SEARCH

SEARCH BY CITATION

Keywords:

  • block copolymers;
  • fluorination;
  • self-assembly;
  • interaction parameter;
  • micelles

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

Fluorinated block copolymers combine the unique properties of fluoropolymers and the intriguing self-assembly of hybrid macromolecules. The preparation of the title molecules by selective fluorination procedures and the effect of fluorine incorporation on the material thermodynamics are presented. We highlight two fluorination schemes developed in our laboratory, difluorocarbene and perfluoroalkyliodide additions to polydienes, that allow for the selective and tunable incorporation of different fluorinated groups into model block copolymers. The fluorination changes the physical properties of the parent materials and leads to interesting changes in the component incompatibilities. The role of fluorination in determining block copolymer thermodynamics in both the solid state and in solution and in ultimately exploiting fluorination to produce novel, higher order structures is central to our research efforts. © 2001 John Wiley & Sons, Inc. J Polym Sci Part A: Polym Chem 40: 1–8, 2002


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

The remarkable physical and chemical properties of molecular systems that result from the incorporation of atomic fluorine are well documented. Refrigerants, surfactants, liquid crystals, and blood substitutes are examples of compounds with particular attributes (e.g., extreme hydrophobicity and high O2 solubility) as a result of covalently bound fluorine atoms. In parallel to these small-molecule examples, the combination of fluorine and synthetic polymers has been fruitful from both commercial and academic perspectives; poly(tetrafluoroethylene) (PTFE) is the canonical example. PTFE has extraordinary physical properties that position this material in a class by itself. The high thermal stability imparted by the strong C[BOND]F and C[BOND]C bonds, the low surface energy resulting from the small polarizability of fluorine and C[BOND]F bonds, and the high melting point resulting in part from the stiffness of the backbone (itself a consequence of fluorine's bulkiness) are all due to the unique properties of fluorine. These attributes have led to the use of PTFE in numerous applications, ranging from slippery dental floss to electrical insulation to noncorrosive gaskets used for the separation of uranium isotopes. Although PTFE and many other commercially available fluoropolymers are useful materials with broad applicability, there are synthetic and processing challenges. For example, although PTFE was discovered more than 60 years ago, only in the last year were conditions identified that refute the notion that this polymer is generally not melt-processible.1

Block copolymers consist of two or more chemically dissimilar chains connected at their termini. The covalent connection of thermodynamically incompatible segments leads to self-assembly of the components into ordered structures with periodicity or compositional heterogeneity on a nanometer length scale. For example, an AB diblock copolymer such as polystyrene-b-polyisoprene (PS–PI) with a specific ratio of the two components will self-assemble at the appropriate temperature into nanoscopic rods of polystyrene (PS) packed on a hexagonal lattice in a polyisoprene (PI) matrix (An example for a PS-PI-PS triblock is shown in Fig. 1).2 Simple AB diblock copolymers have been thoroughly studied, and both their thermodynamic and dynamic properties have been well documented.3 The uniting of disparate polymers in a block copolymer offers distinct property advantages in comparison with the related admixture of two polymers in the corresponding blend. For example, a polystyrene-b-polybutadiene-b-polystyrene (ABA) triblock copolymer exhibits elastomeric characteristics at low temperatures and thermoplastic-like behavior at high temperatures (i.e., it becomes melt-processable), whereas the analogous blends are prone to macrophase separation, leading to poor mechanical properties. These ABA triblock copolymers, termed thermoplastic elastomers, are commercially available and enjoy a wide array of applications.

thumbnail image

Figure 1. Transmission electron micrograph of an aligned PS–PI–PS block copolymer that formed a cylindrical structure. The PI domain was stained with OsO4.

Download figure to PowerPoint

Although the physical properties of block copolymers have been utilized since their discovery, research that exploits the unique molecular assemblies found in ordered block copolymers for templating, delivery of functional groups, or nanoscopic confinement has been very active only over the past decade or so. Our interests center on the thermodynamics of self-assembly in fluorinated block copolymers for the purpose of revealing and designing novel nanoscopic structures with both fundamental and technological relevance. The aforementioned unique properties of fluorine are predicted to have significant consequences on the self-assembly of block copolymers when contained in one or more of the segments. As in any experimental program of this nature, the preparation of suitable molecules precedes all other characterization protocols.

