Weak Affinity Chromatography Using Organic Solvents on Cyanopropyl- and Arene-Modified Silica-Gel Columns—Structural Demands for Host and Guest Moieties and Effect of Solvents

Abstract

Weak affinity chromatography (WAC) is effective for isolating target compounds from analogous compounds with similar functional groups. Previously, a few chromatographic behaviors based on WAC using organic solvents were observed in a series of cyclic multiporphyrin systems on cyanopropyl-modified silica gel (CN-MS). Here, three cyclic porphyrin trimers with various rigidity were examined on CN-MS to understand the mechanism of the specific interactions between porphyrin derivatives and functional groups on modified silica gel. In addition to CN-MS, six modified silica-gel columns were tested to compare their retention abilities for a cyclic nickel porphyrin dimer (C₄Ni₂MsCP2). We examined the cosolvent effects of the pyridine eluents for C₄Ni₂MsCP2. Apparent dissociation constants of C₄Ni₂MsCP2 with functional groups on the MS columns and effective amounts of the functional groups were estimated by frontal affinity chromatography (FAC). ¹H NMR titrations of acetonitrile and nitrobenzene to C₄Ni₂MsCP2 were conducted to compare their association constants with movable guest molecules to the dissociation constants with immobilized functional groups obtained in FAC. We found rigidity of cyclic porphyrin derivatives and immobilization of functional groups on silica gel is necessary for significant retentions using WAC. The affinity interaction does not occur at the center of C₄Ni₂MsCP2, but probably occurs on the surface composed of a bipyridyl moiety and the adjacent edges of the two porphyrins. Polar solvents, such as nitrobenzene, acetonitrile, and methanol, weakened the interaction. Although C₄Ni₂MsCP2 dissolves well in chloroform, the interactions between C₄Ni₂MsCP2 and the MS columns are considerably strengthened in the presence of chloroform. The competitiveness of solvents and cosolvents with the interaction of the porphyrin on WAC is independent of the solubility of the analyte.

Porphyrin, Non-bonding interaction, Molecular recognition

1. Introduction

Liquid chromatography (LC) is a major purification method used to separate nonvolatile organic compounds from mixtures under ambient conditions. The scope of organic compounds to be separated is extremely expanded into low-to-high molecular weights, charged and uncharged, polar and less polar compounds by using various methods and columns. The methods are classified as normal and reverse-phase chromatography, ion exchange chromatography, gel permeation chromatography (GPC), affinity chromatography, and others based on their principles.¹ In each method, various columns composed of various kinds of stationary phase are available. Because a combination of columns and eluents provides different chromatograms from a mixed analyte, the variation of chromatography is vast. Reverse-phase chromatography using silica gel modified with organic functional groups is one of the most popular methods. Using a gradient of aqueous eluents, efficient separation of a mixed analyte is often observed. For example, separations of constituent isomers of disubstituted benzene derivatives are reported on some functionalized columns, such as naphthylethyl- and pyrenylethyl-modified columns.² Because the nonbonding interactions between the aromatic groups on the MS and analytes are strong in aqueous eluent systems, reverse-phase systems with a gradient are used frequently in analytical LC. In a gradient LC system, regeneration of the used column is required before the next separation. By contrast, under isocratic elution conditions, in which the same eluent system is used throughout the separation, repeated application of analytes is possible, and the eluted solvent can be easily recycled because the solvent composition is constant. Therefore, isocratic elution conditions have a substantial advantage in time and economy in preparative chromatography. Isocratic elution conditions are generally used in GPC, and recycling GPC systems³ are useful to separate a mixture difficult to separate by one chromatographic run.⁴ Another type of chromatography operated under isocratic elution conditions is weak affinity chromatography (WAC).⁵ In WAC, nontarget materials are eluted at the first fraction without retention, then the target compounds are eluted specifically with some retention. WAC is applied to relatively weak specific interactions (with dissociation constants, K_d, ranging in the order of from 10⁻⁴ to 10⁻⁶ M) between analytes and ligands immobilized on the stationary phase.⁶ Biomaterials, such as proteins, are used for either analyte or immobilized material in reported WAC, and the eluent systems are aqueous or buffer solutions. To our knowledge, WAC systems using only synthetic materials and organic solvent eluents are unknown except for our previous cyclic porphyrin systems.

Previously, unique retention behaviors of some cyclic multiporphyrin derivatives were observed on cyanopropyl- and arene-MS columns with an isocratic nonaqueous pyridine eluent system. The first example was discovered in a project to construct multicofacial multinuclear complexes.⁷ Reductive coupling of bis(chloropyridyl)porphyrinatozinc gave a mixture of acyclic porphyrin oligomers and the target cyclic porphyrin trimer (Scheme 1). In the early periods of the project, a recycling GPC system was used to isolate the target cyclic compound. However, the isolated yield substantially decreased because eluted fractions contaminated with acyclic oligomers were eliminated during the purification. Normal silica gel (as used in thin layer chromatography) and various MS columns (as used for high-performance LC, HPLC) were tested to improve the chromatographic separation. We found only the target cyclic trimers were retained on cyanopropyl- and arene-modified silica-gel (MS) columns with a mixture of pyridine and toluene as an eluent, whereas monomer and acyclic oligomers were not retained at all. Preparative isolation of the target materials was conducted by using cyanopropyl MS for flash chromatography. The isolation yield was extremely high because of no overlapping with other porphyrin derivatives during the elution. In addition, the used cyanopropyl MS can be reused for another separation. In our following reports, the WAC method was applied to isolate four cyclic bisporphyrins, C₄Ni₂MsCP2, Zn₂MsCP2_o, Zn₂MsCP2_m, and Zn₂MsCP2_p (Figure 1).^8,9 In the case of Zn₂MsCP2_o, the WAC method was ideally efficient. Thus, the target compound was isolated quantitatively from the reaction mixture, in which Zn₂MsCP2_o was only included as 3%.

Scheme 1.

Figure 1.

Structures of cyclic porphyrins retained on cyanopropyl-modified silica-gel columns.

