**Preface to "Radically Different—A Themed Issue in Honor of Professor Bernd Giese on the Occasion of His 80th Birthday"**

#### Dear Bernd,

Upon the editor's prompt, your colleagues and friends, took steps and would like to dedicate this issue of Chemistry to you for your 80th birthday. We all felt that your birthday was a unique opportunity to convey our deep appreciation to an exceptional scientist, a great teacher and a man of culture. Many more would have liked to join the Special Issue but a nasty combination of ARN, proteins and lipids called COVID-19 prevented them from meeting the deadline.

Three words come to our mind when we look at each step of your scientific career: eclecticism, curiosity, and rigor.

Eclecticism: Your research covers a wide range of subjects, from bridged cations, selectivity–reactivity correlations of reactive intermediates, and the polar and steric effects of radical addition reactions, stereoselectivity of radical carbon–carbon bond formations and conformation determinations of chiral radicals by ESR to important problems of chemical biology such as radical-induced DNA strand cleavage, electron transfer through DNA, peptides, and, more recently, the electron transfer mechanism used by bacteria to adapt to the presence of metal ions in their environment. We should not forget important interludes such as the total synthesis of macrolides or the development of photocleavable protective groups.

Curiosity: Your desire to learn by exploring the unknown has obviously been the driving force of your research. It was and is still served by your unique ability to select important and fundamental questions.

Rigor: The way you approach scientific projects obviously originates from your education in Munich, where you were nurtured in the principles of physical organic chemistry by your former mentor Rolf Huisgen. You certainly belong to a small group of creative physical chemists who use the deep understanding of molecular properties to devise new reactions or new molecules with important properties. The acclaimed "Giese Reaction" is a textbook example of this interplay between "understanding" and "making".

You once said to me, "Whatever we have done in research is probably less important than our contribution as teacher." Many young chemists owe you thanks for the stimulating and sound mentorship you provided. There is undoubtedly a Giese school in the community of chemists. Each of us has enjoyed listening to your stimulating lectures and chatting with you about chemical problems, not only because of your knowledge of the field but also because you are a man with an original and profound vision of modern society.

For this, your colleagues and friends who took part in this Special Issue would like to give thanks.

Congratulations and happy birthday!

#### Léon

Prof. Emeritus Dr. L. Ghosez

Visiting scientist at the Institut de Chimie et Biologie, Université de Bordeaux

#### **Professor Giese is a pioneer in selective radical chemistry, electron transfer through biomolecules, and, recently, electron transfer through bacterial membranes**

Professor Bernd Giese is Professor Emeritus of Organic Chemistry at the University of Basel, Switzerland, and is now "postprof" in the group of Katharina M. Fromm at the University of Fribourg, Switzerland.

He was born in Hamburg, Germany, in 1940, studied in Heidelberg, Hamburg, and Munich, and received his Ph.D. in 1969 while working in the group of the late Rolf Huisgen. After two years in a pharmaceutical research group at the BASF, Ludwigshafen, he started his independent research at the University of Münster and received his Habilitation at the University of Freiburg in 1976. One year later, he became a Full Professor at TU Darmstadt, Germany, and accepted the position of Chair at the University of Basel in 1989. He served as dean at TU Darmstadt and as head of the department at the University of Basel. He is a member of the Editorial Advisory Board of several journals and institutes and has acted as a regional editor of SYNLETT from its beginning. Professor Giese has published more than 300 papers and has authored or co-authored three books on radical chemistry. He is a member of the Deutsche Akademie der Naturforscher Leopoldina and the American Academy of Arts and Sciences. His awards are numerous and include the Gottfried Wilhelm Leibniz Prize in 1987, the Tetrahedron Prize in 2005, the Emil Fischer Medal in 2006, and the Paracelsus Prize of the Swiss Chemical Society in 2012. In 2019, on the occasion of his 50th Ph.D. anniversary, his Ph.D. diploma from the Ludwig Maximilian University was renewed in the presence of his Ph.D. supervisor Rolf Huisgen, then aged 99.

From left to right: Katharina M. Fromm, Rolf Huisgen, and Bernd Giese on the occasion of Bernd Giese's 50th Ph.D. jubilee and renewal at LMU on June 18th, 2019.

Professor Giese's research encompasses studies on bridged cations, selectivity–reactivity correlations of reactive intermediates, polar and steric effects of radical addition reactions, the stereoselectivity of radical C–C bond formations, conformation determinations of chiral radicals by ESR, the total synthesis of macrolides, radical-induced DNA strand cleavage, photocleavable protecting groups, as well as electron transfer through DNA, peptides, and proteins.

He developed a new synthetic method that involves alkyl halides, metal hydrides, and alkenes. This three-component radical chain reaction was one of the starting points of modern synthesis with carbon-centered radicals. He applied this method—referred to today as the "Giese Reaction" in textbooks and much of the recent literature—to the synthesis of several target molecules. Bernd has thus developed important concepts for the understanding of kinetics and the selectivity of complex reactions. He has pioneered the introduction of radical reactions as powerful synthesis methods and contributed substantially to the area of physical– organic chemistry.

Today, modern physical–organic chemistry plays a major role in biochemistry. In his bioorganic studies, Bernd Giese's experiments were crucial in elucidating the controversial problem of long-distance electron transfer through DNA. He showed that electrons migrate through DNA in a multistep hopping reaction, where each single hopping step depends strongly on the distance, using appropriate DNA bases as stepping stones. He also proposed new mechanisms for DNA strand breaks via intermediate radicals. Another topical case is the study of electron transfer through proteins that connect distant molecule parts and enable redox reactions, for example, ribonucleotide reductase—the only enzyme that makes deoxyribonucleotide (DNA) available from ribonucleotide (RNA). The production site of the reactive intermediate is 3.5 nm from the reduction site, and the intervening protein is the medium for long-distance electron transfer. Bernd Giese was able to show here that amino acid side groups are used as stepping stones in this process by generating radical cations in the ground state at one end of a peptide model and studying the kinetic of electron transfer as a function the amino acid sequences and further charges at the end groups.

From model systems, Bernd Giese recently moved to living microorganisms, in particular, Geobacter sulfurreducens, which is able to reduce metal ions outside of the cell and can produce metal nanoparticles in aqueous solutions. Here, electrons can migrate from the inside to the outside of the cell using either filaments (pili) of aggregated proteins or c-type cytochromes, which transport electrons through the periplasm and the inner and outer membrane. These studies are important for understanding basic processes in life. They can also lead to enzyme inhibitors and nanoelectronic devices or help to clean polluted water. The interplay between the understanding of molecular behavior and the creation of new materials or devices is crucial, and Bernd Giese will further contribute to these exciting research activities.

Bernd Giese's seminal contributions have thus not only shaped organic synthesis but also had a profound impact on chemical biology research. Happy Birthday, and many more important results to come!

> **Katharina M. Fromm**  *Guest Editor*

### *Review* **Before Radicals Were Free – the** *Radical Particulier* **of de Morveau**

#### **Edwin C. Constable \* and Catherine E. Housecroft**

Department of Chemistry, University of Basel, BPR 1096, Mattenstrasse 24a, CH-4058 Basel, Switzerland; catherine.housecroft@unibas.ch

**\*** Correspondence: edwin.constable@unibas.ch; Tel.: +41-61-207-1001

Received: 31 March 2020; Accepted: 17 April 2020; Published: 20 April 2020

**Abstract:** Today, we universally understand radicals to be chemical species with an unpaired electron. It was not always so, and this article traces the evolution of the term radical and in this journey, monitors the development of some of the great theories of organic chemistry.

**Keywords:** radicals; history of chemistry; theory of types; valence; free radicals

#### **1. Introduction**

The understanding of chemistry is characterized by a precision in language such that a single word or phrase can evoke an entire back-story of understanding and comprehension. When we use the term "transition element", the listener is drawn into an entire world of memes [1] ranging from the periodic table, colour, synthesis, spectroscopy and magnetism to theory and computational chemistry. Key to this subliminal linking of the word or phrase to the broader context is a defined precision of terminology and a commonality of meaning. This is particularly important in science and chemistry, where the precision of meaning is usually prescribed (or, maybe, proscribed) by international bodies such as the International Union of Pure and Applied Chemistry [2]. Nevertheless, words and concepts can change with time and to understand the language of our discipline is to learn more about the discipline itself. The etymology of chemistry is a complex and rewarding subject which is discussed eloquently and in detail elsewhere [3–5]. One word which has had its meaning refined and modified to an extent that its original intent has been almost lost is *radical*, the topic of this special issue.

This article has two origins: firstly and most importantly, on the occasion of his 80th birthday, it is an opportunity to express our gratitude and thanks for the friendship and assistance of Bernd Giese in our years together in Basel, and secondly to acknowledge a shared interest with Bernd in the history of our chosen discipline.

#### **2. Modern Understanding**

It seems relevant to present the IUPAC definition of a radical in full at this point in the text as it both provides a precision for modern usage and also contains hints of the historical meaning:

"A molecular entity such as ·CH3, ·SnH3, Cl· possessing an unpaired electron. (In these formulae the dot, symbolizing the unpaired electron, should be placed so as to indicate the atom of highest spin density, if this is possible.) Paramagnetic metal ions are not normally regarded as radicals. However, in the 'isolobal analogy', the similarity between certain paramagnetic metal ions and radicals becomes apparent. At least in the context of physical organic chemistry, it seems desirable to cease using the adjective 'free' in the general name of this type of chemical species and molecular entity, so that the term 'free radical' may in future be restricted to those radicals which do not form parts of radical pairs. Depending upon the core atom that possesses the unpaired electron, the radicals can be described as carbon-, oxygen-, nitrogen-, metal-centered radicals. If the unpaired electron occupies an orbital having

considerable s or more or less pure p character, the respective radicals are termed σ- or π-radicals. In the past, the term 'radical' was used to designate a substituent group bound to a molecular entity, as opposed to 'free radical', which nowadays is simply called radical. The bound entities may be called groups or substituents, but should no longer be called radicals" [6].

To summarize, in accepted modern usage, a radical possesses an unpaired electron.

#### **3. A Radical Birth**

#### *3.1. de Morveau's Introduction*

The word radical was introduced by the French politician and chemist, Louis-Bernard Guyton, Baron de Morveau (1737–1816, prudently identified after the French revolution without the aristocratic rank as Louis-Bernard Guyton-Morveau, Figure 1) [7]. In 1782, de Morveau published an article entitled *Sur les Dénominations Chymiques, La nécessité d'en perfectioner le systême, et les règles pour y parvenir* in which he identified the need for a new systematic nomenclature in chemistry [8]. In this paper, he not only formulated his five principles of nomenclature which later became embodied in the *Méthode de Nomenclature Chimique* [9,10], but also introduced the word *radical* to describe a multiatomic entity; in his own words "Having found the adjectives arsenical and acetic consecrated by usage, it was necessary to preserve them and form only such close nouns to the radicals of these terms that they could be understood without explanation. Arseniates and acetates seemed to me to fulfil this condition." He makes no further comment on the term in this paper, which also includes a table which lists acids, the generic names of salts derived from these acids, bases or substances that bind to acids. This table also confirms that he was still a phlogistonist [11,12] in 1782, as phlogiston is listed amongst the bases or substances that bind to acids. The word *radical* itself seems to derive from the Latin word radix (root).

By the time of the publication of the *Méthode*, the concept of radicals was embedded in the core of the model in five classes of substances which had not been decomposed into simpler materials (the second class includes all the acidifiable bases or radical principles of the acids) [9,10]. In this work, the "radical of the acid" was precisely defined as "the expression of acidifiable base". The explanations given in the text are difficult for the modern reader to follow as the conversion of the radical (such as nitrate or acetate) to the parent acid did not involve the addition of protons but rather oxygen. Although the credit for the discovery of oxygen should be shared between William Scheele, Joseph Priestley and Antoine Laurent de Lavoisier [13,14], Lavoisier's contribution included the name *oxygène*, from the Greek ὀξύς (acid, sharp) and -γενής (producer, begetter), on the basis of his belief that oxygen was a constituent of all acids. On this basis, the *Méthode* continues to clarify the nomenclature of radicals defining known acids as arising from the addition of oxygen to "pure charcoal, carbon or carbonic radical ... Sulphur or sulphuric ... radical and phosphorus or phosphoric radical". The identification of oxygen as the essential component of an acid was not without its difficulties and for elements such as sulfur, with variable oxidation states, it was necessary to state that "it is evident that the sulphur is at the same time sulphuric radical, and sulphureous radical". Additional problems arose with nitrogen derivatives, with de Morveau using both *Azote* and *Radical Nitrique* for the parent radical. It took Jean Antoine Chaptal [15] to introduce the name *nitrogène* in his 1790 work *Eléments de chimie* [16,17].

The text of the *Méthode* uses the term radical extensively to describe acids and their salts and the construction of the names is illustrated in the extensive tables correlating the old names with the ones which are newly proposed. One of the most important features of the *Méthode* was the folding table of substances in which the core radicals are identified.

One aspect of the establishment of the concept of radicals is reminiscent of the later work of Mendeleev, who proposed missing elements from the periodic table and identified their likely properties. In the same way, the *Méthode* recognizes that muriatic acid (modern name hydrochloric acid) contained an unknown radical, described as muriatic radical or muriatic radical principle. The extention of the radical concept to organic chemistry was also pre-empted by de Morveau when he noted that the reaction of sucrose with nitric acid to give ethanedioic acid (*acide saccharin*), which is a combination of oxygen and *radical saccharin*.

#### *3.2. Lavoisier's Adoption*

The use of the term radical in the original sense of de Morveau was broadly adopted by Antoine-Laurent de Lavoisier and his wife Marie-Anne Pierrette Paulze Lavoisier [18–21] in a number of subsequent and influential texts (Figure 2). The *Méthode* was republished and expanded [22], but the most influential was the *Traité Élémentaire de Chimie, Présenté dans un Ordre Nouveau, et d'Après des Découvertes Modernes* [23–25]. This also served to further bring the changes in nomenclature and philosophy to the attention of the anglophone world, which received the first translation of the *Méthode* in 1788 and was able to delight in the English translation of the *Traité* from 1791 onwards [22,26–30]. The radical concept is intrinsic to the book and is also clearly defined "The word acid, being used as a generic term, each acid falls to be distinguished in language, as in nature, by the name of its base or radical. Thus, we give the generic names of acids to the products of the combustion or oxygenation of phosphorus, of sulphur, and of charcoal; and these products are respectively named, phosphoric acid, sulphuric acid, and carbonic acid". In his list of elements in the Traité, Lavoisier lists *Radical muriatique*, *Radical fluorique* and *Radical boracique* (the elements chlorine, fluorine and boron respectively) as unknown (*Inconnu*). In the context of organic chemistry, Lavoisier recognized that organic compounds contained compound radicals which could combine with oxygen to form more complex substances, such as ethanol or ethanoic acid. We are fortunate that not only was Marie-Anne Pierrette Paulze Lavoisier an enthusiastic and gifted co-worker (and according to the *mores* of the times, not listed as a co-author), but that she also actively contributed to the *Traité* and preserved many of Antoine Lavoisier's writings, including his notebooks, for the benefit of future generations.

