DIPEA through the two-step C–N coupling route as well as ΔEdef for the associated transition states.

Table 2

Reaction Barriers for the Transition States TS3-4 (kJ mol–1)

	TS3-4c (Mg/ZnCl2)		TS3-4d (Zn/MgCl2)		TS3-4e (Zn/ZnCl2)		TS3-4f (Mg/MgCl2)
without solvent effect	403.9	160.6	268.2	184.6	262.0	192.2	310.9	143.6
with solvent effecta	392.9	147.5	248.7	168.2	243.5	163.2	291.1	122.4

Solvent effect[47] was simulated using dipropylamine as the solvent as the associated parameters for diisopropylamine are lacking; numbers in the left columns are the barriers for forward reactions, and the ones in the right columns are for reverse barriers (the same for the tables below).

Thus, the Grignard reaction turned out not to play a role in acceleran class="Chemical">ting the reactions mediated by metal–Lewis acid. Then, we went back to the pathway featured by TS1-2. Indeed, the addition of pure metal may decrease the barrier for this transition state. For example, as shown in Figure , when adding an “innocent” Mg atom to IM1a to form IM1c, the subsequent reaction barrier (80 kJ mol–1, for TS1-2c) is lowered by 48 kJ mol–1 compared to TS1-2a (see Figure ). The analogous four-membered ring structure formed by Zn–Cl–Mg–Cl can also be found in lithium chloride solubilized 2,2,6,6-tetramethylpiperidide base (TMPZnCl·LiCl), which was a well-known TMP base for converting a wide range of aromatic and heterocyclic substrates into valuable organometallic reagents.[46] This finding holds true for the Zn/MgCl2 system. In contrast, adding Zn to the ZnCl2 system slightly increases the reaction barrier; a transition state similar to TS1-2c was not located for the Mg/MgCl2 system, as during the optimization, the Mg atom always escapes from the reaction center and thus not playing a role therein. With these computational results, the functions of the pure metals in the C–N coupling processes are addressed, which are in line with the experimental findings. As suggested by both computation and experiments, Mg/ZnCl2 is the optimal combination to mediate the C–N coupling reaction between diisopropylamine and chloroethane. In addition, we also considered computationally the possibility that the Lewis acid serves as a chloride-acceptor to invoke a comparison by abstracting the chloride atom from C2H5Cl (see Figure S5, Supporting Information). Compared to the direct C–N coupling with a barrier of 191.5 kJ mol–1, the barrier is slightly lowered when ZnCl2 or MgCl2 serves as a chloride-acceptor, even much lower for FeCl3; here, most likely, the chemical inertness of chloroethane matters.

Figure 6

PESs for the formation of DIPEA via TS1-2 after adding Zn/Mg as well as ΔEdef for the associated transition states.

To obtain an even better understanding of the C–N coupling processes, an analysis on the frontier molen class="Chemical">cular orbitals (FMOs) associated in the pathway A was performed. In general, formation of the C–N bond via pathway A proceeds in an ethylium-transfer mode. For example, as shown in Figure , in the sequence IM1c → TS1-2c, the elongation of the C–Cl bond of chloroethane relocates an electron pair from the σ(C–Cl) orbital to a 3p(Cl) orbital; meanwhile, delocalization of a lone electron pair from the 2p(N) orbital to the σ*(C–Cl) orbital produces an elongated σ(C–N) bond; consequently, a delocalized, empty 2p(C) orbital forms. Further revolution of the electronic structures completes the cleavage of the C–Cl bond and the formation of the C–N bond. Thus, for the barrier of a C–N coupling step via pathway A, the energy gap between the empty σ*(C–Cl) orbital and the doubly occupied 2p(N) orbital (ΔE2p(N)→σ*(C–Cl)) may matter. According to the calculation, however, the values of ΔE2p(N)→σ*(C–Cl) for different systems are close to each other. Thus, it is not the more activated bond/orbitals to be reorganized in the transformation that results in a low barrier in the Mg/ZnCl2 system, rather, some other factors matter.

Figure 7

Schematic FMOs for the C–N bond making process via pathway A as catalyzed by Mg/ZnCl2.

