1
Phosphines and Related Tervalent Phosphorus Systems: Organophosphorus Compounds as Ligands in Organometallic Catalysis

Piet W. N. M. van Leeuwen

LPCNO, INSA‐Toulouse, 135 Av. de Rangueil, 31077 Toulouse, France

1.1 Introduction

1.1.1 History

Phosphines and related phosphorus‐containing molecules play a major role in homogeneous catalysis. The history of homogeneous metal complex catalysis, as we know it today, started in the 1960s, although there had been even industrial applications long before that. In the 1920s, a catalytic process was used for the addition of water to acetylene. The metal mercury was used in a sulfuric acid solution. The reaction was very slow and large volumes were needed; thus, this was far from attractive. A related process still in operation is the zinc‐salt‐catalyzed addition of carboxylic acids to acetylene. With the introduction of petrochemistry, the feedstock for acetaldehyde production changed to ethene. The reaction used until today is a stoichiometric oxidation of ethene by palladium, the so‐called Wacker process, in which palladium is reoxidized with oxygen and a copper catalyst. Carbonylation catalysis came on stream in the 1930s and 1940s, although its application was retarded by World War II (). Initially, the metals of choice were nickel, e.g. work by Reppe, and cobalt, especially hydroformylation by Roelen, and methanol carbonylation. Probably, Reppe (1948) was the first to use triphenylphosphine as a modifying ligand in a catalytic reaction, which concerned the addition of nickel‐cyanide‐catalyzed carbonylative alcohol addition to alkynes, leading to acrylates [1]. He used nickel cyanide also in the synthesis of polyketone from carbon monoxide and ethene in those early years. As of the 1960s, all these “leads” were greatly improved by ligand effects and by changing to the more active second‐row transition metals palladium and rhodium. Cobalt was also modified by phosphine ligands, and in this instance, the catalyst produced more of the linear oxygenate product, which now is mainly the alcohol rather than the aldehyde (Shell) [2]. Early examples of triphenylphosphine‐modified group 10 hydrogenation catalysis are due to Bailar and Itatani [3]. Ever since, more publications have appeared that reported phosphine effects on catalytic reactions.

1.1.2 Alternative Ligands

Before concentrating only on phosphorus ligands, we should mention that in the past three decades, ligands based on other donor atoms have become equally important and, in some areas, even more important than phosphines. In the mid‐1980s, the metallocene era started for the early transition metals especially in polymerization catalysis, followed by alkoxides, amides, and salen ligands. Meanwhile, metallocene catalysts have found industrial applications. In the late transition metal area, the diimine ligands stand out together with a shift to the first‐row metals for alkene polymerization, and they almost made it to a replacement of the nickel catalyst in the oligomerization of ethene. They were followed a little later by the outburst of the NHC ligands, which have beaten, in several instances, the best phosphines used so far in certain reactions. They have found commercial applications in metathesis reactions. A combination of all donor ligands in bidentates has further enriched the toolbox of homogeneous catalysis. One should not forget that the “ligand‐free” systems are attractive, as they do not suffer from ligand decomposition, but their life can still be limited because of precipitation or formation of a compound with the wrong valence state. The stabilizing ligands in these cases are, for example, carbon monoxide, alkenes, halides, and other anions, for example, the Wacker process, cobalt‐catalyzed hydroformylation (Exxon), nickel‐catalyzed oligomerization of butene to 3‐methylheptane (IFP, Dimersol process), rhodium‐catalyzed carbonylation of methanol (Monsanto, now BP), and ditto for iridium (BP, Cativa process).

