Exceptional activity of gallium(III) chloride and chlorogallate(III) ionic liquids for Baeyer–Villiger oxidation

Abstract

Baeyer–Villiger oxidation of cyclic ketones, using H₂O₂ as the oxidising agent, was systematically studied using a range of metal chlorides in different solvents, and in neat chlorogallate(III) ionic liquids. The extremely high activity of GaCl₃ in promoting oxidation with H₂O₂, irrespective of solvent, was reported for the first time. The activity of all other metal chlorides was strongly solvent-dependent. In particular, AlCl₃ was very active in a protic solvent (ethanol), and tin chlorides, SnCl₄ and SnCl₂, were active in aprotic solvents (toluene and dioxane). In order to eliminate the need for volatile organic solvent, a Lewis acidic chlorogallate(III) ionic liquid was used in the place of GaCl₃, which afforded typically 89–94% yields of lactones in 1–120 min, at ambient conditions. Raman and ⁷¹Ga NMR spectroscopic studies suggest that the active species, in both GaCl₃ and chlorogallate(III) ionic liquid systems, are chlorohydroxygallate(III) anions, [GaCl₃OH]⁻, which are the products of partial hydrolysis of GaCl₃ and chlorogallate(III) anions; therefore, the presence of water is crucial.

---

Introduction

Baeyer–Villiger (BV) oxidation, in which linear or cyclic ketones are oxidised to esters or lactones, has been known for over 110 years.¹ It is particularly useful to generate lactones, used as intermediates for pharmaceuticals, herbicides and polymers, as well as solvents, flavours and fragrance additives.^1,2 Common oxidants in BV oxidation are peroxyacids, such as peroxyacetic or m-chloroperoxybenzoic. They are very active oxidants, but also intrinsically unstable and shock-sensitive, and thus expensive to store and transport, hazardous to use, and they produce acidic waste. All these factors drive the search for more sustainable oxidising agents.

Aqueous hydrogen peroxide, a much ‘greener’ alternative: it is more stable, therefore safer to use, and generates water as the oxidation by-product. Although H₂O₂ is characterised by a high content of active oxygen, it is kinetically inert, and requires an acidic catalyst to oxidise ketones.^2,3

Lewis acids catalyse BV oxidation by coordinating to the carbonyl group of a ketone, thereby activating it towards the nucleophilic attack of hydrogen peroxide by increasing the polarisation of the C=O double bond.^2,4 A number of Lewis acidic metal chlorides (AlCl₃, SnCl₄, FeCl₃, ZrCl₄, TiCl₄ and ZnCl₂) have been used as catalysts in Baeyer–Villiger oxidation with H₂O₂ (as aqueous solution) or with silyl peroxides (which is a masked version of 100% H₂O₂).^3,5–10 However, the literature data are fragmentary, and it is not possible to directly compare the catalysts due to the use of different solvents, substrates and conditions. The most studied reaction is the oxidation of ketones with H₂O₂ catalysed by AlCl₃, both homogenously (in ethanol), and heterogeneously (on alumina or silica supports).^5–7 Less encouraging results were found using tin chlorides: SnCl₄ in methyl t-butyl ether was very weakly active in these oxidation of 2-adamantanone with H₂O₂, yielding only 3% conversion,⁸ whereas SnCl₂ in butanol was found inactive in the oxidation with H₂O₂.⁹ In contrast, SnCl₄, AlCl₃ and FeCl₃ were all active catalysts in BV oxidation with bistrimethylsilyl peroxide as an oxidising agent.¹⁰

Additionally, several organoselenium compounds, like bis[3,5-bis(trifluoromethyl)phenyl]diselenide,¹¹ 3,5-bis(perfluorooctyl)phenyl butylselenide,¹² diselenide bearing bistriflate¹³ and dibenzyl diselenide¹⁴ were used as catalysts for the Baeyer–Villiger oxidation with H₂O₂ (30 or 60% aq.). The oxidation of ketones were carried out generally at room temperature from 2 (for cyclobutanone) to 24 h (for larger cyclic ketones) giving high yields of lactones.

The very promising results with H₂O₂ as an oxidant were also achieved using molecular sieves modified with tin (Sn-Beta) or Sn(IV) centres incorporated into mesoporous MCM-41 (Sn–MCM-41) as catalysts, in dioxane at 90 °C.^8,15 Surprisingly, homogenous BV oxidation in dioxane with Lewis acidic chlorides has not been reported.

In the search for a sustainable BV oxidation, we have studied the use of ionic liquids, both as solvents and, more recently, as catalysts. In studying the solvent effect, it has been demonstrated that both conversion and reaction kinetics depended on the structure of ionic liquid or molecular solvent used.^16,17 Mildly Brønsted acidic, supported ionic liquids based on [HSO₄]⁻, were active and recyclable catalysts for BV oxidation with H₂O₂ in dichloromethane.¹⁶ Strongly Lewis acidic ionic liquids with chloroaluminate(III) anions were very potent catalysts for cyclobutanone oxidations with silyl peroxides.¹⁷ Despite promising results, this work remains the only example of BV oxidation catalysed with Lewis acidic ionic liquids reported to-date.

Here, we aim to directly compare the catalytic performance of various metal chlorides in BV oxidation of model cyclic ketones with H₂O₂, and to investigate whether the performance of the most active metal chlorides could be further enhanced by using them in the form of chlorometallate ionic liquids, rather than as a solution in a molecular solvent.