Model block copolymers are materials with discrete overall and relative block sizes. Furthermore, architectural integrity and narrow distribution in molecular weight are tandem criteria. These aspects are important for the systematic structure–property studies crucial to understanding block copolymer self-assembly, especially in a comparison of experimental observations and theoretical predictions. Since the discovery of living anionic polymerization in the 1950s, the preparation of model block copolymers by the sequential addition of monomers with this methodology has been widespread.4 Although this approach is extremely useful, the high reactivity of the propagating carbanionic centers has generally limited the scope of monomers that can be polymerized in a controlled manner. Several other techniques have been developed for the controlled or living polymerization of a broad range of monomer structures, including cationic, ring-opening metathesis, radical, and metal-catalyzed polymerizations. By now, a veritable cornucopia of block copolymer structures have been prepared. This fact notwithstanding, synthetic routes to fluorinated block copolymers (especially by the sequential addition of monomers) are rare. Therefore, we have focused on synthetic approaches to fluorinated block copolymers that rely on the time-tested capabilities of living anionic polymerization combined with the flexibility of polymer modification chemistry to prepare new block copolymers containing fluorinated segments.

Polymer modification schemes for the preparation of macromolecular materials have certain advantages over traditional polymerization strategies. For example, the extent of modification can often easily be modulated by reaction stoichiometry or kinetics. This results in the controlled incorporation of functionality, often with a random distribution along the chain. Additionally, polymer modification routes can lead to backbone structures that cannot be prepared by a direct polymerization technique [the preparation of poly(vinyl alcohol) by the hydrolysis of poly(vinyl acetate) is a standard example]. For the purposes of model block copolymer synthesis, the modification route must not compromise the molecular integrity of the starting materials. For instance, the molecular weight distribution of the parent block copolymer should be preserved upon modification. The selectivity of the modification must also be considered because in most cases the incorporation of functional groups into a block copolymer should occur on only one chemical moiety in the hybrid macromolecule. When it comes to fluorination, challenges to meet the requirements of an ideal polymer modification procedure are evident. Although there are certainly fluorination reactions that can be applied to macromolecular substrates, they are often chemically severe, such as treatments with F2 gas5 and the action of SF6 under electrochemical conditions.6 Furthermore, most of the published reactions dealing with polymeric fluorinations are surface modifications and are unsuitable for bulk transformations. Recently, the fluorination of polymers by chemical modification has been reviewed,7 and a sampling of modification chemistries are depicted in Figure 2. Of the known routes, only a few are suitable for block copolymer modifications, with the work of Ober et al.8 and Antonietti et al.9 being exemplary. In the course of our investigations, we have developed two distinct fluoro-modification chemistries that can be readily applied to block copolymers and that closely approach the ideal criteria previously identified.

thumbnail image

Figure 2. Representative examples of polymer fluorination.

Download figure to PowerPoint

CASE STUDY 1: DIFLUOROCARBENE INCORPORATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

The reaction of polydienes such as PI with halogenated carbenes was first reported by Pinazzi and Levesque in 1965.10 Since that time, several examples of polymer modification with dichloro, chlorofluoro, and difluoro carbene have been reported.11 We demonstrated that a model PS–PI diblock copolymer could be modified by the addition of difluorocarbene (generated from the thermolysis of hexafluoropropylene oxide) while retaining the molecular parameters of the parent material (e.g., polydispersity and degree of polymerization).12 The reaction scheme is given in Figure 3. The key features of this modification are that despite the 180 °C reaction temperature, the addition of difluorocarbene is extremely selective for the PI segments over the PS repeat units and the formation of the gem-difluorocyclopropane structure is exclusively observed. Additionally, there is evidence for random incorporation of the difluorocarbene, and control of the modification extent can be achieved by either stoichiometry or reaction kinetics. Furthermore, the reaction is technically straightforward and involves only one step. Using this essentially ideal fluoro-modification procedure, we evaluated the effect of this type of fluorine incorporation on the block copolymer thermodynamics.13

thumbnail image

Figure 3. Reaction scheme for the fluorination of a model PS–PI block copolymer.