**Figure 2.** Antoine-Laurent de Lavoisier (1743–1794, subsequently Antoine Lavoisier) popularized the use of the term radical (Public domain image. Source https://commons.wikimedia.org/wiki/File: Antoine\_Laurent\_de\_Lavoisier.png).

#### **4. From** *Radical Particulier* **to the Radical Theory and the Theory of Types**

#### *4.1. Gay-Lussac and the CN Radical*

The next player in our drama of radicals should be Joseph Louis Gay-Lussac [31] (Figure 3a) and, in particular, his work on cyanides. Although HCN (hydrocyanic acid, prussic acid) was a known compound, Gay-Lussac established its formula and showed that it contained no oxygen, another of the nails in the coffin of Lavoisier's theory that all acids contained oxygen. By 1815, he had prepared metal cyanide salts as well as ClCN and cyanogen and correctly identified that the CN unit was retained throughout chemical transformations. His publication *Recherches sur l'acide prussique*, repeatedly refers to the *radical de l'acide prussique* [32–35]. This, in turn, necessitates a subsequent and consequent linguistic distinction between "simple radicals" (iron, sulphur, nitrogen, phosphorus and carbon) and "compound radicals"; containing multiple elements bonded together but which behave as distinct (and inseparable) units. As Gay-Lussac wrote "Here, then, is a very great analogy between prussic acid and muriatic and hydriodic acids. Like them, it contains half its volume of hydrogen; and, like them, it contains a radical which combines with the potassium, and forms a compound quite analogous to the chloride and iodide of potassium. The only difference is, that this radical is compound, while those of the chloride and iodide are simple" [36]. In isolating cyanogen, Gay-Lussac claimed to have isolated the first compound radical (actually the dimer, (CN)2).

The identification of compound radicals was further expanded by Jöns Jacob Berzelius in 1817. Berzelius (Figure 3b) was the leading exponent of the electrochemical dualism theory which considered that all compounds are salts derived from basic and acidic oxides [37,38]. As one of the most respected chemists of the time, Berzelius' support for this model resulted in its widespread acceptance. For example, Berzelius would regard the compound potassium sulfate, K2SO4, as arising from the combination of the positively charged metal oxide K2O and negatively charged SO3. The radical theory as applied to inorganic compounds meshed well with his views, but he had difficulties in extending these to organic species. Nevertheless, he considered that the new concept of simple and compound radicals would clarify the differences between the inorganic acids with simple radicals and the organic acids with compound radicals "In inorganic nature all oxidized bodies contain a simple radical, while all organic substances are oxides of compound radicals. The radicals of vegetable substances consist generally of carbon and hydrogen, and those of animal substances of carbon, hydrogen and

nitrogen" [39]. In reality, Berzelius refused to accept the possibility that a radical could contain oxygen and this, ultimately, led to the discrediting of the theory. In the intermediate period, however, the compound radical model was the origin of a new radical theory for organic chemistry and ultimately the modern functional group model.

**Figure 3.** (**a**) Joseph Louis Gay-Lussac (1778 – 1850) showed that CN was a compound radical and opened the doors to the Radical Theory of organic chemistry. (Public domain image. Source https: //en.wikipedia.org/wiki/Joseph\_Louis\_Gay-Lussac#/media/File:Gaylussac.jpg) (**b**) Jöns Jacob Berzelius (1779 – 1848) was one of the leading chemists of his age and in 1817 he laid the basis for the Radical Theory in organic chemistry. (Public domain image. Source https://en.wikipedia.org/wiki/Jöns\_Jacob\_ Berzelius#/media/File:Jöns\_Jacob\_Berzelius.jpg).

#### *4.2. The General Radical Theory*

The stage is now set for the generalization of the radical theory. The major players in this were Friedrich Wöhler (Figure 4a) [40], Justus Freiherr von Liebig (Figure 4b) [41,42] and (at least for a period) Jean Baptiste André Dumas (Figure 4c) [43]. The three had a vision of radicals as collections of atoms that behaved like elements and persisted through chemical reactions, although Dumas subsequently shifted his allegiance to the theory of types (Section 4.3).

**Figure 4.** (**a**) Friedrich Wöhler (1800–1882) showed that CN was a compound radical and opened the doors to the Radical Theory of organic chemistry. (Public domain image. Source https://en.wikipedia. org/wiki/Friedrich\_Wöhler#/media/File:Friedrich\_Wöhler\_Litho.jpg) (**b**) Justus Freiherr von Liebig (1803–1873) was one of the leading chemists of his age and in 1817 he laid the basis for the Radical Theory in organic chemistry. (Public domain image. Source https://en.wikipedia.org/wiki/Justus\_von\_ Liebig#/media/File:Justus\_von\_Liebig\_NIH.jpg) (**c**) Jean Baptiste André Dumas (1800–1884).

One of the critical publications was *Untersuchungen über das Radikal der Benzoesäure* by Liebig and Wöhler in 1832 [44], which introduces synthetic chemistry in a manner that we rarely see today "If it is possible to find a bright point in the dark area of organic nature, which seems to us to be one of the entrances through which we can perhaps reach true paths of exploration and recognition. From this point of view, one may consider the following attempts, which, as far as their extent and their connection with other phenomena is concerned, leave a wide, fertile field to cultivate". In a way, this publication was somewhat heretical, at least in the eyes of Berzelius, as Wöhler and Liebig maintained that a radical could be more than just the base of an acid. Specifically, Wöhler and Liebig showed that the benzoyl radical (C6H5CO in modern formulation) persisted in the compounds C6H5CO-H, C6H5CO-OH, C6H5CO-Cl, C6H5CO-I, C6H5CO-NH2, C6H5CO-Br, and (C6H5CO-)2S. The conclusion was that the benzoyl radical behaved in a similar manner to an inorganic radical and persisted unchanged through multiple reactions.

The impact of this publication on the organic chemistry community cannot be underestimated and resulted in an explosive reporting of new radicals over the next few years, including acetyl, methyl, ethyl, cacodyl (Me2As), cinnamoyl (C6H5CH=CH), and *n*-C16H33. Originally, Dumas was opposed to the radical theory but eventually became convinced by Liebig's arguments. Dumas was responsible for the recognition of the methyl, cinnamoyl and *n*-C16H33 radicals. Although the radical theory has not survived, the nomenclature introduced is still in use today. Berzelius himself was responsible for the identification of the ethyl radical [37,45]. The state-of-the-art in radical theory in the Berzelius spirit is found in another publication of Liebig which interprets a large number of experimental results on ethers in terms of the Berzelius radical model [46].

By 1837, although Dumas and Liebig still disagreed in detail on which groups of atoms were to be considered radicals, they were sufficiently confident in the universality of their radical model, that they published their "Note on the present state of organic chemistry", which is a comprehensive overview of the radical theory at that time [47]. It appears that Liebig was given to flights of purple prose "and that, we are convinced, is the whole secret of organic chemistry. Thus, organic chemistry possesses its own elements which at one time play the role belonging to chlorine or to oxygen in mineral chemistry and at another time, on the contrary, play the role of metals. Cyanogen, amide, benzoyl, the radicals of ammonia, the fatty substances, the alcohols and analogous compounds—these are the true elements on which organic chemistry is founded and not at all the final elements, carbon, hydrogen, oxygen, and nitrogen elements which appear only when all trace of organic origin has disappeared. For us, mineral chemistry embraces all substances which result from the direct combination of the elements as such. Organic chemistry, on the contrary, should comprise all substances formed by compound bodies functioning as elements would function. In mineral chemistry, the radicals are simple; in organic chemistry, the radicals are compound; that is all the difference One year later, in 1838, Liebig clearly defined what he understood by the term radical, in the context of the CN radical: "So we call cyanogen a radical, because 1) it is the non-changing constituent in a series of compounds, because 2) it can be replaced in them by other simple bodies, because 3) it can be found in its connections with a simple body of the latter, and represented by equivalents of other simple bodies. Of these three main conditions for the characteristic of a composite radical, at least two must always be fulfilled if we are to regard it in fact as a radical" [48].

The proposals of Liebig were not universally accepted. Robert Hare in the United States of America published a number of articles dismissing the commonality of the oxoacids and "simple" acids such as the hydrogen halides, well summarized in his monograph "An attempt to refute the reasoning of Liebig in favor of the salt radical theory" [49]. Berzelius, in particular, came to have difficulties with the radical theory of Wöhler and Liebig because it directly challenged his electrochemical dualism theory [50]. For example, the relationship between benzaldehyde C6H5CO-H and benzoyl chloride C6H5CO-Cl could not possibly be correct because the hydrogen which has a positive charge cannot be replaced by a negative chlorine.

Not only were ever more radicals being identified, but they were also being isolated as chemical species. A few highlights serve to exemplify this. Robert Wilhelm Bunsen (1811–1899) reinvestigated some arsenic compounds first reported by Cadet and obtained a foul-smelling and highly toxic liquid which he called *Alkarsin*, although Berzelius suggested that cacodyl (or kakodyl) was more appropriate. The compound, formulated (CH3)2As [51] was obtained from the reaction of (CH3)2AsCl with zinc and was widely thought to be the free cacodyl radical. This compound was subsequently shown to be the dimer, (CH3)2AsAs(CH3)2. Similarly, Kolbe isolated the free methyl radical [52] and Frankland the free ethyl radical [53], although both were actually the dimers (ethane and butane, respectively).

#### *4.3. The Theory of Types*

The theory of types is rather a difficult concept for the modern chemist to appreciate. Put simply, it retains the fundamentals of the radical theory, but allows the replacement of elements and groups within a radical. With hindsight, it is possible to see the origins of the functional group model of organic chemistry within this approach. The development leading to the theory of types came from Dumas, who in 1838 described the chlorination of acetic acid to give trichloroacetic acid [54–57]. The substitution of hydrogen by chlorine generated a new radical (trichloroacetyl or trichloromethyl rather than acetyl or methyl) but did not change the molecular *type*. The chemical properties of acetic acid and trichloroacetic acid were very similar, indicating the same molecular type. Dumas published two papers which enunciated his theory of types [55,56] The level of vitriol and animosity in the debate is well exemplified by the spoof publication by S. C. H. Windler (actually written by Wöhler) in *Annalen* in which he rather wickedly parodies the substitution theories of Dumas and collagues [58]. He describes sequentially replacing atoms in manganese(II) acetate (his formulation, MnO + C4H6O3) with chlorine, initially producing manganese(II) trichloroacetate and eventually, Cl2Cl2 + Cl8Cl6Cl6 (i.e., Cl24). This compound was a yellow solid resembling the original manganese(II) acetate, because "hydrogen, manganese, and oxygen may be replaced by chlorine, there is nothing surprising in this substitution". In a footnote, he adds "I have just learned that there is already in the London shops a cloth of chlorine thread, which is very much sought after and preferred above all others for night caps, underwear, etc."

By 1853, primarily due to the work of Charles Adolphe Wurtz, Hoffman, Williamson and Gerhardt, four different types had been identified; the water type, the hydrogen type, the hydrogen chloride type and the ammonia type. The water type included water, alcohols, ethers and carboxylic acids, the hydrogen type, dihydrogen, and alkanes, the hydrogen chloride type included organohalogen compounds such as C2H5Cl and finally, the ammonia type which included all primary, secondary and tertiary amines [59].

#### *4.4. Laurent and the Theory of Types*

Auguste Laurent (1807–1853) also studied substitution reactions and from 1834 onwards described numerous examples in which hydrogen atoms within radicals were replaced by halogens or oxygen [60–62]. Probably, the credit for the theory of types should be shared by Laurent with Dumas, because the former clearly recognized that the fundamental properties of the compound were not significantly changed by the substitution [63–65]. His theories are clearly stated in his book *Méthode de Chimie* from 1854 [66] but the ideas are clearly formulated (and seen to be almost identical to those of Dumas) as early as 1836 "All organic compounds are derived from a hydrocarbon, a fundamental radical, which often does not exist in its compounds but which may be represented by a derived radical containing the same number of equivalents" [67]. It appears that Dumas deliberately underplayed the importance of Laurent and over-emphasized the relevance of his protegé Henri Victor Regnault. On occasion, Laurent expressed his feelings in plain rather than scientific language " ... others, pretend that I have taken some ideas of M. Dumas. M. Dumas. ... has done much for the science; his part is sufficiently great that one should not snatch from me the fruit of my labors and present the offering to him" [68]. And concerning radicals, he wrote "I claim with a conviction most

profound that to me belongs, and to me alone, the most part of the ideas developed by M. Dumas" [69]. The arguments continued!

In 1837, Laurent developed a theory of fundamental and derived radicals, subsequently known as his nucleus theory, which was based upon an obscure geometrical argument and attempted to rationalize the carbon core of radicals undergoing substitution. Like much of his work, this was an interesting and novel attempt to bring order to organic chemistry [70]. Nevertheless, the theory of Laurent was anathema to Liebig, who in his usual offensive manner discussed it "not because he found something in it worthy of mention, not in order to admit its having an influence on the development of chemistry but in order to demonstrate that it is unscientific, good for nothing".

#### *4.5. Dualities, Inconsistencies and Ambiguities within the Radical Theory*

Even at the time of its greatest success, there were many inconsistencies and dualities within the radical theory. Today, we would understand the term acetyl radical to refer to the species CH3CO. Unfortunately, this was not the case in the 19th Century CE. In 1835, Henri Victor Regnault (1810–1878) [71,72] reported a new radical C2H3 (formulated C4H6 at the time) which he termed *aldehydène* [73]. This radical was present in the compounds H2C=CHCl, H2C=CHBr, BrCH2CH2Br and many others that he isolated. He also linked the radical aldehydène to ethanal and ethanoic acid, which Regnault formulated as {C4H6O + H2O} and {C4H6O3 + H2O}, respectively. In 1839, Liebig suggested that the radical C2H3 should be called acetyl, in accord with his own system of nomenclature [74]. This 1839 paper of Liebig served to link together in a more-or-less coherent manner the various radicals and radical theories which had been proposed for C2 compounds (although with the atomic weight confusion at the time many of these were formulated C4 species). The Aetherin (or etherin) theory was proposed by Dumas and Boullay in 1828 and considered that C2H4 (formulated C4H8 at the time) was the common radical in C2 compounds: thus, C2H5OH, C2H5OC2H5 and C2H5Cl were the aetherin radical with water, ethanol and HCl, respectively [75]. In contrast, Berzelius formulated these compounds in terms of the C2H5 (ethyl) radical [37,45].