1.1.3 Aim of the Chapter

The aim of this chapter is to give an introduction to the use of tervalent phosphorus compounds as ligands in homogeneous catalysis. Several chapters in this work refer to that area and have their own introductions. We have tried to avoid overlap and provide some basic concepts in a nutshell while referring to those chapters that deal in more detail with this topic. In Section 1.2, we deal with the most common elementary steps used for the synthesis of phosphorus ligands. In more specific chapters, synthesis will be dealt with in much more detail than what we were able to cover here. The overview is very limited, as, for example, in our laboratories students are introduced to phosphorus ligand synthesis with a series of about 200 synthetic steps of which we think they are worthwhile for a starter in this area! In Section 1.3, the properties of phosphorus ligands will be discussed by presenting the most common yardsticks used, such as Tolman's χ and θ values for the electronic and steric parameters, respectively, and Casey's βn, the natural bite angle for bidentate ligands. For the steric and electronic parameters, several alternatives have been developed, and all the parameters have found use particularly in catalysis [4]. Studies on the use of parameters in Linear Free Energy Relations and QUALE will be mentioned.

In Section 1.4, chiral phosphorus ligands will be introduced focusing on the types of chiral ligands available, involving the most typical phosphorus and diphosphorus ligands, and heterobidentate ligands.

The next two sections will deal with two examples of ligand effects, namely a few highlights in hydroformylation and the next one on modern cross‐coupling chemistry. As both are huge areas, these parts also serve as a brief introduction to the fields. We will highlight the crucial issues concerning monodentate and bidentate ligands.

Section 1.6 includes the main decomposition pathways of phosphorus ligands, which are also discussed in dedicated chapters in books and reviews.

1.2 Synthesis of Phosphorus Ligands

1.2.1 Introduction

Clearly, the synthesis of phosphorus ligands involves a library of organic phosphorus chemistry to which one cannot do justice in just a few pages. Chapter 7 by Stevens deals with the most important routes for the introduction of phosphonates into complex organic molecules, and more details and references can be found there. Phosphonates can be converted into phosphines, of which there exist many examples. Here, we will deal with a simple summary of the common elementary steps for making phosphorus ligands. Phosphaalkenes will not be discussed as they are not yet of proven interest in catalysis. Although phosphinines have been exploited occasionally in catalysis and have shown interesting properties, for instance, in rhodium‐catalyzed hydroformylation [5], we will not discuss their synthesis. We will confine ourselves to a series of elementary steps thought to be useful for our purposes. Even that will rather be a short list of less than 40, as, for example, in my group, the students acquainted themselves in phosphine synthesis using a set of about 150–200 reactions. Although one could bring down the number as there are less reaction types, it would still be too large to list for the present purposes. Below we have ordered the reactions according to the main reaction types, which are still feasible, because the number of ways to make a PC bond is far less numerous than that for making CC bonds!

1.2.2 Nucleophilic Substitution by Carbanions at Pδ+

The ionic approach to the formation of a carbon–phosphorus bond has two possibilities, namely the use of phosphorus as a nucleophile or as an electrophile. The latter seems more in accord with the electronic properties of phosphorus, as a slightly positively charged phosphorus species is common and stable, whereas phosphido anions (Section 1.2.3) tend to show electron transfer reactions in addition to and before entering a nucleophilic attack. Indeed, nucleophilic substitution is the most common reaction used, although both routes have their pros and cons. One might not often directly see potential cons for a certain route. For example, an attack of a benzylic anion at a phosphorus electrophile will proceed smoothly, but the product formed, PhCH2PR2, has an acidic proton at the benzylic position, and the remaining benzyl anion may be consumed via a simple proton abstraction and, at best, the yield is only half of the expected amount. The nucleophile will often be a relatively simple Grignard or a hydrocarbyllithium reagent, and it can be used in excess, as during the workup by hydrolysis, it can be easily removed. If the nucleophile is a more complicated molecule, of which the remains can be removed only with difficulty, the use of a large excess must be avoided. Also, the carbonucleophile may substitute other hydrocarbyl groups at phosphorus (Scheme 1.3). Reaction 3 shows a less common nucleophilic substitution, but it shows its versatility.

Perfluoroalkyl groups are far less frequently used than simple aromatics, which we will encounter below. The first example shows pentafluoroethyl groups, which are not the most common substituents at phosphorus, but they are highly desirable in the studies...