We have observed that distances between two porphyrins and the rigidity of cyclic compounds seem important for their retention on MS. Although the acyclic oligomers are composed of the same units of porphyrin and pyridyl moieties, no retention was observed. It appears evident that conformationally restricted cyclic structures are essential for retention on MS columns. A spacer to connect two porphyrins is also important. A rigid cyclic C-(Zn-Zn-Zn)₃ (Figure S1),¹⁰ in which porphyrins are connected through m-phenylene linkers, was not retained on CN-MS. At present, 6,6′-porphyrin-substituted 2,2′-bipyridyl is a specific spacer for the WAC observation. In the previous paper, phthalic (o), isophthalic (m), and terephthalic (p) diamides were used as the other linkers in Zn₂MsCP2_o, Zn₂MsCP2_m, and Zn₂MsCP2_p, respectively.⁹ Among the three, Zn₂MsCP2_o, in which the two porphyrins are located most closely to each other, is retained most strongly. By contrast, relationships between the porphyrin distances and retention abilities were inverted in Zn₂MsCP2_m and Zn₂MsCP2_p. Because the conformational restriction of Zn₂MsCP2_p is higher than that of Zn₂MsCP2_m, the rigidity of cyclic porphyrins seems to be another important factor.

Determining the scope and mechanism of the unique WAC operating under organic solvent elution will contribute substantially to developing separation science and technology as well as fundamental chemistry concerning nonbonding interactions and solvation. In the present report, we clarify the hypothesis proposed above, and we have first compared three cyclic porphyrin trimers having zinc, metal-free, and nickel ions (Scheme 2). The ring sizes are almost the same, but their rigidities will vary because the sizes of the chelated ions are different. In addition to studying differences in rigidity, we conducted the following systematic experiments to understand the mechanism of the characteristic interaction on MS with a rigid compound, C₄Ni₂MsCP2, as an analyte, whose structural analyses were previously achieved by both nuclear magnetic resonance (NMR) and X-ray diffraction.⁸ Relationships between the ratio of retention times (RRT) and structures of functional groups on the MS columns were examined, along with the effects of the solvent and cosolvent used as eluents. Estimations of apparent dissociation constants with functional groups on MS and effective ligand amounts on the MS columns were determined by frontal affinity chromatography (FAC). Association constants were estimated by NMR spectroscopy using acetonitrile and nitrobenzene as movable ligands.

Scheme 2.

2. Experimental

General Procedure

All chemicals and solvents were of commercial reagent quality, and used without further purification unless otherwise stated. Syntheses of ZnMsP1, FbMsP1, C₄NiMsP1, ZnP, Zn₃MsCP3, Fb₃MsCP3, and C₄Ni₂MsCP2 were previously reported.^7–9Ni₃MsCP3 was prepared by treatment of Ni(acac)₂ with Fb₃MsCP3. Reactions were monitored on silica gel 60 F₂₅₄ TLC plates (Merck). The silica gel utilized for column chromatography was purchased from Kanto Chemical Co. Inc.: Silica Gel 60N (Spherical, Neutral) 60–210 µm and 40–210 µm (Flash). Medium pressure liquid chromatography was performed on a Yamazen Pump540, UV-detector PREP UV-10V, and fraction collector FR50N. HPLC was performed on a JASCO PU-2080plus and MD-2018plus (photodiode array detector, PDA) system, or a JASCO PU-2089plus and UV-2075plus (UV-vis detector) system with various GPC and modified silica-gel columns. For GPC analysis, two TSK gel G2500H_HR (Tosoh company, exclusion limit: 20,000) columns and one G2000H_HR (Tosoh company, exclusion limit: 10,000) column were connected successively, and pyridine was used as an eluent. Modified silica-gel (MS) columns, COSMOSIL^® series (4.6 mm I.D. × 15 cm, 4.6 mm I.D. × 5 cm, 2.0 mm I.D. × 15 cm): C₁₈-MS-II, Cholester, PYE, π-NAP, NPE, Br, PFP, PE-MS and CN-MS, were purchased from Nacalai Tesque company. For flush preparative chromatography by use of modified silica-gel, Cyanogel was purchased from Yamazen Corporation and packed in a glass column (2 cm I.D. × 6.5 cm), and a mixture of toluene and pyridine (15/85 (vol/vol)) were used as an eluent. NMR spectra were recorded on a JEOL ECA-300, JEOL ECX-400 or ECA-500, and chemical shifts were recorded in parts per million (ppm) relative to tetramethylsilane. Hi-resolution MALDI–TOF mass spectra were collected on a JEOL JMS S-3000 with dithranol or trans-2-[3-(4-tert-butylphenyl)-2-methyl-2-propenylidene]malononitrile (DCTB) as a matrix with and without sodium iodide (NaI). A mixture of polyethylene glycol and NaI was used for internal standards in the hi-resolution MALDI–TOF mass spectra. Mass data analyses were performed on msTornado Analysis (JEOL) and mMass ver.5.5. (Open Source Mass Spectrometry Tool, http://www.mmass.org/). UV-vis spectra were collected on a JASCO V-650 spectrometer. Fluorescence spectra were collected on a Hitachi F-4500 spectrometer. Curve fitting analyses of chromatography were performed using ORIGIN 2020 software.

Ni₃MsCP3

A mixture of Fb₃MsCP3 (2.3 mg, 1.2 µmol) and nickel(II) bis(acetylacetonate) (3.2 mg, 12 µmol) was heated in toluene at 90 °C for 72 h. The mixture was concentrated, and the residue was dissolved in CHCl₃. The organic layer was washed with 0.1 M ethylenediaminetetraacetic acid aqueous solution, dried over anhydrous Na₂SO₄, and concentrated under reduced pressure to give Ni₃MsCP3 (2.4 mg). ¹H NMR (500 MHz, CDCl₃). δ 8.95 (d, 6H, J = 1.5 Hz), 8.76 (d, 6H, J = 1.5 Hz), 8.45 (d, 12H, J = 5.0 Hz), 7.29 (d, 12H, J = 5.0 Hz), 6.99 (s, 12H), 4.01 (s, 18H), 2.50 (s, 18H), 1.37 (s, 36H). ¹³C NMR (125 MHz, CDCl₃). δ166.0 (C), 162.3 (C), 156.5 (C), 142.8 (C), 142.7 (C), 139.0 (C), 137.4 (C), 136.9 (C), 131.7 (CH), 130.9 (CH), 128.3 (CH), 127.5 (CH), 121.0 (CH), 117.6 (C), 116.7 (C), 53.1 (CH₃), 21.4 (CH₃), 21.2 (CH₃). MALDI-TOF MS (matrix: dithranol) calcd. for [M (C₁₅₆H₁₂₀N₁₈Ni₃O₁₂) + H]⁺: 2611.7466, obs. 2611.7478.