#### **5. Valency Displaces Radicals**

The real death of the old radical theory and the theory of types came in 1852 when Edward Frankland formulated what was to become the concept of valency, "When the formulae of inorganic chemical compounds are considered, even a superficial observer is struck with the general symmetry of their construction ... it is sufficiently evident ... no matter what the character of the uniting atoms may be, the combining power of the attracting element, if I may be allowed the term, is always satisfied by the same number of these atoms" [76]. Frankland's combining power was the first formulation of the basic idea of valence and the entry to the electronic view that has dominated chemistry ever since.

A few years later, in 1858, Kekulé proposed a fixed valence for elements; although he did not equate the combining power with valence [77]. Kekulé successfully rationalized the structures of organic compounds by assuming a fixed valence of four for carbon, and extended this fixed valence idea to the elements nitrogen and oxygen which had fixed valences of three and two, respectively. The fixed valence of four for carbon necessitated multiple bonds (or free valences) in appropriate compounds. And so modern organic chemistry was born—or rather, as we have seen on a number of occasions in this article, we can testify to another of its births!

It is one of the pleasures associated with the study of the development of chemistry in the 19th Century CE, to read not only the contemporary primary literature, but also the textbooks and monographs of the period. These often provide a unique view of the way in which views changed and also give an understanding of the tensions and controversies in the science of the time. One of the lesser known works of this period is "A Short History of the Progress of Scientific Chemistry in Our Own Times" by William Tilden, which gives a detailed account of the evolution of chemistry to the last year of the 19th Century CE. The sections on the development of the Theory of Types and the subsequent Valency Model are excellent and also document a number of the poorly documented highways and by-ways associated with the scientific journey to the Valency Model [78]. An excellent contemporary (1867) overview of the Theory of Types and the relationship to the atomicity of the radicals is given by Adolphe Wurtz [79].

An interesting historical overview of the development of the subject written after the triumph of valency theory is to be found in the books by von Meyer [80] and Venable [81].

#### **6. The Freeing of the Radical—the First Modern Radicals**

Although transition metal compounds with unpaired electrons were well-known, and "simple" inorganic substances, such as Frémy's salt (K4[ON(SO3)2]2) [82], NO or NO2, which fulfill our modern definition of a radical had been long established, the dominance and success of the valence theory in organic chemistry, based upon the invariable and inviolable tetravalency of carbon led to the widely accepted opinion that organic radicals (modern sense) could not exist. The confidence in the tetravalency of carbon and the complacency of the organic community was shattered in 1900, when Moses Gomberg at the University of Michigan reported the preparation of triphenylmethyl radical, Ph3C, as the product from the attempted preparation of hexaphenylethane from the reaction of chlorotriphenylmethane with zinc [83]. The title of the paper, "An instance of trivalent carbon: triphenylmethyl" hints at the supremacy of the "tetravalent carbon" dogma [84].

The rest, dear reader, is history.

#### **7. Final Words**

In this short article, we have presented a story which describes the evolution of organic chemistry and which laid the basis for our modern understanding based on the electronic, molecular orbital and functional group approaches. Perhaps surprising for the modern reader is the passion with which the debate was conducted and the manner in which the personalities of the individual involved come though and, indeed, the personalization of the rhetoric. The well-known *Schwindler* article has already been referred to. The correspondence between Berzelius, Liebig, Dumas and Wöhler is a wonderful introduction to the art and science of denigrating your rivals in language that is rarely found in the scientific literature [85]. The discourse was not limited to scientific matters, but also to the character and nationality of the players, for example, Liebig described Dumas on various occasions as a swindler, charlatan, tightrope dancer, Jesuit, highwayman, and a thief, like "nearly all Frenchmen" [86]. As he became older, Berzelius became increasingly cantankerous, and writes of Liebig "I will say nothing of Liebig's ruthless, thoughtless and unjustified criticism, ... it just disappoints and saddens me ... with the manner of a dictator, who wishes to abolish an old constitution and create a new one ... I hold it unlikely that he will take the slightest notice of my advice" [85]. Berzelius again, talking of Liebig "Either Liebig is mad, which I already began to painfully fear a year ago, in which case he deserves the pity of everyone and needs to be treated accordingly, or he is an unwise, inflated fool" [85]. Wentrup has recently published an assessment of some aspects of the debate in the context of Zeise's discovery of his eponymous salt, K[Pt(C2H4)Cl3] which also documents the acrimonious exchanges between the players [86].

**Author Contributions:** This article was conceived and written jointly by E.C.C. and C.E.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work received no external funding.

**Acknowledgments:** As always, we give our thanks to the various library and abstracting services which have aided us in identifying and sourcing material and, in particular, to the online sources of historical materials which have made the life of scientists interested in the origins of their subject so much easier in the 21st Century C.E. Finally, we acknowledge the friendship and scholarship of Bernd Giese on the occasion of his 80th birthday.

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Towards a Real Knotaxane**

#### **Torben Duden and Ulrich Lüning \***

Otto-Diels-Institut für Organische Chemie, Christian-Albrechts-Universität zu Kiel, Olshausenstr. 40, D-24098 Kiel, Germany; tduden@oc.uni-kiel.de

**\*** Correspondence: luening@oc.uni-kiel.de

Received: 12 March 2020; Accepted: 10 April 2020; Published: 26 April 2020

**Abstract:** Two classes of mechanically interlocked molecules, [3]rotaxanes and knotted [1]rotaxanes, were the subject of this investigation. The necessary building blocks, alkyne-terminated axles containing two ammonium ions and azide-terminated stoppers, and azide-containing substituted macrocycles, have been synthesized and characterized. Different [3]rotaxanes were synthesized by copper-catalyzed "click" reactions between the azide stoppers and [3]pseudorotaxanes formed from the dialkyne axles and crown ethers (DB24C8). Methylation of the triazoles formed by the "click" reaction introduced a second binding site, and switching via deprotonation/protonation was investigated. In preliminary tests for the synthesis of a knotted [1]rotaxane, pseudorotaxanes were formed from azide-containing substituted macrocycles and dialkyne substituted diammonium axles, and copper-catalyzed "click" reactions were carried out. Mass spectral analyses showed successful double "click" reactions between two modified macrocycles and one axle. Whether a knotted [1]rotaxane was formed could not be determined.

**Keywords:** mechanically interlocked molecules; knot; rotaxane; macrocycle; click reaction; switching; shuttle

#### **1. Introduction**

Chemical elements bind each other through metallic, ionic and covalent bonds. Furthermore, molecules may bind each other supramolecularly, and in addition, for half a century, the mechanical bond has been known of [1]. The latter is responsible for the formation of rotaxanes [2], catenanes [3,4], molecular knots [5] and other mechanically interlocked molecules (MIM). Although only known since the second half of last century, a plethora of unusual MIMs have been synthesized and investigated [1], and some MIM types have even been combined. In 2003, Vögtle and co-workers described a MIM they called knotaxane because it is a rotaxane with molecular knots as stoppers [6].

In this work, we would like to discuss a "real" knotaxane. In contrast to Vögtle's MIM, it is not a combination of a rotaxane and a knot but a molecule in which the rotaxane property arises from a knotting. The fundamental feature of a "normal" rotaxane is the fact that the stoppers prevent the ring from slipping off the axle. However, from a mathematical-topological point of view, a rotaxane is topologically not special because by deformation (shrinking of the stoppers or enlargement of the ring), the ring may slip off—as it does in pseudorotaxanes [7–9]. Figure 1 (right) shows a sketch of a "real" knotaxane. The rings sit on a central axle as in a [3]rotaxane (Figure 1, left), but connections between the rings and the ends of the axle rather than stoppers prevent their slipping-off. This special MIM is a knotted molecule but also contains rings on an axle, as in a rotaxane. However, in contrast to standard rotaxanes, *all* atoms in this MIM are covalently connected. It may therefore be called a "knotted [1]rotaxane".

**Figure 1.** Comparison of the geometry of a [3]rotaxane with that of a knotted [1]rotaxane.

By introducing different binding sites into the axle of a rotaxane, the position of rings on the central axle may be switched, as demonstrated first by Stoddart [10]. Introduced into the knotaxane, this switching will result in a breathing of the molecule, or if suspended at its ends, it may act as a type of spring. We chose crown ethers as macrocyclic rings and ammonium and triazolium ions as binding sites (see below; for instance, **15** in Figure 6 and **17** in Figure 7). Such a pair of binding sites allows the controlled shuttling of a ring [11].

In acidic media, the ammonium nitrogen atoms are protonated. In comparison to ammonium ions, triazolium ions are poorer binding sites, since the charge is not localized and there are no hydrogen bonds. Therefore, a crown ether binds to an ammonium ion preferentially due to stronger Coulomb interactions and hydrogen bonds. Upon deprotonation, the interactions of the positive charge and one hydrogen bond vanish. In basic media, the triazolium ion is the only positively charged site and is, therefore, the better binding site, and the ring binds there after deprotonation.

Retrosynthetically, triazolium ions call for a "click" reaction between an azide and an alkyne (copper-catalyzed alkyne-azide cycloaddition, CuAAC) followed by alkylation. We chose to place the azides on the stopper side. Therefore, bis-alkyne terminated, diammonium-containing central parts of the axles were needed. In order to study the CuAAC and to allow the investigation of the switching, we first synthesized [3]rotaxanes by using simple azide stoppers. For the synthesis of the knotaxane, the azide function had to be connected to the crown ether. Therefore, the following tasks had to be accomplished: synthesis of the central axles, the alkyne terminated diammonium ions and crown ethers, each with an azide-terminated side chain; syntheses of [3]rotaxanes from the central axles, crown ethers and azide stoppers; alkylation of the triazoles; subsequent study of switching of the [3]rotaxanes; and finally, the study of the connection between the central axles with the azide-terminated crown ethers to give the knotted [1]rotaxane (Figure 2). As depicted in Figure 2, after formation of the [3]pseudorotaxane, there are two possibilities for "click" reactions. Either the azide connected to ring **a** reacts with alkyne end **b** or it reacts with **b**- . In the latter case, the desired knotaxane is formed (Figure 2, top right); in the other case a different topology is produced in which the rings may even slip off each end to form a handcuff topology (Figure 2, bottom right). The exact dimensions of the axle and the lengths of the tethers at the macrocycles needed to obtain the knotaxane cannot be forecasted. Therefore, different axles and tethers had to be synthesized.

**Figure 2.** A double click reaction between macrocyclic rings carrying an azide function **a** and **a**- **,** and the alkyne ends **b** and **b** of the axle in a [3]pseudorotaxane (**left**) may generate a knotted [1]rotaxane (**top**, **right**) by reaction of **a** with **b** and **a** with **b**, respectively, or a handcuff molecule (**bottom**, **right**) by reaction of **a** with **b** and **a** with **b**- , respectively.

#### **2. Materials and Methods**

#### *2.1. Methods*

1H and 13C NMR spectra were recorded with Bruker DRX 500 MHz or Bruker Avance 600 MHz spectrometers (Bruker, Billerica, MA, USA) Mass spectrometric analysis was performed with AccuTOFGCv4G (HR-MS, electron ionization) from Jeol (Tokyo, Japan); a Q Exactive Plus mass spectrometer (HR-MS, electrospray ionization, positive mode) from Thermo Scientific (Waltham, MA, USA); and Autoflex speed (MALDI) from Bruker (Billerica, MA, USA). HPLC-MS experiments were carried out with VWR-Hitachi HPLC system Elite LaChrom coupled to an expression CMS mass spectrometer (electrospray ionization, positive mode) from Advion (Ithaca, NY, USA). IR spectra were recorded with Perkin–Elmer Spectrum100 FT-IR spectrometer (Perkin-Elmer, Waltham, MA, USA). The spectra were recorded using a MKII Golden GateTM Single Reflection ATR A531-G system from Specac (Orpington, UK). The elemental analyses were performed with the CHNS-O elemental analyzer EURO EA 3000 Series from Euro Vector (Pavia, Italy) and vario MICRO CUBE from Elementar at the Institut für Anorganische Chemie of the Christian-Albrechts-Universität zu Kiel (Germany). For this purpose, the samples were burned in zinc containers in a stream of oxygen.

#### *2.2. General Synthetic Procedure*

The detailed synthetic procedures and all analytical data for molecules not known in the literature are described in the Supplementary Materials. General procedures:

#### 2.2.1. Reductive Amination of Aldehydes

Under a nitrogen atmosphere, a diamine (1,8-diaminooctane, 1,10-diaminodecane) and two equivalents of aldehyde **1** were dissolved in dry methanol. Sodium borohydride was added in portions while cooling with an ice bath followed by stirring at room temperature. Water was added and the solvent was removed in vacuo. The aqueous layer was extracted three times with dichloromethane. The combined organic layer was dried with magnesium sulfate and the solvent was removed in vacuo. Pale yellow solid diamines **2a**/**b** were obtained in good yields. Diamine **6** was synthesized analogously by starting with one equivalent of **4** and two equivalents of amine **5**.

#### 2.2.2. Protonation and Ion Exchange

The secondary amines **2a**/**b** or **6** were dissolved in ethanol and then mixed with concentrated aqueous hydrochloric acid. The suspensions were stirred at room temperature and then filtered. The hydrochlorides were suspended in acetone and mixed with a saturated aqueous ammonium hexafluorophosphate solution. The solvent was removed in vacuo and the residue was stirred in water. The colorless solids **3a**/**b** and **7** were filtered and dried in vacuo.

#### 2.2.3. Click Reaction to Form [3]rotaxanes

Under nitrogen atmosphere, an axle **3a**/**b** or **7** and dibenzo-24-crown-8 (DB24C8, **13**) were suspended in dry dichloromethane. The mixture was stirred at room temperature until a clear solution was obtained. Then, stopper **11**, tetrakis(acetonitrile)copper(I) hexafluorophosphate and 2,6-dimethylpyridine were added. The solution was stirred for 1 d at room temperature and then water was added. The aqueous layer was extracted with dichloromethane and the solvent was removed in vacuo. The residue was filtered through silica gel (dichloromethane:methanol (80:20)). The solvent of the filtrate was removed in vacuo and the residue was purified by chromatography on silica gel (dichloromethane:methanol (1:0) → (24:1)). The resulting yellow solid was dissolved in ethyl acetate and excess DB24C8 (**13**) precipitated. The solid was filtered off and the process was repeated. Finally, after concentration, the residue was again purified by chromatography on silica gel (dichloromethane:methanol (1:0) → (24:1)). Pale yellow solid rotaxanes **14a**/**b** or **16** were obtained.

