*Article* **Isothiourea-Catalyzed Enantioselective** α**-Alkylation of Esters via 1,6-Conjugate Addition to** *para***-Quinone Methides**

**Jude N. Arokianathar <sup>1</sup> , Will C. Hartley <sup>1</sup> , Calum McLaughlin <sup>1</sup> , Mark D. Greenhalgh 1,2, Darren Stead <sup>3</sup> , Sean Ng <sup>4</sup> , Alexandra M. Z. Slawin <sup>1</sup> and Andrew D. Smith 1,\***


**Abstract:** The isothiourea-catalyzed enantioselective 1,6-conjugate addition of *para*-nitrophenyl esters to 2,6-disubstituted *para*-quinone methides is reported. *para*-Nitrophenoxide, generated in situ from initial *N*-acylation of the isothiourea by the *para*-nitrophenyl ester, is proposed to facilitate catalyst turnover in this transformation. A range of *para*-nitrophenyl ester products can be isolated, or derivatized in situ by addition of benzylamine to give amides at up to 99% yield. Although low diastereocontrol is observed, the diastereoisomeric ester products are separable and formed with high enantiocontrol (up to 94:6 er).

**Keywords:** isothiourea; ammonium enolate; aryloxide; quinone methide; ester functionalization; 1,6-conjugate addition

### **1. Introduction**

Quinone methides (QMs) are electrophilic compounds composed of a cyclohexadiene core bearing a carbonyl either *ortho* or *para* to an exocyclic alkylidene unit [1,2]. Due to their electrophilicity [3–6], QMs have been used in a variety of biological and medicinal processes [1,2,7], are present within natural products and pharmaceuticals [1,2,8–11], and have been applied as electrophiles in a variety of synthetic reactions [1,2,12–18]. While *ortho*-QMs have been used extensively in enantioselective catalysis [19], particularly as components in formal [4+2] cycloaddition reactions, the use of *para*-QMs has only recently received increased attention [19–23]. The majority of enantioselective organocatalytic methods that involve *para*-QMs have utilized Brønsted acid [24–33] or hydrogen bonding catalysts [34–38], with only a relatively small number of examples using Lewis base catalysis [39–51].

C(1)-Ammonium enolate intermediates [52–55], generated by the reaction of a tertiary amine Lewis base catalyst with a ketene, anhydride or acyl imidazole [56], have found widespread application for the synthesis of heterocyclic scaffolds in high yield and with excellent enantiocontrol. Traditionally these approaches have been limited by the requirement for the electrophilic reaction partner to contain a latent nucleophilic site to facilitate catalyst turnover. This conceptual obstacle has resulted in catalysis via C(1)-ammonium enolates being mostly applied for formal cycloaddition reactions. More recently, aryl esters have emerged as alternative C(1)-ammonium enolate precursors [55,57,58]. Significantly, following acylation of the tertiary amine catalyst by the aryl ester, a nucleophilic aryloxide is liberated, which may be exploited again in the catalytic cycle to facilitate ammonium enolate formation and catalyst turnover (Scheme 1a) [55,59]. This strategy offers a potentially

**Citation:** Arokianathar, J.N.; Hartley, W.C.; McLaughlin, C.; Greenhalgh, M.D.; Stead, D.; Ng, S.; Slawin, A.M.Z.; Smith, A.D. Isothiourea-Catalyzed Enantioselective α-Alkylation of Esters via 1,6-Conjugate Addition to *para*-Quinone Methides. *Molecules* **2021**, *26*, 6333. https://doi.org/ 10.3390/molecules26216333

Academic Editors: Igor Djerdj and Alejandro Baeza Carratalá

Received: 31 August 2021 Accepted: 14 October 2021 Published: 20 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

general solution to allow the expansion of electrophile scope within catalytic processes using C(1)-ammonium enolate intermediates. In 2014, we applied this concept for the isothiourea-catalyzed[2,3]-rearrangement of allylic ammonium ylides (Scheme 1b) [60–64]. More recently, this approach has been used by Snaddon (Scheme 1c) [65–72], Hartwig (Scheme 1d) [73] and Gong [74,75] for co-operative isothiourea/transition metal-catalyzed α-functionalization of pentafluorophenyl esters. In both cases, an isothiourea-derived C(1) ammonium enolate is intercepted by an electrophilic transition metal complex to affect an allylation or benzylation reaction. We have expanded the scope of electrophiles applicable within this catalyst turnover strategy to include iminium ions generated under either photoredox conditions or Brønsted acid catalysis, as well as bis-sulfone Michael acceptors, and pyridinium salts (Scheme 1e) [76–79]. Recently, Waser also reported an elegant example of this turnover strategy for the enantioselective α-chlorination of pentafluorophenyl esters [80], while Zheng and co-workers reported a related approach using diphenyl methanol as an external turnover reagent for the fluorination of carboxylic acids [81,82]. A significant challenge within this area is the identification of electrophilic reaction partners that react with the catalytically-generated C(1)-ammonium enolate, but are compatible with the nucleophilic tertiary amine catalyst and aryloxide, which is essential for catalyst turnover. Building upon this conceptual platform, it was envisaged that *para*-QMs may be suitable electrophiles to apply in formal 1,6-conjugate additions. within catalytic processes using C(1)-ammonium enolate intermediates. In 2014, we applied this concept for the isothiourea-catalyzed[2,3]-rearrangement of allylic ammonium ylides (Scheme 1b) [60–64]. More recently, this approach has been used by Snaddon (Scheme 1c) [65–72], Hartwig (Scheme 1d) [73] and Gong [74,75] for co-operative isothiourea/transition metal-catalyzed α-functionalization of pentafluorophenyl esters. In both cases, an isothiourea-derived C(1)-ammonium enolate is intercepted by an electrophilic transition metal complex to affect an allylation or benzylation reaction.We have expanded the scope of electrophiles applicable within this catalyst turnover strategy to include iminium ions generated under either photoredox conditions or Brønsted acid catalysis, as well as bis-sulfone Michael acceptors, and pyridinium salts (Scheme 1e) [76–79]. Recently, Waser also reported an elegant example of this turnover strategy for the enantioselective α-chlorination of pentafluorophenyl esters [80], while Zheng and co-workers reported a related approach using diphenyl methanol as an external turnover reagent for the fluorination of carboxylic acids [81,82]. A significant challenge within this area is the identification of electrophilic reaction partners that react with the catalytically-generated C(1)-ammonium enolate, but are compatible with the nucleophilic tertiary amine catalyst and aryloxide, which is essential for catalyst turnover. Building upon this conceptual platform, it was envisaged that para-QMs may be suitable electrophiles to apply in formal 1,6-conjugate additions.

nucleophilic aryloxide is liberated, which may be exploited again in the catalytic cycle to facilitate ammonium enolate formation and catalyst turnover (Scheme 1a) [55,59]. This strategy offers a potentially general solution to allow the expansion of electrophile scope

a) Isothiouronium enolate generation and turnover strategy

Molecules 2021, 26, x FOR PEER REVIEW 2 of 13

Scheme 1. Isothiourea-catalyzed enantioselective processes using aryloxide-facilitated catalyst **Scheme 1.** Isothiourea-catalyzed enantioselective processes using aryloxide-facilitated catalyst turnover.

turnover.

