Amine and Titanium (IV) Chloride, Boron (III) Chloride or Zirconium (IV) Chloride-Promoted Baylis-Hillman Reactions

The Baylis-Hillman reactions of various aryl aldehydes with methyl vinyl ketone at temperatures below -20oC using Lewis acids such as titanium (IV) chloride, boron (III) chloride or zirconium (IV) chloride in the presence of a catalytic amount of selected amines used as a Lewis bases afford the chlorinated compounds 1 as the major product in very high yields. Acrylonitrile can also undergo the same reaction to give the corresponding chlorinated product in moderate yield. A plausible reaction mechanism is proposed. However, if the reaction was carried out at room temperature (ca. 20oC), then the Z-configuration of the elimination product 3, derived from 1, was formed as the major product.


Introduction
The Baylis-Hillman reaction and related processes, typically catalyzed by DABCO or tertiary phosphines, have become increasingly important in organic synthesis because the resulting adducts may have several functional groups available for numerous further transformations [1][2][3][4][5]. The major drawbacks of the Baylis-Hillman reaction are its slow reaction rate and a limited range of useful substrates. To overcome these shortcomings many variations have been devised, such as the use of Lewis acids or various other additives to activate the carbonyl electrophiles [6][7][8][9][10]. Among those Lewis acids examined, TiCl 4 has been successfully used to promote the Baylis-Hillman reaction in the presence of Lewis base catalysts [9,11,12]. During our own investigations of the Baylis-Hillman process we found that many amines are very effective Lewis bases in this interesting reaction and the reaction products differ considerably from those reported so far [13]. Herein we wish to report the full details of the titanium (IV) chloride, boron (III) chloride or zirconium (IV) chloride and amine promoted Baylis-Hillman reactions, along with a plausible reaction mechanism based on the previous findings and our own results.

Results and Discussion
We initially attempted the reaction of p-nitrobenzaldehyde with methyl vinyl ketone in the presence of TiCl 4 (1.0 eq) at -78 o C. No reaction occurred ( Table 1, entry 1). However, after adding 20 mol % (0.20 eq) of triethylamine (Et 3 N) as a Lewis base, the reaction took place smoothly to give the chlorinated product 1a, rather than 2a and 3a (usually considered the Baylis-Hillman olefin and trisubstituted alkene) as reported by Kataoka and Li [9,12], respectively (Scheme 1, Table 1, entry 2). By careful investigation, we found that this reaction was very sensitive to the amounts of both TiCl 4 and amines present in the reaction mixture (Table 1). By means of catalytic amounts of amine and excess amounts of TiCl 4 the reaction proceeded very well. However, using large excesses of amine as a Lewis base, the reaction was completely stopped (Table 1, entry 6). This result suggested that the amine could coordinate to TiCl 4 and that free TiCl 4 acting as a Lewis acid was required to promote the reaction. The amount of TiCl 4 was also crucial for this reaction because using a catalytic amount of TiCl 4 , the reaction became very slow and gave low yields of 1a (Table 1, entry 7 and 8). The best reaction conditions were found to be the use of 20 mol % of amine as a Lewis base and 1.4 eq of TiCl 4 as a Lewis acid ( For many arylaldehydes having strongly electron-withdrawing group on the phenyl ring, the reactions proceed quickly to give compounds 1 in high yields using a catalytic amount of Lewis base (20 mol %) at -78 o C (Scheme 2, Table 2). However, other arylaldehydes or aliphatic aldehydes needed higher temperatures (-20 o C) to give the corresponding chlorinated product 1 in high to moderate yields. Moreover, we found that besides triethylamine, 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) and diethylamine were also very effective Lewis bases for this reaction in the presence of TiCl 4 (  Table 1 only needed a catalytic amount of amine (20 mol %) to bring the reaction to completion in the presence of TiCl 4 ( Table 2). It should be emphasized that in all cases, only one diastereomer was formed during the reaction process based on the 1 H-NMR spectral data evidence. Their relative configurations were confirmed as the syn-form by analysis of the X-ray crystal structure of 1a [13] (Figure 1).  Compound 1 can be easily and completely transformed to the compound 2 [14] by treating with an excess amount (2.0 eq) of triethylamine or 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) (Scheme 3). The purification of 1 by preparative thin layer chromatography (TLC) was noted to also cause the transformation of 1 to 2, therefore quick flash column chromatography is required in order to obtain the pure product 1. Besides methyl vinyl ketone, acrylonitrile underwent the same reaction to give the corresponding chlorinated product 1h in moderate yield in CH 2 Cl 2 at 10 o C for 5 days (Scheme 4, Table 3, entries 1 and 2), but at -78 o C, no reaction occurred (entry 3). Raising the reaction temperature to reflux (45 o C) caused a decrease of the yield of 1h (  isolated yield