#### 2.2.4. Methylation of the Triazole

A rotaxane **14a**/**b** or **16** was dissolved in methyl iodide and stirred for 4 days at room temperature. The solvent was removed in vacuo and the residue was dissolved in dichloromethane. An identical volume of saturated aqueous ammonium hexafluorophosphate solution was added and the mixture was stirred vigorously. The aqueous layer was extracted with dichloromethane. The organic layer was dried with sodium sulfate and filtered, and the solvent was removed in vacuo. Pale yellow solids **15a**/**b** or **17** were obtained in good yields.

#### 2.2.5. Switching of the Rotaxanes (Deprotonation and Protonation)

*Deprotonation*: The rotaxane **15a**/**b** or **17** was dissolved in chloroform (5 mL), mixed with 1 n sodium hydroxide solution (5 mL) and shaken vigorously. The layers were separated, the organic layer was dried with sodium sulfate and the solvent was removed in vacuo. Pale yellow solids were obtained.

*Protonation:* The residue of the deprotonation experiment was dissolved in chloroform (5 mL) and treated with a 0.2 n solution of trifluoroacetic acid in chloroform (3 mL). The solution was shaken vigorously and was then mixed with a saturated, aqueous ammonium hexafluorophosphate solution (3 mL). The mixture was again shaken vigorously followed by separation of the layers. The organic layer was dried with sodium sulfate, and the solvent was removed in vacuo. Pale yellow solids were obtained.

#### 2.2.6. Etherification of 4-Benzyloxyphenol (**25**)

Under nitrogen, 4-benzyloxyphenol (**25**) and potassium carbonate were suspended in acetonitrile and a dibromoalkane was added. The suspension was stirred under reflux for 20 h and then filtered. The solvent was removed in vacuo and the residue was dissolved in dichloromethane. The organic layer was washed with an aqueous sodium hydroxide solution (10%) several times, dried with magnesium sulfate and filtered. The solvent was removed in vacuo and the residue was recrystallized from *n*-hexane. Colorless solids **26a**–**d** were obtained.

#### 2.2.7. Reductive Deprotection of the Benzyl Protecting Group

Under a hydrogen atmosphere, the benzyl-protected hydroquinone derivative (**26a–d**) was dissolved in chloroform, and palladium on activated carbon (10% Pd content, 0.1 equivalent) was added. The suspension was stirred for 16 h at room temperature, then filtered through Celite, and the solvent was removed in vacuo. Pale grey solids **27a**–**d** were obtained.

#### 2.2.8. Substitution of Bromide by Azide

A bromide (**27a–d**) was dissolved in dimethyl sulfoxide, sodium azide was added and the mixture was stirred at room temperature for 18 h. Water was added, and the aqueous layer was extracted with diethyl ether. The organic layer was dried with magnesium sulfate and filtered, and the solvent was removed in vacuo. The crude product was purified by chromatography on silica gel [dichloromethane → dichloromethane:methanol (93:7)]. Colorless oils **28a**–**d** were obtained.

#### **3. Results and Discussion**

#### *3.1. Syntheses of the Axles*

All dialkyne substituted central axles contain two secondary amine functions. These were synthesized by condensation of aldehydes with primary amines followed by reduction.

Alkyne substituted aldehyde **1** was synthesized following a literature-known synthesis [12]. In a ratio of 2:1, it was reacted with commercially available diamines (1,8-diaminooctane and 1,10-diaminodecane) (Figure 3). Quickly, the diimines precipitated as colorless solids. The imines were not isolated, but were directly reduced using an excess of sodium borohydride to obtain the secondary amines **2a** or **2b**. **2a**/**b** were obtained in very good yields after aqueous work-up and without further purification. Next, the secondary amines **2a**/**b** were dissolved in ethanol and mixed with concentrated hydrochloric acid. The hydrochlorides precipitated directly. In a mixture of acetone/water, the chloride ions were then exchanged by hexafluorophosphate ions in good yields.

**Figure 3.** Synthesis of the axles **3a**/**b**. (a) 1. 1,8-diaminooctane or 1,10-diaminodecane, MeOH, 2 h, r.t.; 2. NaBH4, 16 h, 95% (**2a**), 90% (**2b**). (b) 1. HCl, EtOH, 2 h, r.t.; 2. NH4PF6, acetone/water, 2 h, r.t., 74% (**3a**), 86% (**3b**).

The starting materials for the synthesis of axle **7**, dialdehyde **4** and amine **5**, were prepared following synthetic procedures from the literature [13–15]. As in the reactions to give **2a**/**b**, the reductive amination of **4** was performed with sodium borohydride in methanol (Figure 4). Due to the poor solubility of dialdehyde **4**, the reaction was carried out under reflux. The secondary amine **6** was obtained after aqueous work-up in a yield of 89%. Chromatographic purification was not possible. Therefore, the crude diamine **6** was directly protonated with hydrochloric acid to give the ammonium salt. The dihydrochloride was obtained after precipitation from ethanol in a yield of 87%. The subsequent ion exchange was again carried out in a solvent mixture of acetone and water with ammonium hexafluorophosphate. The bishexafluorophosphate salt **7** was obtained with a yield of 99%.

**Figure 4.** Synthesis of axle **7**. (a) MeOH, 3 h, reflux; 2. NaBH4, 16 h, 89%. (b) 1. HCl, EtOH, 2 h, r.t.; 2. NH4PF6, acetone/water, 2 h, r.t., 99%.

#### *3.2. Syntheses of the [3]rotaxanes*

In order to synthesize a rotaxane by the capping method, an axle, a macrocycle and stoppers are needed. In this work, the capping was performed by a "click" reaction between the dialkyne terminated central axles and stoppers carrying an azide function. Azide **11**, whose synthesis is literature-known [16,17], was chosen as the stopper (Figure 5). It was synthesized from the respective trityl alcohol **8**, first connecting it with phenol, and then forming the aryl alkyl ether **10**. The synthesis of the phenol is possible following two different routes: direct reaction of the trityl alcohol **8** with phenol under acidic conditions or generation of the trityl chloride **12** first. It turned out that the route via the chloro intermediate **12** gave better yields [18]. Finally, bromide **10** was converted to azide **11** in 90% yield by reaction with sodium azide.

**Figure 5.** Synthesis of stopper **11**. (a) Phenol, HCl, 24 h, 160 ◦C, 63%; (b) 3-bromopropanol, PPh3, DIAD, THF, 17 h, r.t., 73%; (c) acetyl chloride, toluene, 1 h, reflux, 92%; (d) phenol, HCl, 4 d, 120 ◦C, 98%; (e) NaN3, DMF, 20 h, 80 ◦C, 90%.

For a successful synthesis of [3]rotaxanes, it is beneficial to generate a pseudorotaxane first before stoppering takes place. For this purpose, the axles **3a**/**b** or **7** and two equivalents of macrocycle **13** were mixed in dichloromethane. Due to the insolubility of the diammonium ions in non-polar solvents, first, a suspension was formed. Upon stirring, the mixture turned into a clear solution, indicating that the pseudorotaxanes formed. Then, two equivalents of stopper **11**, a copper(I) salt as a catalyst for the CuAAC "click" reaction and a catalytic amount of 2,6-dimethylpyridine were added. All reactions were carried out in the non-polar solvent dichloromethane to allow strong ion-dipole interactions and hydrogen bonds between the ammonium ions of the axles and the macrocycles. After hydrolytic work-up and extraction, the copper salt was removed by filtration and the product was purified by chromatography on silica gel. Remaining stopper **11** could be removed, indicating that the conversion had not been complete. By addition of ethyl acetate, excess crown ether precipitated. This process was repeated several times until no more solid precipitated. Finally, the products were purified again by chromatography on silica gel.

[3]Rotaxane **14a** was isolated as a pale yellow solid in a yield of 34% (Figure 6). [3]Rotaxane **14b** was synthesized analogously. NMR analysis of **14b** showed more than two crown ethers per rotaxane. Two sets of NMR signals for the macrocycle were found in an approximate ratio of 2:1, both sets of signals differing from those of the free macrocycle DB24C8 (**13**). Due to overlapping signals, the amount of the additional macrocycle cannot be determined precisely. As discussed below, additional macrocycle was also found in the next product, the dimethylated [3]rotaxane **15b**. After deprotonation, the amount of excess **13** was determined to be one equivalent per rotaxane. A possible explanation for additional macrocycle **13** in the rotaxanes **14b** and **15b** could be that this macrocycle adheres to the outside of the [3]rotaxanes **14b** and **15b.** The interaction to rotaxane **14b** must have been strong enough to survive the chromatography. [3]Rotaxane **14b** was obtained in a yield of 39%.

**Figure 6.** Syntheses of the rotaxanes **15a**/**b**. (a) CH2Cl2, Cu(MeCN)4PF6, 2,6-dimethylpyridine, 1 d, r.t., 34% (**14a**), 39% (**14b**); (b) MeI, 4 d, r.t,. quantitative (**15a**), 98% (**15b**).

Due to the two positive charges of the two ammonium functions, the [3]rotaxanes **14a**/**b** possess one binding site for each of the two macrocycles. Since, as discussed in the introduction, the rotaxanes should be switched by addition of a base or an acid, additional binding sites had to be introduced; i.e., by methylation of the triazole rings. The resulting triazolium ions are permanently positively charged and will serve as secondary binding sites, and the only ones after deprotonation. The bis(triazolium ion) **15a** was formed by reacting [3]rotaxane **14a** with excess methyl iodide. After aqueous work-up, methylated [3]rotaxane **15a** was obtained in quantitative yield. The same conditions were used for the analogous [3]rotaxane **14b**. The methylated [3]rotaxane **15b** was obtained in a yield of 98%. Unfortunately, the excess crown ether in the **14b** sample could also not be separated from the dimethylated product **15b**.

Axle **7** differs from the axles **3a**/**b** in the substitution pattern of the ammonium ions: dibenzyl versus benzyl-alkyl. Using the same synthetic procedure as for the rotaxanes **14a**/**b**, [3]rotaxane **16** with two benzyl units in the proximity of each of the ammonium ions was obtained in a yield of 6% after purification (Figure 7). NMR spectroscopy showed again that the ring starting material, crown ether **13**, also could not be completely separated from **16**. The subsequent methylation of **16** was carried out identically to the methylation of **14a**/**b**. After aqueous work-up, the methylated [3]rotaxane **17** was obtained with a quantitative yield. Again, the excess crown ether was detected and could not be removed.

**Figure 7.** Synthesis of rotaxane **17**. (a) CH2Cl2, Cu(MeCN)4PF6, 2,6-dimethylpyridine, 1 d, r.t., 6%; (b) MeI, 4 d, r.t. quantitative.

#### *3.3. Switching of the [3]rotaxanes*

Having introduced the additional binding sites by methylation of the triazoles, the base/acid switching of the methylated [3]rotaxanes **15a**/**b** and **17** was investigated. By addition of sodium hydroxide, the ammonium ions were deprotonated, leaving only the triazolium ions as positively charged binding sites. Protonation by trifluoroacetic acid regenerated the ammonium functions. The switching process was followed by NMR spectroscopy.

#### 3.3.1. Switching of [3]rotaxane **15a**

In the aromatic region of the 1H NMR spectrum of rotaxane **15a**, deprotonation caused a low field shift of almost 0.6 ppm for triazolium proton at 8.51 ppm (Figure 8, green). Binding of the crown ether to the triazolium ion leads to a deshielding. All other aromatic signals show much smaller shift changes, but no signal has an identical chemical shift as in the previous spectrum (Figure 8, blue). The multiplet at 6.8 ppm has split to become three individual signals. In the aliphatic range, more distinct differences can be observed. The two methylene groups (5.22 ppm and 4.76 ppm) near the triazolium unit exchange their positions. Additionally, the methylene groups at 3.0 ppm and 2.5 ppm switch positions. In this region, most signals experience a high field shift after deprotonation, except for the signal at 3.4 ppm, which has low field shifted. In the deprotonated [3]rotaxane (Figure 8, green), the signals for the central methylene groups appear between 1.2 and 1.5 ppm, as expected for oligomethylene H atoms. In the protonated cases (Figure 8, blue and red), however, these signals are high field shifted (1.3, 0.8 and 0.75 ppm), indicating the vicinity of the aromatic benzene rings of the macrocycles to these methylene groups.

**Figure 8.** 1H NMR spectra (500 MHz, CDCl3) of the base/acid switching of rotaxane **15a**. Blue: before; green: after deprotonation; red: after protonation.

#### 3.3.2. Switching of [3]rotaxane **15b**

Analogous to the previous experiment, a pH-dependent NMR investigation was also carried out with rotaxane **15b** (Figure 9). In the aromatic range, the triazolium proton (8.51 ppm) shows a low field shift of almost 0.4 ppm after deprotonation. Compared to the switching of the analogous [3]rotaxane **15a,** this shift is smaller by 0.2 ppm. Reasons for this difference are unclear. But overall, the same trend can be observed in the aromatic sector as for [3]rotaxane **15a**. In the aliphatic sector, changes similar to those observed with rotaxane **15a** can be observed as well. Additionally, the two signals of the methylene

groups near the triazolium group exchange positions. It is noticeable, however, that signals which can be assigned to the free macrocycle **13** are observed after deprotonation (two centered multiplets at 4.14 and 3.92, and a singlet at 3.84 ppm). This supports the postulated formation of a complex that was formed during the synthesis of rotaxane **14b**. Deprotonation released the ring from the complex, resulting in the observation of signals for the free ring **13**. After reprotonation, the signals of the free macrocycle **9** cannot be detected anymore. This supports the assumption that the free macrocycle **13** is attached to [3]rotaxane **15b** in the protonated form. After protonation, all other signals are also found at their original positions. Repeated deprotonation/protonation cycles resulted in the identical changes of the chemical shifts.

**Figure 9.** 1H NMR spectra (500 MHz, CDCl3) of the base/acid switching of rotaxane **15b**. Blue: before; green: after deprotonation (the arrows highlight the signals of additional crown ether **13**); red: after protonation.

