Download PDF
Synlett 2020; 31(19): 1862-1877
DOI: 10.1055/s-0040-1707168
account
© Georg Thieme Verlag Stuttgart · New York

Bumpy Roads Lead to Beautiful Places: The Twists and Turns in Developing a New Class of PN-Heterocycles

Jeremy P. Bard  
,
Darren W. Johnson  ∗
,
Michael M. Haley  ∗

The authors thank the National Science Foundation (Grant No. CHE-1607214) for funding our research on PN-heterocycle chemistry. This work was also supported by the Bradshaw and Holzapfel Research Professorship in Transformational Science and Mathematics to DWJ.
Further Information

Publication History

Received: 07 May 2020

Accepted after revision: 27 May 2020

Publication Date:
13 July 2020 (online)

 
 PDF Download Permissions and Reprints All articles of this category


This Account is dedicated to our postdoctoral mentors, Profs. Julius Rebek Jr. and Peter Vollhardt, who shaped our research interests by supporting the exploration of areas where one follows one’s nose – letting serendipity take you where it may

Abstract

The Haley and Johnson labs at the University of Oregon have been collaborating since 2006, combining skillsets in synthetic organic, physical organic, and supramolecular chemistries. This joint project has produced many examples of host molecules that bind anionic guests and give chemical, photophysical, and/or electrical responses. Many of these receptors utilize two-armed arylethynyl backbones that have a variety of hydrogen- or halogen-bonding functional groups appended. However, in attempts to produce a bisamide-containing host using a peptide-coupling protocol with P(OPh)3 present, we isolated something unexpected – a heterocycle containing neighboring P and N atoms. This ‘failed’ reaction turned into a surprisingly robust synthesis of phosphaquinolinones, an unusual class of PN-heterocycles. This Account article tells the rollercoaster story of these heterocycles in our lab. It will highlight our key works to this field, including a suite of fundamental studies of both the original PN-naphthalene moiety, as well as a variety of structural modifications to the arene backbone. It will also discuss the major step forward the project took when we developed a phosphaquinolinone-containing receptor molecule capable of binding HSO4 – selectively, reversibly, and with recyclability. With these findings, the project has gone from hospice care to making a full, robust recovery.

1 Introduction

2 Initial Discovery

3 Setbacks Breathe New Life

4 A New Dynamic Duo Develops Dozens of Derivatives

5 Physicochemical Characterization

5.1 Fluorescence

5.2 Molecular Structures

5.3 Solution Dimerization Studies

6 Applying What We Have Learned

6.1 Development of Supramolecular Host

6.2 Use of PN Moiety as an Impressive Fluorophore

7 Conclusions and Outlook


#

Keywords

fluorophore - PN-heterocycles - hydrogen bonding - supramolecular chemistry - anion binding

Biographical Sketches

Zoom Image

Jeremy P. Bard is a 5th year PhD candidate at the University of Oregon under the guidance of Professors Darren W. Johnson and Michael M. Haley. He graduated from Eastern Oregon University (La Grande, OR) in 2016 with a BSc in chemistry and a minor in mathematics. Jeremy has performed a variety of physical organic studies, focusing mainly on modifying both fluorescence properties and dimerization strengths of various phosphorus- and nitrogen-containing heterocycles. He enters into his last year of graduate studies having recently been awarded a UO Doctoral Dissertation Research Fellowship for 2020–2021.

Zoom Image

Darren W. Johnson is a Professor of Chemistry, Director of the Materials Science Institute, and the Bradshaw and Holzapfel Research Professor in Transformational Science and Mathematics at the University of Oregon. After obtaining a PhD from UC-Berkeley in 2000 with Prof. Ken Raymond, he was an NIH postdoctoral researcher in the lab of Prof. Julius Rebek Jr. at the Scripps Research Institute from 2001–2003. His lab uses supramolecular chemistry as a tool to tackle problems in environmental, biomedical, and sustainable chemistry. Specifically, the group explores topics in dynamic covalent chemistry, self-assembly, molecule/ion recognition, and inorganic cluster synthesis. He was co-founder of SupraSensor Technologies, a UO spinout that sought to commercialize sensors for nutrient management in precision agriculture and was acquired by The Climate Corporation in 2016. He has co-authored over 140 publications, is co-inventor on over 15 pending and issued patents, and is an inaugural Senior Member of the National Academy of Inventors.

Zoom Image

Michael M. Haley received his BA (1987) and PhD (1991) degrees from Rice University in Houston, Texas working for Prof. Ed Billups. After a postdoctoral stay at the UC-Berkeley with Prof. Peter Vollhardt, he joined the faculty at the University of Oregon in 1993 where he currently is the Richard M. and Patricia H. Noyes Professor of Chemistry. Haley also served as head of the department from 2008–2014. He has co-authored over 200 research articles and is co-inventor on 13 pending and issued patents. Like Johnson, Haley is also a co-founder of SupraSensor Technologies and a senior member of the National Academy of Inventors. In addition to PN-hetero­cycles, his current research interests focus on the synthesis and properties of phenylacetylene-based molecular scaffolds for anion sensing and on organic semiconductors with pronounced antiaromatic and/or diradicaloid character.

1

Introduction

The field of supramolecular chemistry continues to grow and diversify, especially as more creative, functional, and predictably dynamic systems are developed. From drug molecule encapsulation within supramolecular cavitands[1] to molecular machines,[2] supramolecular chemistry has applications in several different areas based on the variety of molecular interactions it entails. One application that has long been a hallmark of supramolecular chemistry is host–guest chemistry. Many disparate fields such as medical,[1] [3] [4] agricultural,[5] and environmental[6–12] have benefited from host–guest chemistry in recent years with the development of drug-containing capsules, nutrient sensors, and noncovalent waste remediation systems, respectively. Because of this broad yet varied applicability, host–guest ‘chemistry’ spans chemical biology, materials sciences, agricultural and food/fragrance research, biology, and biomedical engineering, among other fields.