### **2. Results**

### *2.1. Reaction Optimization*

Initial studies focused on the isothiourea-catalyzed 1,6-conjugate addition of *para*nitrophenyl (PNP) ester **1** to 2,6-di-*tert*-butyl *para*-QM **5** (Table 1). Benzylamine was added at the end of the reaction to convert the PNP ester product to the corresponding amide. Based on our previous experience, the amide was expected to be more stable to chromatographic purification. Using tetramisole HCl **6** (20 mol%) as the catalyst, *i*-Pr2NEt as the base and MeCN as the solvent gave a 55:45 ratio of diastereoisomeric amide products **7** and **8** in high yield (82%) and with excellent enantioselectivity (**7**: 97:3 er; **8**: 94:6 er) (Entry 1). Chromatographic separation of the diastereoisomers was not possible, however, the enantioenrichment of both **7** and **8** could be reliably determined by chiral stationary phase (CSP)-HPLC analysis of the mixture. A control reaction in the absence of the catalyst showed no conversion (Entry 2). The use of six alternative solvents was investigated (PhMe, CH2Cl2, CHCl3, THF, 1,4-dioxane and DMF), (see the Supporting Information for details) however, only the use of DMF provided any conversion to the product, indicating that solvent polarity may be significant for the success of this transformation. A control reaction in DMF in the absence of the catalyst, however, also led to comparable conversion to the product, consistent with the operation of a competitive Brønsted base-promoted reaction (see the Supporting Information for details). Taking MeCN as the optimal solvent, the use of eight different organic and inorganic bases was investigated (see the Supporting Information for details). Of those tested, Et3N provided an improved yield of 98%, whilst maintaining comparable diastereo- and enantioselectivity (Entry 3). The use of alternative aryl esters was next probed, with pentafluorophenyl ester **2** and bis(trifluoromethyl)phenyl ester **3** giving amide products **7** and **8** in high yield, but with lower enantioselectivity than when using PNP ester **1** (Entries 4 and 5). The use of 2,4,6-trichlorophenyl ester **4** resulted in only 31% yield (Entry 6), which is consistent with previous studies in this field [65,72,76,79], and most likely reflects the increased steric hindrance of the aryloxide attenuating its nucleophilicity. Finally, using PNP ester **1**, the catalyst loading could be reduced to 5 mol% with only a small drop in stereoselectivity (Entry 7), while heating the reaction to 40 ◦C provided a slight improvement in yield (Entry 8). The reaction could also be performed in the absence of a base (Entry 9), however, slightly lower yield was obtained and therefore, during investigation of the substrate scope, Et3N was routinely used as an auxiliary base.

### *2.2. Reaction Scope and Limitations*

Due to the low diastereoselectivity observed using 2,6-di-*tert*-butyl *para*-QM **5**, the alternative use of 2,6-disubstituted *para*-QMs were investigated (Scheme 2). 2,6-Dimethyl, dibromo and diphenyl *para*-QMs **9**–**11** bearing a phenyl substituent at the exocyclic olefin were applied under optimized conditions. In all cases significantly lower conversion was observed (≤43%), and the amide products **14**–**16** were difficult to isolate. Based on analysis of the crude reaction mixture by <sup>1</sup>H NMR spectroscopy, the 2,6-dimethyl and dibromo-substituted analogues **14** and **15** were obtained with marginally improved diastereoselectivity (~70:30 dr), indicating that alternative substituents in these positions could prove beneficial if the products were isolable. Next, variation of the exocyclic substituent was probed. Incorporation of a methyl group at this position provided no improvement in dr, but the amide product **17** was isolated in a 50% yield and with moderate enantioenrichment for both diastereoisomers. Finally, incorporation of a 2-naphthyl substituent at this position was well tolerated, with amide **18** obtained in an 83% yield, 70:30 dr and excellent enantiocontrol (94:6 er) for the major diastereoisomer.


**Table 1.** Reaction optimization.

used as an auxiliary base.

Table 1. Reaction optimization.

Molecules 2021, 26, x FOR PEER REVIEW 3 of 13

Initial studies focused on the isothiourea-catalyzed 1,6-conjugate addition of paranitrophenyl (PNP) ester 1 to 2,6-di-tert-butyl para-QM 5 (Table 1). Benzylamine was added at the end of the reaction to convert the PNP ester product to the corresponding amide. Based on our previous experience, the amide was expected to be more stable to chromatographic purification. Using tetramisole HCl 6 (20 mol%) as the catalyst, i-Pr2NEt as the base and MeCN as the solvent gave a 55:45 ratio of diastereoisomeric amide products 7 and 8 in high yield (82%) and with excellent enantioselectivity (7: 97:3 er; 8: 94:6 er) (Entry 1). Chromatographic separation of the diastereoisomers was not possible, however, the enantioenrichment of both 7 and 8 could be reliably determined by chiral stationary phase (CSP)-HPLC analysis of the mixture. A control reaction in the absence of the catalyst showed no conversion (Entry 2). The use of six alternative solvents was investigated (PhMe, CH2Cl2, CHCl3, THF, 1,4-dioxane and DMF), (see the Supporting Information for details) however, only the use of DMF provided any conversion to the product, indicating that solvent polarity may be significant for the success of this transformation. A control reaction in DMF in the absence of the catalyst, however, also led to comparable conversion to the product, consistent with the operation of a competitive Brønsted base-promoted reaction (see the Supporting Information for details). Taking MeCN as the optimal solvent, the use of eight different organic and inorganic bases was investigated (see the Supporting Information for details). Of those tested, Et3Nprovided an improved yield of 98%, whilst maintaining comparable diastereo- and enantioselectivity (Entry 3). The use of alternative aryl esters was next probed, with pentafluorophenyl ester 2 and bis(trifluoromethyl)phenyl ester 3 giving amide products 7 and 8 in high yield, but with lower enantioselectivity than when using PNP ester 1 (Entries 4 and 5). The use of 2,4,6-trichlorophenyl ester 4 resulted in only 31% yield (Entry 6), which is consistent with previous studies in this field [65,72,76,79], and most likely reflects the increased steric hindrance of the aryloxide attenuating its nucleophilicity. Finally, using PNP ester 1, the catalyst loading could be reduced to 5 mol% with only a small drop in stereoselectivity (Entry 7), while heating the reaction to 40 °C provided a slight improvement in yield (Entry 8). The reaction could also be performed in the absence of a base (Entry 9), however, slightly lower yield was obtained and therefore, during investigation of the substrate scope, Et3N was routinely

2. Results

2.1. Reaction Optimization


All reactions were carried out on a 0.25 mmol scale; isolated yields are a mixture of diastereoisomers **7** and **8**; dr was determined by <sup>1</sup>H NMR spectroscopic analysis of the crude reaction mixture; er was determined by CSP-HPLC analysis: **7** (2*S*,10*R*:2*R*,10*S*) and **8** (2*S*,10*S*:2*R*,10*R*). [a] Reaction performed at 40 ◦C.

Although structural variation of the *para*-QM provided marginal improvements in dr, the use of 2,6-di-*tert*-butyl *para*-QM **5** was considered most convenient for further investigations due to its stability and ease of synthesis, and the higher yields of product obtained from catalysis. To investigate if the dr obtained in these reactions was a manifestation of a kinetic or thermodynamic preference, isolation of diastereoisomeric PNP esters **19** and **20** and resubjection to catalysis conditions was attempted. Although the diastereoisomeric amides **7** and **8** had proved difficult to separate, the corresponding PNP esters **19** and **20** were chromatographically separable and displayed high stability. Epimerization studies were conducted using each diastereoisomer through sequential treatment with *i*-Pr2NEt, (*S*)-TM HCl **6**, *para*-nitrophenoxide and benzylamine (Scheme 3). These experiments were followed by in situ <sup>1</sup>H NMR spectroscopic analysis and revealed no epimerization in either case. This indicates the dr obtained in the catalytic reaction most likely reflects the inherent diastereoselectivity of the transformation. Following separation of the diastereoisomers, the absolute configuration of the major diastereoisomer could also be confirmed as (2*S*,10*R*) by single crystal X-ray crystallographic analysis [83,84]. Based on literature precedent, the (*S*)-configuration at C(2) was expected to be generated under catalyst-control, and therefore the absolute configuration of the minor diastereoisomer was predicted to be (2*S*,10*S*).

Scheme 2. Scope: Variation of para-quinone methide—0.25 mmol scale; only the structure of the major diastereoisomer is shown; isolated yield is a mixture of diastereoisomers; dr determined by 1H NMR spectroscopic analysis of the crude reaction product mixture; er determined by CSP-HPLC analysis. [a] Product not isolable: conversion based on 1H NMR analysis of the crude reaction product mixture; ers could not be determined. **Scheme 2.** Scope: Variation of *para*-quinone methide—0.25 mmol scale; only the structure of the major diastereoisomer is shown; isolated yield is a mixture of diastereoisomers; dr determined by <sup>1</sup>H NMR spectroscopic analysis of the crude reaction product mixture; er determined by CSP-HPLC analysis. [a] Product not isolable: conversion based on <sup>1</sup>H NMR analysis of the crude reaction product mixture; ers could not be determined. Scheme 2. Scope: Variation of para-quinone methide—0.25 mmol scale; only the structure of the major diastereoisomer is shown; isolated yield is a mixture of diastereoisomers; dr determined by 1H NMR spectroscopic analysis of the crude reaction product mixture; er determined by CSP-HPLC analysis. [a] Product not isolable: conversion based on 1H NMR analysis of the crude reaction product mixture; ers could not be determined.