#### 3.3.3. Switching of [3]rotaxane **17**

Additionally, dimethylated [3]rotaxane **17** was deprotonated and reprotonated, and the switching was investigated by 1H NMR spectroscopy (Figure 10). Additionally, in this case, the spectra are in accordance with a base/acid shuttling of the rings from the ammonium ions to the triazolium ions and back. For the proton of the triazolium ion, a low field shift of the signal from 8.49 to 9.10 ppm was detected. The changes in the aromatic and the aliphatic regions of the spectra are comparable to what was observed when rotaxanes **15a** and **15b** were switched. For example, again, the methylene groups near the triazolium ions reverse positions, from 5.21 or 4.75 ppm to 4.73 or 5.16 ppm, respectively. The signals of the methylene groups near the ammonium ions exhibit a high field shift from 4.48 to 3.72 ppm. Additionally, in this case, signals for free macrocycle (**13**) can be observed after deprotonation (4.16, 3.90 and 3.83 ppm).

**Figure 10.** 1H NMR spectra (500 MHz, CDCl3) of the base/acid switching of rotaxane **17**. Blue: before; green: after deprotonation (the arrows highlight the signals of additional crown ether **13**); red: after protonation.

After reprotonation and counter ion exchange, all signals are back at their original positions, and no more free macrocycle **13** can be detected, arguing for a strong binding of an additional macrocycle to the "outside" of the [3]rotaxane **17**, its release in basic media and re-binding in acidic media.

#### *3.4. Synthesis of Crown Ether DB24C8-CH2Br (***24***)*

In the envisaged knotted [1]rotaxane, no separate stoppers are needed, as the macrocycles themselves prevent the slipping-off. For this purpose, the macrocyclic rings must be connected to azide functions which are then reacted with the alkynes by the CuAAC "click" reaction. An obvious position to connect an azide tether to the macrocycle is position 4 of one of the benzene rings of DB24C8 (**13**). The synthesis of a respective hydroxymethylated DB24C8 **23** is described in the literature [19–21]. Tosylation of triethylene glycol (**18**), connection with catechol, tosylation of the remaining hydroxyl groups of **20**, macrocyclization with methyl 3,4-dihydroxybenzoate and reduction of the ester group of **22** were reproduced in good yields (total yield: 68% over five steps) (Figure 11). Finally, the hydroxysubtituted macrocycle **23** was converted to bromide **24** in quantitative yield using phosphorus tribromide.

**Figure 11.** Synthesis of the bromomethyl macrocycle **24**. (a) *p*TsCl, NaOH, THF/H2O, 18 h, r.t., 91%; (b) 1,2-dihydroxybenzene, K2CO3, MeCN, 60 h, reflux, 98%; (c) *p*TsCl, NaOH, THF/H2O, 18 h, r.t., 90%; (d) methyl 3,4-dihydroxybenzoate, K2CO3, MeCN, 60 h, reflux, 86%; (e) LiAlH4, THF, 2 h, reflux, 99%; (f) PBr3, CH2Cl2, 2 h, r.t., quant.

#### *3.5. Syntheses of Azide Substituted Phenols* **28a–d**

Next, the azide tethers had to be synthesized. The three step syntheses of azides **28a–d** begin with an ether formation to give mono-protected hydroquinone **25** (Figure 12). For this purpose, conditions from the literature were chosen [22]. The resulting 4-(benzyloxy)phenol (**25**) was dissolved in acetonitrile and was reacted under alkaline conditions with an excess of dibromoalkanes. After alkaline work-up, bromoalkyl ethers **26a–d** were obtained with yields of 14% to 75%. Deprotection of the benzyl ethers **26a–d** adapting a literature protocol [23] yielded the phenols **27a–d** in yields of 65% to quant. The work-up was easy. Filtration of the suspensions through Celite and evaporation of the solvent yielded pure products **27a–d**; no further purification was needed. In the last step, the phenols **27a–d** were dissolved in dimethyl sulfoxide and sodium azide was added. After stirring at room temperature for 18 h and chromatographic purification, azides **28a–d** were obtained in yields of 65%–93% which is comparable to the literature yield of the already known azide **28a** (67%) [24].

**Figure 12.** Syntheses of the azide compounds **28a–d**. (a) Dibromoalkane, MeCN, K2CO3, 20 h, reflux, 14% (**26a**), 46% (**26b**), 36% (**26c**), 75% (**26d**); (b) Pd/C, H2, CHCl3, 18 h, r.t, 94% (**27a**), 98% (**27b**), 65% (**27c**), quant. (**27d**); (c) NaN3, DMSO, 18 h, r.t., 93% (**28a**), 69% (**28b**), 65% (**28c**), 90% (**28d**).

All azides **28a–d** possess oligomethylene chains. To allow a different rotational flexibility in the chain, an additional azide **32** with a triethylene glycol chain could be made. The length of the tether in **32** is identical to that of the octamethylene derivative **28d,** but two methylene groups are replaced by oxygen atoms. The first three reactions were described in the literature [25–27]. In the last step, a substitution reaction with sodium azide gave azide **32** in a yield of 84% (Figure 13).

**Figure 13.** Synthesis of the azide **32**. (a) **19**, DMF, 90 ◦C, 24 h, 61%; (b) *p*TsCl, NEt3, DMAP, CH2Cl2, 24 h, r.t., 86%; (c) Pd/C, H2, CHCl3, 16 h, r.t., 97%; (d) NaN3, DMF, 2 d, 70 ◦C, 84%.

#### *3.6. Syntheses of the Azide Containing Crown Ethers*

The final reaction to give the azide substituted macrocycles **33a–e** was the connection of the macrocyclic bromide precursor **24** with the azide-terminated phenols **28a–d** and **32**. Cesium carbonate in acetone was used as deprotonating reagent. The yields varied strongly. For the aliphatic azides **33a–d**, the yields were between 26% and 75% (Figure 14). It shall be noted that in all four reactions a double purification had to be performed to obtain the crown ethers in crystalline form. Chromatography was performed first followed by crystallization. Macrocycle **34** containing a triethyleneglycol chain was obtained after recrystallization in a yield of 37% (Figure 15).

**Figure 14.** Synthesis of azide-terminated macrocycles **33a–d**. (a) DB24C8-CH2Br (**24**), Cs2CO3, KI, acetone, 16 h, reflux, 57% (**33a**), 34% (**33b**), 75% (**33c**), 26% (**33d**).

**Figure 15.** Synthesis of the azide-terminated macrocycle **34**. (a) DB24C8-CH2Br (**24**), Cs2CO3, KI, acetone, 16 h, reflux, 37%.

#### *3.7. Knotting Attempts*

For first orientational experiments, two of the azides (**33c**/**d**) were selected, primarily due to their better availabilities. First, an axle and the macrocycles were dissolved in dichloromethane and pseudorotaxane formation was allowed. Then, the resulting solution was added very slowly to a highly diluted solution of copper(I) hexafluorophosphate and 2,6-dimethylpyridine. The resulting suspension was filtered through silica gel. After evaporation of the solvent, the residue was analyzed by MALDI mass spectrometry. The results for azide **33c** are shown in Table 1. No mass signal matching a double click reaction could be found when axles **3a** or **3b** were used. But with axle **7**, signals in the mass spectrum could be observed which were in accordance to a successful double click reaction. But please note that mass spectra cannot differentiate between the desired knotted [1]rotaxane and the alternative product in which the azides and alkynes have reacted "wrongly" with one another (handcuff, see Figure 2).

**Table 1.** MALDI mass spectra signals for the "click" reactions with azide-terminated macrocycle **33c**.


Azide-terminated macrocycle **33d** was also investigated. The knotting attempts were performed using the optimized conditions of the previous experiments. After purification on silica gel, mass spectra were also recorded (Table 2). In all experiments, signals for doubly charged ions without counter ions for 2:1 adducts could be observed. This indicates a successful reaction, and thus the formation of a [1]rotaxane. Again though, the question of whether the [1]rotaxane is knotted or not cannot be answered from these data.

**Table 2.** ESI mass spectral signals for the experiments with the macrocycle **33d**.


Next, the triethyleneglycol analog of **33d**, **34**, was tested by applying the same reaction conditions. Additionally, with **34**, signals matching a [1]rotaxane were found in the MS (**3a**•2 **34** − 2 PF6 <sup>−</sup> <sup>−</sup> H+, *m*/z = 1889.2).

In summary, with azide-terminated macrocycle **33c**, a mass signal corresponding to a [1]rotaxane was only found in one case, but the longer tethers (**33d** and **34**) showed respective signals in all tested combinations.

The reaction mixtures were then investigated by reverse-phase HPLC coupled with ESI-MS. The HPLC runs revealed numerous reaction products probably also arising from intermolecular "click" reactions. The fact that no mass spectral signals were found for larger parts of the HPLC traces argues for oligomers. However, also in the HPLC runs, mass spectral signals corresponding to the products of a double "click" reaction could be found.

#### **4. Conclusions**

Dialkyne-substituted axles which contain two ammonium ions as primary binding sites for macrocyclic crown ethers can be used to form [3]pseudorotaxanes with these rings followed by locking of the rotaxane structure by two-fold copper-catalyzed "click" reaction between the alkynes and azides. In the [3]rotaxanes, the resulting triazoles can be methylated, which introduces a second type of binding site. Upon deprotonation and reprotonation, the macrocycles can be shuttled between the two binding sites, the ammonium ions and the triazolium ions. The click reaction is also possible when the azide function is connected to the DB24C8 macrocycle. But the isolation of the coupling products

from by-products is still a challenge. Preparative HPLC might be the solution. Then, whether the applied dimensions of the tethers and the axles were allowing the formation of the desired knotaxane or whether axles and tethers have to be modified (other lengths and rigidity) can be studied.

**Supplementary Materials:** Supplementary material with synthetic procedures and analyses is available online at http://www.mdpi.com/2624-8549/2/2/305\T1\textendash321/s1.

**Author Contributions:** Conceptualization, U.L. and T.D.; methodology, U.L. and T.D.; validation, U.L. and T.D.; formal analysis, T.D.; investigation, T.D.; resources, U.L.; writing—original draft preparation, U.L. and T.D.; writing—review and editing, U.L. and T.D.; visualization, T.D.; supervision, U.L.; project administration, U.L.; funding acquisition, U.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We would like to thank Tobias Paschelke, and Sven Schultzke, for their synthetic support in the production of the azide containing phenols. Furthermore, we thank Dennis Stöter, for the support in the syntheses of the axles. We would also like to thank Vanessa Nowatschin, for her support in the optimization of the reproduction of the stopper.

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

#### **References and Note**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Addition of Heteroatom Radicals to** *endo***-Glycals** †

#### **Torsten Linker**

Department of Chemistry, University of Potsdam, Karl-Liebknecht-Str. 24-25, 14476 Golm, Germany; linker@uni-potsdam.de; Tel.: +49 331 9775212

† Dedicated to Bernd Giese on the occasion of his 80th birthday and his pioneering work on radicals in carbohydrate chemistry.

Received: 6 February 2020; Accepted: 18 February 2020; Published: 20 February 2020

**Abstract:** Radical reactions have found many applications in carbohydrate chemistry, especially in the construction of carbon–carbon bonds. The formation of carbon–heteroatom bonds has been less intensively studied. This mini-review will summarize the efforts to add heteroatom radicals to unsaturated carbohydrates like *endo*-glycals. Starting from early examples, developed more than 50 years ago, the importance of such reactions for carbohydrate chemistry and recent applications will be discussed. After a short introduction, the mini-review is divided in sub-chapters according to the heteroatoms halogen, nitrogen, phosphorus, and sulfur. The mechanisms of radical generation by chemical or photochemical processes and the subsequent reactions of the radicals at the 1-position will be discussed. This mini-review cannot cover all aspects of heteroatom-centered radicals in carbohydrate chemistry, but should provide an overview of the various strategies and future perspectives.

**Keywords:** radicals; carbohydrates; heteroatoms; synthesis

#### **1. Introduction**

Radical reactions of carbohydrates are important for chemistry, biology, and medicine. For example, free radicals are involved in several biosynthetic pathways or are used for cancer-treatment [1–4]. On the other hand, radiation can cause DNA strand break by H atom abstraction and radical generation at the sugar backbone [5–7]. Bernd Giese's group proved that radical cations are involved in the mechanism [8] and investigated how the positive charge is transferred through the DNA [9]. For synthetic applications, Bernd Giese's group also made carbohydrate radicals available for the formation of carbon–carbon bonds under mild conditions [10]. Due to steric interactions, carbohydrates provide high stereoselectivities [11,12], and the importance of such reactions has been reviewed many times [13–18]. Furthermore, under appropriate conditions, 2-deoxy sugars can be synthesized from bromo sugars in only one step [19].

To develop efficient radical reactions, it is important to understand the reactivities of the corresponding radicals [12,20,21]. Thus, such reactive species can have a nucleophilic or an electrophilic character [22], which controls their addition to alkenes. Applied to the anomeric center of carbohydrates, the radicals **1** exhibit a nucleophilic character due to the adjacent oxygen atom, and add preferentially to electron poor double bonds with electron withdrawing (EWG) and nondonating (EDG) groups (Scheme 1a). If the carbohydrate is used as radical acceptor, unsaturated carbohydrates like *endo*-glycals **2** become attractive substrates, which can be easily synthesized on a large scale [23]. However, once the double bond becomes electron rich the reaction proceeds only with electrophilic radicals (Scheme 1b).

**Scheme 1.** Examples for radical reactions in carbohydrate chemistry: (**a**) starting from a carbohydrate radical **1** and (**b**) radical additions to glycals **2**.

During the last 25 years, we developed C-C bond formations by the addition of electrophilic radicals, mainly derived from malonates, to glycals **2** with various further synthetic transformations [24–27]. However, the addition of heteroatom radicals to glycals is very attractive as well, since such radicals exhibit the required electrophilic character [22,28]. On the other hand, heteroatom radicals are prone to undergo H atom abstraction, which is problematic in carbohydrate chemistry with various functional groups. Thus, although alkoxyl radicals [29] can be generated from carbohydrates and undergo fast fragmentations [30,31], of the way in which such radicals can be added to glycals is unknown. Since halogen atoms, nitrogen-, phosphorus-, and sulfur-centered radicals are less prone to undergo H atom abstraction, this mini-review will focus on the addition of such radicals to *endo*-glycals.

#### **2. Addition of Halogen Atoms**

The halogenation of *endo*-glycals **2** is one of the oldest transformations of such unsaturated carbohydrates, already described by Lemieux in 1965 [32]. Thus, tri-*O*-acetyl-d-glucal (**2a**) or the corresponding isomer tri-*O*-acetyl-d-galactal (**2b**) reacted with chlorine or bromine in high yields to the main products **3a** and **3b** (Scheme 2).