Of the variety of potential guests, anionic species represent a specifically challenging target in this area due to their high solvation energies, diffuse charges, and lack of strong attractions between anions and most hosts.[13] Anions are bound through hydrogen bonding or other similar reversible electrostatic interactions. Upon the binding of these guests, many hosts are monitored through chemical, photophysical, or physical methods, and the reversible response can then dynamically report anion concentration in the medium tested. In the lab, this typically is observed via NMR, UV/Vis, and/or fluorescence spectroscopies, while in real-world applications reporter devices such as electrochemical sensors, chemically sensitive field-effect transistors (ChemFETs), or fluorescence microscopy are preferred. There exist innumerable hosts capable of providing signal transduction or reporting guest binding, yet as more and more new and interesting analytes, cellular pathways, problematic contaminants, and chemical markers are discovered, the demand for new responsive hosts will continue to rise as well.

Towards this aim, our lab has developed several aryl­ethynyl frameworks (Figure [1]) capable of binding a variety of anionic guest molecules, including halides,[14] [15] [16] [17] [18] oxoanions,[19–21] and hydrochalcogenides.[22] [23] [24] These hosts take advantage of both their multidentate binding pockets formed from the meta-substitution of the phenyl or pyridyl cores with two arylethynyl arms as well as the fluorescent nature of the π-conjugated backbone.

Zoom Image
Figure 1 Hydrogen- and halogen-bonding supramolecular hosts prepared and studied in the Haley/Johnson collaboration. Atoms involved in anion binding are bolded.

The original hosts in this series were designed in 2006 by Orion Berryman in the Johnson lab and Charles Johnson in the Haley lab and were based on bis-sulfonamide binding units.[25] Our first graduate student to be formally co-advised on this project, Calden Carroll, quickly discovered that bis­ureas such as 1 were superior as these could offer four or five hydrogen bonds to the anionic guest in reproducible 1:1 host–guest binding stoichiometries, depending on the protonation state of the pyridine.[14] This opened a floodgate of 13 eventual co-advised students and postdocs over the next 14 years who explored anion–π-interactions, the effect of increasing or reducing the number of urea binding units, the effects of different heterocyclic cores, and the physical organic chemistry of these receptors and their guests.[16] , [19] [20] [21] [22] [23] [24] Along the way, graduate student Blake Tresca further improved upon the early designs by replacing the pH-sensitive pyridine core with a benzene as in 2 and showed that the C–H···anion interaction could be tuned by installing electron-donating/electron-withdrawing groups para to the CH unit.[26] Very recently, postdoctoral researcher Jessica Lohrman extended this receptor class to include halo­gen-bonding recognition units by showing that we could switch out the urea groups with polarized iodine atoms (e.g., 3).[18]

One less explored binding motif in our studies is that of bisamides, such as shown in 4. One of our first examples of anion binding in the arylethynyl scaffolds was a derivative of 4 (n = 1) where Orion and Charles reduced the disulfide motif and protected it as the thioacetate; nonetheless, the receptor bound chloride weakly.[27] Several years later graduate student Chris Vonnegut and undergraduate Airlia Shonkwiler prepared a series of disulfide macrocycles based on 4 (n = 1–3).[18] [28] Of the molecules, thin films of 4 (n = 3) showed a turn-on fluorescence response in the presence of HCl, accompanied with a red shift in emission. X-ray crystallography revealed that the chloride anion sat nicely within the binding pocket of protonated 4 (n = 3). In contrast, when the film was treated with trifluoroacetic acid (TFA), it showed a different emission response than with HCl, suggesting a potential for discrimination between different anionic guests with host 4 (n = 3). We never guessed at the time that this simple modification would ultimately result in the surprising discovery of new PN-heterocycles.


# 2

Initial Discovery

The original purpose of the disulfide macrocycles was to function as cellular, redox-active probes between the fluorescent, reduced dithiol form and the nonfluorescent, oxidized disulfide state. Despite the best efforts of Calden and Chris,[28] [29] along with those of visiting Japanese graduate student Daisuke Inokuchi, we could never get this project over the proverbial goal line: one methylene (4, n = 1) was incredibly reactive and oxidized to the 13-membered disulfide ring in a matter of minutes, whereas formation of the larger macrocycles (4, n = 2–3) required a strong chemical oxidant (iodine). As a final effort, Chris and Airlia decided to target dithiol 5/disulfide 6 (Scheme [1]) with the belief that it would be easier to reduce/oxidize. Attempts to condense the starting bisaniline with thiosalicylic acid using a variety of peptide-coupling conditions only gave polymeric products. In one last-ditch attempt they investigated a procedure for aromatic polypeptide synthesis that uses triphenylphosphite (P(OPh)3) in the presence of pyridine at high temperature;[30] however, as if Monty Python themselves dictated the formed product, it was now time for something completely different.

Zoom Image
Scheme 1 Proposed synthesis of dithiol 5 and macrocyclic disulfide 6 redox pair along with the actual PN-heterocyclic product 7 formed from the reaction, which was confirmed by X-ray crystallography. ORTEP ellipsoids drawn at 50% probability; CCDC 1976534.

Analysis of the reaction mixture by TLC revealed the presence of a polar, highly fluorescent material as the main product. The 1H NMR spectrum of the yellow solution showed that there were clearly too many aromatic signals, meaning the material almost certainly was not 5 or 6. After sitting on the benchtop for a few days, yellow crystals suitable for X-ray diffraction had formed inside the NMR tube. Much to our surprise, the crystal structure revealed formation of the meta-terphenyl-like structure 7 where a 2,6-disubstituted pyridine ring linked two independent ‘phosphaquinoline’ moieties (Scheme [1]).[28] This molecule had little to no precedent in the literature, yet was clearly the main product of the reaction.

A quick literature search into similar systems dating back to the 1950s showed that while there are many known phosphorus heterocycles,[31] [32] [33] [34] [35] [36] [37] [38] [39] along with reviews on similar conjugated organophosphorus materials,[40,41] ‘PN-heterocycles’ (phosphorus- and nitrogen-containing) are rather rare as their synthesis and isolation are often plagued with difficulties.[42–52]

Realizing we had potentially uncovered an efficient route to PN-heterocycles, we went back to the drawing board and simplified the synthetic sequence to start from ortho-ethynylanilines, which in turn are easily prepared from commercially available or previously published ortho-haloanilines. As shown in Scheme [2], treatment of ethynylaniline 8 with only P(OPh)3 at 100 °C (i.e., omitting the thiosalicylic acid) afforded phosphonimidate 9, which we soon discovered could be hydrolyzed in wet THF to furnish phosphonamidate 10. Both structures were confirmed by X-ray crystallography (vide infra). With this simplified synthesis, Vonnegut and Shonkwiler prepared a series of congeners with aryl groups ranging from electron-rich to electron-poor (Scheme [2] and Table [1]).