Frontal Chromatography

On the HPLC system equipped by MS columns (4.6 mm I.D. × 150 mm), a long stainless sample tube (I.D. 1 mm, 40 m long, inner volume ca. 31.4 mL) was attached to the 6-way valve (Rheodyne). A sample solution was continuously introduced into a column. On the chromatogram, a frontal curve appears at later than RT 1.5 min. When the chromatograph reached a plateau, introduction of the sample was shut. The chromatogram was analyzed as shown in Figure 4.

Computational Methods

Materials Studio 2020 software (BIOVIA) was used to construct molecular models. Semi-empirical molecular orbital method (PM6) was used to obtained local minimized structures. The DFT calculation was carried out using the Gaussian 09 program.¹¹ The structure was optimized using the B3LYP functional with the 6-31G(d) basis set for all atoms except for Ni and with the LANL2DZ basis set for Ni. The stationary point was verified using vibrational analysis.

3. Results

Comparison of Zn₃MsCP3, Fb₃MsCP3, Ni₃MsCP3, and C₄Ni₂MsCP2

The demetalation of Zn₃MsCP3 with trifluoroacetic acid followed by neutralization gave the free base Fb₃MsCP3, and the reaction of Fb₃MsCP3 with Ni(OAc)₂ afforded Ni₃MsCP3 (Scheme 2). ¹H NMR spectra of Zn₃MsCP3, Fb₃MsCP3, Ni₃MsCP3, and C₄Ni₂MsCP2 at 298 K are presented in Figure 2a–d, respectively. Only two kinds of bipyridyl protons and two kinds of β-pyrrole protons are observed in the range between 9.3 and 8.1 ppm in all the cyclic compounds, indicating that the cyclic compounds are apparently symmetric concerning the cyclic parts composed of bipyridines and porphyrins. Differences among them are observed in the signals of mesityl substituents around 7 ppm (filled circles) and 1–2 ppm (filled triangles). The inside aryl-H and inside methyl protons in the mesityl groups appear in higher shield regions compared with those of the corresponding outside protons (Figure 2e). In Zn₃MsCP3 and C₄Ni₂MsCP2, the inside and outside protons are independently observed as sharp peaks. By contrast, the peaks are broadened in Fb₃MsCP3. In Ni₃MsCP3, the inside and outside protons are apparently identical at 298 K. Because signals were separately observed at 195 K in CD₂Cl₂ (Figure S2), mesityl substituents in Ni₃MsCP3 smoothly rotate around the C–C bond connected through the porphyrin and the mesityl groups at 298 K. The rotation of mesityl groups is related to the rigidities of the porphyrin skeletons. In general, single nickel porphyrins take various conformations.¹² In the case of the less strained cyclic nickel porphyrin trimer, Ni₃MsCP3, such conformational changes occur in each nickel porphyrin moiety. By contrast, the rigid cyclic nickel porphyrin dimer, C₄Ni₂MsCP2, is severely strained. Because no conformational change occurs in each nickel porphyrin moiety, the rotation of mesityl parts is suppressed in C₄Ni₂MsCP2. Chromatograms of Zn₃MsCP3, Fb₃MsCP3, and Ni₃MsCP3 on a cyanopropyl-modified silica-gel (CN-MS) column using pyridine as an eluent are shown in Figure 2f–h. In this figure, R₀ at a retention time (RT) of 2.2 min indicates the peak position of chloroform and the corresponding porphyrin monomers shown in Figures S1 and S3, which were not retained on the column under the same conditions. The ratio of the RTs of analyte and monomer (R/R₀) is defined as RRT. The observed RTs and RRTs are tabulated in Table 1. Here, a larger RRT value indicates that the analyte is retained more strongly on the CN-MS column, whereas RRT = 1 indicates no retention. Zn₃MsCP3 and Fb₃MsCP3 were moderately retained to a similar extent, whereas Ni₃MsCP3 showed almost no retention. Because C₄Ni₂MsCP2 was significantly retained, the presence of nickel ion does not directly determine the retention behavior of Ni₃MsCP3. A comparison of the dynamics of mesityl groups on Ni₃MsCP3 and the other two derivatives (Zn₃MsCP3 and Fb₃MsCP3) concludes that the rigidity of the cyclic structure greatly affects the retention on the CN-MS column. This conclusion is consistent with the results observed in Zn₂MsCP2_o, Zn₂MsCP2_m, and Zn₂MsCP2_p.⁹

Figure 2.

¹H NMR spectra in CDCl₃, (a) Zn₃MsCP3 (300 MHz), (b) Fb₃MsCP3 (300 MHz), (c) Ni₃MsCP3 (500 MHz), (d) C₄Ni₂MsCP2 (400 MHz). (e) Assignments of characteristic signals. Asterisks in (a)–(c) indicate impurities. (f)–(h) Chromatogram charts (column: CN-MS (4.6 mmID × 15 cm), eluent: pyridine, flow rate: 1 mL/min, monitored at 536 nm), (f) Zn₃MsCP3, (g) Fb₃MsCP3, (h) Ni₃MsCP3. Arrows (R₀) in (f)–(h) indicate the peak positions of corresponding monomeric porphyrins in Figure S1. Chloroform is also eluted at R₀.

Table 1.

RT and ratio of retention times (RRT)

Run	Porphyrin	RT/min	RRT
1	C4NiMsP1	2.18	1
2	Zn3MsCP3	3.87	1.8
3	Fb3MsCP3	4.10	1.9
4	Ni3MsCP3	2.22	1.0
5	C4Ni2MsCP2	8.70	4.0

Conditions: CN-MS column ID 4.5 mm, 15 cm, eluent: pyridine 100%, 1.0 mL/min. Pyridine solutions of porphyrin analytes were injected.