**Scheme 2.** Halogenation of glycals **2a** and **2b**.

Although the authors proposed an ionic pathway via halonium ions, the 1,2-*cis*-configurations might be explained by homolysis of the labile halogen bonds and addition of the resulting electrophilic radicals. To distinguish between a radical or an ionic pathway, halogen azides are attractive precursors because they easily undergo homolysis and are used in regio- and stereoselective syntheses [33]. Thus, reaction of tri-*O*-acetyl-d-glucal (**2a**) with chlorine azide afforded regioisomers **4** and **5**, depending on the reaction conditions (Scheme 3) [34].

**Scheme 3.** Reaction of tri-*O*-acetyl-d-glucal (**2a**) with chlorine azide.

In the dark, 2-chloro-2-deoxy sugars **4** were isolated as main products, whereas irradiation gave the 2-azido-2-deoxy isomer **5**. Such different regioselectivities were explained by an ionic pathway via chloronium ions **6a** in the dark and a radical mechanism during irradiation via radical **7**. However, it was not possible to add the generated chlorine atom to the glucal, because the azide radical is more reactive (see Section 3).

More recently, Vankar developed a reagent system based on oxalyl chloride and silver nitrate to activate the carbon-chlorine bond [35]. An intermediate **8** was proposed, which cleaves into nitrate and carbon monoxide and transfers chlorine to the double bond of tri-*O*-acetyl-d-galactal (**2b**). The chloronium ion **6b** is subsequently trapped by the solvent acetonitrile/water to afford the 2-chloro-2-deoxy sugar **9** in high yield (Scheme 4).

**Scheme 4.** Reaction of tri-*O*-acetyl-d-galactal (**2b**) with oxalyl choride/silver nitrate.

Compared to 2-chloro derivatives, the corresponding iodides are even more attractive because the carbon-iodine bond can be easily reduced to 2-deoxy sugars, important building blocks for carbohydrate chemistry. Thus, various strategies have been developed by oxidation of iodides by hypervalent iodine(iii) [36,37] or sodium periodate [38] in the presence of *endo*-glycals **2**, affording 2-iodo-2-deoxy sugars **10** in very good yields (Scheme 5). The mechanism proceeds by oxidation of iodide to iodine in the first step, formation of an iodonium ion similar to intermediate **6a** (Scheme 3), and trapping of the 1-position with the carboxylate with high 1,2-*trans* selectivity. In summary, halogen atoms can be easily introduced at the 2-position of glycals. However, the reactions proceed mainly by ionic pathways, and irradiation of halogen azides results in the formation of C-N bonds in the 2-position.

**Scheme 5.** Synthesis of 2-iodo-2-deoxy sugars **10** from *endo*-glycals **2**.

#### **3. Addition of Nitrogen-Centered Radicals**

In contrast to halogenations (chapter 2), the oxidative addition of azides to *endo*-glycals clearly proceeds by a radical pathway. The azidohalogenation was one example (Scheme 3); however, it is more attractive to generate the radicals by electron transfer. Cerium(iv) ammonium nitrate (CAN) is a very versatile single-electron oxidant, which can oxidize anions efficiently and has found many applications in organic synthesis [39]. We used this reagent for the generation of malonyl radicals and investigated the mechanism of their reactions with glycals **2** [25,26,40]. The pioneer Lemieux described the first application of azide oxidation in carbohydrate chemistry by addition of sodium azide to tri-*O*-acetyl-d-galactal (**2b**) in the presence of CAN (Scheme 6) [41].

**Scheme 6.** Addition of sodium azide to tri-*O*-acetyl-d-galactal (**2b**) in the presence of cerium(iv) ammonium nitrate (CAN).

In the first step, CAN oxidizes the azide anion to the corresponding radical, which has electrophilic character and adds to the double bond exclusively at the 2-position. The preferential formation of the equatorial product (only 8% of the *talo* isomer is formed as well) can be explained by the steric demands of the substituents in the 3- and 4-position. The resulting C-1 radical **11** is further oxidized by CAN to the carbenium ion **12**, which is finally trapped by the nitrate ligand from both faces to afford the 2-azido-2-deoxy sugar **13** in 75% yield. Thus, this addition is not a typical radical chain-reaction [13] because more than two equivalents of CAN are required.

The azidonitration of glycals was later extended to tri-*O*-acetyl-d-glucal (**2a**), but with lower stereoselectivity because not all substituents shield the same face. Furthermore, Paulsen [42] and Schmidt [43] found with this glucal **2a** different selectivities depending on the reaction conditions and temperature. However, up to now the azidonitration of glycals has been the best method to synthesize 2-amino sugars by simple reduction of the azide group, and has found many applications in carbohydrate chemistry, like in a very recent synthesis of a bisphosphorylated trisaccharide [44].

However, a disadvantage of the azidonitration of glycals is the lability of the nitrate group at the anomeric center, which can be easily hydrolyzed. Although it is possible to use glycosyl nitrates directly for glycosidations [45], they usually have to be transformed into suitable glycosyl donors. To overcome this problem, an interesting azidophenylselenylation has been developed [46–48]. Now, sodium azide is oxidized by (diacetoxyiodo)benzene to the corresponding radical, which adds regioselectively to glycals like tri-*O*-acetyl-d-galactal (**2b**) (Scheme 7). In the presence of diphenyldiselenide, the C-1 radical is trapped to afford directly selenoglycoside **14** in high yield and steroeselectivity [47].

**Scheme 7.** Azidophenylselenylation of tri-*O*-acetyl-d-galactal (**2b**).

An interesting intramolecular version of a radical C-N bond formation was developed by Rojas (Scheme 8) [49] In the first step, azidoformate **2c** reacts with FeCl2 under extrusion of nitrogen to intermediate **15**, which can be discussed as a Fe-complexed nitrogen-centered radical. Addition to the double bond affords C-1 radical **16**, which is trapped by chlorine to the labile complex **17**; after work-up, tricycle **18** is formed in moderate yield.

**Scheme 8.** Intramolecular addition of a nitrogen-centered radical **15**.

A similar intermolecular addition of hydroxylamines as radical precursors to glycals was described recently as well [50] In summary, the formation of C-N bonds in the 2-position of carbohydrates can be easily accomplished by the addition of nitrogen-centered radicals to glycals. The best method is azidonitration in the presence of cerium(iv) ammonium nitrate, or azidophenylselenylation, which has found many applications in carbohydrate chemistry.

#### **4. Addition of Phosphorus-Centered Radicals**

The reaction of phosphorus-centered radicals is well-established and has many synthetic applications, summarized in several reviews [28,51,52]. Because phosphorus can exist in different oxidation states, it is possible to generate phosphinyl, phosphinoyl, or phosphonyl radicals. Furthermore, the lability of the phosphorus-hydrogen bond allows for efficient chain-reactions with only catalytic amounts of radical initiator or under photochemical conditions. However, in contrast to nitrogen, only a few examples of the addition of phosphorus-centered radicals to glycals have been described in literature. Already in 1969, Inokawa demonstrated that diethyl thiophosphite reacts with unprotected glucal **2d** under UV irradiation with a high-pressure mercury lamp to the 2-deoxy-2-phosphorus analogue **21** in high yield and stereoselectivity (Scheme 9) [53].

**Scheme 9.** Addition of diethyl thiophosphite to d-glucal (**2d**) under irradiation.

After the radical initiation step, the thiophosphonyl radical **19** adds regioselectively to the 2-position of the carbohydrate, due to its electrophilic character. The resulting C-1 radical **20** abstracts a hydrogen atom from diethyl thiophosphite, regenerating the phosphorus-centered radical **19**, closing the chain.

A very similar approach with protected tri-*O*-acetyl-d-glucal (**2a**) was published more recently (Scheme 10) [54]. This time, the radical chain was initiated by triethylborane/air, which generates ethyl radicals, and the additions of diethyl thiophosphite and diethyl phosphite were realized. However, the reactions afforded products **22a** and **22b** in somewhat lower yields compared to the photochemical process.

**Scheme 10.** Addition of diethyl phosphites to tri-*O*-acetyl-d-glucal (**2a**) initiated by BEt3/air.

Recently, phosphinoyl radicals were added to tri-*O*-acetyl-d-glucal (**2a**) by a similar mechanism. The radicals were generated from diphenylphosphine oxide and manganese(ii) acetate and air, affording the 2-deoxy-2-phosphorus analogue **23** in high yield and stereoselectivity (Scheme 11). The authors could extend this reaction to various other *endo*-glycals **2** as well [55].

**Scheme 11.** Addition of diphenylphosphine oxide to tri-*O*-acetyl-d-glucal (**2a**) in the presence of manganese(ii) acetate and air.

However, all methods have the disadvantage that the 1-position is reduced under the reaction conditions. Therefore, we investigated the addition of dimethyl phosphite to various benzyl-protected glycals **2e** in the presence of cerium(iv) ammonium nitrate (CAN) (Scheme 11) [56]. Now, the C-1 radical is further oxidized to a carbenium ion (see Scheme 6), which is trapped by the solvent methanol, generating the anomeric center of carbohydrates. The yields of the 2-deoxy-2-phosphorus analogues **24** are good, but stereoisomers had to be separated. Subsequent Horner–Emmons reaction with benzaldehyde afforded unsaturated carbohydrates **25** as E/Z isomers in only one step (Scheme 12) [56].

**Scheme 12.** Addition of dimethyl phosphite to benzyl-protected glycals **2e** in the presence of CAN and subsequent Horner–Emmons reaction.

#### **5. Addition of Sulfur-Centered Radicals**

Sulfur-centered radicals can be easily generated from thiols by chemical or photochemical processes, because the S-H bond is much weaker than the corresponding O-H bond [28]. Subsequent addition to alkenes can initiate efficient chain reactions by hydrogen atom abstraction (thiol-ene reaction) or polymerizations. Indeed, the application of thiyl radicals in organic synthesis [52,57] or polymer chemistry [58] has been reviewed extensively. Even thio sugars are suitable radical precursors, and have been used for cyclizations and additions to other unsaturated carbohydrates at various positions [59,60] Therefore, this mini-review will focus only on the additions of sulfur-centered radicals to *endo*-glycals.

The first example of a C-S bond formation by radical addition to glycals was published in 1970 [61]. Thus, the chain-reaction was initiated by cumene hydroperoxide (CHP) with thioacetic acid as radical precursor. The 2-thiocarbohydrates **26** were isolated in high yields with the *manno* isomer **26a** as main product (Scheme 13).

**Scheme 13.** Addition of thioacetic acid to tri-*O*-acetyl-d-glucal (**2a**), initiated by CHP.

The addition of alkyl thiols to tri-*O*-acetyl-d-glucal (**2a**) was realized by photochemical initiation with acetone as sensitizer [62]. The 2-S analogues **27a** and **27b** were isolated in even higher yields but with lower stereoselectivities (Scheme 14).

**Scheme 14.** Photochemical addition of alkyl thiols to tri-*O*-acetyl-d-glucal (**2a**).

More recently, 2,2-dimethoxy-2-phenylacetophenone (DPAP **28**) became more attractive as radical initiator, which was developed for polymerizations and fragments under UV irradiation by an interesting mechanism (Scheme 15) [58]. Thus, in the first step a carbon–carbon bond is cleaved to generate a benzoyl radical **29**, which can abstract hydrogen atoms from thiols to initiate the chain reaction. The second dimethoxybenzyl radical **30** can fragment into benzoate **31** and methyl radicals **32**, which act as initiators as well.

 **Scheme 15.** Mechanism of the decomposition of 2,2-dimethoxy-2-phenylacetophenone (DPAP **28**).

Dondoni applied this initiator for the synthesis of *S*-disaccharides **33** [63]. Starting from thiosugar **34** and tri-*O*-acetyl-d-glucal (**2a**), the products **33** were isolated in high yield as a 1:1 mixture of epimers (Scheme 16).

**Scheme 16.** Addition of thiosugar **34** to tri-*O*-acetyl-d-glucal (**2a**), initiated by DPAP **28**.

In all reactions described above (Schemes 13–16), the C-S bond is formed selectively at the 2-position of the carbohydrates, due to the enol structure of the glycals. To obtain this bond at the 1-position of sugars, another strategy was developed by Borbas [64,65]. Thus, 2-acetoxy-3,4,6-tri-*O*-acetyl-d-glucal (**2f**) was used as radical acceptor, which reacted with various thiols in the presence of DPAP. Because of the additional oxygen substituent in the 2-position, orbital interactions allow the attack of electrophilic radicals from the 1- and 2-position. However, steric interactions result in the sole formation of 1-thiosugars **35** (Scheme 17, only one example with thiosugar **34** is shown).

**Scheme 17.** Addition of thiosugar **34** to 2-acetoxy-3,4,6- tri-*O*-acetyl-d-glucal (**2f**).

The addition of thiols by radical chain reactions to the 2-position of glycals has only one disadvantage: that the 1-position is reduced under the reaction conditions. Therefore, we investigated the oxidation of ammonium thiocyanate by cerium(iv) ammonium nitrate (CAN) and addition of the generated sulfur-centered radicals to various benzyl-protected glycals **2e** (Scheme 18) [66]. Similarly to the reaction of dimethyl phosphite (Scheme 12), the C-1 radical is further oxidized to a carbenium ion, which is trapped by the solvent methanol, generating the anomeric center of carbohydrates. The yields of the 2-deoxy-2-sulfur analogues **36** are moderate to good, but stereoisomers have to be separated. The thiocyanate groups can be cleaved to the corresponding thiols, which can bind to concanavalin A [66] or gold nanoparticles [67].

**Scheme 18.** Addition of thiocyanate to benzyl-protected glycals **2e** in the presence of CAN.

#### **6. Summary and Perspectives**

The addition of heteroatom radicals to glycals has been known for more than 50 years and has found various applications. The aim of this mini-review was to highlight early examples and discuss recent developments. Heteroatom radicals can be easily generated by initiators, photochemical processes, or by electron transfer. They exhibit electrophilic character and add regioselectively to the 2-position of the electron-rich double bond of *endo*-glycals. On the other hand, they are prone to H atom abstraction, which limits, especially for alkoxyl radicals, their applications in carbohydrate chemistry. The simple reaction of unsaturated sugars with halogens is possible, but proceeds mainly by ionic pathways. Nitrogen-centered radicals can be generated by oxidation of azides with cerium(iv) ammonium nitrate and add readily to glycals, which is still the best method to synthesize glycosamines. Phosphorus-centered radicals have been less intensively studied in carbohydrate chemistry, but addition products can be used for further transformations. On the other hand, the addition of sulfur-centered radicals to glycals has become very attractive for the synthesis of thio-disaccharides [68–72]. Photochemical initiators based on ketones have been developed for thiol-ene-reactions with unsaturated sugars, affording products in high yields. Finally, simple 2-thio sugars were synthesized by oxidation of thiocyanate and addition to glycals.