Scheme 3. Control studies and confirmation of absolute configuration of products. The majority of hydrogen atoms are omitted for clarity within X-ray crystal structure representation of 19. Scheme 3. Control studies and confirmation of absolute configuration of products. The majority of hydrogen atoms are omitted for clarity within X-ray crystal structure representation of 19. **Scheme 3.** Control studies and confirmation of absolute configuration of products. The majority of hydrogen atoms are omitted for clarity within X-ray crystal structure representation of **19**.

Having established the stability of the PNP ester products and demonstrated the potential to separate the diastereoisomers by column chromatography, the scope of the catalytic transformation was investigated through variation of the PNP ester substrate. In each case, the PNP ester product diastereoisomers were at least partially separable, enabling unambiguous characterization. To test the applicability of the procedure, *p*-tolyl-substituted PNP ester product **19** was prepared on a larger scale (1.25 mmol) (Scheme 4). A combined 85% yield of both diastereoisomers was obtained, with comparable stereoselectivity to that observed when the reaction was conducted on an analytical scale (Table 1, Entry 7). The generality of the procedure was further probed using five electronically- and stericallydifferentiated PNP esters. Introduction of an electron-donating 4-methoxy substituent was well tolerated, with PNP ester **26** obtained in quantitative yield, 60:40 dr and with high enantioselectivity for both diastereoisomers. Under the optimized conditions, the introduction of an electron-withdrawing 4-trifluoromethyl group resulted in low enantioselectivity (**27**: 63:37 er), which was attributed to a competitive Brønsted base-promoted background reaction. Consistent with this hypothesis, repeating the reaction in the absence of Et3N, and using the free base of the isothiourea catalyst **6**, provided PNP ester product **27** in 94% yield and significantly improved enantioselectivity (88:12 ermaj; 79:21 ermin). A similar effect was observed when using a 2-naphthyl-substituted PNP ester, with optimal enantioselectivity obtained in the absence of an auxiliary base (**28**: 91:9 ermaj; 85:15 ermin). Introduction of a sterically-imposing 1-naphthyl or a heteroaromatic thienyl substituent was also tolerated, with **29** and **30** obtained in excellent yield and with high enantioselectivity. potential to separate the diastereoisomers by column chromatography, the scope of the catalytic transformation was investigated through variation of the PNP ester substrate. In each case, the PNP ester product diastereoisomers were at least partially separable, enabling unambiguous characterization. To test the applicability of the procedure, p-tolylsubstituted PNP ester product 19 was prepared on a larger scale (1.25 mmol) (Scheme 4). A combined 85% yield of both diastereoisomers was obtained, with comparable stereoselectivity to that observed when the reaction was conducted on an analytical scale (Table 1, Entry 7). The generality of the procedure was further probed using five electronicallyand sterically-differentiated PNP esters. Introduction of an electron-donating 4-methoxy substituent was well tolerated, with PNP ester 26 obtained in quantitative yield, 60:40 dr and with high enantioselectivity for both diastereoisomers. Under the optimized conditions, the introduction of an electron-withdrawing 4-trifluoromethyl group resulted in low enantioselectivity (27: 63:37 er), which was attributed to a competitive Brønsted basepromoted background reaction. Consistent with this hypothesis, repeating the reaction in the absence of Et3N, and using the free base of the isothiourea catalyst 6, provided PNP ester product 27 in 94% yield and significantly improved enantioselectivity (88:12 ermaj; 79:21 ermin). A similar effect was observed when using a 2-naphthyl-substituted PNP ester, with optimal enantioselectivity obtained in the absence of an auxiliary base (28: 91:9 ermaj; 85:15 ermin). Introduction of a sterically-imposing 1-naphthyl or a heteroaromatic thienyl substituent was also tolerated, with 29 and 30 obtained in excellent yield and with high enantioselectivity.

Having established the stability of the PNP ester products and demonstrated the

Molecules 2021, 26, x FOR PEER REVIEW 6 of 13

Scheme 4. Scope: Variation of para-nitrophenyl ester—0.25 mmol scale; only the structure of the major diastereoisomer is shown; isolated yields are given for the combination of diastereoisomers; dr determined by 1H NMR spectroscopic analysis of the crude reaction product mixture; er determined by CSP-HPLC analysis. [a] 1.25 mmol scale. [b] Conducted at r.t., in the absence of Et3N, and using free base of 6. **Scheme 4.** Scope: Variation of *para*-nitrophenyl ester—0.25 mmol scale; only the structure of the major diastereoisomer is shown; isolated yields are given for the combination of diastereoisomers; dr determined by <sup>1</sup>H NMR spectroscopic analysis of the crude reaction product mixture; er determined by CSP-HPLC analysis. [a] 1.25 mmol scale. [b] Conducted at r.t., in the absence of Et3N, and using free base of **6**.

#### *2.3. Proposed Mechanism* 2.3. Proposed Mechanism

The mechanism of this transformation is proposed to begin with *N*-acylation of the free base isothiourea catalyst **6** by PNP ester **31** to generate the corresponding acyl ammonium *para*-nitrophenoxide ion pair **32** (Scheme 5). Subsequent deprotonation leads to (*Z*)-ammonium enolate **33**. Based on previous mechanistic studies [78,85], and the catalytic activity observed in the absence of an auxiliary base (Table 1, Entry 9), deprotonation is likely to be affected by the *para*-nitrophenoxide counterion. 1,6-Conjugate addition of ammonium enolate **33** to *para*-QM electrophile **34**, followed by protonation, gives acyl ammonium intermediate **35**. Finally, regeneration of catalyst **6**, and concurrent release of product **37**, is proposed to be facilitated by *para*-nitrophenoxide [59–79,86,87]. Although not essential for reactivity, the addition of Et3N as an auxiliary base may be beneficial as a proton shuttle, and to maintain the isothiourea catalyst in its non-protonated form **6** [77,86]. The enantioselectivity of the transformation indicates the C–C bond forming event takes place on the *Si*-face of the ammonium enolate. This selectivity can be rationalized through preferential formation of the (*Z*)-ammonium enolate [76–79,85], which is conformationally-restricted by an intramolecular 1,5-O· · · S interaction [61,88–104] and results in the phenyl stereodirecting group of the catalyst blocking the enolate *Re*-face. The observed poor diastereoselectivity can be tentatively rationalized by a simple stereochemical model that assumes a favored, open pre-transition state assembly where steric interactions are minimized about the forming C–C bond. Minimal differentiation between the aryl- and quinone substituents of the *para*-QM quinone leads to the two transition state assemblies **38** and **39** that give the major and minor diastereoisomers, respectively. The mechanism of this transformation is proposed to begin with N-acylation of the free base isothiourea catalyst 6 by PNP ester 31 to generate the corresponding acyl ammonium para-nitrophenoxide ion pair 32 (Scheme 5). Subsequent deprotonation leads to (Z) ammonium enolate 33. Based on previous mechanistic studies [78,85], and the catalytic activity observed in the absence of an auxiliary base (Table 1, Entry 9), deprotonation is likely to be affected by the para-nitrophenoxide counterion. 1,6-Conjugate addition of ammonium enolate 33 to para-QM electrophile 34, followed by protonation, gives acyl ammonium intermediate 35. Finally, regeneration of catalyst 6, and concurrent release of product 37, is proposed to be facilitated by para-nitrophenoxide [59–79,86,87]. Although not essential for reactivity, the addition of Et3N as an auxiliary base may be beneficial as a proton shuttle, and to maintain the isothiourea catalyst in its non-protonated form 6 [77,86]. The enantioselectivity of the transformation indicates the C–C bond forming event takes place on the Si-face of the ammonium enolate. This selectivity can be rationalized through preferential formation of the (Z)-ammonium enolate [76–79,85], which is conformationally-restricted by an intramolecular 1,5-O•••S interaction [61,88–104] and results in the phenyl stereodirecting group of the catalyst blocking the enolate Re-face. The observed poor diastereoselectivity can be tentatively rationalized by a simple stereochemical model that assumes a favored, open pre-transition state assembly where steric interactions are minimized about the forming C–C bond. Minimal differentiation between the aryland quinone substituents of the para-QM quinone leads to the two transition state assemblies 38 and 39 that give the major and minor diastereoisomers, respectively.

