Multicomponent azide – alkyne cycloaddition catalyzed by impregnated bimetallic nickel and copper on magnetite †

A new bimetallic catalyst derived from nickel and copper has been used successfully for the first time in the multicomponent reaction of terminal alkynes, sodium azide, and benzyl bromide derivatives. The presence of both metallic species on the surface of magnetite seems to have a positive and synergetic effect. The catalyst loading is the lowest ever published for a catalyst of copper anchored on any type of iron support. The catalyst could be easily removed from the reaction media just by magnetic decantation and it could be reused up to ten times without any negative effect on the initial results.


Introduction
Although the 1,3-dipolar cycloaddition of azide derivatives and alkynes dates back to the nineteenth century, 1 the pioneer and seminal works of the Medal 2 and Sharpless 3 groups on the copper-catalyzed process were the denitive push for the blossoming of this process.This process allowed access to different 1,2,3-triazoles of great interest to different areas of chemistry and pharmacy, in short reaction times, under mild conditions, and as only one regioisomer. 4he tremendous success of the homogenous copper(I) complexes as catalysts has eclipsed the activity of others, such as those derived from ruthenium, platinum, palladium, 5 silver 6 or nickel, 7 as well as the use of other heterogeneous catalysts.However, very recently some heterogeneous catalysts have emerged as an alternative.Thus, the particles of copper, 8 or its oxide derivatives, 9 different copper salts supported on charcoal, 10 on organic materials, 11 as well as on inorganic supports 12 have been tested for this transformation, with copper loading of these catalysts ranging from 0.5 to 12 mol%.Interestingly, some of the inorganic supports were based on iron, which permitted the development of magnetic catalyst and separation, as it was for the case of copper supported on iron (5 mol%), 13 copperferrite (5 mol%), 14 or ligand-graed copper on magnetite (2 mol%). 15he intrinsic instability of organic azides, mainly those of low molecular weight, has been an important drawback in the generalization of this approach for the synthesis of interesting polyvalent structures.However, the use of a multicomponent approach, generating the azide derivative in situ by reaction of sodium azide and the corresponding organic reagent, 9c,10b,d-f,11d,f,h,12f,h,14,15b,16 has permitted us to overcome this problem.On the other hand, we have recently developed a new, simple and robust method to immobilize different metal oxides 17 onto magnetite, 18 and we initially decided to apply the copper impregnated on magnetite catalyst 19 to the multicomponent azide-alkyne cycloaddition reaction.

Simple azide-alkyne cycloaddition
Although our ultimate goal was to get a heterogeneous and recyclable catalyst for the multicomponent version of azidealkyne cycloaddition, the study was started with the standard two-component reaction between ethynylbenzene (1a) and (azidomethyl)benzene (2a) catalyzed by impregnated copper on magnetite (Table 1).
The initial reaction was conducted in absence of catalyst at 110 C in water, obtaining aer 7 days a 1 : 1 mixture of both possible isomers.Then, the reaction was repeated in the presence of copper catalyst in toluene at 70 C giving exclusively 1-benzyl-4-phenyl-1H-1,2,3-triazole (3a) in a modest yield (entry 2).Both the decrease and the increase of temperature led to the formation of a mixture of regioisomers (entries 3 and 4).Then, the inuence of solvent was examined, nding that the highest yield was reached in water (entry 10).Under these conditions, the role of magnetite support was studied and high activity of the supposed inert material was found (entry 12).
Once the activity of copper catalyst was examined, its recycling was studied.Aer the rst trial, the magnetite was collected with a magnet, washed with toluene and ethanol, and dried.The recycled catalyst could be re-used three fold with similar results (82-78%).However, the yield dropped to 35% in the forth use, keeping this level of results during the following 5 cycles (Fig. 1).The phenomenon of leaching was studied by ICP-MS analysis of the resulting reaction solution mixture, and 1.1% of the initial amount of copper was detected (0.007% of iron), which could explained the lost of activity.Moreover, the TEM images of the recycled catalyst showed a small change in copper particle size from 7.1 AE 6.5 nm of the fresh prepared catalyst to 6.4 AE 5.2 nm for the recycled one, which would not affect the reactivity of the recycled catalyst.Finally, it should be pointed out that the BET surface area did not suffer a great change, from 6.2 m 2 g À1 for the initial catalyst to 8.4 m 2 g À1 for the used one, which is practically the same specic area.