Experimental

Materials and methods

All materials (metal chlorides, 1-butyl-3-methylimidazolium chloride, ketones) were purchased from Sigma-Aldrich. Metal chlorides were anhydrous (purity 99.99%, sealed under argon in glass ampoules).

Substituted cyclobutanones (3-phenylcyclobutanone and 3-butylcyclobutanone) were synthesised following a standard two-step procedure: a [2 + 2] cycloaddition of dichloroketene to the vinyl derivative, followed by the reduction of the resulting dichloroketones with zinc in acetic acid (see ESI† for the NMR spectra).¹⁸

GC analyses were performed using a PerkinElmer Clarus 500 gas chromatograph equipped with an SPB™-5 column (30 m × 0.2 mm × 0.2 μm).

¹H NMR spectra were recorded at 300 MHz or 600 MHz and ¹³C NMR at 75 or 150 MHz (Varian system), respectively.

⁷¹Ga NMR spectra were recorded at 300 K using a Bruker AvanceIII 400 MHz spectrometer. Samples were analysed neat, using d₆-dimethylsulfoxide capillaries as an external lock. A solution of Ga(NO₃)₃ hydrate in D₂O was used as an external reference (δ = 0 ppm).

Raman analyses were performed using a Perkin-Elmer Raman Station 400F, with a 785 nm focussed laser beam. All samples were studied in quartz cuvettes, with forty 2 second scans recorded for each sample.

Synthetic procedures

Chlorometallate(III) ionic liquids. Chlorometallate(III) ionic liquids were synthesised in a nitrogen-filled glovebox (MBraun labmaster dp, <0.3 ppm of H₂O and O₂). In a typical procedure,¹⁹ 1-butyl-3-methylimidazolium chloride, [C₄mim]Cl, was placed in a vial (10 cm³) equipped with a stirring bar. Then, metal(III) chloride (M = Al or Ga) was added slowly, in portions, to achieve the desired composition, and allowed to react until a homogenous liquid was formed (3 h, 60 °C, 1000 rpm). Exact amounts of reactants are shown in Table 1.

Table 1 Compositions of chlorometallate ionic liquids

Ionic liquid	χMClx^a	[C4mim]Cl		MClx
		mmol	g	mmol	g
a χMClx – molar fraction of metal chlorides in ionic liquids.
[C4mim]Cl–GaCl3	0.50	5.70	0.9956	5.70	1.0037
[C4mim]Cl–GaCl3	0.67	3.80	0.6637	7.59	1.3364
[C4mim]Cl–GaCl3	0.75	2.85	0.4978	8.54	1.5037
[C4mim]Cl–AlCl3	0.67	4.53	0.7913	9.06	1.2081

---

Oxidation of ketones in the presence of metal chlorides. Into a two-necked round-bottomed flask equipped with a septum, a condenser with a balloon filled with nitrogen, and a magnetic stirring bar, 2-adamantanone (0.1 g, 0.67 mmol) and metal chloride (0.008–0.017 g, 0.067 mmol) were placed. Then, a molecular solvent (5 cm³, see Table 2) was added, followed by the slow addition of 30% aq. H₂O₂ (0.152 g, 1.34 mmol). The reactions were carried out at 40–90 °C, and monitored by GC.

Table 2 The influence of the metal chloride MCl_x structure on the oxidation of 2-adamantanone^a

Solvent	MClx	Yield of lactone^b/%
a Reaction conditions: 2-adamantanone (0.67 mmol), H2O2 30% aq. (1.34 mmol), solvent 5 cm^3 catalyst 10 mol%, reaction time 6 h.b Yields determined by GC.
Dichloromethane 40 °C	GaCl3	96
	AlCl3	32
	SnCl2	57
	SnCl4	65
	InCl3	43
	TiCl3	25
	TiCl4	12
	ZnCl2	1
	FeCl3	7
Toluene 90 °C	GaCl3	99
	AlCl3	12
	SnCl2	98
	SnCl4	96
	InCl3	71
	TiCl3	0
	TiCl4	7
	ZnCl2	66
	FeCl3	1
Ethanol 70 °C	GaCl3	75
	AlCl3	94
	SnCl2	57
	SnCl4	78
	InCl3	6
	TiCl3	18
	TiCl4	22
	ZnCl2	23
	FeCl3	4
1,4-Dioxane 90 °C	SnCl2	83
	SnCl4	83