Scheme 5. Proposed mechanism (only the pathway for the formation of the major diastereoisomer is shown) and transition state assemblies. ArOH corresponds to either PNPOH or the reaction **Scheme 5.** Proposed mechanism (only the pathway for the formation of the major diastereoisomer is shown) and transition state assemblies. ArOH corresponds to either PNPOH or the reaction product.

product.

### **3. Materials and Methods**

### *3.1. General Procedure for the Enantioselective 1,6-Addition*

In a flame-dried vial, the requisite *para*-quinone methide (1.0 equiv.), aryl ester (1.5 equiv.), (*S*)-TM HCl (5 mol%), Et3N (1.0 equiv.) and anhydrous MeCN (0.6 M) was added and stirred at r.t. for 24 h. The reaction was then quenched with benzylamine (5.0 equiv.) and stirred at r.t. for a further 12 h before being concentrated in vacuo. The residue was diluted with EtOAc (20 mL) and washed successively with 10% citric acid (20 mL × 1), aqueous NaOH (20 mL × 3) and brine (20 mL × 1). The organic layer was extracted, dried over MgSO<sup>4</sup> and the filtrate was concentrated in vacuo. The crude material was purified by flash silica column chromatography to give the desired product.

### *3.2. Representative Synthesis and Characterization of Compounds* **7** *and* **8** *(Entry 8)*

Following the general procedure above, 4-benzylidene-2,6-di-*tert*-butylcyclohexa-2,5 dien-1-one **5** (74 mg, 0.25 mmol), 4'-nitrophenyl 2-(*p*-tolyl)acetate **1** (102 mg, 0.375 mmol), (*S*)-TM HCl **6** (3 mg, 5 mol%) and Et3N (35 µL, 0.25 mmol) were dissolved in anhydrous MeCN (0.42 mL). The reaction mixture was stirred at 40 ◦C for 24 h before being quenched with benzylamine (137 µL, 1.25 mmol) at r.t. to give a crude mixture containing the title compound in 60:40 dr. The mixture was purified by flash silica column chromatography (petroleum ether/EtOAc, 85:15) to afford diastereoisomers **7** and **8** (60:40 dr) (132 mg, 99%) as a pale yellow solid.

mp 126–128 ◦C; [*α*] 20 *D* +10.0 (*<sup>c</sup>* 1.0, CHCl3); IR <sup>ν</sup>max (film)/cm−<sup>1</sup> 3638 (O−H) 3304 (N−H), 2955 (C−H), 1645 (C=O); HRMS (ESI<sup>+</sup> ) C37H43NO<sup>2</sup> ([M + H]<sup>+</sup> ), found 534.3359, requires 534.3367 (−1.4 ppm).

*Data for major diastereoisomer* (7): Chiral HPLC analysis, Chiralpak AD-H (10% *i*-PrOH/hexane, flow rate 1.5 mLmin−<sup>1</sup> , 211 nm, 40 ◦C), t<sup>R</sup> 8.5 min and 29.3 min, 92:8 er; <sup>1</sup>H NMR (500 MHz, CDCl3) δH: 1.42 (18H, s, (C(30000)C(C*H*3)3, C(50000)C(C*H*3)3), 2.24 (3H, s, C(400)C*H*3), 3.90–4.05 (2H, m, C(2)*H*, C*HA*Ph), 4.44 (1H, dd, *J* 15.0, 6.8, C*HB*Ph), 4.82 (1H, d, *J* 11.7, C(1')*H*), 5.14 (1H, s, O*H*), 5.55 (1H, t, *J* 5.6, N*H*), 6.71–6.76 (2H, m, *Ar*), 6.96–7.05 (2H, m, *Ar*), 7.09–7.15 (2H, m, *Ar*), 7.15–7.22 (4H, m, *Ar*), 7.23–7.30 (4H, m, C(4000)*H*, C(20000)*H*, C(60000)*H, Ar*), 7.34 (1H, t, *J* 7.6, *Ar*), 7.43–7.51 (1H, m, *Ar*); <sup>13</sup>C{1H} NMR (126 MHz, CDCl3) δC: 21.0 (C(400)*C*H3), 30.4 (C(30000)C(*C*H3)3, C(50000)C(*C*H3)3), 34.4 (C(30000)*C*(CH3)3, C(50000)*C*(CH3)3), 43.6 (*C*H2Ph), 54.1 (*C*(1')), 59.5 (*C*(2)), 124.7 (*C*(20000), *C*(60000)), 125.8 (*C*(4000)), 128.2 (*Ar*), 127.2 (*Ar*), 127.3 (*Ar*), 128.0 (*Ar*), 128.4 (*Ar*), 128.5 (*Ar*), 128.6 (*Ar*), 129.0 (*Ar*), 133.8 (*C*(1000)), 135.2 (*C*(100)), 135.6 (*C*(30000), *C*(50000)), 136.4 (*C*(400)), 137.9 (*i*-*Ph*), 142.4 (*C*(10000)), 152.4 (*C*(40000)), 172.1 (*C*(1)).

*Selected data for minor diastereoisomer* (8): Chiral HPLC analysis, Chiralpak AD-H (10% *i*-PrOH/hexane, flow rate 1.5 mLmin−<sup>1</sup> , 211 nm, 40 ◦C), t<sup>R</sup> 3.8 min and 18.6 min, 85:15 er; <sup>1</sup>H NMR (500 MHz, CDCl3) δH: 1.27 (18H, s, (C(30000)C(C*H*3)3, C(50000)C(C*H*3)3), 2.27 (3H, s, C(400)C*H*3), 3.90–4.05 (2H, m, C(2)*H*, C*HA*Ph), 4.50 (1H, dd, *J* 15.0, 6.8, C*HB*Ph), 4.69 (1H, d, *J* 11.7, C(10 )*H*), 4.89 (1H, s, O*H*), 5.69 (1H, t, *J* 5.6, N*H*), 6.81–6.84 (2H, m, *Ar*); <sup>13</sup>C{1H} NMR (126 MHz, CDCl3) δC: 21.0 (C(4")*C*H3), 30.2 (C(30000)C(*C*H3)3, C(50000)C(*C*H3)3), 34.2 (C(30000)*C*(CH3)3, C(50000)*C*(CH3)3), 43.4 (*C*H2Ph), 54.7 (*C*(1<sup>0</sup> )), 59.5 (*C*(2)), 125.3 (*C*(20000), *C*(60000)), 126.3 (*C*(4000)), 127.5 (*Ar*), 128.2 (*Ar*), 128.4 (*Ar*), 128.8 (*Ar*), 132.1 (*C*(1000)), 134.9 (*C*(30000), *C*(50000)), 135.5 (*C*(100)), 136.4 (*C*(400)), 138.2 (*i*-*Ph*), 143.6 (*C*(10000), 151.7 (*C*(40000)), 172.0 (*C*(1)).

### **4. Conclusions**

An isothiourea-catalyzed enantioselective 1,6-conjugate addition of *para*-nitrophenyl (PNP) esters to *para*-quinone methides (QMs) has been developed. Variation of the arylacetic ester and *para*-QM substrates has provided a range of functionalized products in generally excellent yields and high enantiocontrol (up to 94:6 er). An inherent limitation of the method is that the products were routinely obtained in ~60:40 dr. This diastereoselectivity was shown to arise from kinetic control, but was relatively insensitive to changes in reaction conditions and structural variation of the substrates. Although the dr could

not be improved, the diastereoisomeric PNP ester products could be separated by column chromatography. The success of this catalytic methodology is proposed to rely upon the *para*-nitrophenoxide, expelled during *N*-acylation of the catalyst, to facilitate catalyst turnover and release the product. Current work in our laboratory is focused on further applications of using in situ-generated aryloxides to promote catalyst turnover in Lewis base catalysis.