Further, according to an analysis of the deformation energies (ΔEdef) required for the geometric distortion of the reactants along the reaction coordinates from IM1 to TS1-2, the ΔEdef for TS1-2c is higher as compared to the ones in the other systems; however, TS1-2c is of the lowest barrier. Thus, much more prominent electronic interactions within the transition structures exert for TS1-2c. According to a symmetry-adapted perturbation theory (SAPT)[48] analysis, the interaction energies between the C2H5Cl unit and the rest moiety at the transition states are −178, −206, −215, −196, and −179 kJ mol–1 for TS1-2a–TS1-2e, respectively (see Figure S7, Supporting Information). Indeed, a more prominent electronic interaction between the reaction center and C2H5 contributes to the low barrier for TS1-2c. Further, in order to assess how much the additional metal center contributes to the lowering of the reaction barrier, an SAPT analysis on the interaction between the second metal center and the remaining part of the transition structure was performed. It turned out that the interaction of Mg with the other part of the transition structure may contribute to the lower barrier for TS1-2c compared to TS1-2c–TS1-2e; as shown in Figure , of all interactions, the electrostatics dominate.

Figure 8

Composition of the interaction energies between the second metal center and the remaining part of the transition structures for TS1-2c–TS1-2e.

Conclusions

In summary, an efficient, magnesium-promoted synthesis of n class="Chemical">DIPEA catalyzed by ZnCl2 has been studied both experimentally and computationally, and the behavior of magnesium was explained using DFT. In general, according to the calculations, the C–N coupling process in all selected systems proceeds in an ethylium-transfer mode. A more prominent electronic interaction between the C2H5 unit and the remaining part of the transition structure in the Mg/ZnCl2 system contributes to a lower reaction barrier compared to the other systems; in addition, the relatively stronger electrostatic interaction between Mg and the reaction center also stabilizes the transition state. Though the addition of Mg to the ZnCl2-mediated reaction brings about the increase of the deformation energy required for distorting the reaction coordinate to the transition structure, the much more prominent electronic interaction within the structure induced by Mg compromises this effect and, moreover, lowers the reaction barrier. The results presented in this work may provide implications for improving the chemical processes catalyzed by Lewis acids. Further efforts on the application scope of the approach in preparing various hindered non-nucleophilic amines are indicated.

Experimental and Computational Section

Experimental Details

General Procedure for Mg/ZnCl2-Catalyzed N-Ethylation of Diisopropylamine with Chloroethane

Diisopropylamine and n class="Chemical">chloroethane were commercial products and were used without further purification. All inorganic reagents were commercial products with a purity of over 99% and were used as received. A mixture of diisopropylamine, magnesium powder, and ZnCl2 solid with a specific ratio were added to a 500 mL autoclave, and the reactions were carried out under nitrogen. After completion of the reaction, the reaction mixture was cooled and poured out. The mixture was then neutralized, and the oil phase was received using a separating funnel.

Analysis Instruments Used

Gas chromatogtaphy–mass spectrometry (GC–MS): Agilent 5973–6890 series (FID detector, weakly polar capillary column SE-30, nitrogen as carrier gas); XPS: Thermo n class="Chemical">Scientific K-Alpha+ system using Al Kα radiation (1486.6 eV) under a base pressure of 2 × 10–7 Torr; XRD: Shimadzu X-ray diffractometer (MAXima XRD-7000, the Japan) with Cu Kα radiation at 40 kV and 40 mA.

Computational Details

The structural optimization, frequency analysis, and analysis on the FMOs were performed using the Gaussian 09 program; For the FMO analysis, the orbitals were localized using the Pipek-Mezey approach with the multiwfn 3.7 package. See the Supporting Information for more technical details.

Table 3

Reaction Barriers for the Two-step C–N Coupling Route (kJ mol–1)

			Mg/ZnCl2		Mg/MgCl2
without solvent effect	TS3-5	step 1	56.7	447.1	23.9	281.9
		step 2	467.8	187.4	298.7	160.8
with solvent effect	TS5-6	step 1	47.0	434.4	11.3	288.2
		step 2	470.1	189.2	313.1	170.8

Table 4

Reaction Barriers for the Transition States TS1-2 (kJ mol–1)

	TS1-2a (ZnCl2)		TS1-2b (MgCl2)		TS1-2c (Mg/ZnCl2)
without solvent effect	128.2	247.2	105.5	249.3	79.5	190.9
with solvent effect	124.0	250.5	100.4	253.5	74.6	189.5

Table 5

Energy Gaps Between σ*(C–Cl) and 2p(N) in Intermediate IM1 (ΔE2p(N)→σ*(C–Cl)) for Different Systems

catalyst used for different systems	ΔE2p(N)→σ*(C–Cl), eV
ZnCl2	9.32
MgCl2	9.64
Mg/ZnCl2	9.49
Zn/MgCl2	9.30
Zn/ZnCl2	9.53