---

Oxidation of ketones in the presence of ionic liquids. Into a two-necked round-bottomed flask equipped with a septum, a condenser with a balloon filled with nitrogen, and a magnetic stirring bar, a ketone (0.094–0.234 g, 1.34 mmol) and ionic liquid (0.470–0.942 g, 1.34 mmol) were placed. Then, 30% aq. H₂O₂ (0.304 g, 2.68 mmol) was slowly added into the flask using a syringe. After the reaction completion (1–120 min), 2 cm³ of water and 2 cm³ of dichloromethane were added, and stirred vigorously (5 min). The organic phase was analysed by GC with decane as an internal standard. For the isolation of product, the reaction mixture was dissolved in water (2 cm³), extracted with dichloromethane (6 × 5 cm³) and concentrated. The yields of lactones after the purification using column chromatography with a short bed of Al₂O₃ (the conversion of ketone was 100%, and only the isolation of traces of ionic liquid was necessary) and dichloromethane as the eluent were 89–93%. 4-Oxatricyclo[4.3.1.13.8]undecan-5-one was isolated by the extraction of the post reaction mixture with toluene (10 × 2 cm³). After vacuum drying, the lactone was purified by crystallisation from hexane–ethyl acetate. The structure and purity of all synthesised ketones and lactones were confirmed by ¹H and ¹³C NMR analysis (ESI) and were essentially identical with published spectral data for authentic samples:
3-Phenylcyclobutanone¹⁸. (¹H-NMR, 300 MHz, CDCl₃, TMS): δ/ppm 3.22–3.27 (m, 2H), 3.44–3.50 (m, 2H), 3.66 (quint., J = 8.2 Hz, 1H), 7.20–7.38 (m, 5H). ¹³C-NMR (75 MHz, CDCl₃): δ/ppm 31.1, 57.4, 129.2, 129.3, 131.4, 146.3, 209.2.
3-Butylcyclobutanone¹⁸. (¹H-NMR, 300 MHz, CDCl₃, TMS): δ/ppm 0.92 (t, J = 6.8 Hz, 3H), 1.20–1.40 (m, 4H), 1.55–1.62 (m, 1H), 2.28–2.43 (m, 1H), 2.50–2.70 (m, 2H), 3.10–3.19 (m, 2H). ¹³C-NMR (75 MHz, CDCl₃): δ/ppm 13.9, 22.4, 23.8, 30.4, 35.9, 52.4, 208.6.
3-Phenyl-γ-butyrolactone¹⁸. (¹H NMR, 300 MHz, CDCl₃, TMS): δ/ppm 2.66 (dd, J = 17.5 Hz, 9.1 Hz, 1H), 2.92 (dd, J = 17.5 Hz, 8.1 Hz, 1H), 3.77 (quint., J = 8.4 Hz, 1H), 4.26 (dd, J = 8.9 Hz, 8.0 Hz, 1H), 4.67 (dd, J = 9.0 Hz, 7.9 Hz, 1H), 7.16–7.27 (m, 2H), 7.27–7.33 (m, 1H), 7.33–7.34 (m, 2H). ¹³C NMR (75 MHz, CDCl₃): δ/ppm 35.7, 41.1, 74.0, 126.7, 127.7, 129.1, 139.4, 176.3.
3-Butyl-γ-butyrolactone¹⁸. (¹H NMR, 300 MHz, CDCl₃, TMS): δ/ppm 0.92 (t, J = 6.9 Hz, 3H), 1.15–1.30 (m, 4H), 1.43–1.52 (m, 2H), 2.14–2.33 (m, 2H), 2.49–2.66 (m, 2H), 3.89–3.95 (m, 1H), 4.39–4.44 (m, 1H). ¹³C-NMR (75 MHz, CDCl₃): δ/ppm 13.8, 22.4, 29.4, 32.7, 34.4, 35.6, 73.3, 177.2.
γ-Butyrolactone²⁰. (¹H-NMR, 300 MHz, CDCl₃, TMS): δ/ppm 2.13–2.24 (m, 2H), 2.38–2.44 (m, 2H), 4.27 (t, J = 7.4 Hz, 2H). ¹³C-NMR (75 MHz, CDCl₃): δ/ppm 22.3, 28.0, 68.8, 178.1.
4-Oxatricyclo[4.3.1.13.8]undecan-5-one²⁰. (¹H-NMR, 300 MHz, CDCl₃, TMS): δ/ppm 1.70–2.15 (m, 12H), 3.01–3.12 (m, 1H), 4.39–4.52 (m, 1H). ¹³C NMR (75 MHz, CDCl₃): δ/ppm 25.8, 30.9, 33.7, 35.7, 41.2, 73.1, 178.2.
2-Oxabicyclo[3.2.1]octan-3-one²⁰. (¹H-NMR, 300 MHz, CDCl₃, TMS): δ/ppm 1.65–2.25 (m, 7H), 2.40–2.80 (m, 2H), 4.82–4.89 (m, 1H). ¹³C NMR (75 MHz, CDCl₃): δ/ppm 29.1, 31.7, 32.4, 35.7, 40.5, 80.9, 170.7.
ε-Caprolactone²⁰. (¹H-NMR, 600 MHz, CDCl₃, TMS): δ/ppm 1.72–1.87 (m, 4H), 1.88 (dd, J = 5.3 Hz, 2.6 Hz, 2H), 2.62–2.67 (m, 2H), 4.21–4.26 (m, 2H). ¹³C-NMR (150 MHz, CDCl₃): δ/ppm 22.9, 28.9, 29.3, 34.5, 69.2, 176.4.
4-Methyl-ε-caprolactone²¹. (¹H-NMR, 300 MHz, CDCl₃, TMS): δ/ppm 0.94 (d, J = 6.6 Hz, 3H), 1.80–1.92 (m, 5H), 2.55–2.71 (m, 2H), 4.15–4.25 (m, 2H). ¹³C-NMR (75 MHz, CDCl₃): δ/ppm 18.8, 21.6, 34.2, 36.7, 40.2, 67.5, 175.5.
4-Ethyl-ε-caprolactone²². (¹H NMR, 600 MHz, CDCl₃, TMS): δ/ppm 0.92 (t, J = 7.5 Hz, 3H), 1.40–1.28 (m, 3H), 1.57–1.43 (m, 2H), 1.93 (dddt, J = 15.1, 7.5, 3.6, 2.0 Hz, 1H), 2.03–1.96 (m, 1H), 2.61 (ddd, J = 14.1, 12.4, 2.1 Hz, 1H), 2.69 (ddd, J = 14.1, 7.6, 1.7 Hz, 1H), 4.18 (dd, J = 12.8, 10.1 Hz, 1H), 4.31 (ddd, J = 12.7, 5.9, 1.6 Hz, 1H). ¹³C NMR (150 MHz, CDCl₃): δ/ppm 11.14, 28.38, 28.98, 33.02, 34.83, 41.73, 68.05, 176.01.
4-Propyl-ε-caprolactone²². (¹H NMR, 600 MHz, CDCl₃, TMS): δ/ppm 0.90 (t, J = 7.2 Hz, 3H), 1.38–1.22 (m, 6H), 1.53–1.44 (m, 1H), 1.67–1.57 (m, 2H), 1.92 (dddd, J = 16.0, 7.6, 3.8, 1.9 Hz, 1H), 2.01–1.95 (m, 1H), 2.61 (ddd, J = 14.1, 12.3, 2.1 Hz, 1H), 2.68 (ddd, J = 14.0, 7.6, 1.7 Hz, 1H), 4.18 (dd, J = 12.9, 10.4 Hz, 1H), 4.30 (ddd, J = 12.9, 6.0, 1.8 Hz, 1H). ¹³C NMR (150 MHz, CDCl₃): δ/ppm 13.97, 19.70, 28.74, 33.04, 35.20, 38.50, 39.75, 68.05, 176.08.
4-tert-Butyl-ε-caprolactone²³. (¹H NMR, 600 MHz, CDCl₃, TMS) δ/ppm 0.90 (s, 9H), 1.41–1.30 (m, 2H), 1.57–1.47 (m, 1H), 2.11–1.99 (m, 2H), 2.62–2.53 (m, 1H), 2.71 (ddd, J = 14.2, 7.5, 1.2 Hz, 1H), 4.16 (dd, J = 12.9, 10.5 Hz, 1H), 4.34 (ddd, J = 12.9, 6.0, 1.8 Hz, 1H). ¹³C NMR (150 MHz, CDCl₃): δ/ppm 23.76, 27.43, 30.33, 32.98, 33.45, 50.77, 68.62, 176.29.
4-Phenyl-ε-caprolactone²⁴. (¹H NMR, 600 MHz, CDCl₃, TMS): δ/ppm 1.89–1.80 (m, 1H), 2.11–1.99 (m, 2H), 2.17–2.11 (m, 1H), 2.88–2.73 (m, 3H), 4.32 (dd, J = 12.8, 10.3 Hz, 1H), 4.39 (ddd, J = 13.0, 5.5, 2.0 Hz, 1H), 7.17–7.20 (m, 2H), 7.25–7.21 (m, 1H), 7.34–7.30 (m, 2H). ¹³C NMR (150 MHz, CDCl₃): δ/ppm 30.33, 33.69, 36.75, 47.25, 68.20, 126.57, 126.84, 128.75, 144.92, 175.58.
Isopropyl propanoate²⁵. (¹H NMR, 600 MHz, CDCl₃, TMS): δ/ppm 1.11 (t, J = 7.6 Hz, 3H), 1.22 (d, J = 6.3 Hz, 6H), 2.28 (q, J = 7.6 Hz, 2H), 5.00 (dt, J = 12.5, 6.2 Hz, 1H). ¹³C NMR (150 MHz, CDCl₃): δ/ppm 9.11, 21.78, 27.93, 67.45, 176.11–174.37.
p-Methoxyphenyl acetate²⁶. (¹H NMR, 600 MHz, CDCl₃, TMS): δ/ppm 2.28 (s, 2H), 3.79 (s, 2H), 6.89 (d, J = 9.1 Hz, 1H), 7.00 (d, J = 9.1 Hz, 1H). ¹³C NMR (150 MHz, CDCl₃): δ/ppm 21.02, 55.55, 114.44, 122.28, 144.17, 157.23, 169.87.