Coordination of pyridine to metal ions in porphyrins may affect the retention because large amounts of pyridine used as an eluent are present. In Figure S3, UV–vis spectra of monomers of free base, zinc and nickel porphyrins, and cyclic compounds, Ni₃MsCP3, Fb₃MsCP3, Zn₃MsCP3, and C₄Ni₂MsCP2 are shown along with their chromatograms. Coordination of pyridine to the metal ions in porphyrins can be confirmed qualitatively by the redshift of the Soret band observed between 400 and 500 nm and Q-band between 500 and 650 nm. The Soret band and Q-band observed at 440 nm and 560 nm, respectively, indicate that the metalloporphyrin is coordinated with pyridine, whereas they are observed at 420 nm and 530 nm, respectively, when the porphyrin is not coordinated. In the case of monomers of zinc and nickel porphyrins, five and six coordination complexes are formed, respectively. Based on these indexes, it is estimated that all three zinc porphyrins in Zn₃MsCP3 are coordinated by pyridine, whereas approximately 60% of nickel porphyrins in Ni₃MsCP3 are coordinated (Figure S4). The coordinated and uncoordinated nickel porphyrins in Ni₃MsCP3 exchange rapidly at 298 K on the NMR timescale by coordination and dissociation of pyridine molecules (Figure S5). In the case of constrained C₄Ni₂MsCP2, no coordinated nickel porphyrin was observed. Free-base compound Fb₃MsCP3, of course, gave no coordinated species. Because Zn₃MsCP3, Fb₃MsCP3, and C₄Ni₂MsCP2 were retained on the CN-MS column, the existence of the coordination by pyridine does not directly determine the retention. In conclusion, in comparisons between Ni₃MsCP3, Fb₃MsCP3, and Zn₃MsCP3, rigidity is crucial as a structural demand for cyclic compounds.

Chromatographic behavior of C₄Ni₂MsCP2–Solvent Effects and Comparison among Various Modified Silica-Gel Columns

In this section, chromatographic conditions and comparison of the MS columns were examined systematically using C₄Ni₂MsCP2 as an analyte, which is the most rigid and the most retained compound synthesized previously in our group.⁴ In general, appropriate solubility and boiling point for evaporation are required for the eluent used for preparative chromatography with an isocratic eluent. In the case of WAC, the eluent should act somewhat competitively for interaction between host and guest, one of which is immobilized on the stationary phase. The degree of competition should be controlled by changing the contents of the eluent. Solubilities of chloroform, pyridine, acetone, toluene, methanol, and acetonitrile were tested (Table 2), which could be evaporated using a conventional evaporator. Chlorinated hydrocarbons, such as chloroform, are good solvents to dissolve C₄Ni₂MsCP2. Except for chlorinated hydrocarbons, pyridine was the best solvent in Table 2. The solubilities of other solvents were too low to use as an eluent. It is noteworthy that butyl ester substituents in C₄Ni₂MsCP2 are important for dissolving it in chlorinated hydrocarbons and pyridine. When purified, the solubility of the corresponding methyl ester C₁Ni₂MsCP2 is low, and it would not dissolve further in any solvent, including chloroform, in which it precipitated.⁸ Therefore, significant self-aggregation of the methyl ester derivative C₁Ni₂MsCP2 occurs, and butyl ester substituents in C₄Ni₂MsCP2 prevent self-aggregation.

Table 2.

Saturation concentrations of C₄Ni₂MsCP2

Solvent	Saturated molar concentration (10−4 M)	Solubility (g/L)
Chloroform	98	19
Pyridine	9.3	1.8
Acetone	3.5	0.67
Toluene	3.3	0.63
Methanol	1.0	0.19
Acetonitrile	0.28	0.054

C₄Ni₂MsCP2 (molar mass 1911.57 g/mol) at 298 K

A chromatogram using chloroform as a dominant eluent on a CN-MS column (5 cm length) is shown in Figure S6. C₄Ni₂MsCP2 was gradually eluted after an RT of 10 min, but no peak was observed within 20 min because of the significant adsorption on CN-MS. When a strong competitor, nitrobenzene, was used as a cosolvent with chloroform, an elution peak of C₄Ni₂MsCP2 was observed. However, 40 vol% of nitrobenzene was required to suppress the adsorption. Therefore, chloroform was not appropriate to use as a dominant eluent. By contrast, chromatograms using pyridine as an eluent gave a peak within acceptable periods, indicating that pyridine acts as an appropriate competitor to the interaction. Because pyridine has moderate solubility for porphyrin derivatives in our experience, pyridine was chosen to be used as a dominant eluent in a series of MS chromatographies. Relationships between the concentration of samples injected (2.4–0.20 mM in CHCl₃) and their elution times under the same chromatographic conditions were studied (Figure S7). The positions of the peak tops are plotted as a function of the concentration in Figure S8. Because increasing the concentration of the samples shifted the peak positions forward, a fixed amount of sample solution having the same concentration must be injected to compare RTs among various columns. In the following chromatographic experiments, 10 µL of a 0.7 mM sample solution in chloroform was injected. To examine the cosolvent effect for host–guest interaction in MS chromatography, 40–10 vol% of acetonitrile was mixed with pyridine eluent (Figure S9). We found 20 vol% of cosolvent significantly affected the RT, and the use of 20 vol% did not precipitate the analyte. Therefore, 20 vol% of cosolvents were used to examine their solvent effect. In Figure S10, chromatograms of C₄Ni₂MsCP2 in CN-MS using 20 vol% of cosolvents, nitrobenzene, acetonitrile, methanol, chlorobenzene, acetone, toluene, and chloroform in pyridine eluent are presented. Compared with that using 100% pyridine, the analyte was eluted earlier with nitrobenzene, acetonitrile, and methanol, suggesting that those cosolvents interfere with the interaction between C₄Ni₂MsCP2 and CN-MS more strongly than pyridine. Chlorobenzene and acetone were comparable to pyridine. With toluene and chloroform, the RTs were longer than that in pyridine alone. In the case of monomer C₄NiMsP1, no cosolvent effect was observed (Figure S11). As an index to compare the strength of interaction between the analyte and MS, RRT is used, where the RT of C₄Ni₂MsCP2 is divided by the RT of C₄NiMsP1 (R₀) under the same chromatographic conditions. The RRTs of C₄Ni₂MsCP2 using 100% pyridine as the eluent on nine kinds of MS columns (Figure 3a) are presented in Figure 3b. RRT > 6 was observed on MS columns having polar functional groups, such as pentafluorophenyl (PFP), cyanopropyl (CN-MS), and nitrophenyl (NPE). By contrast, nonpolar aromatics, such as phenyl (PE), naphthyl (π-NAP), and pyrenyl (PYE) groups on MS gave moderate RRT values (3–4). Although the sizes and shapes of phenyl, naphthyl, and pyrenyl are substantially different, their RRT values are comparable, indicating that there is little substrate specificity among nonpolar aromatics. Because no retention was observed (RRT = 1.0) on octadecyl modified- (C₁₈-MS-II)) or cholesterol modified (Cholester) silica gels,^¶ cyano- or aromatic groups on MS seem to be essential. Except for C₁₈-MS-II and Cholester columns, a cosolvent effect was examined. All the chromatograms are presented in Figures S12–19. Although the elution curves are sometimes not simple, showing shoulders, the peak tops were assigned as the RTs. The RRTs are tabulated in Table S1, and all the values are represented as a bar graph in Figure 3d. In Figure 3c, cosolvent effects on a CN-MS column are extracted. From Figure 3b–d, the tendencies of cosolvent effects on all columns are similar. Thus, polar solvents, such as nitrobenzene, acetonitrile, and methanol, eluted the analyte earlier, whereas toluene and chloroform retain the analyte for longer. These results indicate that the principle behind the interaction between the analyte and functional groups on the MS columns is the same for all the columns. Because little substrate specificity was observed among nonpolar aromatics, phenyl, naphthyl, and pyrenyl groups, the formation of an inclusion complex between the functional groups and C₄Ni₂MsCP2 was excluded. Indeed, C₄Ni₂MsCP2 has limited inside space, in which only one diethyl ether molecule can be included (vide infra, Figure 5j and k).⁸ Therefore, a part of the functional group must interact with the surface or enclosed space of C₄Ni₂MsCP2. This feature is of interest as molecular recognitions on the surface in solution.