Zoom Image
Scheme 2 Synthetic pathway for the various PN frameworks

Table 1 Photophysical Properties for Heterocycles 10

Entry

Compd 10

Yield (%)a

R

λabs (nm)b

λem (nm)b

Stokes shift (nm/cm–1)b

 1

10a

39

3,5-(F3C)2C6H3

349

434

85/5600

 2

10b

43

4-NCC6H4

349

430

81/5400

 3

10c

40

4-(EtO2C)C6H4

351

432

81/5300

 4

10d

80

4-ClC6H4

359

422

63/4200

 5

10e

32

Ph

342

418

76/5300

 6

10f

52

4-MeC6H4

341

413

72/5100

 7

10g

37

4-MeOC6H4

339

410

71/5100

 8

10h

–c

4-Me2NC6H4

–c

–c

–c

 9

10i

73

n-pen

318

383

65/5300

10

10j

22

2-pyridyl

363

442

79/4900

a Yields reported over two-step cyclization–hydrolysis.

b Values collected in CHCl3 at room temperature.

c Not determined as the imidate did not hydrolyze to the amidate form.

Focusing on heterocycles 10, nearly every derivative shared a common absorbance peak around 350 nm, yet emission wavelengths ranged from 383–442 nm and Stokes shifts ranged from 63–85 nm/4200–5600 cm-1 (Table [1]). Our initial communication on this class of compounds also reported details of the emission spectra, showing a correlation between more withdrawing substituent groups and more redshifted emissions.[53]

Chris then performed preliminary mechanistic studies to determine a rough reaction pathway for the cyclization (Scheme [3]).[28] He found that the reaction does not proceed using a non-nucleophilic base such as diisopropylethylamine, suggesting that the relatively nucleophilic pyridine first adds to the P(OPh)3. The aniline nitrogen then can add to the activated phosphorus center to afford the respective phosphoramidite upon deprotonation. The alkyne is next likely attacked by the phosphorus center before a series of proton transfers to produce the heterocyclic imidate 9.

Zoom Image
Scheme 3 Proposed cyclization mechanism for phosphonimidate 9 formation

# 3

Setbacks Breathe New Life

The initial excitement and momentum generated by our PN-heterocycle discovery was tempered by the fact both Vonnegut and Shonkwiler graduated in early 2016. Nonetheless, the project would be given continuity and new life – or so that was the plan – by two new junior research team members and an undergraduate researcher in the group, Noah Takaesu, who Chris and Airlia had trained. Unfortunately, unforeseen personnel issues and re-evaluation of career goals by the junior members left only Noah to push the project forward by the end of winter 2017. At the same time, the project had been injected with its first generous funding through the US NSF ‘INFEWS’ program in the summer of 2016. As a result, a serious reassessment of the project direction, research personnel, and continuity was needed.

While ‘hope springs eternal’ serves as far more powerful guidance in the context of the current global pandemic, the metaphor fit the timing of Spring 2017 for this project – we were fortunate to breathe new life into the project by hiring postdoctoral researcher Chun-Lin Deng and (somehow, miraculously) convincing first-year graduate student and the conceiver of this Account, co-author Jeremy Bard, to join the project. Though there was some overlap with Noah, he too soon graduated and left that summer for graduate school. Unfortunately, there was minimal knowledge transfer, something that everyone who has worked in a research lab knows is critical. Chun-Lin and Jeremy essentially had to start from scratch and figure out what to do with the project, and because of this, the rate of progress diminished greatly. Every research talk given on the project over the following year had a different introduction and scope, as we were trying to find the most compelling story we could tell and find how our work fit into the bigger picture of the grant and the over-arching project. Though a large amount of effort was being exerted, the barrier of no knowledge transfer with two new researchers to the project was almost too much to surmount. Some interesting results were generated over the next year, yet very few significant discoveries were made, resulting in no papers for close to 18 months. After an arduous group meeting at the end of spring 2018, we elected to pull the plug on the PN-heterocycle studies and allow the project to sunset. This is probably one of the last things that a graduate student wants to hear right after they have advanced to candidacy (i.e., Bard), having just put so much thought and effort into the future of the project. Admittedly that meeting was a letdown, although we left with enough inspiration for the rest of the year to wrap things up and try to be creative, as sunsetting a project can provide a certain sense of freedom perhaps to try new ideas. And as many theologians, musicians, and even a butler once said, ‘the darkest hour is before the dawn’.


# 4

A New Dynamic Duo Develops Dozens of Derivatives

As Jeremy and Chun-Lin worked to tie up loose ends and finish their ongoing projects, Bard went back to the beginning to build off the initial results from Vonnegut on the PN-naphthalene system analogous to 10. That initial report hinted that there was a lot that could be learned about this new moiety and begged the question as to whether phosphaquinolinone optoelectronic properties could be modified in a similar fashion as coumarin and carbostyril, two well-known, well-studied chromophores (Figure [2]).[54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] There is an obvious structural relationship between the latter two structures in Figure [2], where the carbonyl in carbostyril is replaced with an isolobal P(OR)=O group to afford the phosphaquinolinone scaffold.

Zoom Image
Figure 2 Structure of the phosphaquinolinone compared to analogous coumarin and carbostyril moieties

Taking inspiration from the structure–property relationships drawn for the coumarin and carbostyril family of fluorophores, Jeremy set out to perform structure–property relationship studies with substituent groups at two different sites upon the backbone. Frontier orbital occupancy calculations by Chun-Lin suggested that studying substituent effects of groups located at carbons 3 and 6 on the backbone (Figure [3, a]) would be the most fruitful.[73] [74] [75] [76] [77] Use of the same synthetic steps given in Scheme [2] furnished a large family of disubstituted heterocycles 11 (Figure [3, b]), the properties of which are discussed in the next section.