**Supplementary Materials:** Full experimental procedures, characterization data, NMR spectra and HPLC chromatograms for all new compounds, as well as crystallographic data for product **19** (CCDC 1992504) are available online.

**Author Contributions:** Conceptualization, A.D.S. and J.N.A.; investigation, J.N.A.; W.C.H.; C.M.; X-ray crystallographic analysis, A.M.Z.S.; writing—original draft preparation, M.D.G.; writing review and editing, A.D.S. and all authors; supervision, D.S., S.N. and A.D.S.; funding acquisition, A.D.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** We thank the ERC under the European Union's Seventh Framework Programme (FP7/2007- 2013)/E.R.C. grant agreement n◦ 279850, AstraZeneca and EPSRC (EP/M506631/1 (J.N.A.)), Syngenta and the EPSRC Centre for Doctoral Training in Critical Resource Catalysis (CRITICAT, EP/L016419/1 (W.C.H.)), and EPSRC (EP/M508214/1 (C.M.)) for funding. A.D.S. thanks the Royal Society for a Wolfson Research Merit Award. We thank the EPSRC UK National Mass Spectrometry Facility at Swansea University.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The research data underpinning this publication can be found at DOI: 10.17630/f6cf6c80-483d-4f16-bd79-80e1537513b2.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** Samples of the compounds are available from the authors on request.

### **References and Notes**


## *Article* **Azetidinium Lead Halide Ruddlesden–Popper Phases**

**Jiyu Tian 1,2, Eli Zysman-Colman 2,\* and Finlay D. Morrison 1,\***


**Abstract:** A family of Ruddlesden–Popper (*<sup>n</sup>* = 1) layered perovskite-related phases, Az2PbCl*x*Br4−*<sup>x</sup>* with composition 0 ≤ *x* ≤ 4 were obtained using mechanosynthesis. These compounds are isostructural with K2NiF<sup>4</sup> and therefore adopt the idealised *n* = 1 Ruddlesden–Popper structure. A linear variation in unit cell volume as a function of anion average radius is observed. A tunable bandgap is achieved, ranging from 2.81 to 3.43 eV, and the bandgap varies in a second-order polynomial relationship with the halide composition.

**Keywords:** layered perovskite; bandgap tuning; azetidinium; Ruddlesden–Popper; structure-property relations

### **1. Introduction**

Ruddlesden–Popper (R–P) phases are composed of layered perovskite structures with alternating layers of AMX<sup>3</sup> perovskite and AX rock salt along the *c*-axis. They are described by the general formula A*n*+1M*n*X3*n*+1 (or A'2A"*n*−1M*n*X3*n*+1 in the case of two distinct A-cations), where *n* is a positive integer representing the number of perovskite layers that are separated by additional 'A-cation excess' rock-salt layers [1,2]. Importantly, the intergrowth rock salt layer means that the octahedra in the perovskite layers are aligned in the successive layers. In 1955, Balz and Plieth reported the first R–P phase layered structure K2NiF<sup>4</sup> (*n* = 1) [3]. In 1957, Ruddlesden and Popper reported a series of layered structures in oxides, such as Sr2TiO<sup>4</sup> and Ca2TiO<sup>4</sup> [4]. Nowadays, the R–P phase is more commonly used to represent this type of layered perovskite structure and, increasingly, in organic–inorganic hybrid perovskites (OIHPs). Several families of layered OIHPs containing alternating layers of AMX<sup>3</sup> perovskite and organic cations with structures similar to R–P phases have been reported. Such examples of layered OIHPs include BA2PbI<sup>4</sup> (BA = C4H9NH<sup>3</sup> + ) [5] and PEA2PbX<sup>4</sup> (PEA = C8H12N<sup>+</sup> , X = Cl, Br, I), [6,7] in which the organic cations are too big to be accommodated in the cuboctahedral cavities of the 3D MX<sup>6</sup> framework. Without the constraint of the size of the cuboctahedral cavities, a wider range of organic A-cations would be available for layered phases. In addition, by mixing large (A') organic cations, such as those mentioned above, and small organic cations such as methylammonium (A" = MA), organic-inorganic hybrid materials with the general formula A'2A"*n*−1M*n*X3*n*+1 can be prepared [5,8]. They show good bandgap tunability by modifying the number of layers (*n*) of A"PbX3. Stoumpos et al. [5] reported orthorhombic crystal structures of BA2MA*n*−1Pb*n*X3*n*+1 (X = Br, I) with bandgaps changing progressively from 2.43 eV (*n* = 1) to 1.50 eV (*n* = ∞), with intermediate values of 2.17 eV (*n* = 2), 2.03 eV (*n* = 3) and 1.91 eV (*n* = 4). The thickness of the perovskite layer, *n*, in (BA)2(MA)*n*−1Pb*n*I3*n*+1 can be reasonably controlled by modifying the ratio of BA/MA cations in the precursor solutions. However, many so-called R–P phases reported in such compounds often do not have the required rock salt-structured interlayer between the 2D perovskite layers, resulting in an offset in the alignment of the perovskite blocks in successive layers. Such examples, therefore, do not conform to the definition of an R–P phase and are more correctly termed R–P-like

**Citation:** Tian, J.; Zysman-Colman, E.; Morrison, F.D. Azetidinium Lead Halide Ruddlesden–Popper Phases. *Molecules* **2021**, *26*, 6474. https:// doi.org/10.3390/molecules26216474

Academic Editors: William T. A. Harrison, R. Alan Aitken and Paul Waddell

Received: 29 September 2021 Accepted: 21 October 2021 Published: 27 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

OIHPs. Such R–P-like layered OIHPs have demonstrated higher stability when exposed to light, humidity and heat stress compared to 3D perovskite analogues, which are prone to unwanted phase transition under these test conditions [9,10]. For example, Ren et al. reported an R–P-like OIHPs solar cell material with general formula (MTEA)2(MA)4Pb5I<sup>16</sup> (*n* = 5) which achieved a power conversion efficiency up to 17.8% [11]. Their cells retained over 85% of the initial efficiency after 1000 h operation time.

Azetidinium (Az<sup>+</sup> , (CH2)3NH<sup>3</sup> + ) is a four-membered ring ammonium cation. In our previous study on mixed halide azetidinium lead perovskites, AzPbBr3−*x*X*<sup>x</sup>* (X = Cl or I), the structure progresses from 6H to 4H to 9R perovskite polytypes with varying halide composition from Cl<sup>−</sup> to Br<sup>−</sup> to I<sup>−</sup> [12]. The fact that AzPbX<sup>3</sup> (X = Cl or Br) forms a hexagonal perovskite rather than a cubic (3C) perovskite led to our study on mix-cation solid solutions of the form AzA"PbBr3, A" = MA<sup>+</sup> or FA<sup>+</sup> (FA<sup>+</sup> = formamidinium). Such systems show only partial solid solutions and phase separation of the hexagonal and cubic forms; the extent of solid solution formation also depends on the synthesis route [13]. These studies also suggest that the cation radius of Az<sup>+</sup> is ~310 pm, which is larger than the calculated cation radius of Az, *r*Az = 250 pm (for comparison the reported radii for FA<sup>+</sup> and MA<sup>+</sup> are *r*FA = 253 pm, *r*MA = 217 pm [14], respectively). MA<sup>+</sup> and FA<sup>+</sup> are commonly used as A-site cations in OIHPs, and that adopt (pseudo-) cubic perovskite structures [15,16]. With our cation radius estimation that Az<sup>+</sup> is larger than MA<sup>+</sup> and FA<sup>+</sup> , Az2PbX<sup>4</sup> (X = Cl, Br) are found to adopt a *n* = 1 R–P phase structure. The fact that Az<sup>+</sup> can form a layered structure indicates that our estimation of its cation radius is more accurate than that from the computational calculation [13,14]. Furthermore, a family of mixed halide R–P phases, Az2PbCl*x*Br4−*<sup>x</sup>* with composition 0 ≤ *x* ≤ 4 were prepared by mechanosynthesis and their structures and optical properties were analysed by powder X-ray diffraction (PXRD) and absorption spectroscopy, respectively. A linear variation in unit cell volume as a function of anion average radius is observed. The band gap was found to range from 2.81 to 3.43 eV, which varies as a second-order polynomial relationship with the halide composition.