Multicomponent cycloaddition processes
Aer nding that copper catalyst was effective in the cycloaddition between azides and terminal alkynes, we faced the problem of the multicomponent version, 9c,10d,e,11d,h,12h,14,15b,16b,c using benzyl bromide (5a), sodium azide (6) and ethynylbenzene (1a) as reaction model (Table 2).The reaction in water gave a mixture of expected heterocycle 3a together with its regioisomer 4a (compare entry 1 in Table 2 and entry 9 in Table 1).This initial trial showed that the change from simple cycloaddition to the multicomponent reaction one was not so simple.Thus, a new optimization process on this multicomponent reaction was carried out, starting by studying the effect of solvent (entries 1-8 in Table 2).The best result was obtained in absence of solvent, but a small amount of the product arising from homocoupling of terminal alkyne was found.19e The optimal temperature seemed to be 50 C (entries 8-11), since at higher temperatures different by-products were formed, and at lower temperatures a modest yield was achieved.Finally, the increase of amount of reagents, 5a and 6, increased the yield (compare entry 12 in Table 2 with entry 9 in Table 1).
Although copper catalysts have been the most used, other metal catalysts have also shown some activity for this reaction.For this reason we tested a series of impregnated metal catalyst in this multicomponent version (Table 3), starting from the uncatalyzed reaction (entry 1).From all ductile metal oxide, only nickel and copper catalysts showed activity (entries 2-14).
Then, a series of bimetallic derivatives were studied, nding that Pd/Cu system 19c could render the expected product 3a (entry 15).Very recently, different bimetallic Ni-Cu/C composite catalysts 20 have been tested in the simple cycloaddition of azides and terminal alkynes and these results prompted us to prepare the corresponding bimetallic one impregnated on magnetite.Its reaction gave the expected product with an excellent result (entry 16).The decrease of the amount of Ni-Cu    catalyst had an important detrimental effect, meanwhile its increase had a marginal benet (compare entries 16-18).Faced with the excellent result obtained with the bimetallic nickel-copper catalyst we wondered if the yield was a result of a simple addition of two independent catalytic sites or was it the result of some type of synergic effect.To answer that question, the reaction was repeated using both catalysts (the copper and the nickel one) with almost the same loading and the result seemed to be the addition of the activity of both catalysts (compare entries 5 and 6 with entry 19).Therefore, we believe that the bimetallic catalyst develops a synergetic effect that makes it superior to the addition of both parts, although the nature of this positive interaction is unknown.
Finally, the unsupported metal catalysts were tested.Thus, the reaction using CuO alone gave the expected product 3a with a good result (Table 3, entry 20), meanwhile the related nickel oxide gave a worse result (entry 21).When the reaction was repeated with the corresponding metal hydroxide derivatives the yields were slightly lower (entries 22 and 23).The equimolecular mixture of both metallic catalysts did not show any improvement of the result obtained by the copper derivative (compare entries 20, 22 and 24, 25, respectively).
The bimetallic Ni-Cu catalyst could be recycled and reused tenfold, just by collection of the catalyst with a magnet, washing with toluene and ethanol, and drying, without any depreciation in its activity (Fig. 2).
The phenomenon of leaching was studied by ICP-MS analysis of the resulting reaction solution mixture, and 1.1, and 0.2% of the initial amount of copper, and nickel, respectively, was detected (0.006% of iron).The TEM images of the recycled catalyst showed a small change in the particle size from 3.1 AE 1.7 nm of the freshly prepared catalyst to 4.7 AE 2.4 nm for the recycled one.Moreover, XPS data analysis of bimetallic catalyst showed only NiO, CuO and Cu 2 O species (Fig. 3), which was conrmed by Auger spectroscopy (see ESI †).However, the recycled one showed the presence of Ni(OH) 2 as well as Cu(OH) 2 .These small changes, in particle size and the nickel species seemed not to affect the activity of the bimetallic catalyst, since it could be reused several times with similar activity.To know if the reaction took place by the leached copper or nickel species to the organic medium, we performed the standard multicomponent reaction (Table 3, entry 16).Aer that, the catalyst was removed carefully by a magnet at high temperature, and washed with toluene.The solvents of the above solution, without catalyst, were removed under low pressure and alkyne 1a, sodium azide (6) and 4-bromobenzyl alcohol were added to the above residue.The resulting mixture was heated again at 50 C for 24 h.The analysis of crude mixture, aer hydrolysis, revealed the formation of compound 3a in 95% (catalyzed process) and product 3b in less than 1% yield by GC-analysis (compare with entry 2 in Table 4).Therefore, we could exclude that the nal leached copper-nickel species were responsible for the reaction results under the standard conditions.
Once the catalytic activity and the recyclability of bimetallic catalyst were proved, the scope of the reaction was tested (Table 4).The reaction gave excellent results independently of substituent position on the aromatic ring of the bromide 5 (entries 2-4).The electronic nature of substituent on the aromatic ring of the bromide 5 seemed not to have inuenced on the results (compare entries 1-7), since the results disagrees with the Hammett constants.Also, the reaction was accomplished with akynes 1 with different groups in the aromatic ring, with no clear correlation of the reached yields with the electronic nature of substituents (entries 8-16).The reaction with 2-(bromomethyl)isoindoline-1,3-dione gave the expected compound 3q in a modest yield (entry 17).However, it should be pointed out that the reactions using less electrophilic reagents such as aliphatic bromide (1-bromododecane) or benzyl chloride, failed aer seven days under standard conditions, recovering unchanged the starting alkyne, as well as in the case of using either an aliphatic substituted alkyne (oct-1-yne).
Then, the initial source of benzyl azide was tested (Scheme 1).The reaction with benzylic alcohols failed aer six days, recovering unchanged the initial alkyne.The reaction also failed using the silyl ether 7b.However, the reaction using benzyl mesylate gave a modest yield (35%) aer 2 days reaction time.When the reaction time was increased up to 6 days a reasonable yield was isolated (75%).When the reaction was performed with benzyl tosylate (7d) the result was very modest.
Once the scope of the reaction was studied, we faced the problem of reaction sequentiality.For this proposal, we carried the reaction with the dibromide derivative 10, and a double amount of sodium azide (6), obtaining aer six days the azide 11 with a moderate yield (Scheme 3).
The GC-MS analysis of crude mixture did not show the presence of corresponding bis-triazole, with the relate bis-azide derivative being the main by-product.The isolated azide 11 was submitted to another cycloaddition process, yielding the unsymmetrical bis-triazole derivative 12 with good yield.This approach highlights the possibilities of the catalyst in the synthesis of different substituted triazoles.