Results and discussion

Screening of metal chlorides

Metal chlorides were used as Lewis acidic catalysts in a model oxidation of 2-adamantanone (Scheme 1) with a two-fold molar excess of H₂O₂ (30% aq.), at 10 mol% metal chloride loading. Each metal chloride was tested in three solvents: dichloromethane, ethanol and toluene. Additionally, tin chlorides were tested in 1,4-dioxane.

Scheme 1 Model Baeyer–Villiger oxidation of 2-adamantanone.

The observed catalytic activity was strongly dependent the metal chloride, the solvent, and the reaction temperature (Table 2). In agreement with the literature, high yield of lactone was obtained using AlCl₃ in ethanol, but not with less polar and aprotic solvents.⁵ In contrast, the previously unstudied GaCl₃ was the most active catalyst under all conditions. Moreover, SnCl₂ and SnCl₄ were very active in toluene and 1,4-dioxane, but not in ethanol and dichloromethane. These observations may explain the previously reported low activity of tin chlorides.^8,9 Other metal chlorides (InCl₃, ZnCl₂, TiCl₃, TiCl₄, FeCl₃) exhibited much lower activity, with an unpredictable dependence on the solvent.

The high catalytic activity of GaCl₃ was most likely related to its high solubility in the reaction mixtures, compared to other metal halides, which resulted in favourable reaction kinetics. Kinetic studies of selected catalytic systems (Fig. 1) revealed superior performance of GaCl₃, even at a lower temperature (40 °C).

Fig. 1 Kinetics of the oxidation of 2-adamantanone (0.67 mmol), with H₂O₂ 30% aq. (1.34 mmol) in solvent (5 cm³), in the presence of MCl_x (10 mol%).

Screening of chlorogallate(III) ionic liquids

Seeking to additionally improve the reaction sustainability by the removal of volatile organic solvents, and in the hope of further enhancement in the reaction rate, chlorogallate(III) ionic liquids were studied as Lewis acidic catalysts.