### **2. Method**

PbBr<sup>2</sup> (98%) and PbCl<sup>2</sup> (98%) were purchased from Alfa Aesar. Hydrobromic acid in water (48%) and AzCl (95%) were purchased from Fluorochem. All other reagents and solvents were obtained from commercial sources and used as received. AzBr were synthesised according to our previous study [17].

Preparation of Az2PbCl*x*Br4−*<sup>x</sup>* solid solutions with 0 ≤ *x* ≤ 4 (in *x* = 0.67 increments) was carried out by mechanosynthesis. Appropriate molar ratios of dry AzX and PbX<sup>2</sup> (AzX:PbX<sup>2</sup> = 2:1, X = Cl or Br) were ground together in a Fritsch Pulverisette planetary ball mill at 600 rpm for 1 h using 60 cm<sup>3</sup> Teflon pots and high-wear-resistant zirconia media (nine 10 mm diameter spheres). Az2PbBr<sup>4</sup> samples could also be obtained by hand grinding AzBr and PbBr<sup>2</sup> in an agate mortar and pestle for 25 min.

PXRD was carried out using a PANalytical Empyrean diffractometer with Cu Kα<sup>1</sup> (λ = 1.5406 Å). Rietveld refinements of PXRD data using GSAS [18] were used to confirm phase formation and for the determination of lattice parameters.

Optical properties were determined from solid-state absorption spectra recorded using a Shimadzu UV-2600 spectrophotometer and bandgaps were calculated by plotting (*αhν*) 2 (cm−<sup>1</sup> ·eV)<sup>2</sup> with *<sup>h</sup>ν*(eV) according to the Tauc method, in which *<sup>α</sup>*, *<sup>h</sup>* and *<sup>ν</sup>* stand for absorbance, Planck's constant and incident light frequency.

### **3. Results**

The PXRD data for Az2PbCl*x*Br4−*<sup>x</sup>* with compositions ranging from 0 ≤ *x* ≤ 4 were prepared by mechanosynthesis and are shown in Figure 1b. The structures of these samples were determined to be R–P *n* = 1 phase in the *I*4/*mmm* space group (Figure 1a). The theoretical diffraction pattern of the tetragonal R–P phase is shown in Figure S1. Characteristic peaks of the R–P phase show systematic peak shifts to higher 2*θ* angle from Az2PbBr<sup>4</sup> to Az2PbCl4, which indicate the lattice parameters decreased with more Cl content in the

solid solution. The Az<sup>+</sup> cations, which are represented as solid spheres situated at the centre of electron density, form rock salt layers with the X− anions. Synthesis from solution is preferred when manufacturing devices because solutions can be easily processed into thin films by spin-coating and blade-coating methods compared to bulk powder [19]. Thus, precipitation synthesis of Az2PbX<sup>4</sup> (X = Cl, Br) were also attempted (synthetic details included in the supporting information) and their PXRD data are shown in Figure S2. Although the precipitated samples contain additional phase(s) associated with additional peaks (e.g., at 6◦ and 11◦ ) and have yet to be assigned to a structure. Ganguli [20] reported an empirical prediction that possible R–P phase structures are associated with a ratio of A-site and metal cation radii (*r*A/*r*M) in the range of 1.7 to 2.4. As discussed in our previous study [12], our estimation of the cation radius of Az<sup>+</sup> (~310 pm) differs from that calculated (250 pm) [14]. The *r*Az/*r*Pb calculated using our estimated radius is 2.60, while that using the literature value [14] is 2.10. solid solution. The Az+ cations, which are represented as solid spheres situated at the centre of electron density, form rock salt layers with the X− anions. Synthesis from solution is preferred when manufacturing devices because solutions can be easily processed into thin films by spin-coating and blade-coating methods compared to bulk powder [19]. Thus, precipitation synthesis of Az2PbX4 (X = Cl, Br) were also attempted (synthetic details included in the supporting information) and their PXRD data are shown in Figure S2. Although the precipitated samples contain additional phase(s) associated with additional peaks (e.g., at 6° and 11°) and have yet to be assigned to a structure. Ganguli [20] reported an empirical prediction that possible R–P phase structures are associated with a ratio of A-site and metal cation radii (*r*A/*r*M)in the range of 1.7 to 2.4. As discussed in our previous study [12], our estimation of the cation radius of Az+ (~310 pm) differs from that calculated (250 pm) [14]. The *r*Az/*r*Pb calculated using our estimated radius is 2.60, while that using the literature value [14] is 2.10.

were determined to be R–P *n* = 1 phase in the *I*4/*mmm* space group (Figure 1a). The theoretical diffraction pattern of the tetragonal R–P phase is shown in Figure S1. Characteristic peaks of the R–P phase show systematic peak shifts to higher 2*θ* angle from Az2PbBr4 to Az2PbCl4, which indicate the lattice parameters decreased with more Cl content in the

*Molecules* **2021**, *26*, x 3 of 8

**Figure 1.** (**a**) *n* = 1 Ruddlesden–Popper (R–P) phase of Az2PbX4 (X = Cl, Br) showing alternating AzPbX3 perovskite and AzX rock salt layers along the *c*-axis, (**b**) PXRD data of mix-halide layered R–P phases: Az2PbCl*x*Br4−*x* with composition 0 ≤ *x* ≤ 4 prepared by mechanosynthesis. **Figure 1.** (**a**) *n* = 1 Ruddlesden–Popper (R–P) phase of Az2PbX<sup>4</sup> (X = Cl, Br) showing alternating AzPbX<sup>3</sup> perovskite and AzX rock salt layers along the *c*-axis, (**b**) PXRD data of mix-halide layered R–P phases: Az2PbCl*x*Br4−*<sup>x</sup>* with composition 0 ≤ *x* ≤ 4 prepared by mechanosynthesis.

Unfortunately, our attempts to synthesise single-phase Az2PbI4 were unsuccessful. The PXRD of mechanosynthesised Az2PbI4 is shown in Figure S3. In addition to the R–P phase, there are evident amounts of 9R AzPbI3 phase [12,21] and the relative intensity of this phase increased with increased ball mill grinding time (1 to 3 h). PXRD of the Az2PbI4 sample obtained from a hand grinding synthesis showed that this method can increase the proportion of R–P phase in the samples, evidenced by the increased relative intensity of peaks associated with the R–P phase, but the presence of the 9R phase persisted across all samples. These results indicate that the 9R phase is the more stable phase compared to the R–P phase for the iodide analogue It is likely that the activation energy for the transformation of azetidinium lead iodide from a layered phase to the 9R phase is low. Unfortunately, our attempts to synthesise single-phase Az2PbI<sup>4</sup> were unsuccessful. The PXRD of mechanosynthesised Az2PbI<sup>4</sup> is shown in Figure S3. In addition to the R–P phase, there are evident amounts of 9R AzPbI<sup>3</sup> phase [12,21] and the relative intensity of this phase increased with increased ball mill grinding time (1 to 3 h). PXRD of the Az2PbI<sup>4</sup> sample obtained from a hand grinding synthesis showed that this method can increase the proportion of R–P phase in the samples, evidenced by the increased relative intensity of peaks associated with the R–P phase, but the presence of the 9R phase persisted across all samples. These results indicate that the 9R phase is the more stable phase compared to the R–P phase for the iodide analogue It is likely that the activation energy for the transformation of azetidinium lead iodide from a layered phase to the 9R phase is low.