Zoom Image
Figure 3 (a) HOMO (gray) and LUMO (purple) occupancy levels on the carbons in the 3- and 6-positions of the phosphaquinolinone backbone as determined at the B3LYP/6-31g(d) level of theory. (b) Structure of disubstituted PN-naphthalene heterocycles 11 and (bottom) specific substituent groups and yields of each derivative for two-step cyclization–hydrolysis.

We were also curious about how extension of the arene π-electron backbone could affect the photophysical properties of the phosphaquinolinone moiety. Noah had originally started examining this idea by using the synthetic steps outlined in Scheme [2], though starting on naphthalene derivative 12 rather than phenyl derivative 8, to first give the PN-anthracene imidates 13 and subsequently amidates 14 in moderate to good yields (Table [2]).[78]

Table 2 Synthetic Yields for Heterocycles 14, 17, 20, and 27

Entry

Heterocycle

R

Yield (%)a

 1

14a

4-MeOC6H4

47

 2

14b

4-MeC6H4

70

 3

14c

4-F3CC6H4

33

 4

14d

4-NCC6H4

51

 5

17a

4-NCC6H4

34

 6

17b

4-F3CC6H4

31

 7

17c

4-MeOC6H4

33

 8

17d

4-MeC6H4

27

 9

20a

3,4-(NC)2C6H3

32

10

20b

4-F3CSO2C6H4

30

11

20c

4-NCC6H4

16

12

20d

4-MeSO2C6H4

39

13

20e

4-F3CC6H4

18

14

20f

4-MeOC6H4

16

15

27a

CN

21

16

27b

CF3

58

17

27c

Cl

31

18

27d

H

61

19

27e

t-Bu

65

20

27f

OEt

31

a Yields reported for two-step cyclization–hydrolysis.

Building upon this linear extension of the aromatic backbone (b ring in Scheme [2]), Chun-Lin’s first study focused on the effect of its nonlinear extension (a and c rings).[79] Two families of PN-phenanthrene derivatives were prepared, where the first started with aminonaphthalenes 15 that were cyclized to ‘bent-up’ PN-phenanthrene imidates 16 and subsequently hydrolyzed to ‘bent-up’ amidates 17. The second was made starting from cyclization of aminonaphthalenes 18 to ‘bent-down’ PN-phenanthrene imidates 19 and hydrolysis to amidates 20 (Scheme [2] and Table [2]).

Zoom Image
Scheme 4 Synthesis of PN-pyrene 21

To further expand the arene backbone, we next prepared PN-pyrene 21 following the steps outlined in Scheme [4]. Reaction of known nitropyrenol 22 with triflic anhydride furnished nitrotriflate 23. Sonogashira cross-coupling of 23 with 4-tolylacetylene furnished nitropyrene 24, which was then reduced using Zn in AcOH to yield aminopyrene 25. Lastly, cyclization of 25 with P(OPh)3 and then hydrolysis gave PN-pyrene 21.[80]

Having studied both substituent effects upon the backbone as well as effects of arene core modification, we then probed potential substituent effects directly upon the phosphorus center. For this, Jeremy, Chun-Lin, and rotation student Hannah Bates performed the cyclization reaction on a smaller set of substituted ethynylanilines 26, except this time using PhP(OPh)2 to afford the P-phenyl heterocycles 27 (Scheme [5]).[81] There were two reasons that we were interested in examining this P-substitution: (a) derivatives 27 have one less degree of freedom than their heterocycle 11 analogues, suggesting a potentially higher quantum yield, and (b) the change in both the steric and electronic nature of the phenyl group in 27 might have substantial effects upon the solid-state properties.


# 5

Physicochemical Properties

5.1

Fluorescence

The photophysical properties of the original PN-heterocycles 10 (Table [1]) achieved decent Stokes shifts and elucidated the relationship between emission energy and the electronic nature of the substituent groups. However, these initial results did not offer a thorough understanding as to why these properties were changing and/or why the general trends were present. In addition, neither complete characterization of the photophysical properties (i.e., absorption coefficients, quantum yields, fluorescence lifetimes) nor a comprehensive rationalization of the trends observed were performed at the time of the discovery.

Zoom Image
Scheme 5 Synthesis of heterocycles 27

Table 3 Photophysical Properties of Disubstituted PN-Naphthalenes 11 a

Entry

Compd

R1

R2

λabs (nm), ε (M–1cm–1)

λem (nm), φ (%)b

Stokes shift (nm/cm–1)

Brightness (ε·φ)

 1

11a

4-CN

CN

346, 12000

430, 18

 84/5646

2160

 2

11b

4-CN

CF3

341, 20000

429, 11

 88/6015

2200

 3

11c

4-CN

Cl

358, 23000

450, 11

 92/5710

2530

 4

11d

4-CN

t-Bu

358, 22000

453, 11

 95/5858

2420

 5

11e

4-OMe

t-Bu

344, 16000

421, 6

 77/5316

 960

 6

11f

4-Me

t-Bu

343, 14000

424, 8

 81/5569

1120

 7

11g

H

t-Bu

343, 17000

430, 8

 87/5899

1360

 8

11h

3,4-(CN)2

t-Bu

371, 19000

477, 41

106/5989

7790

 9

11i

4-Cl

t-Bu

345, 19000

431, 8

 86/5784

1520

10

11j

H

CN

337, 12000

410, 13

 73/5283

1560

11

11k

4-OMe

CN

345, 15000

471, 28

126/7754

4200

12

11l

H

CO2Et

335, 12000

411, 10

 76/5519

1200

13

11m

4-CN

OEt

379, 25000

490, 24

111/5977

6000

a Values collected in CHCl3 at room temperature.

b Excited at 365 nm and collected using a quinine sulfate (in 0.1 M H2SO4) standard.