Chlorogallate(III) ionic liquids are synthesised by the reaction of anhydrous GaCl₃ with an organic chloride salt. Various molar fractions of GaCl₃, expressed as χ_GaCl3, can be used to prepare homogenous ionic liquids.¹⁹ For χ_GaCl3 < 0.50, a neutral [GaCl₄]⁻ anion coexists with Lewis basic chloride anion; for χ_GaCl3 > 0.50, multinuclear chlorogallate(III) anions: [Ga₂Cl₇]⁻ and [Ga₃Cl₁₀]⁻ are formed, which impart strong Lewis acidity to these systems.²⁷ At the ideally neutral composition, χ_GaCl3 = 0.50, only [GaCl₄]⁻ is present. Compared to chloroaluminates(III), chlorogallates(III) offer equal or stronger Lewis acidity, as quantified by acceptor number (AN) approach,²⁷ and are more resistant to hydrolysis. Neutral chlorogallates(III) (χ_GaCl3 = 0.50) have been used as solvents and co-catalysts in palladium-catalysed hydroethoxycarbonylation²⁸ and acetal formation,²⁹ whereas Lewis acidic systems (χ_GaCl3 = 0.67, 0.75) catalysed arene carbonylation,³⁰ and oligomerisation of alkenes to lubricant base oils.³¹ In general, however, applications of chlorogallate(III) systems are relatively poorly explored compared to their chloroaluminate(III) analogues.

Three compositions of chlorogallate(III) ionic liquids were used: χ_GaCl3 = 0.50, 0.67 and 0.75, which corresponds to three different chlorogallate(III) anions, dominant in the respective compositions: [GaCl₄]⁻ (neutral, as a benchmark), [Ga₂Cl₇]⁻ (Lewis acidic), and [Ga₃Cl₁₀]⁻ (Lewis acidic). All ionic liquids were based on 1-butyl-3-methylimidazolium cation, [C₄mim]⁺. Results of a model oxidation of 2-adamantanone with a two-fold molar excess of H₂O₂ (30% aq.) are presented in Table 3. In order to form a homogenous reaction medium, equimolar amounts of ketone and ionic liquid (based on cation) were typically required. Depending on the anionic speciation, this resulted in molar excess of gallium metal to substrate between 1 and 3. Hence, the reactions in the ionic liquids contained at least ten times the concentration of metal than the equivalent reactions in molecular solvents. Nevertheless, this does not account for the significantly greater activity of the ionic liquid.

Table 3 The influence of chlorometallate(III) ionic liquids as catalysts for 2-adamantanone oxidation with H₂O₂ 30% aq., at ambient temperature^a

Ionic liquid	IL^b/mol%	Time/min	Yield of lactone^c/%
a Reaction conditions: 2-adamantanone (0.67 mmol), H2O2 30% aq. (1.34 mmol), RT.b Relative to ketone.c Yields determined by GC.
[C4mim]Cl–GaCl3, χ = 0.75	25	1	99
	50	1	99
	100	1	99
[C4mim]Cl–GaCl3, χ = 0.67	100	1	93
[C4mim]Cl–GaCl3, χ = 0.50	100	1	0
		60	15
		300	43
[C4mim]Cl–AlCl3, χ = 0.67	100	1	0
		60	11
		300	11

---

Using the ionic liquid with the highest GaCl₃ concentration (χ_GaCl3 = 0.75) and the highest Lewis acidity (expressed as acceptor number, AN = 107.5), the oxidation was remarkably fast, reaching 99% yield of lactone and ketone conversion, in less than 1 min at ambient temperature. This contrasts with solutions of GaCl₃ in molecular solvents, which took more than two orders of magnitude longer, at elevated temperatures, to produce comparable yields.

The chlorogallate(III) ionic liquid with lower GaCl₃ concentration (χ_GaCl3 = 0.67) and slightly lower Lewis acidity (AN = 99.5) was also slightly less active. Nevertheless, it still displayed extremely high catalytic activity, yielding 93% lactone yield after 1 min at ambient temperature. In contrast, no conversion was detected under the same reaction conditions in the presence of the neutral chlorogallate(III) ionic liquid (χ_GaCl3 = 0.50). Upon prolonged reaction time some reactivity occurred (43% lactone yield after 300 min), possibly due to hydrolysis of [GaCl₄]⁻, resulting in catalysis by the evolved hydrogen chloride.

For comparison, strongly Lewis acidic chloroaluminate(III) ionic liquids were used as catalysts (χ_AlCl3 = 0.67, AN = 95.6), at the same temperature. Very low yields of lactone (11%) were obtained, even after 300 min. This is in agreement with some of the results recorded for catalysis with metal chlorides (Table 2), where the performance of AlCl₃ was inferior to that of GaCl₃.

Screening of oxidising agents

The catalytic activity of [C₄mim]Cl–GaCl₃ (χ_GaCl3 = 0.75) was tested in the presence of a range of oxidising agents. Results for two concentrations of H₂O₂ in water (30 and 60%) were compared to oxidations using the anhydrous urea·H₂O₂ complex (UHP), cumyl hydroperoxide (neat), and tert-butyl hydroperoxide, used as solutions in water and in decane (Table 4). Equimolar amounts of ionic liquid (based on the cation) and ketone were used in all cases.