Figure 3.

(a) Structures of functional groups on modified silica-gel (MS) columns. (b) Ratio of retention times (RRT) of C₄Ni₂MsCP2 on various columns (eluent: 100% pyridine). (c) RRT of C₄Ni₂MsCP2 on CN-MS with various cosolvents (20 vol%). (d) RRT of C₄Ni₂MsCP2 on various columns with various cosolvents (20 vol%). HPLC conditions: columns (ID 4.5 mm, 15 cm) were used except for the 5CN-MS column (ID 4.5 mm, 5 cm). 1.0 mL/min. C₄Ni₂MsCP2 (0.7 mM dissolved in chloroform, 10 µL) was injected.

Experiments using cosolvents reveal the competitiveness of the cosolvents relative to pyridine on the interaction between MS and C₄Ni₂MsCP2. When RRTs on the PFP column using cosolvent are plotted as a function of the dielectric constants of used cosolvents (Figure S20a), they seem to be weakly correlated. Thus, polar solvents having large dielectric constants tend to be competitive to the interaction, suggesting that the interaction is based on electrostatic interaction between polarized functional groups, such as pentafluorophenyl, nitrile, aromatic C-H, and CH₂ on benzylic and cyanopropyl group in MS, and aromatic π-electrons and peripheral C-H as well as lone pairs on pyridyl groups in C₄Ni₂MsCP2.^§ However, the correlation is ambiguous, and the plot of chloroform is far from the correlation. The exceptional behavior of chloroform is discussed later. Another interesting result is that large amounts of toluene used as cosolvent is not competitive, even on a PE-MS column composed of phenyl groups. This finding indicates that C₄Ni₂MsCP2 interacts with the phenylethyl groups on PE-MS much more strongly than with toluene, which has a similar electronic structure and size to the phenylethyl group on PE-MS. Not only the structural factor of functional groups but also other factors, such as immobility and orientation of functional groups, on MS should also be considered.

Quantitative Analyses Using Frontal Affinity Chromatography

To estimate the apparent association or dissociation constants of C₄Ni₂MsCP2 with the functional groups on MS, FAC was conducted on seven MS columns (internal diameter 4.5 mm, 150 mm length). To avoid a chloroform effect making the interaction strong, the analyte dissolved at various concentrations in pyridine alone was continuously injected into an MS column. When the frontal curve plateaued, the injection of analyte solution was stopped, and only pyridine eluent was introduced. In Figure S21, a frontal curve of C₄NiMsP1, which is not retained on the column, is presented. In the early trials, some experiments using another narrow column (ID 2 mm, 150 mm length) with various flow rates were conducted to reduce the amounts of sample used. However, no ideal frontal curve was obtained with a column of ID 2.0 mm (Figure S22). Therefore, seven MS columns (ID 4.5 mm, 150 mm length) were used with a flow rate of 1 mL/min, which is the flow rate recommended for these columns. Ideally, FAC should be conducted under slow flow rate to keep the equilibrium between host and guest. Therefore, the present FAC conditions might be different from the ideal equilibrium. PDA detector calibration was conducted without a column. Various concentrations of C₄Ni₂MsCP2 were introduced into the apparatus as the frontal conditions. Relationships between PDA intensities and concentration of C₄Ni₂MsCP2 solution in pyridine were obtained in the absence of a column (Figure S23). FAC experiments were conducted using more than five different concentrations of samples on seven MS columns. The purity of injected C₄Ni₂MsCP2 was confirmed as more than 98%. In Figure 4a, the chromatograms obtained under FAC conditions on an NPE column are presented. In the inset of Figure 4a, the normalized charts are represented. R₀ indicates the concentration-independent frontal line of monomeric C₄NiMsP1, which was not retained on the same column.

Figure 4.

(a) Frontal chromatograms of various concentrations of C₄NiMsCP2 on an NPE column (4.6 mm ID, 15 cm), R₀: The concentration-independent retention time of C₄NiMsP1, which was not retained on the column. See Figure S21. Inset: the intensities were normalized for analysis. (b) Relationships between 1/[A]₀ and 1/([A]₀(V − V₀)). See eq 2. (c) A curve fitting analysis (k_d = 1.0 × 10⁻² s⁻¹) for the dissociation curve in a chromatogram in (a).

The apparent dissociation constants, K_d, of the 1:1 complex of A and B are represented as eq 1.⁶ Because the present FAC conditions might be different from the ideal dynamic equilibrium conditions, “apparent” is used to distinguish them from the “real” dissociation constants.

$$\begin{align}K_{\text{d}} &= \frac{[A][B]}{[AB]} = \frac{[A]_{0}\{ [B]_{0} - [A]_{0}(V - V_{0})/v\} }{[A]_{0}(V - V_{0})/v} \\&= \frac{B_{t}}{V - V_{0}} - [A]_{0}\end{align}$$

(1)

where the concentrations of injected A and the amounts of functional groups (B) per unit volume of column used are set as [A]₀ and [B]₀, respectively, and v is the volume of the column. Therefore, v[B]₀ is the total amount of functional groups in the column, which is represented as B_t. V and V₀ are elution volumes of A and samples not retained on the column, respectively. A and B correspond to C₄Ni₂MsCP2 and the functional group on MS, respectively. V₀ was determined by using monomeric C₄NiMsP1. Equation 1 can be transformed as in eq 2.

$$\begin{equation}\frac{1}{[A]_{0}(V - V_{0})} = \frac{K_{\text{d}}}{B_{t}} \times \frac{1}{[A]_{0}} + \frac{1}{B_{t}}\end{equation}$$

(2)

Preparing relationships between 1/[A]₀ and 1/([A]₀(V − V₀)) gives 1/B_t and K_d/B_t as the intercept and slope, respectively. The relationships from the FAC data in Figure 4a are plotted in Figure 4b. All the FAC charts and relationships obtained in seven MS columns are represented in Figures S24 and S26a. Because all the plots were fitted as straight lines, a 1:1 complex was formed in all the columns.