Understanding these initial trends, building upon the initial family of molecules, and developing more complete sets of optoelectronic properties were the primary objectives Jeremy and his coworkers wanted to address with disubstituted heterocycles 11. Upon successful synthesis, their photophysical properties were measured and are listed in Table [3]. With this family, two trends for the effect of various substituent groups can be seen, where both more donating C6-substituent groups and more withdrawing C3-substituent groups led to redshifting in the emission. This correlates nicely with the frontier orbital occupancy calculations, which showed a predominance of the LUMO on C3 and a predominance of the HOMO on C6. These substituent effect trends support the hypothesis that the emission is redshifting due to a lowering of the LUMO or a raising of the HOMO, respectively. To further support this idea, Chun-Lin calculated the geometries and energy levels of the HOMOs and LUMOs for each derivative, uncovering an interesting predictive tool. Not only did we see the expected trends within the HOMOs and LUMOs of analogous derivatives 11a–m, but we also found a very nice correlation between the computationally predicted HOMO–LUMO energy gap and the experimental emission energies (Figure [4], blue). With this relationship, we realized that we potentially had an excellent way to predict the emission energies of future heterocycles based solely on their computationally derived HOMO–LUMO energy gaps.

Zoom Image
Figure 4 Relationship between the HOMO–LUMO gaps and the experimental emission energies of (blue) heterocycles 10b, 11a–d and (red) heterocycles 10b, 11a–d, and 11m

Based on this equation, we ‘screened’ many different theoretical congeners to see if we could achieve even more redshifted emission. Calculations suggested that an ethoxy group on the C6-position would have an immense effect on the emission energy, predicted to be 2.50 eV (496 nm). This was somewhat surprising, as the σpara value of the –OEt substituent was not that much more donating than the previously most donating tert-butyl substituent of 11d. Upon the synthesis of ethoxy-substituted 11m, the emission was determined to be at 490 nm (2.53 eV), giving a nearly perfect match (Table [3]). This demonstrated that we were able to reliably predict the emission energy of future congeners, as well as gave us a new predictive relationship between HOMO–LUMO energy gaps (Figure [4], red).

In addition to these two relationships, we also determined the complete set of photophysical properties (Table [3]). These values were all suboptimal for their potential use as fluorescent dyes in most cases and begged the question of how could they be improved. This question served as a foundation for many aims of the project mentioned below.

As shown by the diversity of syntheses in Schemes 2, 4, and 5, we spent a large amount of time and effort examining the variety of effects that different backbone modifications have on the fluorescence of these heterocycles. From Noah’s work with the PN-anthracenes, he found that 14 exhibited much more redshifted emissions and significantly larger Stokes shifts likely due to the increased conjugation of the backbone (Table [4]). While 14 possessed significantly larger absorption coefficients, the molecules still retained low quantum yields, leading to low brightness values.

Table 4 Photophysical Properties of PN-Anthracenes 14, PN-Phenanthrenes 17 and 20, PN-Pyrene 21, and P-Phenyl Heterocycles 27 a

Entry

Compd

λabs (nm), ε (M–1cm–1)

λem (nm), φ (%)b

Stokes shift (nm/cm–1)

Brightness (ε·φ)

 1

14a

339, 27000

464, 7

125/7900

 1900

 2

14b

331, 27000

469, 9

138/8900

 2400

 3

14c

327, 21000

492, 2

165/10000

  420

 4

14d

336, 31000

506, 4

170/10000

 1200

 5

17a

384, 13000

461, 12

 77/4300

 1600

 6

17b

375, 9800

450, 10

 75/4400

  980

 7

17c

374, 25000

442, 16

 68/4100

 4000

 8

17d

371, 15000

440, 9

 69/4200

 1400

 9

20a

404, 12000

493, 93

 89/4500

11000

10

20b

397, 5200

490, 80

 93/4800

 4100

11

20c

389, 10000

469, 36

 80/4400

 3600

12

20d

388, 7400

470, 66

 82/4500

 4900

13

20e

380, 6900

457, 29

 77/4400

 2000

14

20f

372, 6800

441, 19

 69/4200

 1300

15

21

395, 26000

465, 70

 70/3800

 8000

16

27a

348, 15000

447, 76

 99/6400

11400

17

27b

343, 18000

449, 38

106/6900

 6840

18

27c

360, 19000

474, 43

114/6700

 8170

19

27d

352, 18000

467, 50

115/7000

 9000

20

27e

360, 22000

475, 51

115/6700

11220

21

27f

381, 22000

515, 35

134/6800

 7700

a Values collected in CHCl3 at room temperature.

b Excited at 365 nm and collected using a quinine sulfate (in 0.1 M H2SO4) standard.

Following this study, Chun-Lin examined both ‘bent-up’ PN-phenanthrenes 17 and ‘bent-down’ PN-phenanthrenes 20, the photophysical properties of which are also given in Table [4]. With both families, we again noticed much more modest absorption coefficients and emission wavelengths yet found that quantum yields of up to 93% with heterocycles 20; however, the absorption coefficients kept brightness values low.

To overcome this issue, Jeremy, with the help of rotation student Jenna Mancuso, prepared PN-pyrene 21. Its emission maximum was 465 nm, affording a Stokes shift of 3800 nm–1. Gratifyingly, we found that we not only achieved a quantum yield of 70% but also attained an absorption coefficient of 26000 M–1cm–1, giving 21 a brightness value of 18000 M–1cm–1 (Table [4]). Further, fluorescence experiments at higher concentration suggested that above 2 mM in CHCl3, excimer formation could still be observed, illustrated by the growth of a very redshifted peak in the emission spectrum at 582 nm.[82] [83] [84] [85] [86] [87] With some modification, this motif could have potential for integration into larger systems that could take advantage of both its photophysical and supramolecular properties.

Another factor that can be controlled with the PN-heterocycles that cannot be altered in either coumarin or carbostyril is substitution at the phosphorus center. In the examples above, all contain a phenoxy group attached to the phosphorus center. To explore any potential substituent effects on this position as well, we prepared the P-phenyl family 27. The photophysical properties of 27 were very exciting, as we saw an almost universal improvement compared to heterocycles 11 (Tables 3 and 4), specifically a dramatic increase in the quantum yields and a slight redshifting in the emissions as well as improvement of the Stokes shifts in most cases. We attributed this to both a slight rigidification of the scaffold as well as an increase in planarity between the aromatic core and the pendent aryl group in the excited state of 27 (based upon TD-DFT calculated geometries). This family is also favorable, as the synthesis, isolation, and solubility all allow for its easier production and use, suggesting that any future fluorophore applications should include this modification.