Table 4 The influence of oxidation agent on the oxidation of 2-adamantanone in the presence of the chlorogallate(III) ionic liquid^a

Oxidant	Oxidant^b/mol%	Time/min	Yield of lactone^c/%
a Reaction conditions: 2-adamantanone (0.67 mmol), [C4mim]Cl–GaCl3, χGaCl3 = 0.75 (0.67 mmol), RT.b Relative to ketone.c Yields determined by GC.d 30% concentration of H2O2 in the solution.
H2O2 30% aq.	200	1	99
	150	1	98
	110	1	87
H2O2 60% aq.	200	1	99
UHP	200	60	80
UHP 30% aq.^d	200	1	92
PhC(CH3)2OOH	200	60	3
^tBuOOH in water	200	60	5
^tBuOOH in decane	200	60	0

---

Using decreasing amounts of H₂O₂ (30% aq.), it was demonstrated that the ratio of oxidising agent could be decreased from two-fold to nearly equimolar, without significant product loss. Decreasing water content (i.e. increasing H₂O₂ concentration to 60%) had no observable effect on conversion and yield, compared to using 30% aqueous solution of H₂O₂. However, replacing aqueous solution with anhydrous UHP complex (with formal H₂O₂ concentration of 36%), led to a dramatic decrease in the lactone yield (80% after 60 min). To confirm the beneficial role of water, the reaction with UHP was repeated in the presence of 0.1 g of water (giving formal H₂O₂ concentration of ca. 30%), with resulted in 92% lactone yield after 1 min. This confirmed that the presence of water is crucial for the observed high reaction rates in gallium(III)-promoted BV oxidation with H₂O₂. Interestingly, this is in direct opposition to earlier views, whereby water was detrimental for metal-catalysed BV oxidation, which was considered suitable only for anhydrous conditions.²

In BV oxidation, H₂O₂ is known to be a less active oxidant than peracids, but more active than hydroperoxides.^2,3 Considering the excellent performance of gallium(III) catalysts in H₂O₂ oxidation, it was anticipated that they may also catalyse oxidation by hydroperoxides. However, irrespective of the solvent (neat, water, aprotic solvent – see Table 4), hydroperoxides remained inactive in the model oxidation catalysed by the chlorogallate(III) ionic liquids.

Mechanistic considerations

Screening studies revealed that gallium(III) chloride and chlorogallate(III) ionic liquids were much more active catalysts in BV oxidation with aqueous H₂O₂ than their aluminium(III) analogues. A combination of Raman and ⁷¹Ga NMR spectroscopic studies was used to shed some light on the activation process. Two active systems were investigated: GaCl₃ and the chlorogallate(III) ionic liquid (χ_GaCl3 = 0.75), both as mixtures with aqueous H₂O₂ (molar ratio 1 : 1). As a comparison, aqueous H₂O₂ solutions of the inactive, neutral chlorogallate(III) ionic liquid (χ_GaCl3 = 0.50), and of the less active AlCl₃, have also been studied.

The inactive mixture of [C₄mim]Cl–GaCl₃, χ_GaCl3 = 0.50, and hydrogen peroxide was biphasic, comprising an aqueous phase and an organic phase. In the organic phase, the only detected gallium(III) species was the neutral tetrachlorogallate(III) anion, confirmed by the presence of a Raman Ga–Cl stretching frequency at 349 cm⁻¹ (Fig. 2A),³² and a narrow ⁷¹Ga NMR signal at 250 ppm (Fig. 2B),¹⁹ both characteristic of the tetrahedral [GaCl₄]⁻. The aqueous phase contained H₂O₂ (Raman O–O stretching frequency at 876 cm⁻¹, Fig. 2A), and the product of complete hydrolysis of the chlorogallate(III) anion, [Ga(H₂O)₆]³⁺, characterised by a narrow ⁷¹Ga NMR signal at 0 ppm (Fig. 2B), as the only observable gallium species.

Fig. 2 (A) Raman spectra of [C₄mim]Cl–GaCl₃ χ_GaCl3 = 0.50 (green), water phase (navy blue) and organic phase (red) after reaction with H₂O₂; (B) ⁷¹Ga NMR spectra of water phase (navy blue) and organic phase (red) of [C₄mim]Cl–GaCl₃ χ_GaCl3 = 0.50 after addition of H₂O₂.

The mixture of AlCl₃ and hydrogen peroxide formed a homogenous liquid, in a very exothermic mixing process. It contained products of AlCl₃ hydrolysis, but no H₂O₂. Hydrogen peroxide was assumed to have undergone a rapid decomposition due to temperature and/or released HCl, which may be the main factor lowering the activity of AlCl₃.

The catalytically active ionic liquid, [C₄mim]Cl–GaCl₃ (χ_GaCl3 = 0.75), in its neat state, contained oligonuclear chlorogallate(III) anions, such as [Ga₂Cl₇]⁻ and [Ga₃Cl₁₀]⁻, as indicated by a group of Raman signals between 500 and 100 cm⁻¹ (Fig. 3A), that correspond to vibrations of terminal and bridging chlorides in oligomeric chlorogallate(III) anions, in agreement with the literature.³³

Fig. 3 (A) Raman spectra of [C₄mim]Cl–GaCl₃ χ_GaCl3 = 0.75 (orange), [C₄mim]Cl–GaCl₃ χ_GaCl3 = 0.75 after reaction with H₂O₂ (blue) and of the mixture GaCl₃–H₂O₂ (green); (B) ⁷¹Ga NMR spectra of [C₄mim]Cl–GaCl₃ χ_GaCl3 = 0.75 after reaction with H₂O₂ (blue) and of the mixture GaCl₃–H₂O₂ (green).