Download figure to PowerPoint

The progressive addition of difluorocarbene to the PI block of PS–PI diblock copolymers leads to an overall increase in the order–disorder-transition (ODT) temperature and to an increase in the microdomain periodicity (L; obtained from the position of the principal small-angle scattering peak). Both features are expected when the modification increases the degree of segregation between the two blocks; that is, the Flory–Huggins interaction parameter (χ) goes up. However, the detailed effect of fluorination on the thermodynamics of PS–PI copolymers is more interesting. Figure 4 shows the ODT temperature for a particular PS–PI/FPI as a function of the degree of fluorination, where FPI denotes difluorocarbene-modified PI. The inset shows the determination of an individual ODT by rheology; the low-frequency dynamic elastic modulus (G′) exhibits an abrupt drop by several orders of magnitude as the sample passes from a layered (lamellar) ordered phase to a disordered melt. With the addition of difluorocarbene, however, the ODT goes down before increasing again. This suggests that FPI has a higher cohesive energy density (CED) than PI; therefore, a small fraction of FPI brings the average PI/FPI block CED closer to that of PS. Continued fluorination moves the CED above that of PS. This scenario is confirmed by an analysis similar to that described for polymer blends, whereby a random copolymer of A and B segments can be miscible with a C homopolymer, even if all three pairwise interaction parameters are positive, as long as AB repulsion is stronger than AC and BC repulsion.14 With this model, it is possible to extract values of χPS/FPI ≈ 0.40, χPI/FPI ≈ 0.46, and χPS/PI ≈ 0.077 at 110 °C. In addition, the temperature dependence of L gives access to the temperature dependence of χ. Figure 5 summarizes the temperature and degree of fluorination dependence of the effective interaction parameter (χeff) between the PS and PI/FPI blocks. These data indicate that the difluorocarbene-modified polydienes are more polar than PS, in contrast to the naive view that fluorinated materials are less polar. The origin of this polarity is the asymmetry of the bridging difluorocarbene group; the repeat unit resembles much more the relatively polar poly(vinylidene fluoride) than PTFE. Interestingly, the difluorocarbene modification also confers enhanced solubility in supercritical CO2, which opens up other possible avenues of research.15

thumbnail image

Figure 4. Order–disorder temperature (TODT) for a difluorocarbene-modified PS–PI block copolymer as a function of difluorocarbene incorporation: (▪) static birefringence measurements and (○) rheological measurements. The inset shows the rheological determination of TODT for the 29% sample.

Download figure to PowerPoint

thumbnail image

Figure 5. Temperature and degree-of-fluorination dependence of χeff between the PS and PI/FPI blocks. x is the molar fraction of difluorocarbene incorporation.

Download figure to PowerPoint

CASE STUDY 2: PERFLUOROALKYL INCORPORATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

The addition of difluorocarbene to polydienes has the limitation that only two fluorine atoms can be incorporated per repeat unit of the polydiene. Experimental and anticipated reactivity problems16 with higher homologues of hexafluoropropylene oxide inspired an approach that enabled the incorporation of perfluoroalkyl groups into a polydiene backbone. We focused on the free-radical addition of perfluoroalkyliodides to pendant double bonds in 1,2-polybutadiene (PBD) homopolymers. Using a boron-based free-radical initiator, we were able to control the addition of C6F13I to PBD.17 The addition of C6F13I to the polymer chain results in an interesting cyclic structure due to the rapid reaction of the intermediate free radical (the product of the initial Rf· addition) with an adjacent double bond. The repeat unit structure that results is shown in Figure 2. The fluorination of PBD–PS block copolymers with this method led to the selective modification of PBD, giving block copolymers with interesting properties.

In Figure 6, the low surface energy of these fluorinated homopolymers and block copolymers is demonstrated in Zisman plots.18 The surface of the thin modified polymer films used for this study appear to be enriched with the fluorinated side chains because their critical surface tensions were all measured to be between 14 and 15 mN/m (which may be compared to the value of 19 mN/m typical of PTFE). In addition to perfluoroalkyl incorporation, the free-radical addition of perfluoroalkyl iodides results in a relatively weak C[BOND]I bond in the backbone of the polymer (see Fig. 2). Although this compromises the thermal stability of the modified polymer, the data presented in Figure 6 suggest that replacement of this labile bond with a C[BOND]H bond does little to the surface energy of the modified polymers.

thumbnail image

Figure 6. Zisman plots for perfluoroalkyl-modified polymers: (○) 1,2-PBD·C6F13I, (□) PS-b-1,2-PBD·C6F13I, and (•) PS-b-1,2-PBD·C6F13.