For simplicity, Rietveld refinements were carried out by replacing the organic Az+ cations with Mn2+, as they have similar electron densities. Figure 2 shows an example of the PXRD data refinement of Az2PbX4 (X = Cl, Br) samples obtained from the ball mill mechanosynthesis. The refined lattice parameters of Az2PbBr4 are *a* = 5.993(6) Å and *c* = For simplicity, Rietveld refinements were carried out by replacing the organic Az<sup>+</sup> cations with Mn2+, as they have similar electron densities. Figure 2 shows an example of the PXRD data refinement of Az2PbX<sup>4</sup> (X = Cl, Br) samples obtained from the ball mill mechanosynthesis. The refined lattice parameters of Az2PbBr<sup>4</sup> are *a* = 5.993(6) Å and *c* = 21.501(1) Å, with goodness-of-fit parameters *χ* <sup>2</sup> = 10.21 and *wR<sup>p</sup>* = 0.115, while those of Az2PbCl<sup>4</sup> are *a* = 5.765(0) Å and *c* = 21.027(2) Å, with goodness-of-fit parameters *χ* <sup>2</sup> = 7.20 and *wR<sup>p</sup>* = 0.102. The difference between the organic moieties and Mn2+, which is associated with their actual atomic position and thermal motion, is one possible reason

for such high *χ* <sup>2</sup> values for both refinements and may be responsible for the differences in the peak shape and intensities shown. Single crystal diffraction analysis is required for detailed structural analysis, including accurate atoms positions (particularly of the Az<sup>+</sup> cation), however, this would require preparation of sufficiently large single crystals which are challenging by this mechanosynthesis route. Nevertheless, it is clear from the rudimentary Rietveld analysis of the PXRD data that all peaks are accounted for and that the PXRD unambiguously show the formation of *n* = 1 R–P materials. In addition, as the peaks positions can be determined accurately the unit cell dimensions are reliable. high *<sup>χ</sup>*2 values for both refinements and may be responsible for the differences in the peak shape and intensities shown. Single crystal diffraction analysis is required for detailed structural analysis, including accurate atoms positions (particularly of the Az+ cation), however, this would require preparation of sufficiently large single crystals which are challenging by this mechanosynthesis route. Nevertheless, it is clear from the rudimentary Rietveld analysis of the PXRD data that all peaks are accounted for and that the PXRD unambiguously show the formation of *n* = 1 R–P materials. In addition, as the peaks positions can be determined accurately the unit cell dimensions are reliable.

21.501(1) Å, with goodness-of-fit parameters *χ*2 = 10.21 and *wRp* = 0.115, while those of Az2PbCl4 are *a* = 5.765(0) Å and *c* = 21.027(2) Å, with goodness-of-fit parameters *χ*2 = 7.20 and *wRp* = 0.102. The difference between the organic moieties and Mn2+, which is associated with their actual atomic position and thermal motion, is one possible reason for such

*Molecules* **2021**, *26*, x 4 of 8

**Figure 2.** Rietveld refinement of PXRD data in *I*4/*mmm* space group of Az2PbX4, X = Br (top) and Cl (bottom) obtained from mechanosynthesis with observed data (open circles), calculated data (red line for Br and magenta line for Cl), background (green lines), reflection positions (black bars) and difference plots (blue lines). **Figure 2.** Rietveld refinement of PXRD data in *I*4/*mmm* space group of Az2PbX<sup>4</sup> , X = Br (top) and Cl (bottom) obtained from mechanosynthesis with observed data (open circles), calculated data (redline for Br and magenta line for Cl), background (green lines), reflection positions (black bars) and difference plots (blue lines).

To study the mixed-halide solid solutions Az2PbCl*x*Br4−*x*, the lattice parameters of each mechanosynthesised composition were determined by Rietveld refinement of PXRD data. The cell volume of these R–P phases varies linearly as a function of the average anion radius, Figure 3a (the average anion radius was calculated using *r*Br = 196 pm and *r*Cl = 181 pm according to Shannon [22]). This linear variation is expected in accordance with Vegard's law. The lattice parameters *a* and *c*, on the other hand, show a nonlinear relationship with the average anion radius (Figure 3b), which suggests anisotropic expansion/contraction along the *a*- and *c*-axis. The larger expansion in *a* is consistent with the increased X anion radius which affords a larger void for the Az+ cation, resulting in less required To study the mixed-halide solid solutions Az2PbCl*x*Br4−*<sup>x</sup>*, the lattice parameters ofeach mechanosynthesised composition were determined by Rietveld refinement of PXRDdata. The cell volume of these R–P phases varies linearly as a function of the averageanion radius, Figure 3a (the average anion radius was calculated using *<sup>r</sup>*Br = 196 pm and*r*Cl = 181 pm according to Shannon [22]). This linear variation is expected in accordance with Vegard's law. The lattice parameters *a* and *c*, on the other hand, show a nonlinear relationship with the average anion radius (Figure 3b), which suggests anisotropic expansion/contraction along the *a*- and *c*-axis. The larger expansion in *a* is consistent with the increased X anion radius which affords a larger void for the Az<sup>+</sup> cation, resulting in less required expansion in the interlayer spacing. Based on the analysis using Mn2+ as a proxy for Az<sup>+</sup> we have no information regarding any orientation or dynamics of the Az<sup>+</sup> cation.

to the molar ratios of the raw materials (nominal composition).

**Figure 3.** (**a**) Cell volume, (**b**) lattice parameters as a function of average halide anion radius for *n* = 1 R–P phases Az2PbCl*x*Br4−*x* (0 ≤ *x* ≤ 4) as determined from Rietveld refinement of PXRD data. **Figure 3.** (**a**) Cell volume, (**b**) lattice parameters as a function of average halide anion radius for *n* = 1 R–P phases Az2PbCl*x*Br4−*<sup>x</sup>* (0 ≤ *x* ≤ 4) as determined from Rietveld refinement of PXRD data.

The optical properties of Az2PbCl*x*Br4−*x* (0 ≤ *x* ≤ 4) solid solutions were studied by absorption spectroscopy (Figure 4a). The absorption onsets are systematically red-shifted from ca. 386 nm (Az2PbCl4) to ca. 457 nm (Az2PbBr4) with increasing average anion size (from Cl− to Br−). The bandgaps of Az2PbCl4 and Az2PbBr4 are calculated to be 3.43 and 2.81 eV, which are the same (within error) as the bandgap of the 6H hexagonal perovskite AzPbCl3 (3.43 eV) and AzPbBr3 (2.81 eV) [12]. However, unlike the linear variation in the 6H AzPbX3 (X− = Cl−, Br−), the bandgap of layered R–P Az2PbX4 (X = Cl, Br) shows a bowing with the average anion radius (Figure 4b). The bowing effect [25,26] simply describes the One of the benefits of mechanosynthesis is that all materials are retained during the reaction, so the overall starting composition must be retained in the post-reaction compound(s). By inference, any product(s) must have the nominal starting composition. While we do not have direct compositional analysis, the PXRD results, Figure 2, clearly show that the product formed is entirely *n* = 1 R–P phase. It has been reported that the actual composition shows a good match with the nominal composition in the mechanosynthesis of OIHPs [23,24]. Thus, the halide compositions of Az2PbCl*x*Br4−*<sup>x</sup>* are calculated according to the molar ratios of the raw materials (nominal composition).

expansion in the interlayer spacing. Based on the analysis using Mn2+ as a proxy for Az+

One of the benefits of mechanosynthesis is that all materials are retained during the reaction, so the overall starting composition must be retained in the post-reaction compound(s). By inference, any product(s) must have the nominal starting composition. While we do not have direct compositional analysis, the PXRD results, Figure 2, clearly show that the product formed is entirely *n* = 1 R–P phase. It has been reported that the actual composition shows a good match with the nominal composition in the mechanosynthesis of OIHPs [23,24]. Thus, the halide compositions of Az2PbCl*x*Br4−*x* are calculated according

we have no information regarding any orientation or dynamics of the Az+ cation.