Several different families of phosphaquinolinones have been built based off the simple cyclization found by Vonnegut and Shonkwiler, all of which having their own pros and cons. From these studies, we have accessed a wide range of emission wavelengths (383–515 nm), Stokes shifts (3800–10000 cm–1), and brightness levels (420–18000 M–1cm–1, Figure [5] and Figure [6]). In each family, we found that withdrawing substituents upon the pendent aryl groups lead to both redshifted emissions and subsequently larger Stokes shifts. We also determined that increasing the conjugation of the backbone leads to redshifting in the emission as well and can lead to increased brightness. Lastly, we found that the replacement of the P-phenoxy group with a P-phenyl group leads to even greater increases in brightness, redshifted emission colors, and larger Stokes shifts (Figure [6]).

Zoom Image
Figure 5 Representative phosphaquinolinone scaffolds colored by their respective experimental emission colors
Zoom Image
Figure 6 Comparison of all measured Stokes shifts and brightnesses

# 5.2

Molecular Structures

An advantage of the PN-heterocycles is their high crystallinity, thus making it relatively easy to obtain molecular structures via single-crystal X-ray diffraction. Figure [7] shows archetypical structures for the imidate (e.g., 9) and amidate (e.g., 10) forms. Based on the bond lengths of the 17 structures obtained to date, the aromatic rings in the backbone as well as the pendent aryl rings on the heterocycles behave as true aromatics with bond lengths ranging from 1.38–1.40 Å. Within the PN-heterocyclic rings, the C=C double bonds are best described as isolated double bonds with bond lengths around 1.34–1.36 Å. The P–C and N–C bond lengths are also tightly clustered in the range of 1.76–1.79 Å and 1.38–1.40 Å, respectively. The greatest difference is the P–N bond length, which changes from 1.55–1.56 Å in 7 and 9j (the only two imidate structures we have secured) to a range of 1.63–1.65 Å in the amidate form (where we have data for over a dozen structures). All of the structures show that the PN-heterocyclic rings have only small deviations from planarity. The dihedral angles between the heterocyclic cores and the pendent aryl rings varying from as little as 2° up to ca. 45°, which indicates very good to excellent communication between the two π-systems.

Zoom Image
Figure 7 Single-crystal structures of (a) 9j, (b) 10j monomer, and (c) 10j–10j dimer as representative structures for typical 9 and 10/11 structures. Ellipsoids drawn at 30% probability. Nonhydrogen bonding hydrogen atoms omitted for clarity.

A hallmark characteristic of the amidate structures is dimerization in the solid state due to the adjacency of the P=O hydrogen-bond acceptor and the N–H hydrogen-bond donor. This permits nearly all of the amidates to form strong centrosymmetric, head-to-tail meso-dimers between one R and one S enantiomer with N–H···O bond lengths ranging from 2.77–2.82 Å (Figure [7, c]). The tetrahedral phosphorus centers give the dimer a pseudo-eight-membered chair arrangement connected by the two head-to-tail N–H···(O)P hydrogen bonds. Interestingly, a decrease in the N–H···O bond lengths is observed when more withdrawing substituents are appended to the scaffold. These two factors hinted that there may be substituent effects in the solution-state dimer strengths as well (vide infra). The two exceptions to this common dimeric orientation involve sterically congested derivatives such as 11f, where a staggered, ‘polymeric’ supramolecular system forms in which a single monomer hydrogen bonds to two separate monomers rather than forming the normal head-to-tail dimer.


# 5.3

Solution Dimerization Studies

Zoom Image
Figure 8 Representative LFER plots for dimerization strengths of heterocycles (a) 11d-11g and (b) 10b, 11b, 11d, and 11m

Another series of studies performed on these scaffolds has been examination of substituent effects in the solution dimerization affinities of these compounds. To measure these dimerization strengths, we used variable-concentration (VC) 1H NMR spectroscopy experiments followed by subsequent data fitting using nonlinear regression analysis to measure the dimerization strengths of each heterocycle.[88] Though more polar solvents, including DMSO and CH3CN, destroy the dimer and/or permit limited solubility, H2O–saturated CHCl3 allowed us to collect consistent measurements across a range of similar compounds.

First, we performed these measurements upon heterocycles 11 as well as heterocycles 10b, 10e, and 10g (Table [5]). With these values, alongside the previously reported value of 130 M–1 for 10b, we determined linear free energy relationships (LFERs) for substituent effects at both C3 and C6. In these, we found that withdrawing groups at either end of the backbone led to an increase in dimer strength, as shown by the two representative linear free energy relationship plots (Figure [8]). Additionally, we determined that the R2 substituents had a larger effect upon the dimer strength than analogous R1 substituents, likely due to their proximity to the phosphaquinolinone hydrogen. Within this series of derivatives, we determined a dimerization constant of 525 M–1 for heterocycle 11j, which is among the highest reported value for similar head-to-tail hydrogen-bonding dimers.[89]

Table 5 Dimerization Strengths of Selected Examples of Heterocycles 10, 11, 17, 20, 21, and 27

Entry

Heterocycles

k dim (M–1)a

 1

10b

130

 2

10e

 62

 3

10g

 56

 4

11b

293

 5

11d

 70

 6

11e

 36

 7

11f

 38

 8

11l

200

 9

11m

 87

10

17a

206

11

17b

150

12

17c

146

13

17d

121

14

20b

306

15

20e

 84

16

20f

 77

17

21

179

18

27b

 82

19

27c

 54

20

27d

 24

21

27e

 22

a Values collected in H2O–saturated CDCl3 at room temperature. Values reported with errors less than 15%.

These findings prompted more questions. Why are these dimers so strong? Why do they prefer to dimerize as meso-dimers between R and S enantiomers? Are there any predictive trends similar to those observed for the emission energy? For answers, we turned to our collaborators 50 miles up the road from us at Oregon State University, the P. H.-Y. Cheong group. In Cheong’s group, graduate student Camille Richardson performed an extensive amount of modeling of our systems and came to a few important conclusions. First, she found that there is likely a stereoelectronic effect aiding both the strength of our dimers as well as directing them to form dimers in the meso-dimer orientation. She also determined that there was indeed a linear relationship that could be used to predict the strengths of each derivative based on the Hammett parameters of the two substituent groups.