Download figure to PowerPoint

In addition to yielding low-surface-energy polymers, the incorporation of perfluoroalkyl groups into a PBD–PS block copolymer significantly increases the incompatibility of this polymer pair. The one-dimensional small-angle X-ray scattering (SAXS) patterns (intensity vs scattering wavevector) in Figure 7 show this effect. In the parent block copolymer, the overall molecular weight was about 10 kg/mol, and the block copolymer contained a majority of PS. As expected, the SAXS data show a broad peak at 120 °C, indicating a disordered polymer melt. After conversion of about 90% of the double bonds in the PBD segments with C6F13I (and subsequent replacement of I with H), the SAXS pattern shows three distinct peaks corresponding to a well-segregated lamellar morphology. The main contribution to the observed ordering behavior upon fluorination is the increase in the interaction parameter between the block copolymer components. In contrast to the difluorocarbene modification previously described, the incorporation of a long-chain perfluoroalkyl group substantially decreases the CED of the parent PBD block.

thumbnail image

Figure 7. One-dimensional SAXS patterns for the PS-b-1,2-PBD block copolymer and its fluorinated analogue (PS-b-1,2-PBD·C6F13I). The triangles show integral multiples of the principal peak (q*).

Download figure to PowerPoint

PERSPECTIVE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

As indicated in the introduction, the principal aim in this research area is to marry the remarkable properties of fluorinated materials with the architectural and structural control of block copolymers. In this section, we discuss two specific manifestations of this general idea: accessing the so-called superstrong segregation regime (SSSR), and developing micellar aggregates with three or more distinct domains. Other fascinating areas of application of fluorinated block copolymers, namely, surface property modification,8 surfactants for polymerizations in liquid and supercritical CO2,19 and associating polymers,20 are not addressed; the interested reader should consult the indicated references.

SSSR

The phase behavior of block copolymers is often categorized into three regimes: weak, intermediate, and strong segregation. These approximately correspond to values of χN of order 1 (or less), 10, and 100 (or more), respectively, where N is the total degree of polymerization. In weak segregation, the effective repulsion between the blocks is a minor perturbation on the normal Gaussian behavior of a molten polymer chain; entropic effects dominate, and disordered liquids result. In strong segregation, energetic effects play a stronger role, leading to separate microdomains that are almost pure A and B, separated by a sharp interface. Most common block copolymers, such as PS–PI and PS–PBD, fall into the intermediate regime; χ is typically 0.05–0.1, and N is 100–1000. The ODT is also typically found in the intermediate regime. In the intermediate and strong segregation regimes, the chain conformation and, therefore, the domain periodicity are dictated by a balance between the interfacial energy, favoring well-separated interfaces and stretched chains, and conformational entropy, favoring coiled blocks and, therefore, more interface/unit volume. The SSSR arises when χ becomes so large that the interfacial tension completely dominates. In this case, the minor block should stretch out completely. Thus the domain size L will scale linearly with N, in contrast to LN2/3, which is characteristic of strong segregation. Figure 8 illustrates the concept of superstrong segregation for an initially spherical micelle. As the interfacial tension increases, transitions to an oblate disk, a hairy hockey puck, and eventually a flat sheet are anticipated.21 In short, new structures are anticipated, whether in solution or in the bulk. There have been no systematic explorations of this concept to date, although various studies are relevant.22, 23

thumbnail image

Figure 8. Transition of a spherical micelle into an oblate micelle, a disklike micelle, and a flat sheet with an increase in χ.

Download figure to PowerPoint

The key step is to develop chemistry in which χ may be varied progressively but over a very large range. For example, χ values as large as 10 or 100 may be necessary. To achieve this, one block should be ionic or at least have an extremely high CED. For the other block, perfluorinated alkane groups offer extremely low CEDs. Nafion is an example of a commercially important material that exploits this idea, but the chain architecture and thus the material morphology are not well controlled; model block copolymers, therefore, offer an appealing possibility. An important practical issue is that of equilibration; the stronger the interactions are at the monomer level, the more likely the system will become trapped in some metastable state. By systematically varying the charge density along one block and the degree of fluorination along the other, we should be able to ramp up the degree of segregation steadily. Although this does not eliminate the problem of metastability, one expects a systematic variation in morphology with χ, and departures from such a progression should be apparent. Furthermore, by varying χ and N independently, we can characterize the crossover from strong segregation (phase behavior dependent on the product χN) to superstrong segregation (phase behavior dependent primarily on χ).

Tryptych Surfactants and Structured Micelles

The micellization properties of small-molecule amphiphiles and block copolymers alike have been extensively studied. Regardless of the particular morphology (e.g., spheres, rods, and vesicles), such materials divide space into two: the micellar core and the micellar corona (the portion compatible with its environment). However, situations in which a tripartite division of space would be desirable are apparent, such as can be obtained with ABC triblock copolymers. As a simple example, consider a core–shell–corona micelle [Fig. 9(a)] in a controlled release application. The corona confers dispersibility in the medium and possibly recognition elements to induce selective binding. The core block would be designed to house the deliverable agent. The middle, shell block could be designed simply to regulate the rate of escape of the agent by a combination of intrinsic permeability and thickness.

thumbnail image

Figure 9. Possible morphologies in an ABC triblock copolymer micelle.