Dissociation curves obtained in the FAC experiment (Figure 4c) were also fitted by using eq 3.

$$\begin{equation}\frac{dC}{dt} = - k_{\text{d}}C\end{equation}$$

(3)

where C is the observed concentration of C₄Ni₂MsCP2 using a PDA detector at t, and k_d is the apparent dissociation rate constant.

In FAC charts, as shown in Figure 4a, minor elution curves were observed from the early region (RT approx. 2–5 min), and corresponding defects having the same area appeared immediately after when injection of sample solution was stopped. Because the amount of impurity in the sample used here is smaller than the areas of minor elution curves, small amounts of C₄Ni₂MsCP2 are eluted without retention under FAC conditions. Therefore, for the dissociation curve analyses, only the later part of the FAC chromatogram was used. By comparing with apparent dissociation rate constants of monomer, relative dissociation rate, k_d (monomer)/k_d, was obtained. All the fitting analyses and normalized dissociation curves are presented in Figures S25 and S26b. The order of the apparent dissociation rates presented in Figure S26b is a good relationship to the degree of slopes in Figure S26a. The order corresponds to the relative strength of interaction between C₄Ni₂MsCP2 and functional groups on the seven MS columns. The total amounts of functional group, B_t, the apparent dissociation constants, K_d, and the apparent dissociation rate constants, k_d, are listed in Table 3. Because the obtained B_t values were extremely small (in the order of 10⁻⁷ to 10⁻⁸ mol), a precise comparison of the K_d values among columns should be avoided. Based on specifications of MS columns supplied by the manufacturer, amounts of functional groups loaded on SiO₂ can be roughly estimated to be approximately 1 mmol in all MS columns, as shown in Table S2. Thus, the amounts of loaded functional groups are larger than B_t determined by FAC by the fifth power of 10, indicating that only trace amounts of functional groups on MS act effectively to retain C₄Ni₂MsCP2 under the FAC conditions. It is noted that actual preparative separation is conducted under different conditions of injection and elution, and the effective amounts of ligand on MS are probably different. Because apparent dissociation constants K_d determined from FAC are in the range between 2 × 10⁻⁵ and 4 × 10⁻⁶ M, the association constants K_a, the inverse of K_d, are represented by 5 × 10⁴ to 3 × 10⁵ M⁻¹. The relatively large values can be detected by other spectroscopic methods, such as NMR titrations.

Table 3.

The values estimated from frontal affinity chromatography

Run	column	Bt/mol	Kd/M	kd/s−1	kd(monomer)/s−1	kd(monomer)/kd
1	PFP	1.1E-07	1.3E-05	9.3E-03	8.3.E-02	8.9
2	NPE	5.4E-08	7.2E-06	1.0E-02	9.2.E-02	9.2
3	π-NAP	2.9E-08	7.3E-06	1.6E-02	8.3.E-02	5.3
4	CN-MS	7.3E-08	2.1E-05	1.9E-02	9.0.E-02	4.7
5	PYE	2.7E-08	8.5E-06	2.6E-02	9.2.E-02	3.5
6	PE-MS	1.2E-08	4.3E-06	2.7E-02	9.5.E-02	3.6
7	PBr	2.9E-08	1.7E-05	2.8E-02	7.5.E-02	2.7

B_t: the estimated total amounts of functional groups on a column, K_d: the estimated apparent dissociation constants, k_d: the apparent dissociation rate constant, k_d(monomer): the apparent dissociation rate constant of monomer.

¹H NMR Titrations

To estimate association constants between C₄Ni₂MsCP2 and acetonitrile and nitrobenzene, ¹H NMR titrations were conducted in CDCl₃. Acetonitrile and nitrobenzene are functional groups on CN-MS and NBE columns, respectively. Chloroform is a suitable solvent with which to dissolve C₄Ni₂MsCP2 at high concentration and not interfere with the interaction, and rather, chloroform is expected to enhance the interaction as observed in its solvent effect in chromatography. In Figures 5a, 5b, S27, and S28, titration spectra with acetonitrile are presented. By adding 50 equiv of acetonitrile, the signal of py1 was low-field shifted by 0.02 ppm, whereas those of outer methyl groups in mesityl groups in C₄Ni₂MsCP2 were high-field shifted by 0.02 ppm and the inner methyl groups low-field shifted by 0.03 ppm. Shifts of other signals induced by the addition of acetonitrile were much smaller. Similarly, titration spectra with nitrobenzene are presented in Figure 5c–e, S29, and S30. By adding 30 equiv of nitrobenzene, the signal of py1 was low-field shifted by 0.1 ppm, and that of py2 was high-field shifted by 0.05 ppm, whereas that of outer methyl groups in mesityl groups in C₄Ni₂MsCP2 were high-field shifted by 0.24 ppm. Although the degrees of the induced shifts were different, the direction of shifts is the same between the cases of acetonitrile and nitrobenzene, suggesting that the position interacting with acetonitrile and nitrobenzene is similar. The approximate association constants of acetonitrile and nitrobenzene were estimated as K = 17 M⁻¹ and 1.4 × 10² M⁻¹, respectively, from Figure 5f and g by assuming the formation of a 1:1 complex.

Figure 5.

(a–e) ¹H NMR (400 MHz) titration of acetonitrile or nitrobenzene to C₄Ni₂MsCP2 in CDCl₃. (a) C₄Ni₂MsCP2 (6.2 mM) and acetonitrile (0 equiv), (b) acetonitrile (50 equiv), (c) C₄Ni₂MsCP2 (2.8 mM) and nitrobenzene (0.5 equiv), (d) nitrobenzene (1 equiv), (e) nitrobenzene (30 equiv). The changes of chemical shifts against added ligands, (f) acetonitrile and (g) nitrobenzene. Schematic structures of C₄Ni₂MsCP2 in (h) top and (i) side views. Structures of C₄Ni₂MsCP2 obtained by X-ray single crystal analysis ((j) top and (k) side view). For clarity, butyl ester and mesityl groups are omitted in (j). One diethyl ether molecule is included, used as a solvent in recrystallization. Reprinted with permission from reference⁸ and ©2020 American Chemical Society.