Upon the addition of the ionic liquid to aqueous hydrogen peroxide, a single phase was formed in a highly exothermic process. The same was observed upon the addition of solid GaCl₃ to aqueous H₂O₂. For both solutions, Raman spectra exhibited a Ga–Cl stretching frequency at 348 cm⁻¹, corresponding to monomeric chlorogallate(III) units, and signal at 876 cm⁻¹, corresponding to the O–O stretch in H₂O₂ (Fig. 3A). The ⁷¹Ga NMR spectra both contained two signals (Fig. 3B). A broad peak ca. 240 ppm corresponded to tetracoordinate gallium(III) anions. The broadening and a slight upfield shift compared to that for [GaCl₄]⁻ may suggests the partial replacement of a chloride with a hydroxide ligand (⁷¹Ga NMR signal for [GaOH₄]⁻ is at 225 ppm).³⁴ It may also indicate the exchange with other gallium species. The signal at 0 ppm was assigned to [Ga(H₂O)₆]³⁺, but again – the broadening indicates exchange with other species. These results are in agreement with prior NMR studies on GaCl₃ hydrolysis.^33,34 In addition, complementary FAB-MS studies on hydrolysis of chlorogallate(III) ionic liquids revealed that the hydrolysis proceeds via chlorohydroxygallate(III) anions, such as [GaCl₃(OH)]⁻ and [GaCl(OH)₃]⁻,³⁵ which is in agreement with observed here small upfield shift in the ⁷¹Ga NMR spectrum.

In summary, both catalytically active systems contain mixtures of [GaCl₄]⁻, chlorohydroxygallate(III) anions, and [Ga(H₂O)₆]³⁺, coexisting in the same phase with H₂O₂. The monomeric [GaCl₄]⁻ anion it is not known to either undergo facile chloride replacement, or to expand its coordination to accommodate a fifth ligand; therefore it is commonly described as ‘neutral’, as opposed to ‘Lewis acidic’.²⁷ In contrast, chlorohydroxygallate(III) anions, such as [GaCl₃(OH)]⁻, are less sterically crowded and have been demonstrated to expand to penta- or hexacoordinate motif in the presence of O-donors, L, e.g. {GaCl₃L₂}.³⁶ Therefore, chlorohydroxygallate(III) complexes are proposed to be the active Lewis acidic species in BV oxidation (Scheme 2a and b), which also explains the beneficial influence of water in this reaction. Finally, a very slow conversion of ketone to lactone (43% after 5 h) detected in the presence of the ‘inactive’ [C₄mim]Cl–GaCl₃ (χ_GaCl3 = 0.50), was probably caused by hydrolytically-generated HCl, protonating the ketone and resulting in its mild activation (Scheme 2c).²

Scheme 2 Electrophilic activation of (a) ketone by Lewis acid, (b) the intermediate by Lewis acid, and (c) ketone by Brønsted acid in Baeyer–Villiger oxidation with hydrogen peroxide. Adapted from ref. 2.

Substrate scope

To determine the catalytic potential of the chlorogallate(III) ionic liquid, [C₄mim]Cl–GaCl₃ (χ_GaCl3 = 0.75), it was used as a catalyst for the synthesis of a range of lactones from the corresponding ketones (Table 5). Numerous cyclic ketones were readily oxidised to the corresponding lactones in high yields (93–99%) under the optimised reaction conditions, which were in general remarkably better than those reported in the literature.^4–7

Table 5 Oxidation of selected cyclic ketones to lactones^a

Entry	Ketone	Lactone	Time/min	Yield^b/%
a Reaction conditions: ketone (1.34 mmol), 30% aq. H2O2 (2.68 mmol), [C4mim]Cl–GaCl3, χGaCl3 = 0.75 (1.34 mmol), RT.b Yields determined using GC, isolated yields given in parenthesis.
1			1	99 (94)
2			5	99 (90)
3			1	99 (92)
4			1	99 (93)
5			1	99 (89)
6			1	26
7			30	99 (91)
8			30	99 (90)
9			15	99 (93)
10			15	99 (92)
11			120	99 (91)
12			120	15
			240	15
13			120	50

---

Very reactive, strained cyclobutanones were oxidised to γ-butyrolactones in 1 min with 99% yield. High yields and short reaction times were also achieved for the oxidation of 2-adamantanone and norcamphor. Non-strained cyclohexanones, which are much less reactive, were also successfully oxidised to corresponding ε-caprolactones at 99% yields. Reaction times varied from 15 to 120 min, depending on the substituent in the para position. The only exception was unsubstituted cyclohexanone, which yielded only 26% of ε-caprolactone after 1 min with full conversion of ketone. This anomaly could be caused by the presence of Brønsted acidic protons, which are known to convert cyclohexanone to the polymeric peroxides in the presence of H₂O₂.^2,37 Unfortunately, isopropyl propanoate and p-methoxyphenyl acetate, which are reactive representatives of linear ketones, were slowly oxidise and did not reached the full conversion.

Conclusions

The gallium(III) chloride systems provide the fastest known homogenous activation of the Baeyer–Villiger oxidation with aqueous hydrogen peroxide to produce near quantitative yields (89–94%) of lactones.² Gallium(III) chloride was found to be a very active catalyst, especially in aprotic solvents. Lewis acidic chlorogallate(III) ionic liquids were also extremely active, and in addition allowed for the elimination of volatile molecular solvents.

It was demonstrated that water was crucial for the efficient process activation. The active species, postulated based on Raman and ⁷¹Ga NMR spectroscopies, were hydroxychlorogallate(III) anions, such as [GaCl₃(OH)]⁻ or [GaCl(OH)₃]⁻, which are the intermediate products of hydrolysis of both GaCl₃ and chlorogallate(III) anions.