Download figure to PowerPoint

The preceding example is particular, but there is a much more general issue of interest here. Biological systems often adopt a hierarchical arrangement of structural units within a given assembly; a cell containing a nucleus, organelles, and so forth is the quintessential example. It is interesting to consider whether rather simple multiblock copolymer architectures can be designed that will generate structures within structures. Figure 9(b–d) illustrates other possibilities for an ABC triblock, where in this case the B block is selected to form the corona. The A and C units could each form separate domains within a core [Fig. 9(b)], form two separate but adjacent domains [Fig. 9(c)], or at higher concentrations form separate domains bridged by B blocks [Fig. 9(d)]. The key regulating factors will be the relative values of the pairwise interaction parameters. To obtain the arguably most interesting structure, two separate domains in one micelle [Fig. 9(b)], making χAC > χAB ≈ χBC would be appropriate. However, it is also essential that the magnitudes of all three χ values be sufficiently large that mixing is prevented for relatively low molecular weight materials. This is where fluorination is advantageous; one fluorinated block, one hydrocarbon (lipophilic) block, and one water-soluble (hydrophilic) block would constitute the necessary threefold philicity to form what we dub a tryptych surfactant. Recent work on ABC triblock copolymers that fall into this category (i.e., they contain lipophilic, fluorophilic, and hydrophilic blocks) has appeared: (1) ABC polyacrylate triblock copolymers containing three dissimilar segments were prepared by anionic polymerization and evidence for aggregation in selective solvents was demonstrated,24 and (2) living cationic polymerization was used to prepare water-soluble polyoxazoline copolymers consisting of a hydrophilic center block capped at one end with a discrete fluorinated group (i.e., C8F17CH2CH2) and at the opposite end with a hydrocarbon group. Aggregation of these copolymers into the structure represented in Figure 9(d) was proposed.25

SUMMARY

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

The incorporation of fluorine can confer unique and highly desirable properties to polymers. Block copolymers constitute a class of self-assembling macromolecules that offer tremendous promise for advanced applications. The combination of structural control, imparted by well-defined block copolymer architectures and the properties of fluorinated moieties, is an appealing strategy for preparing novel materials. This goal requires advances in polymer modification chemistry, whereby tunable amounts of fluorine can be incorporated selectively and mildly into premade copolymers; two particular examples from our laboratory have been described. As we have outlined, understanding the role of fluorination in determining the self-assembled structure of copolymers and in ultimately exploiting fluorination to produce novel, higher order structures forms the current focus of this research.

REFERENCES AND NOTES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

Biographical Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information
Thumbnail image of

Marc A. Hillmyer received his Ph.D. in chemistry from the California Institute of Technology in 1994. After a postdoctoral appointment in the Department of Chemical Engineering & Materials Science at the University of Minnesota, he joined the Chemistry faculty at Minnesota in 1997, where he is currently an assistant professor. In 2000, he was awarded a Packard Fellowship in Science and Engineering, a Camille Dreyfus Teacher–Scholar Award, and a McKnight Land-Grant Professorship. He has co-authored over 50 articles and 5 patents in the field of polymer chemistry. His research interests include selective polymer modifications, renewable resource polymers, and nanostructured macromolecular materials.

Biographical Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CASE STUDY 1: DIFLUOROCARBENE INCORPORATION
  5. CASE STUDY 2: PERFLUOROALKYL INCORPORATION
  6. PERSPECTIVE
  7. SUMMARY
  8. REFERENCES AND NOTES
  9. Biographical Information
  10. Biographical Information

Timothy P. Lodge received his Ph.D. in chemistry from the University of Wisconsin in 1980. After 20 months as a National Research Council Postdoctoral Fellow at National Bureau of Standards, he joined the Chemistry faculty at Minnesota. In 1991, he was promoted to full professor, and in 1995, he became Professor of Chemical Engineering & Materials Science. In 2001, he was named a McKnight Distinguished University Professor. He was co-recipient of the 1993 George Taylor Alumni Award for excellence in research, given by the Institute of Technology at the University of Minnesota, and in 1994, he was named a fellow of the American Physical Society. He received the Arthur K. Doolittle Award from the Polymeric Materials Science & Engineering Division of the American Chemical Society in 1998. In 2001, he became the editor of Macromolecules. He has authored or co-authored over 130 articles in the field of polymer science and advised or co-advised 27 Ph.D. theses.