Though solubility limited our ability to measure dimerization values for PN-anthracenes 14, we determined dimerization strengths for nearly all other π-extended PN derivatives (Table [5]), obtaining dimerization values from 121–206 M–1 for 17, 77–306 M–1 for 20, and 179 M–1 for 21. For PN-phenanthrenes 20, Chun-Lin performed electrostatic potential (ESP) mapping, dipole calculations, and noncovalent interaction (NCI) plots to help explain the strengths and orientations of the dimers. The ESP maps showed that the N–H and P=O were the most electron-poor and electron-rich sites upon the scaffold, respectively. When dimerized, these two sites interact, which leads to a cancelling of the net dipole of the two monomers while in the R-S meso-dimer form. When looking at the analogous S-S dimer, less cancelling of the net dipole exists, and hugely repulsive domains are seen between the two adjacent –OPh groups of each monomer.

Zoom Image
Figure 9 Comparison of all collected phosphaquinolinone dimerization strengths

The dimerization constants of the P-phenyl heterocycles 27 (Table [5]) range from 22–82 M–1, which are roughly three times weaker than the analogous P-phenoxy heterocycles 10 or 11. To explain this, we again turned to NCI analysis, as well as natural bonding orbital (NBO) analysis. When comparing the NCI plots of analogous 11 and 27 derivatives, a larger repulsive interaction between the phenyl ring of one monomer and the backbone of the other monomer is seen in 27 compared to the phenoxy in 11. Additionally, a stronger interaction is observed within the NBO plot between the no of the P=O moiety of one monomer and the σ*N–H of the other in 11 when compared to 27.

With all these derivatives, we can achieve very strong and self-specific dimer formation with the phosphaquinolinone moiety. A subtle interplay between sterics and electronics gives us many potential pathways to improve, modify, and tune the strengths of our dimers. The P-phenoxy families serve as more stable hydrogen bond dimers than the P-phenyl family, and we discovered that even the bulkier PN-phenanthrenes 17 and 20 can rival some of the strongest PN-naphthalene 11 dimer strengths (Figure [9]).


#
# 6

Applying What We Have Learned

6.1

Development of Supramolecular Host

With the breadth of fundamental studies now in hand, we were finally ready to apply what we had learned. Through some careful design, thoughtful choice of substituent groups for the PN moiety guided by Jeremy’s results, and many, many chromatography columns, Chun-Lin developed a two-armed host framework containing both PN-heterocycle and urea arms (28 in Scheme [6]).[90] The design of this unsymmetrical host featuring one urea ‘arm’ was critical: preliminary studies of bifunctional PN-compounds showed that the molecules had too strong a tendency to form hydrogen-bonded polymers rather than serve as monomeric hosts for other molecular guests (a tendency that might be attractive in its own right to researchers in supramolecular polymers). The synthetic design for these ‘hybrid hosts’ started with iodination of 11d with ICl to afford iodinated PN coupling partner 29 (Scheme [6]). The other coupling partner was built via sequential Sonogashira cross-coupling reactions starting with terminal acetylene 30 and the respective dihalobenzene to give asymmetric intermediates 31. Anilines 31 were then coupled with trimethylsilylacetylene (TMSA) to afford 32. Finally, desilylation in basic MeOH, cross-coupling with 29, and condensation with an isocyanate furnished hybrid hosts 28.

Zoom Image
Scheme 6 Synthesis of hybrid hosts 28. Relevant protons on 28 labeled for identification purposes for Figure [10].

Based on some simple geometric models, we believed that 28 would possess an excellent binding pocket for HSO4 – due to the tetrahedral phosphorus center allowing for complementary numbers of hydrogen-bond donors and acceptors as well as a nonplanar binding pocket. Upon initial tests, we were excited to see significant peak shifts in the 1H NMR and 31P NMR spectra of the host upon treatment with TBAHSO4 (Figure [10], bottom), from which we calculated a moderately strong association constant of 9600 M–1 in 10% DMSO-d 6 in CDCl3. This result became even more exciting when we ran similar experiments with several other similar anionic guests, and none showed significant shifting of the phosphaquinolinone N–H signal (Ha) (Figure [10]).

Zoom Image
Figure 10 Partial 1H NMR spectra of a 1.13 mM solution of 28b in 10% DMSO-d 6 in CDCl3 by itself and in the presence of 10 equiv of each guest

Chun-Lin quickly assembled a manuscript with these results yet missing was one essential piece of data – an X-ray structure of the host–guest complex. After hundreds of attempts and ‘one more try’ being said to the bosses countless times, Chun-Lin finally succeeded in growing single crystals suitable for X-ray crystallography, which clearly confirmed our design hypothesis. The structure showed seven hydrogen-bonding interactions between the host and the guest (Figure [11]), and provided an ideal binding pocket for this highly acidic protic oxoanion.

Zoom Image
Figure 11 Coordination environment of HSO4 – in the binding pocket of 28b in the solid state. Ellipsoids drawn at 30% probability. Non-hydrogen-bonding hydrogens omitted for clarity.

In addition to this nearly ideal geometric complementarity, the host exhibited reversible binding of HSO4 – as well. Upon simple liquid–liquid extraction conditions with sulfuric acid solutions, we observed that our host both survived and bound a significant amount of HSO4 –. After extraction, water washing of the organic layer of the host–guest complex showed a complete reversal of binding of the HSO4 – guest and complete recovery of the host. This also highlights one unusual feature of this class of compounds for molecule and ion recognition: the PN unit, while slightly acidic itself, is quite robust in acid and can enable binding of highly acidic guests. The P=O hydrogen-bond acceptor motif is particularly privileged for this attribute, as it is an excellent hydrogen-bond acceptor while paradoxically being a poor base (pK aH+ << 0).


# 6.2

Use of PN Moiety as an Impressive Fluorophore

Taking advantage of all we had learned about the fluorescence of these heterocycles, Chun-Lin developed the π-extended tricyclic PN-pyrenes 33 (Scheme [7]).[91] Starting from aminopyrenes 34, cyclization with PhP(OPh)2 and subsequent iodination furnished intermediates 35. Lastly, Sonogashira cross-coupling accompanied by Pd-mediated indole formation gave 33a–d. As this system includes the P-phenyl modification seen in 27, the increased π-system of 21, and the appropriate placement of substituent groups guided by our previous studies, we were able to access improved photophysical properties (large Stokes shifts, high brightness, and redshifted emission; Table [6]) With these compounds we achieved both emission wavelengths up to 622 nm as well as the highest brightness values we have measured for the phosphaquinolinone moiety (9700–22000 cm–1M–1). Through a series of computations that modeled the geometries and interactions in the excited state and spectroscopic experiments, we deduced that this was due to an increase in planarity upon excitation, as well as a suppression of any excimer formation. Additionally, we found that using a mixture of 33a, 33c, and 33d we were able to produce nearly perfect white light emission in solution (Figure [12]).