Acknowledgements

This work was financed by the Polish Ministry of Science and Higher Education, National Science Centre (grant no. UMO-2012/06/M/ST8/00030). MM would like to acknowledge the COST Action (CM1206 EXIL – Exchange on Ionic Liquids), which allowed for the Short Term Scientific Mission in The QUILL Research Centre.

Notes and references

1. M. Renz and B. Meunier, Eur. J. Org. Chem., 1999, 737.
2. G. Brink, I. Arends and R. A. Sheldon, Chem. Rev., 2004, 104, 4105.
3. G. Krow, Org. React., 1993, 43, 251.
4. C. Jimenez-Sanchidrian and J. R. Ruiz, Tetrahedron, 2008, 64, 2011.
5. Z. Lei, G. Ma, L. Wei, Q. Yang and B. Su, Catal. Lett., 2008, 124, 330.
6. Z. Lei, L. Wei, R. Wang and G. Ma, Catal. Commun., 2008, 9, 2467.
7. G. Ma, Z. Ma, Z. Zhang, Z. Yang and Z. Lei, J. Porous Mater., 2012, 19, 597.
8. A. Corma, M. T. Navarro, L. Nemeth and M. Renz, Chem. Commun., 2001, 2190.
9. Z. Lei, G. Ma and C. Jia, Catal. Commun., 2007, 8, 305.
10. S. Matsubara, K. Takai and H. Nozaki, Bull. Chem. Soc. Jpn., 1983, 56, 2029.
11. G. J. ten Brink, J. M. Vis, I. W. C. E. Arends and R. A. Sheldon, J. Org. Chem., 2001, 66, 2429.
12. G. J. ten Brink, J. M. Vis, I. W. C. E. Arends and R. A. Sheldon, Tetrahedron, 2002, 58, 3977.
13. H. Ichikawa, Y. Usami and M. Arimoto, Tetrahedron Lett., 2005, 46, 8665.
14. X. Zhang, J. Ye, L. Yu, X. Shi, M. Zhang, Q. Xu and M. Lautens, Adv. Synth. Catal., 2015, 357, 955.
15. A. Corma, L. T. Nemeth, M. Renz and S. Valencia, Nature, 2001, 412, 423.
16. A. Chrobok, S. Baj, W. Pudło and A. Jarzębski, Appl. Catal., A, 2009, 366, 22.
17. S. Baj, R. Słupska, A. Chrobok and A. Drożdż, J. Mol. Catal. A: Chem., 2013, 376, 120.
18. C. Bolm, O. Beckmann, T. Kuhn, C. Palazzi, W. Adam, P. B. Rao and C. R. Saha-Moller, Tetrahedron: Asymmetry, 2001, 12, 2441.
19. C. Hardacre, R. W. Murphy, K. R. Seddon, G. Srinivasan and M. Swadźba-Kwaśny, Aust. J. Chem., 2010, 63, 845.
20. M. E. Gonzalez-Nunez, R. Mello, A. Olmos and G. Asensio, J. Org. Chem., 2006, 71, 6432.
21. Z. Yang, L. Niu, X. Jia, Q. Kang, Z. Ma and Z. Lei, Catal. Commun., 2011, 12, 798.
22. J. W. Peeters, O. van Leeuwen, A. R. A. Palmans and E. W. Meijer, Macromolecules, 2005, 38, 5587.
23. K. Žmitek, S. Stavber, M. Zupan, D. Bonnet-Delponb and J. Iskraa, Tetrahedron, 2006, 62, 1479.
24. M. Y. Ríos, E. Salazar and H. F. Olivo, Green Chem., 2007, 9, 459.
25. Z. Jedlinski, A. Misiolek and P. Kurcok, J. Org. Chem., 1989, 54, 1500.
26. D. Qiu, Z. Zheng, F. Mo, Q. Xiao, Y. Tian, Y. Zhang and J. Wang, Org. Lett., 2011, 13, 4988.
27. J. Estager, A. A. Oliferenko, K. R. Seddon and M. Swadźba-Kwaśny, Dalton Trans., 2010, 39, 11375.
28. G. Rangits and L. Kollar, J. Mol. Catal. A: Chem., 2006, 246, 59.
29. Y. J. Kim and R. S. Varma, Tetrahedron Lett., 2005, 46, 7447.
30. E. J. Angueira and M. G. White, J. Mol. Catal. A: Chem., 2007, 277, 164.
31. M. P. Atkins, K. R. Seddon and M. Swadźba-Kwaśny, Pure Appl. Chem., 2011, 83, 1391.
32. S. P. Wicelinski, R. J. Gale, S. D. Williams and G. Mamantov, Spectrochim. Acta, Part A, 1989, 45, 759.
33. A. Fratiello, R. E. Lee and R. E. Schuster, Inorg. Chem., 1970, 9, 82.
34. P. Sipos, T. Megyes and O. Berkesi, J. Solution Chem., 2008, 37, 1411.
35. S. P. Wicelinski, R. J. Gale, K. M. Pamidimukkala and R. A. Laine, Anal. Chem., 1988, 60, 2228.
36. Cambridge Structural Database structures: COHPEU, COHPEU01, XITXUT.
37. A. Berkessel, M. R. M. Andreae, H. Schmickler and J. Lex, Angew. Chem., Int. Ed., 2002, 41, 4481.

---

Footnote

† Electronic supplementary information (ESI) available: NMR spectra of cyclic ketones and lactones. See DOI: 10.1039/c6ra03435g

---