In conclusion, many methods for the introduction of heteroatoms in the 2-position of carbohydrates by radical processes exist in the literature. However, the addition to *endo*-glycals has been limited to nitrogen-, phosphorus-, or sulfur-centered radicals until now. Therefore, there is still space for new developments for other heteroatom additions, like boryl radicals, which can be easily generated [73–75], or future applications of such radical reactions in carbohydrate chemistry.

**Funding:** We acknowledge the support of the Open Access Publishing Fund of the University of Potsdam.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References and Notes**


© 2020 by the author. 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **OH Group E**ff**ect in the Stator of** β**-Diketones Arylhydrazone Rotary Switches** †

**Silvia Hristova 1, Fadhil S. Kamounah 2, Aurelien Crochet 3, Nikolay Vassilev 1, Katharina M. Fromm <sup>3</sup> and Liudmil Antonov 1,\***


Received: 14 April 2020; Accepted: 30 April 2020; Published: 4 May 2020

**Abstract:** The properties of several hydrazon-diketone rotary switches with OH groups in the stators (2-(2-(2-hydroxy-4-nitrophenyl)hydrazono)-1-phenylbutane-1,3-dione, 2-(2-(2-hydroxyphenyl)hydrazono)-1-phenylbutane-1,3-dione and 2-(2-(4-hydroxyphenyl)hydrazono)- 1-phenylbutane-1,3-dione) were investigated by molecular spectroscopy (UV-Vis and NMR), DFT calculations (M06-2X/TZVP) and X-ray analysis. The results show that, when the OH group is in *ortho* position, the E' and Z' isomers are preferred in DMSO as a result of a stabilizing intermolecular hydrogen bonding with the solvent. The availability, in addition, of a nitro group in *para* position increases the possibility of deprotonation of the OH group in the absence of water. All studied compounds showed a tendency towards formation of associates. The structure of the aggregates was revealed by theoretical calculation and confirmed by X-ray analysis.

**Keywords:** rotary switch; UV-Vis spectroscopy; NMR; DFT; X-ray; aggregation

#### **1. Introduction**

The hydrazone functional group has found extensive use in medicine [1–6], supramolecular chemistry (molecular switches and chelate ligands) [7–10] and in combinatory chemistry [11–13]. One important facet of hydrazine-group-containing compounds is the fact that upon appropriate substitution they can exist in solution as a mixture of isomers. 1,2,3-tricarbonyl-2-arylhydrazones are a typical example—they are presented in solution as an equilibrated mixture of intramolecularly H-bonded *E* and *Z* isomers [14–16]. The position of the isomerization equilibrium can be altered by catalytic amounts of acid or base. Upon external stimulation, a controlled switching between the isomers is possible through C-N bond rotation, giving the name "rotary switches". The position of the equilibrium and the switching can be strongly affected by structural modifications, as has already been shown [17–21].

Recently, the spectral properties of **1** (Schemes 1 and 2) were studied in respect of the possible tautomerism and *E*/*Z* isomerization in solution [17,22]. The results show that the availability of the OH group in the stator does not lead to azo-hydrazone tautomerism as could be expected at a first glimpse. The compound exists as a mixture of isomers of the single ketohydrazone tautomer, as shown in Scheme 1. In DMSO, due to the specific stabilizing effect of the solvent, only the *E'* and *Z'* forms are presented [22,23]. Moreover, the availability of the OH group leads to some side effects, according to the spectral and crystallographic data [22,24]: compound **1** deprotonates at low concentrations in DMSO and aggregates at high concentrations (10−<sup>4</sup> M and higher), forming linear (*E'*-*E'*) aggregates.

**Scheme 1.** Conformational isomers of **1**.

**Scheme 2.** Sketch of the investigated compounds.

It is an interesting question whether the lack of tautomerism, the existence of the"'" isomers (indicating *E'* and *Z'* forms) and the side effects could be attributed to the strong electron acceptor ability of the nitro group. To answer it, we have studied compounds **2** and **3,** in which two effects could be clarified: the role of the existence of a nitro group (**2** vs. **1**) and of the position of the OH group by itself (**2** vs. **3**). A combined approach (theoretical calculation, UV-Vis and NMR investigations in solution, and X-ray analysis in solid state) was applied, and the results obtained are presented in the current communication. To best of our knowledge, no such comparative study of **1**–**3** has been performed before.

#### **2. Results and Discussion**

The 1,2,3-tricarbonyl-2-arylhydrazones are potentially tautomeric compounds even without an OH group in the stator. The possible tautomers include ketohydrazone, azoketone and azoenol forms, depending on the substitution [25]. The availability, in addition, of an *ortho* or *para* OH group in the stator makes the tautomeric situation even more complex. The possible tautomers of **2** are sketched in Scheme 3 as an example. Theoretical prediction of the stabilities of the individual forms is complicated by the large number of possible conformers. For instance, the tautomer **I** of **2** can be presented as 24 possible isomers (Scheme SI).

**Scheme 3.** Sketch of the possible tautomeric forms of **2** (The same is valid for **1** and **3**).

We have shown theoretically that, in the case of **1**, only tautomer **I** could be present [22]. As shown in Table S1, the same conclusion can be drawn for **2** and **3**. The theoretical calculations are in agreement with the NMR data in solution [22] and the recent crystallographic data [21]. In analogy, compounds **2** and **3** also should exist as the same single tautomeric form, stabilized as a mixture of *E* and *Z* conformers. The most stable isomers of **2** and **3,** as predicted in DMSO as an environment, are shown in Figures 1 and 2, respectively.

**Figure 1.** Relative energies (in kcal/mole units) and predicted positions of the long-wavelength bands of the most stable isomers of **2**(**I**) in DMSO. The corresponding relative energies for **1**, taken from [24], are given in brackets.

**Figure 2.** Relative energies (in kcal/mole units) and predicted positions of the long-wavelength bands of the most stable isomers of **3**(**I**) in DMSO. The corresponding relative energies for **5**, taken from [24] and **4,** are given in brackets.

As seen from the figures, the stabilization is a result of the strength of the formed intramolecular hydrogen bonds. While better proton-attracting ability of MeCO through the NH..OMe determines a better stabilization of the *E* isomer, additional stabilization through OH..N bonding makes the *E*/*Z* pairs more stable compared to *E'*/*Z'* in **1** and **2**. The effect of the nitro group in **1** leads to an overall weak stabilization in the *E*/*Z* forms and a more pronounced stability of the"'" isomers. In the case of **3**, the effect of the OH group is limited to a non-hydrogen bonding substituent and leads to stabilization of the *Z* isomer. The predicted stabilization effect in the series **5** [22], **4** and **3** follows the experimentally observed trend for a destabilization of the *Z* isomer (molar fractions of 15%, 10% and 5%, correspondingly) going from electron acceptor to electron donor substituents in *para* position in the stator [19]. Most probably, the absence of the OH..N hydrogen bonding in E'/Z' of **2** and in **3** reduces the steric hindrance between the rotor and the stator, leading to an overall stabilization of the corresponding isomers.

The solvation model used so far describes the solvent as a dielectric medium and does not take into account the possible specific solute–solvent interactions. As known previously in the case of **1**, the proton of the OH group interacts with proton-acceptor solvents (such as DMSO), when it is not involved in the intramolecular hydrogen bond with the rotor part, which leads to strong additional stabilization of the " ' " isomers. The model of this specific solvent effect is illustrated in Figures 3 and 4, showing the most stable complexes with DMSO. As can be seen, the interaction between the solvent molecule and the free OH proton in *E'* and *Z'* leads to their stabilization. Moreover, in **2**E and **2**Z, there are no conditions for the formation of any OH..O = SMe2 hydrogen bond, and the formed NH ... O = SMe2 is weak due to the low acidity of the NH proton and steric effect from adjacent functional groups (Figure 3). The changes in the case of **3** are caused by reducing the electron donor ability of the OH group and hence to a rise in the polarization of the N-H bond, leading finally to the stabilization of the *Z* isomer.

**Figure 3.** Relative energies (in kcal/mole units) and predicted positions of the long-wavelength bands of the most stable isomers of **2**(I) in DMSO, accounting for the specific solute–solvent interactions.

**Figure 4.** Relative energies (in kcal/mole units) and predicted positions of the long-wavelength bands of the most stable isomers of **3**(I) in DMSO, accounting for the specific solute–solvent interactions.

In addition to the relative stability of the isomers, the predicted positions of the long-wavelength bands in the absorption spectra are shown in Figures 1–4 as well. The absolute values should be considered with care due to the systematic blue shift of the used M06-2X functional. The relative changes indicate, as expected, that it is practically impossible to distinguish between the most stable isomers by means of UV-Vis spectroscopy. The absorption spectra of **2** and **3**, shown in Figure 5, indicate that there are no substantial changes in the spectral shape upon changing the solvent. This figure strongly supports the hypothesis, in analogy to **1**; there is no tautomeric equilibrium, because the tautomers of **1**–**3** have different conjugated systems, and substantially different spectra could be expected upon changing the solvent [26]. It is seen that the observed long-wavelength absorption band consists of two sub-bands, whose intensity slightly varies with the solvent. They can be associated with the most stable isomers according to the theoretical predictions. However, the strong overlapping between them does not allow either precise estimation of the positions of the bands by derivative spectroscopy (Figure S1) nor a quantitative estimation of the isomers' molar fractions [27].

**Figure 5.** Experimental absorption spectra of **2** (**a**) and **3** (**b**) in various solvents.

The conformational isomers existing in solution can be identified and quantified using NMR. Due to their low solubility in acetonitrile, the investigations were performed in DMSO-*d*6. The corresponding 1H NMR spectra of **2** and **3** in DMSO-*d*<sup>6</sup> are shown in Figures S2 and S3. The data from the NMR measurements are summarized in Table 1. The obtained data for chemical shifts can be compared with those for **1** [22]. From the 1H NMR spectra of **2** and **3**, it can be seen that, in both cases, two isomers are present in DMSO-*d*6. The chemical shifts in DMSO-*d*<sup>6</sup> for NH for the major and minor form of **2** are at 14.55 ppm and 12.62 ppm, and for **3** at 14.48 ppm and 11.42 ppm. Based on the NH signals, the ratio between the isomers is 80%/20% and 45%/55%, respectively for **2** and **3** (65%/35% for **1**), which corresponds to ΔG values (RT) of 0.36, 0.82 and −0.11 kcal/mol for **1**–**3**.




**Table 1.** *Cont.*

<sup>a</sup> Could not be assigned due to overlap; <sup>b</sup> Could not be assigned due to low intensity; <sup>c</sup> The hydroxyl signals of the two forms are overlapped; <sup>d</sup> The hydroxyl signals of two forms and the NH signal of minor form are overlapped; <sup>e</sup> May be interchanged; <sup>f</sup> Numbering according to the Scheme 4.

**Scheme 4.** Numbering of the carbon atoms in **1** (X1=OH, X2=NO2), **2** (X1=OH, X2=H) and **3** (X1=H, X2=OH).

In analogy to **1** [22,24] and following the theoretically predicted relative stabilization, it can be concluded that in DMSO there is an equilibrium between *E'* (major) and *Z'* (minor) forms in **2** and between *E* and *Z* of **3**. The theoretically predicted values for the chemical shifts of the NH proton of the major and minor form at **2** and **3**, respectively, are 14.61 ppm (**2***E'*) and 13.67 ppm (**2***Z'*) and 14.51 ppm (**3***E*) and 13.44 ppm (**3***Z*), i.e., the theoretical results are consistent with the experimental ones. Although the NMR determined Gibb's free energies are lower comparing the predicted relative energies (ΔE), the latter correctly predict the general trend of stabilization of the *Z'* isomers (**3** > **1** > **2**) with a good linearity (Δ<sup>G</sup> <sup>=</sup> 1.28 <sup>×</sup> <sup>Δ</sup><sup>E</sup> <sup>−</sup> 0.98, *<sup>R</sup>*<sup>2</sup> <sup>=</sup> 0.92).

Three additional factors influence the conformational equilibrium in **2** and **3** in solution, namely the temperature, the concentration and the water content of the used solvent. As previously shown in the case of **1**, a spontaneous deprotonation (loss of OH proton) occurs in diluted solutions of dry proton acceptor solvents [22], leading to a new red-shifted band. Actually, **1** is almost fully deprotonated in dry DMSO (Figure S4), while, as seen in Figure 5, deprotonation in **2** is weak and negligible in **3**. Obviously, the effect of the nitro group is decisive in this case. Upon addition of water, the equilibrium is fully shifted towards the neutral isomers (Figure S5). According to the theoretical calculations (Table S2), deprotonation does not substantially change the isomers' ratios.

The concentration is an essential factor determining the deprotonation of the investigated compounds. In **2**, as in **1**, the increase in the concentration leads to a decrease in the content of the deprotonated form, as shown in Figure 6. No such effect is observed in **3**.

**Figure 6.** Experimental absorption spectra of **2** (**left**) and **3** (**right**) in DMSO as a function of the concentration, keeping the cell thickness (b) × concentration (c) constant.

The results above indicate that the association is a possible reason for the observed changes. The theoretical calculations and X-ray data (Figure 7 and Scheme 5) suggest that cyclic aggregates are formed in the case of **2**. In the solid state, compound **2** exists as an *E'* conformer, stabilized via a cyclic dimer. The major difference with **1** is in the shape of the aggregate—again, the *E'* form is stabilized in **1**, but in form of a linear aggregate [24]. The stability of the *E'*-*E'* cyclic dimer in **2** is probably due to the stronger proton acceptor properties of the CH3CO group compared to the PhCO moiety of **1**. The formation of aggregates, as in **1** [22,24], limits the deprotonation, which explains the observed concentration effects.

**Figure 7.** Relative energies (in kcal/mol) of the most stable dimers of **2** in DMSO.

**Scheme 5.** Crystal structure of **2** and a cyclic dimer model via an intermolecular hydrogen bond. Ellipsoids are drawn with a probability of 50% and H-bonds are represented as dashed blue lines, #1: 2 − x, 1 − y, 2 − z.

The increase in the concentration did not lead to significant spectral changes in **3** (Figure 6). This is also expected since the OH group is not polar enough in this particular case, but some linear aggregation cannot be excluded. According to the theoretical simulations (Figure S6) and the crystal structure (Scheme 6), linear *E*-*E* aggregates are expected in solution.