NMR spectra at various temperatures (VT) were collected in the range between 0 and −60 °C after adding 0.5 and 1.0 equiv of nitrobenzene (Figures S31 and S32). High-field shifted signals of the nitrobenzene were observed around 7–5 ppm as broad peaks at −40 °C. The high-field shifts induced indicate that the nitrobenzene does not exist inside the cofacial porphyrins. If inside, larger high-field shifts would be observed.^13–15 When pyrene was added to C₄Ni₂MsCP2 in CDCl₃, no spectral change was observed in either signal of pyrene or C₄Ni₂MsCP2 (Figure S33). The association constant of pyrene and C₄Ni₂MsCP2 must be smaller than those with acetonitrile and nitrobenzene. No interaction was observed between monomeric porphyrin C₄NiMsP1 and acetonitrile (or nitrobenzene) (Figures S34 and S35). In pyridine-d₅, no spectral change was observed between C₄Ni₂MsCP2 and nitrobenzene (Figure S36), suggesting that the association constant in pyridine must be much smaller than that in chloroform. This lack of change is consistent with the cosolvent effect observed in MS chromatography. Because the solvent systems are different, the association constants obtained by NMR in CDCl₃ (10¹–10² M⁻¹) and those obtained by FAC in pyridine (10⁴–10⁵ M⁻¹) cannot be compared with each other. Because the interaction was enhanced significantly in chloroform on MS columns, the difference of actual association constants in chloroform between NMR and FAC would be more widely spread. These results indicate that the interaction of C₄Ni₂MsCP2 with immobilized ligands is much larger than that with movable ligands.

4. Discussion

Structural Demand for the Host

Based on experimental results, including the previous work, structural demands for hosts in the WAC are illustrated in Figure 6a. It is an essential common cyclic structure in which a 2,2′-bipyridyl moiety substituted on the 6,6′-positions by two porphyrins at meso-positions of the porphyrins is tightly connected via rigid spacer shown as C at the opposite meso-positions to the 2,2′-bipyridyl moiety. Although the rigidity of the cyclic structure is important, there seems to be some tolerance in their derivatives. Alkyl ester groups on A positions are fully optional for the WAC, but substitution by butyl ester groups from the corresponding methyl ester on the A positions was effective in preventing self-aggregation of C₄Ni₂MsCP2. Therefore, the substitution at A positions can increase solubility and connect to other materials for further application. As substituents on B positions, ethyl and mesityl derivatives were prepared previously as Zn₃EtCP3 and Zn₃MsCP3 (Scheme 1). Both the cyclic trimers were well retained on WAC columns,⁷ but Zn₃EtCP3 is retained more strongly under the same conditions on the same columns. The ¹H NMR spectra of Zn₃EtCP3 in the presence and absence of pyridine are shown in Figure S37. Although sharp signals of β-protons are observed at 9.3 and 9.0 ppm in the presence of pyridine with signals of the coordinated pyridines in the higher field region (5–3 ppm), those of β-protons and ethyl substituents are significantly broadened in the absence of pyridine. These results suggest that three porphyrin moieties in pyridine-free Zn₃EtCP3 randomly rotate to give various nonsymmetric conformations, and exchange rates among the conformational isomers are comparable to the NMR timescale. Such conformational isomers were not observed in pyridine-free Zn₃MsCP3 having bulky mesityl substituents. The NMR observations and retention behaviors of Zn₃EtCP3 and Zn₃MsCP3 indicate that the rotation of rigid zinc porphyrins in the cyclic compounds does not interfere with the interaction on WAC columns. Probably, arbitrary substituent groups can be introduced into B positions while retaining their properties on WAC. As described in the former section, the roles of metal ions in M positions in porphyrins were various. Rigid porphyrins, such as zinc and free-base porphyrins, should be used in unstrained cyclic structures, such as Zn₃MsCP3 and Fb₃MsCP3. A flexible nickel porphyrin derivative used as a component of Ni₃MsCP3 was not appropriate for the WAC, whereas a flexible and bendable nickel porphyrin is necessary to construct a strained cyclic compound, C₄Ni₂MsCP2, which was significantly retained in WAC. Namely, porphyrins suppressing their bending motions could be used as components in cyclic structures. The tolerance of the distance between two porphyrins, R, seems to be relatively wide for retention on a WAC. As shorter examples, C₄Ni₂MsCP2 and Zn₂MsCP2_o are listed, whereas the distances in Zn₃MsCP3 and Zn₃EtCP3 are longer. All the cyclic porphyrins are retained on CN-MS columns having cyanopropyl groups, which are the smallest functional groups in the seven columns used for WAC (Figure 3a). The wide tolerance of R also supports the hypothesis that the interaction does not occur between two porphyrins in cyclic porphyrins. Most probably, the interaction occurs around the space illustrated as a red globe in Figure 6b–d. Electron density maps of C₁Ni₂MsCP2 obtained by DFT calculation are shown in Figure 6e–g, which correspond to the illustrations of Figure 6b–d. The red globe is faced with polarized π-electrons on 2,2′-bipyridine, the neighboring porphyrins, and positively polarized hydrogen atoms on the porphyrins. On the “hot spot,” cooperative multiple electrostatic interactions with ligands on MS, such as cyano propyl and aryl groups, are expected. In the solid state, such multiple electrostatic interactions were observed between enclosed spaces and small molecules, such as benzene¹⁶ and acetonitrile.^17,18 In the organic solution phase, however, direct observation of host–guest complexes between cages or capsules and such small molecules are difficult because their association constants are small, and the guest molecules can easily diffuse into the solution. Only highly electron-deficient aromatic compounds, such as 1,3,5-trinitrobenzene, and positively charged guests were observed as inclusion or sandwich-manner host–guest complexes in chloroform, dichloromethane, and toluene.^15,19,20 The proposed motif shown in Figure 6a is unique compared with the previous hosts. Thus, it is a rigid but uncaged host with open interaction sites. One of the merits of such a host having an open space is the only slight size specificity for guest molecules. Various sizes of guest molecules can potentially interact with the motif. Indeed, low size specificity was observed among PE, π-NAP, and PYE columns in WAC. Although association constants of host–guest complexes using such an open host motif are small, successive interaction through a column can detect such relatively weak interactions. The present results provide a new method to examine weak host–guest interactions, which cannot be detected by various spectroscopic methods, such as NMR.