Zoom Image
Scheme 7 Synthesis of 3-ring PN pyrenes 33
Zoom Image
Figure 12 Emission color coordinates in the CIE 1931 chromaticity diagram of a 2:1:10 v/v/v mixture of 33a/33c/33d excited at 430 nm

Table 6 Photophysical Properties of 3-Ring PN-Pyrene Derivatives 33 a

Entry

Compd

λabs (nm), ε (M–1cm–1)

λem (nm), φ (%)b

Stokes shift (nm/cm–1)

Brightness (ε·φ)

1

33a

430, 38000

459, 57

 29/1500

22000

2

33b

431, 32000

551, 62

120/5100

20000

3

33c

450, 32000

521, 61

 71/3000

19000

4

33d

437, 42000

622, 23

185/6800

 9700

a Values collected in CHCl3 at room temperature.

b Excited at 365 nm and collected using a quinine sulfate (in 0.1 M H2SO4) standard.


#
# 7

Conclusions and Outlook

In the first three years of this new project, it went through quite a rollercoaster ride – as many projects do – that might have derailed it had it not been for the tenacity of Jeremy and Chun-Lin. The project went from an exciting accidental discovery, to something that had great potential during a period of research staff turnover, to a project which was slated for termination. However, over the last 18 months, a new light has shined on this fluorophore class (or was it emitted?) through both fundamental studies and initial application-based projects. After a middle of the night stay of execution, what does this new metaphorical morning hold for this project?

With all we have learned from our original studies, and the variety of skills, instruments, and connections under the project’s toolbelt, new directions are not lacking. We think some of the unusual properties of this receptor will enable new applications: the receptor is a fluorophore highly stable in acidic media, it has a strong hydrogen-bond donor adjacent to a powerful yet nonbasic hydrogen-bond acceptor, and the composition of matter is new while maintaining similarities to venerable compound classes of use in chemical biology, medicine, and supramolecular polymers among others. Work is currently underway looking for how this new recognition motif might target other oxoanionic guests, including phosphates due to their relevance in many areas. Additionally, perhaps following inspiration from D.W.J.’s academic family, work into utilization of this unusually strong dimer-forming moiety in a supramolecular capsule system is under consideration. Finally, with what we know about functionalizing this fluorophore, either through postsynthetic modification or careful derivatization from the beginning, integration into both biological and other imaging applications might be around the corner.


#
#

Acknowledgment

We gratefully acknowledge the important contributions of our UO co-workers and OSU collaborators cited in in the text and throughout the Haley/Johnson group references.



   Permissions and Reprints All articles of this category
Zoom Image
Zoom Image
Zoom Image
Zoom Image
Figure 1 Hydrogen- and halogen-bonding supramolecular hosts prepared and studied in the Haley/Johnson collaboration. Atoms involved in anion binding are bolded.
Zoom Image
Scheme 1 Proposed synthesis of dithiol 5 and macrocyclic disulfide 6 redox pair along with the actual PN-heterocyclic product 7 formed from the reaction, which was confirmed by X-ray crystallography. ORTEP ellipsoids drawn at 50% probability; CCDC 1976534.
Zoom Image
Scheme 2 Synthetic pathway for the various PN frameworks
Zoom Image
Scheme 3 Proposed cyclization mechanism for phosphonimidate 9 formation
Zoom Image
Figure 2 Structure of the phosphaquinolinone compared to analogous coumarin and carbostyril moieties
Zoom Image
Figure 3 (a) HOMO (gray) and LUMO (purple) occupancy levels on the carbons in the 3- and 6-positions of the phosphaquinolinone backbone as determined at the B3LYP/6-31g(d) level of theory. (b) Structure of disubstituted PN-naphthalene heterocycles 11 and (bottom) specific substituent groups and yields of each derivative for two-step cyclization–hydrolysis.
Zoom Image
Scheme 4 Synthesis of PN-pyrene 21
Zoom Image
Scheme 5 Synthesis of heterocycles 27
Zoom Image
Figure 4 Relationship between the HOMO–LUMO gaps and the experimental emission energies of (blue) heterocycles 10b, 11a–d and (red) heterocycles 10b, 11a–d, and 11m
Zoom Image
Figure 5 Representative phosphaquinolinone scaffolds colored by their respective experimental emission colors
Zoom Image
Figure 6 Comparison of all measured Stokes shifts and brightnesses
Zoom Image
Figure 7 Single-crystal structures of (a) 9j, (b) 10j monomer, and (c) 10j–10j dimer as representative structures for typical 9 and 10/11 structures. Ellipsoids drawn at 30% probability. Nonhydrogen bonding hydrogen atoms omitted for clarity.
Zoom Image
Figure 8 Representative LFER plots for dimerization strengths of heterocycles (a) 11d-11g and (b) 10b, 11b, 11d, and 11m
Zoom Image
Figure 9 Comparison of all collected phosphaquinolinone dimerization strengths
Zoom Image
Scheme 6 Synthesis of hybrid hosts 28. Relevant protons on 28 labeled for identification purposes for Figure [10].
Zoom Image
Figure 10 Partial 1H NMR spectra of a 1.13 mM solution of 28b in 10% DMSO-d 6 in CDCl3 by itself and in the presence of 10 equiv of each guest
Zoom Image
Figure 11 Coordination environment of HSO4 – in the binding pocket of 28b in the solid state. Ellipsoids drawn at 30% probability. Non-hydrogen-bonding hydrogens omitted for clarity.
Zoom Image
Scheme 7 Synthesis of 3-ring PN pyrenes 33
Zoom Image
Figure 12 Emission color coordinates in the CIE 1931 chromaticity diagram of a 2:1:10 v/v/v mixture of 33a/33c/33d excited at 430 nm