Conclusions
We have demonstrated that the new bimetallic catalyst derived from nickel and copper supported on magnetite was a good catalyst for the multicomponent reaction of terminal alkynes, sodium azide, and benzyl bromide derivatives.The presence of both metallic species on the surface of magnetite seemed to have a positive and synergetic effect.The catalyst loading was the lowest ever published for a catalyst of copper anchored on any type of iron support, and being in the lowest level for any type of the heterogeneous ones.The catalyst could be easily removed from the reaction media just by magnetic decantation, and it could be reused up to ten times without any negative effect on the initial results.

Experimental
General XPS analyses were carried out on a VG-Microtech Mutilab.XRD analyses were obtained on a BRUKER D-8 ADVANCE diffractometer with Göebel mirror, with a high temperature chamber (up to 900 C), with a X-ray generator KRISTALLOFLEX K 760-80F (3 kW, 20-60 kV and 5-80 mA).TEM images were obtained on a JEOL, model JEM-2010 equipped with an X-ray detector OXFORD INCA Energy TEM 100 for microanalysis (EDS).XRF analyses were obtained on a PHILIPS MAGIX PRO (PW2400) X-ray spectrometer equipped with a rhodium X-ray tube and a beryllium window.BET isotherms were carried out on a AUTOSORB-6 (Quantachrome), using N 2 .Melting points were obtained with a Reichert Thermovar apparatus.NMR spectra were recorded on a Bruker AC-300 (300 MHz for 1 H and 75 MHz for 13 C) using CDCl 3 as a solvent and TMS as internal standard for 1 H and

General procedures for the preparation of the products
To a stirred solution of sodium azide (6, 2 mmol) and benzyl halide (5, 2 mmol) were added NiO/Cu-Fe 3 O 4 (50 mg, 0.9 mol% of Ni and 0.9 mol% of Cu) and the corresponding alkyne (1 or 8, 1 mmol).The resulting mixture was stirred at 50 C until the end of the reaction.The catalyst was removed by a magnet and the resulting mixture was quenched with deionized water and extracted with AcOEt (3 Â 5 mL).The organic phases were dried over MgSO 4 , followed by evaporation under reduced pressure to remove the solvent.The product was usually puried by chromatography on silica gel (hexane-ethyl acetate) to give the corresponding products 3 or 9 (see ESI †).The chromatographic analyses were determined with a ame ionization detector and a 30 m capillary column (0.32 mm diam.0.25 mm lm thickness, HP-5 stationary phase), using nitrogen (2 mL min À1 ) as carrier gas, P ¼

a
Isolated yield aer column chromatography.b Powder < 5 mm.c Powder < 50 nm.d Reaction performed during 24 h.Fig. 2 Recycling of NiO/Cu-Fe 3 O 4 catalyst for the multicomponent reaction.
General procedure for the preparation of NiO/Cu-Fe 3 O 4 catalystTo a stirred solution of CuCl 2 (1 mmol, 130 mg) and NiCl 2 $H 2 O (1 mmol, 130 mg) in deionized water (120 mL) was added commercially available Fe 3 O 4 (4 g, 17 mmol, powder < 5 mm, BET area: 9.86 m 2 g À1 ).Aer 10 minutes at room temperature, the mixture was slowly basied with NaOH (1 M) until pH around 13.The mixture was stirred during one day at room temperature in air.Aer that, the catalyst was ltered and washed several times with deionized water (3 Â 10 mL).The solid was dried at 100 C during 24 h in a standard glassware oven, obtaining thereaer the expected catalyst.

Table 1
Optimization of cycloaddition reaction conditions a Isolated yield aer column chromatography.b Reaction carried out in absence of catalyst.c Reaction performed using only nanomagnetite (21 mol%).

Table 3
Optimization of catalyst for multicomponent cycloaddition

Table 4
Multicomponent cycloaddition a Isolated yield aer column chromatography.b Reaction performed during 4 days.Scheme 1 Multicomponent cycloaddition with benzyl derivatives.Scheme 2 Multicomponent cycloaddition with internal alkynes.