Figure 6.

(a) The common structure retained on a CN-MS column. (b)–(d) Red globe: a candidate position of ligands on modified silica gels interacting with Ni₂MsCP2. (e)–(g) Electrostatic potential maps of C₁Ni₂MsCP2 optimized with density functional theory calculations (B3LYP/LANL2DZ/6-31G(d)). The maps were generated with GaussView at a 0.05 isodensity value and a color scale of 126 (blue) to −13 (red) kcal/mol.

Effect of Chloroform

Solvent effects in molecular recognition have been studied over more than four decades,²¹ and most cases could be followed by a linear free energy relationship (LFER) in which relationships between Gibbs free energy changes of formation of host–guest complexes and solvent indexes, such as E_T(30) or π* are proportional.^22,23 Based on LFER, the Gibbs free energy change of the host–guest complexes is unfavorable in chloroform compared with that in toluene.¹⁵ However, in the present WAC study using C₄Ni₂MsCP2, the cosolvent effect of chloroform was exceptional from the LFER (Figure S20c and d). Instead, the use of chloroform increased the interaction in the WAC system. One of the explanations for the contradictory result is that LFER is only applicable in the cases of enough solvation space over host and guest molecules, and it becomes exceptional on limited solvation space inside rigid small hosts.^20,24 Although the “hot spot” in Figure 6b–d is an open interaction site, multiple chloroform molecules could not solvate the “hot spot” efficiently. Chlorinated hydrocarbons, such as chloroform, 1,2-dichloroethane, and 1,1,2,2-tetrachloroethane, have high polarizability, and they are empirically known as good solvents to dissolve various organic compounds, including aromatic derivatives. However, in general, it is extremely difficult to estimate how many solvent molecules are needed to solvate the molecules and how large the solvation sphere is. In covalent molecular nanocage synthesis, the clustering of chlorinated solvent could be indirectly observed as yields of different sizes of nanocages.²⁵ Larger cages were produced in CHCl₃ and CH₂Cl₂ solvent compared with those in THF.²⁶ These findings suggest that chlorinated hydrocarbons, such as chloroform, need more space to solvate arene-rich components by continuum solvent structures of chloroform. In the cocrystal of an adaptive cage compound and small molecules, chloroform molecules avoid isolated situations in the cage. Two chloroform molecules are incorporated in the adaptive cage, whereas one molecule of nitrobenzene is included.²⁷ The chloroform molecules seem to pair up to cancel their dipole moments in such a limited space. When chloroform cannot solvate enough the “hot spot” in Figure 6b–d, a less-solvated vacant space can be produced on the “hot spot” in chloroform. This is favorable to interact strongly with functional groups on MS. In the present case, the high solubility of C₄Ni₂MsCP2 in chloroform can be achieved by solvation of the outside of C₄Ni₂MsCP2. This is reasonable because porphyrin derivatives can be solvated well by chlorinated solvents.^28–30 If we assume that chloroform molecules cannot solvate the “hot spot,” a part of the phenomena observed that the strong interactions appear in chloroform can be explained. However, an unresolved problem concerning the chloroform effect remains. Why do only small amounts of chloroform enhance the interaction between C₄Ni₂MsCP2 and MS in 100% pyridine eluent? The unique chloroform effect was observed in CN-MS column chromatography using C₄Ni₂MsCP2 samples injected in pyridine or chloroform solution. RRT values were significantly different at 4.0 and 9.8, respectively, as shown in Tables 1 and S1. Because 100% pyridine was used as an eluent in both cases (1 mL/min), only 10 µL of chloroform introduced with the sample significantly delayed the retention of C₄Ni₂MsCP2. Intuitively, such tiny amounts of chloroform would be diffused at the initial stage on the column. However, the results suggest that the 10 µL of chloroform acts as a cosolvent and considerably enhances the interaction between C₄Ni₂MsCP2 and functional groups on CN-MS. A part of the chloroform seems to be always accompanied by C₄Ni₂MsCP2 as a solvated solvent during its passage through the column. Although the reason remains unresolved, the injection of a chloroform solution of the sample was useful for preparative WAC.

5. Conclusion

Mechanisms of the specific retentions of several cyclic compounds composed of 2,2′-bipyridine and porphyrins (Schemes 1 and 2, Figures 1 and 3) on cyano- and arene-MS columns are unveiled. The present chromatography was classified as WAC operated with organic solvent eluents containing pyridine. The interaction between ligands (guests) on MS and the cyclic compounds (hosts) occurs with multiple electrostatic interactions on the “hot spots” in the host formed by 2,2′-bipyridine and adjacent porphyrins in the rigid cyclic compounds. Because the “hot spots” are open spaces, less size specificity for guests was observed. The rigidity of cyclic porphyrin derivatives and immobilization of functional groups on silica gel are necessary for significant retentions. Polar solvents, such as nitrobenzene, acetonitrile, and methanol, weaken the interaction. They can be used as cosolvents to adjust the RT of the host. The roles of chloroform are noteworthy. It dissolves porphyrin derivatives very well, but it does not work as a competitor and, instead, considerably enhances the interaction. Although these findings appear contradictory, they can be explained as that chloroform solvates the outside of the host with a continuum solvent structure, but it is challenging to solvate the inside similarly. WAC using organic eluents will be a new method to assess such nonbonding interactions in solution.

Supporting Information

Figures S1–S37 and Tables S1–S2 are available as supporting information. This material is available on https://doi.org/10.1246/bcsj.20220317.

Akiharu Satake

Akiharu Satake received his Doctoral Degree (Engineering) from Waseda University in 1995. After working at RIKEN (1997–1999), he worked at Nara Institute of Science and Technology (NAIST) as an assistant professor with Prof. Yoshiaki Kobuke from 1999 to 2010. He moved to Tokyo University of Science in 2010 as an associate professor, and he was promoted to full professor in 2015.

Author notes

On unmodified silica gel as a TLC plate, R_f = 1 with tailing was observed for C₄Ni₂MsCP2 with use of pyridine as an eluent.

The induction force caused by the polarized functional groups and the dispersion force between MS and C₄Ni₂MsCP2 may also contribute the interaction cooperatively.