**Scheme 6.** Crystal structure of **3** and a cyclic dimer model via an intermolecular hydrogen bond. Ellipsoids are drawn with a probability of 30% and H-bonds are represented as dashed blue lines, #1: x − 1, 1 + y, z; #2: x − 2, y + 2, z; #3: x + 1, y − 1, z.

The availability of crystallographic data for the series **1**–**8** allows qualitative estimation of the strength of the existing intra- and intermolecular hydrogen bonding according to Steiner [28] and Jeffrey [29]. The corresponding bond length and angles are collected in Table 2. According to the classification given in [28], the existing N-H ... O hydrogen bonds are classified as moderate ones using the H ... A and D ... .A distances and the D-H ... A angle. It seems that the strength (at least in solid state) of this bond is almost independent on the substitution in the stator. In the cases where this bond is bifurcated, namely **1**, **2** and **6**, the contribution from N-H ... O(=C) is the dominant one. This explains why, according to the theoretical calculations using the solvent only as media, the *E* and *Z* isomers are always more stable compared to the " ' " ones. The data in Table 2 show clearly that the formation of associates through intramolecular hydrogen bonding with strong directionality (D-H ... A angle > 160◦) has a noticeable stabilizing effect.


**Table 2.** Parameters of the hydrogen bonds of the studied compounds, taken from their crystallographic data.

The absorption spectra of **2** and **3** in DMSO in the temperature range of 20–70 ◦C are shown in Figure 8 and Figure S7. In both compounds, it can be seen that with increasing temperature, the absorption maximum of the neutral form (~400 nm) decreases slightly, while the amount of the deprotonated form absorbing at ~500 nm increases.

The result can be interpreted in analogy to **1** [22]. Increasing the temperature leads to the destruction of existing aggregates, which subsequently facilitates the deprotonation of the monomers. The temperature effect was strongest in **1** (Figure S8), followed by **2** (Figure 8) and lowest for **3** (Figure S7). This indicates that **1** has the highest tendency to aggregate again related to the effect of the nitro group. The percentage contents (in %) of the two neutral isomers were determined by measuring NMR spectra as a function of the temperature (Table S3). It could be thereby shown that the ratio does not change significantly (Figures S9–S11), which probably indicates that they exist in the form of associates in the concentration range used in NMR.

**Figure 8.** Experimental absorption spectra of **2** in dry DMSO at different temperatures.

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

#### *3.1. Synthesis*

Reagents and solvents were analytical grade, purchased from Sigma-Aldrich Chemical Co. and used as received unless otherwise stated. Fluka silica gel/TLC-cards 60778 with fluorescence indicator 254 nm were used for TLC chromatography. Merck silica gel 60 (0.040–0.063 mm, Merck, Darmstadt, Germany) was used for flash chromatography purification of the products. LC/MS was carried out on a Bruker MicrOTOF-QIII-system with an ESI source with nebulizer 1.2 bar, dry gas 10.0 L/min, dry temperature 220 ◦C, capillary 4500 V, and end plate offset −500 V, Bruker, Hamburg, Germany. The title 2-(2(2-Hydroxy-4-nitrophenyl)hydrazono)-1-phenylbutane-1,2,3-trione **1**, 2-[2-(2-Hydroxyphenyl)diazenyl]-1-phenyl-1,3-Butanedione **2** and 2-[2-(4-Hydroxyphenyl)diazenyl]-1 phenyl-1,3-Butanedione **3** were synthesized as follows.

#### 3.1.1. 2-(2(2-Hydroxy-4-nitrophenyl)hydrazono)-1-phenylbutane-1,2,3-trione **1**

The synthesis of the compound **1** has been described previously [22].

#### 3.1.2. 2-[2-(2-Hydroxyphenyl)diazenyl]-1-phenyl-1,3-butanedione **2**

A solution of 2-aminophenol (0.2728 g, 2.5 mmol) in concentrated hydrochloric acid (1.5 mL) was cooled in an ice bath for 30 min with stirring. Sodium nitrite (0.1725 g, 2.5 mmol) was added gradually in small portions over 30 min. The diazonium salt was left stirring in the cold for 45 min and added slowly over 30 min into a well-cooled stirred mixture of sodium acetate (1.5 g, 18.3 mmol) and 1-phenylbutane-1,3-dione (0.4061 g, 2.5 mmol) in absolute ethanol (15 mL). The mixture was

stirred in an ice-bath for 3 h, and then left to warm to room temperature over 12 h. The solid was filtered off and washed with water (3 × 25 mL) to give a light brown solid. The ethanol portion was evaporated under reduced pressure to give dark brown solid. The combined solids were dissolved in dichloromethane (30 mL) and washed with water (3 × 50 mL) and the organic portion dried over anhydrous sodium sulfate, filtered and the solvent evaporated under reduced pressure to obtain a crude brown solid. This was purified by flash column chromatography using a silica flash column and ethyl acetate-dichloromethane (1:20) as eluent to give the pure compound as a light brown solid (0.48 g, 68%). HRMS-ESI (*m*/*z*): (M + H) calculated for C16H14N2O3, 282.1004; found 282.1084.

#### 3.1.3. 2-[2-(4-Hydroxyphenyl)diazenyl]-1-phenyl-1,3-butanedione **3**

A solution of 4-aminophenol (0.2728 g, 2.5 mmol) in concentrated hydrochloric acid (1.5 mL) was cooled in an ice bath for 30 min with stirring. Sodium nitrite (0.1725 g, 2.5 mmol) was added gradually in small portions over 30 min. The diazonium salt was left stirring in the cold for 45 min and added slowly over 30 min into a well-cooled stirred mixture of sodium acetate (1.5 g, 18.3 mmol) and 1-phenylbutane-1,3-dion (0.4061 g, 2.5 mmol) in absolute ethanol (15 mL). The mixture was stirred in an ice bath for 3 h, and then warmed to room temperature over 12 h. The solid was filtered off and washed with water (3 × 25 mL) to give a yellow solid. The ethanol portion was evaporated under reduced pressure to give a dark yellow solid. The combined solids were dissolved in dichloromethane (30 mL) and washed with water (3 × 50 mL) and the organic portion was dried over anhydrous sodium sulfate, filtered, and the solvent evaporated under reduced pressure to obtain a crude brown solid. This was purified by flash column chromatography using a silica flash column and ethyl acetate-dichloromethane (1:20) as eluent to give the pure compound as a light-yellow solid (0.52 g, 74%). HRMS-ESI (*m*/*z*): (M + H) calculated for C16H14N2O3, 282.1004; found 282.1066.

#### *3.2. Spectral Measurements*

Spectral measurements were performed on a Jasco V-570 UV-Vis-NIR spectrophotometer, equipped with a thermostatic cell holder (using Huber MPC-K6 thermostat with precision 1 ◦C), in spectral grade solvents.

The 1H-NMR and 13C-NMR spectra were recorded at 600 MHz and 150 MHz or 400 MHz and 100 MHz on a Bruker Avance II+ 600 or Bruker Avance III 400 spectrometer using CDCl3 or DMSO-*d*<sup>6</sup> as a solvent and TMS as internal standard.

#### *3.3. X-ray Measurements*

Single crystals of C16H14N2O3 (**2**) were obtained from a mixture of ethanol:water (5:1) by slow evaporation. A suitable single crystal was selected and mounted on a loop with oil and measured on a STOE IPDS 2T diffractometer. The crystal was kept at 200 K during data collection. Using Olex2 [33], the structure was solved with the ShelXT [34] structure solution program using intrinsic phasing and refined with the ShelXL [35] refinement package using least squares minimization. All the crystal structures have been deposited at the CCDC 1858058 (**1**), 1993960 (**2**) and 1993961 (**3**).

Crystal Data for C16H14N2O3 (*M* = 282.29 g/mol): triclinic, space group *P*-1 (no. 2), *a* = 5.8853(12) Å, *b* = 8.0633(16) Å, *c* = 15.944(3) Å, α = 77.760(15)◦, β = 88.887(16)◦, γ = 68.978(15)◦, *V* = 688.9(2) Å3, *<sup>Z</sup>* <sup>=</sup> 2, *<sup>T</sup>* <sup>=</sup> 200 K, <sup>μ</sup>(Cu Kα) <sup>=</sup> 0.785 mm<sup>−</sup>1, *Dcalc* <sup>=</sup> 1.361 g/cm3, 5325 reflections measured (14.428◦ <sup>≤</sup> <sup>2</sup><sup>θ</sup> ≤ 138.228◦), 5365 unique (*R*int = 0.0571, Rsigma = 0.0392) which were used in all calculations. The final *R*<sup>1</sup> was 0.0712 (I > 4 u(I)) and *wR*<sup>2</sup> was 0.2505 (all data).

Single crystals of C16H14N2O3 (**3**) were obtained from a mixture of ethanol:chloroform:water (10:1:2) by slow evaporation. A suitable single crystal was selected and mounted on a loop with oil and measured on a STOE IPDS 2T diffractometer. The crystal was kept at 298(2) K during data collection. Using Olex2 [33], the structure was solved with the ShelXT [34] structure solution program using intrinsic phasing and refined with the ShelXL [35] refinement package using least squares minimization.

Crystal Data for C16H14N2O3 (*M* = 282.29 g/mol): triclinic, space group *P*-1 (no. 2), *a* = 6.3843(4) Å, *b* = 8.0018(5) Å, *c* = 15.0129(10) Å, α = 96.235(5)◦, β = 101.298(5)◦, γ = 107.252(5)◦, *V* = 706.78(8) Å3, *<sup>Z</sup>* <sup>=</sup> 2, *<sup>T</sup>* <sup>=</sup> 298(2) K, <sup>μ</sup>(MoKα) <sup>=</sup> 0.093 mm<sup>−</sup>1, *Dcalc* <sup>=</sup> 1.326 g/cm3, 8242 reflections measured (2.812◦ <sup>≤</sup> 2θ ≤ 52.618◦), 2808 unique (*R*int = 0.0164, Rsigma = 0.0146) which were used in all calculations. The final *R*<sup>1</sup> was 0.0390 (I > 2σ(I)) and *wR*<sup>2</sup> was 0.1142 (all data).

#### *3.4. Quantum-Chemical Calculations*

Quantum-chemical calculations were performed using the Gaussian 09 program suite [36]. M06-2X functional [37,38] was used with the TZVP basis set [39]. This fitted hybrid meta-GGA functional with 54% HF exchange has especially been developed to describe main-group thermochemistry and non-covalent interactions, showing very good results in the prediction of the position of the tautomeric equilibrium in azonaphthols possessing intramolecular hydrogen bonds [40] and in the description of the proton transfer reactions in naphthols [41,42]. All structures were optimized without restrictions, using tight optimization criteria and an ultrafine grid in the computation of two-electron integrals and their derivatives, and the true minima were verified by performing frequency calculations in the corresponding environment. Solvent effects are described by using the Polarizable Continuum Model (the integral equation formalism variant, IEFPCM, as implemented in Gaussian 09) [43]. The absorption spectra of the compounds were predicted using the TD-DFT formalism. TD-DFT calculations were carried out at the same functional and basis set, which is in accordance with conclusions about the effect of the basis set size and the reliability of the spectral predictions [44–46].

#### **4. Conclusions**

The effect of the position of the OH group and the availability of a nitro group substitution in the stator was investigated in solution by means of DFT calculations, NMR and UV-Vis spectroscopy. The results indicate that, when the OH group is in *ortho* position, the E' and Z' isomers are present in DMSO, stabilized by intermolecular hydrogen bonding with the solvent. The availability, in addition, of a nitro group in *para* position increases the possibility for deprotonation of the OH group in the absence of water. In all studied compounds, a clear tendency towards formation of associates is evident. The obtained X-ray data explain the types of the possible homo-aggregates and correspond very well to the theoretical predictions. The obtained results, revealing the effect of the structural modifications in the stator and the influence of the environment, could be useful in the design of new rotary switches.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2624-8549/2/2/374\T1\ textendash389/s1. Figure S1: Second derivative spectra of a) **2** and b) **3** in various solvents, Figure S2: 1H NMR spectrum of **2** in DMSO-*d*6, Figure S3: 1H NMR spectrum of **3** in DMSO-*d*6, Figure S4: Absorption spectra of **1**, **2** and **3** in dry DMSO and upon base (TEA) addition, Figure S5: Experimental absorption spectra of **1** and **2** upon water addition, Figure S6: Relative energies (in kcal/mol) of the most stable dimers of **3** in DMSO, Figure S7: Absorption spectra of **3** in DMSO in different temperatures, Figure S8: Absorption spectra of **1** in DMSO in different temperatures, Figure S9: 1H NMR spectrum of **1** in DMSO at a temperature range of 20 ◦C to 70 ◦C, Figure S10: 1H NMR spectrum of **2** in DMSO at a temperature range of 20 ◦C to 70 ◦C, Figure S11: 1H NMR spectrum of **3** in DMSO at a temperature range of 20 ◦C to 70 ◦C, Table S1: The most stable isomers of I-VII of **2** in gas phase, Table S2: Relative energies (M06-2X/TZVP) of the most stable neutral and deprotonated forms of **1, 2, 3** in kcal/mol units, Table S3: Ratio between *E*(*E'*)/*Z*(*Z'*) forms in **1**, **2** and **3** in the temperature range of 293 K–343 K, Scheme SI: Possible conformers of **I** and their relative energies (in kcal/mol, according to quantum-chemical calculations) in acetonitrile.

**Author Contributions:** Conceptualization, L.A.; methodology, L.A. and N.V.; validation, S.H. and A.C.; formal analysis, S.H. and L.A.; investigation, S.H., F.S.K., N.V. and A.C.; resources, L.A., F.S.K. and K.M.F; data curation, L.A. and K.M.F.; writing—original draft preparation, S.H.; writing—review and editing, L.A., N.V. and K.M.F; supervision, L.A. and K.M.F.; project administration, L.A.; funding acquisition, K.M.F. and L.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by The Swiss National Science Foundation, Institutional Partnership project IZ74Z0\_160515, Bulgarian National Science Fund, project DN09/10 MolRobot, and Bulgarian Ministry of Educations, project DFNP-17-66/26.07.2017.

**Acknowledgments:** The financial support by Bulgarian Ministry of Educations (project DFNP-17-66/26.07.2017 for support of young scientists), Bulgarian National Science Fund (project DN09/10 MolRobot) and The Swiss National Science Foundation (SCOPES Institutional Partnership project IZ74Z0\_160515 SupraMedChem@Balkans.Net) is gratefully acknowledged.

**Conflicts of Interest:** No conflict of interest.

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).