All cases shown in
To the best of our knowledge, this represents a novel Baylis-Hillman reaction system because use of a catalytic amount of amine as a Lewis base has never been reported to date for titanium(IV) chloride promoted Baylis-Hillman reactions. Recently, Aggarwal has reported that using stoichiometric amounts of amine such as DABCO and a catalytic amount of titanium (IV) chloride gave reduced reaction rates, but that use of stoichiometric amounts of amine and a catalytic amount of lanthanide triflates (5 mol%) gave increased rates in the Baylis-Hillman reaction [7]. Our system shows that using excess amounts of titanium (IV) chloride and catalytic amount of amines, the major reaction products in moderate to high yields are the β-chlorinated compounds 1, which can be readily transformed to the Baylis-Hillman olefin 2. Thus the reaction rate of Baylis-Hillman reaction can be greatly accelerated by means of this reagent system.
We also examined many other metal halides such as PdCl 2 , RhCl 3 , Cp 2 ZrCl 2 , ZrCl 4 , AlCl 3 , TMSCl, SiCl 4 , BF 3 , BCl 3 and found that BCl 3 and ZrCl 4 also worked as Lewis acids for this reaction, although they are not as effective as TiCl 4 . For example, we carried out the Baylis-Hillman reaction using BCl 3 and ZrCl 4 as Lewis acid with Et 3 N as a Lewis base under the same reaction conditions as those shown in Scheme 1. The syn-chlorinated products can be also obtained (Scheme 6), but this required longer reaction times (40 h) at -78 o C. These results are summarized in Table 4.  isolated yields In Scheme 7, we propose a tentative mechanism to explain the formation of product 1. In fact, the reactions of trimethylamine and dimethylamine with titanium (IV) chloride had been investigated by Antler and Laubengayer in 1955 [15a]. Chloride ion was detected although the system was complicated. Based on his findings, Periasamy gave a mechanism for the reaction of tertiary amines with

TiCl 4 (Et 3 N Et 3 N TiCl 3 ) + Cl -+
The reaction mechanism proposed in Scheme 7 is based on those previous findings and the results of our own investigations as shown in Table 1. We believe that amine can strongly coordinate to the Ti metal center of TiCl 4 to give an ionic metal complex containing chloride ion. This reaction is related with the attack of chloride ion on the methyl vinyl ketone in a Michael addition fashion (Scheme 7). Using BCl 3 or ZrCl 4 as a Lewis acid, the reactions would proceed via the same mechanism. Thus, the formation of chlorinated compound 1 is a major reaction process in the TiCl 4 , BCl 3 , and ZrCl 4 and Lewis base amine promoted Baylis-Hillman reaction. show the chiral Lewis bases used for this reaction. These chiral ligands (A-L) were either prepared by us according to the known synthetic methods or purchased from Aldrich. For sterically bulky amines, the reaction is relatively slow and longer reaction times are required, but in all cases, the achieved enantiomeric excesses were only about 10~20%. We believe that this is related to the mechanism shown in Scheme 7 because the reaction is via separated ionic intermediates and the chiral centers are far away from the aldol reaction center. These results partially support our proposed reaction mechanism (Scheme 7 On the other hand, by carrying out this reaction at room temperature (about 20 o C), we confirmed that the elimination product 3 was the only product under the same reaction conditions (Scheme 9,  This reaction was first disclosed by Li and coworkers. They reported that in the presence of stoichiometric or nonstoichiometric TiX 4 , compound 3 could be formed in its Z-configuration [12]. Later we also reported that 3 could be exclusively obtained in the TiCl 4 and chalcogenide promoted Baylis-Hillman reaction at room temperature (20 o C) [16]. The Z-configuration has been confirmed by X-ray analysis ( Figure 3) [16].

Figure 3: Crystal Structure of 3a
We now wish to report that in the initial reaction stage, the presence of Lewis base can still significantly speed up this reaction ( Table 5, entry 1 and 2). In order to clarify the formation of 3, we treated 1a and 2a directly with TiCl 4 in dichloromethane at room temperature. We found that 1a can be transformed to 3a within 6 h, whereas the reaction of 2a with TiCl 4 was much slower (Scheme 10). These results strongly suggest that 3a is derived directly from 1a formed first in the reaction. Thus, we conclude that, at room temperature, the chlorinated products 1 could be formed either in the absence or in the presence of Lewis base, but they are rapidly transformed to the elimination product 3 exclusively.

Conclusions
We have found that TiCl 4 , BCl 3 or ZrCl 4 and amine promoted Baylis-Hillman reaction is a very efficient reaction system for producing chlorinated compounds 1. The amine compounds are good Lewis bases and TiCl 4 , BCl 3 and ZrCl 4 are good Lewis acids for this reaction. The reaction temperatures, the amount of Lewis acid, and the amount of Lewis base can drastically affect the reaction products and reaction rates. The relative activities of different Lewis acids for this reaction are TiCl 4 > BCl 3 > ZrCl 4 and the best combination of Lewis acid and Lewis base for this reaction is TiCl 4 (1.4 eq) with NEt 3 (0.2 eq). The relative configuration of 1 was not affected by the Lewis acids (TiCl 4 , BCl 3 or ZrCl 4 ) used at all. The reaction was initiated by chloride ion attacking at the Michael acceptor of the α,β-unsaturated ketone. The chloride ion was produced by coordination of Lewis bases (NEt 3 ) to Lewis acids such as TiCl 4 , BCl 3 , and ZrCl 4 . Undoubtedly compound 3 was derived mainly from 1. Efforts are underway to elucidate the full mechanistic details of this reaction and to disclose the scope and limitations of this reaction. Work along this line are currently in progress.

General
Melting points are uncorrected. 1 H-and 13 C-NMR spectra were recorded on a Bruker AMX-300 spectrometer at 300 MHz and 75 MHz, respectively. Mass spectra were recorded by the EI method and HRMS was measured by a Finnigan MA+ mass spectrometer. Organic solvents used were dried by standard methods when necessary. All solid compounds reported in this paper gave satisfactory CHN microanalyses. Commercially obtained reagents were used without further purification. All reactions were monitored by TLC with Huanghai GF 254 silica gel coated plates. Flash Column Chromatography was carried out using 300-400 mesh silica gel at increased pressure.