deviation of the measured band gap in continuous solid solutions from the values expected by linear interpolation of the end member values. Band gap bowing is often fitted to a second-order polynomial to account for the divergence from linearity, with a bowing parameter *b* as the binominal coefficient of the fitting Equation (1): [26] ܧሺݔሻ ൌ ሺ1െݔሻܧ|ሺ௫ୀሻ ܧݔ|ሺ௫ୀଵሻ െ ܾݔሺ1 െ ݔሻ (1) The bowing parameter, *b*, of the mechanosynthesised mixed halide layered Az2PbCl*x*Br4−*x* (0 ≤ *x* ≤ 4) is 0.47 with a goodness-of-fit R2 value of 0.995. The bowing parameter of mixed halide OIHPs are usually smaller, variously reported as 7 × 10−4 to 0.33 for MAPbBr3*−x*X*<sup>x</sup>* (X = Cl or I), [27,28] compared to the bowing parameters (0.4 to 1.33) found for other mixed metal perovskite systems such as MA3(Sb1−*x*Bi*x*)I9 (0.4 for Bi rich region and 1.3 for Sb rich region) and 1.06 for MA(Pb1−*x*Sn*x*)I3 [25,26,29]. The optical properties of Az2PbCl*x*Br4−*<sup>x</sup>* (0 ≤ *x* ≤ 4) solid solutions were studied by absorption spectroscopy (Figure 4a). The absorption onsets are systematically red-shifted from ca. 386 nm (Az2PbCl4) to ca. 457 nm (Az2PbBr4) with increasing average anion size (from Cl− to Br−). The bandgaps of Az2PbCl<sup>4</sup> and Az2PbBr<sup>4</sup> are calculated to be 3.43 and 2.81 eV, which are the same (within error) as the bandgap of the 6H hexagonal perovskite AzPbCl<sup>3</sup> (3.43 eV) and AzPbBr<sup>3</sup> (2.81 eV) [12]. However, unlike the linear variation in the 6H AzPbX<sup>3</sup> (X− = Cl−, Br−), the bandgap of layered R–P Az2PbX<sup>4</sup> (X = Cl, Br) shows a bowing with the average anion radius (Figure 4b). The bowing effect [25,26] simply describes the deviation of the measured band gap in continuous solid solutions from the values expected by linear interpolation of the end member values. Band gap bowing is often fitted to a second-order polynomial to account for the divergence from linearity, with a bowing parameter *b* as the binominal coefficient of the fitting Equation (1): [26]

$$E\_{\mathcal{S}}(\mathbf{x}) = (1 - \mathbf{x})E\_{\mathcal{S}|(\mathbf{x}=0)} + \mathbf{x}E\_{\mathcal{S}|(\mathbf{x}=1)} - b\mathbf{x}(1 - \mathbf{x}) \tag{1}$$

**Figure 4.** (**a**) Absorption spectra; (**b**) bandgap determination from the absorption spectra of samples Az2PbCl*x*Br4−*x* with composition 0 ≤ *x* ≤ 4 plotted as a function of average halide anion radius. **Figure 4.** (**a**) Absorption spectra; (**b**) bandgap determination from the absorption spectra of samples Az2PbCl*x*Br4−*<sup>x</sup>* with composition 0 ≤ *x* ≤ 4 plotted as a function of average halide anion radius.

**4. Conclusions**  *n* = 1 Ruddlesden–Popper (R–P) layered perovskite phases were successfully obtained by mechanosynthesis in the mixed halide solid solution Az2PbCl*x*Br4−*x* with composition 0 ≤ *x* ≤ 4. Az2PbX4(X = Cl, Br) was determined to be the conventional R–P *n* = 1 (K2NiF4) structure with a space group of *I*4/*mmm*. A linear variation in unit cell volume as a function of anion average radius is observed. The band gap of the R–P phases Az2PbCl4 The bowing parameter, *b*, of the mechanosynthesised mixed halide layered Az2PbCl*x*Br4−*<sup>x</sup>* (0 <sup>≤</sup> *<sup>x</sup>* <sup>≤</sup> 4) is 0.47 with a goodness-of-fit R<sup>2</sup> value of 0.995. The bowing parameter of mixed halide OIHPs are usually smaller, variously reported as 7 <sup>×</sup> <sup>10</sup>−<sup>4</sup> to 0.33 for MAPbBr3−*x*X*<sup>x</sup>* (X = Cl or I), [27,28] compared to the bowing parameters (0.4 to 1.33) found for other mixed metal perovskite systems such as MA3(Sb1−*x*Bi*x*)I<sup>9</sup> (0.4 for Bi rich region and 1.3 for Sb rich region) and 1.06 for MA(Pb1−*x*Sn*x*)I<sup>3</sup> [25,26,29].

#### and Az2PbBr4 are determined to be 3.43 and 2.81 eV, which is the same (within error) as **4. Conclusions**

the bandgap of 6H hexagonal perovskite AzPbCl3 (3.43 eV) and AzPbBr3 (2.81 eV) [12]. A bowing effect with a bowing parameter of 0.47 is observed in the band gap-composition relationship of R–P layered mixed halide solid solutions, compared to the linear relationship observed in the 6H hexagonal perovskite. **Supplementary Materials:** Supporting Information data include synthetic details of precipitation synthesis of Az2PbX4(X = Cl, Br) (Figures S1 and S2) and synthesis of Az2PbI4 (Figure S3). Also, include selected crystallographic data obtained powder X-ray diffraction of samples prepared by mechanosynthesis (Table S1). **Author Contributions:** Conceptualization, J.T., E.Z.-C. and F.D.M.; methodology, J.T.; validation, J.T., E.Z.-C. and F.D.M.; formal analysis, J.T.; data curation, J.T.; writing—original draft preparation, *n* = 1 Ruddlesden–Popper (R–P) layered perovskite phases were successfully obtained by mechanosynthesis in the mixed halide solid solution Az2PbCl*x*Br4−*<sup>x</sup>* with composition 0 ≤ *x* ≤ 4. Az2PbX4(X = Cl, Br) was determined to be the conventional R–P *n* = 1 (K2NiF4) structure with a space group of *I*4/*mmm*. A linear variation in unit cell volume as a function of anion average radius is observed. The band gap of the R–P phases Az2PbCl<sup>4</sup> and Az2PbBr<sup>4</sup> are determined to be 3.43 and 2.81 eV, which is the same (within error) as the bandgap of 6H hexagonal perovskite AzPbCl<sup>3</sup> (3.43 eV) and AzPbBr<sup>3</sup> (2.81 eV) [12]. A bowing effect with a bowing parameter of 0.47 is observed in the band gap-composition relationship of R–P layered mixed halide solid solutions, compared to the linear relationship observed in the 6H hexagonal perovskite.

J.T.; supervision, E.Z.-C. and F.D.M. All authors have read and agreed to the published version of the manuscript. **Funding:** This research received no external funding. The publication of this work received support from the St Andrews Institutional Open Access Fund. **Supplementary Materials:** The following are available online. Supporting Information data include synthetic details of precipitation synthesis of Az2PbX<sup>4</sup> (X = Cl, Br) (Figures S1 and S2) and synthesis of Az2PbI<sup>4</sup> (Figure S3). Also, include selected crystallographic data obtained powder X-ray diffraction of samples prepared by mechanosynthesis (Table S1).

**Institutional Review Board Statement:** Not applicable. **Informed Consent Statement:** Not applicable. **Data Availability Statement:** The research data supporting this publication can be accessed at https://doi.org/10.17630/fd5aab9b-fced-4926-afee-5eb56e2e6a5e (accessed on 15 October 2021). **Author Contributions:** Conceptualization, J.T., E.Z.-C. and F.D.M.; methodology, J.T.; validation, J.T., E.Z.-C. and F.D.M.; formal analysis, J.T.; data curation, J.T.; writing—original draft preparation, J.T.; supervision, E.Z.-C. and F.D.M. All authors have read and agreed to the published version of the manuscript.

**Acknowledgments:** We thank the Chinese Scholarship Council for support to JT (CSC No. 201603780020). **Funding:** This research received no external funding. The publication of this work received support from the St Andrews Institutional Open Access Fund.

**Conflicts of Interest:** The authors declare no conflict of interest. **Institutional Review Board Statement:** Not applicable.

**Sample Availability:** Samples of the Az2PbCl*x*Br4−*x* with composition 0 ≤ *x* ≤ 4 are available from the **Informed Consent Statement:** Not applicable.

authors. **Data Availability Statement:** The research data supporting this publication can be accessed at https://doi.org/10.17630/fd5aab9b-fced-4926-afee-5eb56e2e6a5e (accessed on 15 October 2021).

**Acknowledgments:** We thank the Chinese Scholarship Council for support to JT (CSC No. 20160378 0020).

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** Samples of the Az2PbCl*x*Br4−*<sup>x</sup>* with composition 0 ≤ *x* ≤ 4 are available from the authors.

### **References**

