Thiourea modified polyacrylnitrile fibers as efficient Pd(II) scavengers

Abstract

A series of thiourea modified fibers were prepared and characterized. The N-(2-aminoethyl)thiourea functionalized polyacrylonitrile fiber (AETU-PANF) was then evaluated for palladium absorption. The absorption process follows a pseudo-second-order model and the equilibrium data fit well to the Langmuir isotherm model. The maximum absorption capacity according to the Langmuir model was 169.2 mg g⁻¹ with an absorption limit of 0.03 ppm. Wetted AETU-PANFs exhibited excellent absorption properties in organic solutions and the sorption rates were higher in organic solvents than in aqueous solutions. This is due to a special microenvironment that is formed inside the AETU-PANFs. The AETU-PANFs were also used to remove palladium from an organic solution containing active pharmaceutical ingredients. The Pd content was reduced from 310 ppm to 0.07 ppm.

---

1 Introduction

During the past several decades, palladium-catalyzed organic reactions, such as hydrogenation reductions, Suzuki reactions and Heck reactions,^1–4 have been widely utilized in the synthesis of fine chemicals, especially active pharmaceutical ingredients (APIs). However, these palladium-catalyzed organic reactions often result in contamination of the final products with palladium. In addition, the permitted limits of metals in APIs are continuing to become more stringent and it is a great challenge to remove Pd to levels that meet these guidelines.⁵

Many of the techniques that have been established to reduce Pd to an acceptable level have serious drawbacks. For example the use of non-immobilized scavengers results in a significant loss of products, involves complicated equipment, and often does not reduce the Pd to the desired level.^6–8 More recent efforts have focused on using immobilized Pd scavengers.^9–22 Supported functional materials including fibers,^12–14 resins,^15–17 silica,^{18–20,22,23} polystyrene,²⁴ and chitosan²¹ have all shown good efficiency for metal purging.

Polyacrylonitrile fiber (PANF) is a common and inexpensive commercial material which can be easily woven into different shapes. In addition, it has large specific surface areas and abundant –CN groups which can be transformed into various other functional groups.^25–27 As a result, various types of modified PANFs have been prepared and investigated as absorbents.^17,25–30 For example, amino,^28,29 thiourea,³⁰ 4-(2-pyridylazo)-1,3-benzenediol (PAR),²⁶ thiadiazole¹⁷ and porphyrin³¹ groups have all been introduced into PANFs for the removal and detection of different metal ions.

Precious metals tend to form stable complexes with ligands containing “soft” donor atoms.¹² Thiourea which contains both nitrogen and sulfur atoms is an excellent ligand for precious metal ions. Therefore, the preparation and application of chelating materials with thiourea groups have attracted a great deal of attention.^12,32–36 Thiourea containing polymers with thiourea groups located on the main chain have been prepared by the polymerization of thiourea and formaldehyde monomers as absorbents.^34,35 However the relatively complicated polymerization process limits their usefulness for practical applications. Most methods used to prepare thiourea containing polymers do so by grafting thiourea groups onto the polymers backbone. Polymers such as polyethyleneglycol,³² resin,³³ and cotton fiber¹² have been used. The application of these grafted-polymers as metal-removers has been limited by the low content of thiourea groups and the complicated functionalizing methods.^12,36 In addition, most research has focused on aqueous solutions and only a few studies have involved the removal of Pd.^30,32,33

Herein, the synthesis, characterization and application of a novel chelating thiourea functionalized polyacrylonitrile fiber for the removal of palladium is reported. During the utilization of palladium catalysts, Pd(0) is usually oxidized to Pd(II).⁶ Therefore, Pd(II) solutions were selected to investigate the Pd-scavenging abilities of the N-(2-aminoethyl)thiourea functionalized polyacrylonitrile fiber (AETU-PANF). The modified fiber has excellent absorption properties for Pd(II).

2 Experimental

The reagents, apparatus and instruments are described in Sections 1 and 2 of the ESI.†

2.1. Preparation of thiourea modified fibers

The preparation of the thiourea-modified fibers is summarized in Scheme 1. Dried PANF (1.0 g), K₂CO₃ (13.1 g, 0.095 mol) and the appropriate quantities (0.17 mol) of thiosemicarbazide (ATU), N-(2-aminoethyl)thiourea hydrochloride (AETU·HCl) or 1-(2-aminoethylamino)thiocarbonyl-2-aminothiocarbonyl hydrazine hydrochloride (AEBTU·HCl) in water (50 mL) were placed in a three-necked flask. After the mixture was refluxed for 4 h, the fibers were filtered and washed with distilled water (60–70 °C) until neutral, and then dried overnight at 60 °C under vacuum to give the modified polyacrylonitrile fibers. The synthesis of AETU·HCl and AEBTU·HCl are described in Section 3 of the ESI.†

Scheme 1 Preparation of the modified fibers.

2.2. Absorption and desorption experiments

Absorption experiments were carried out to investigate the ability of the modified PANFs to absorb Pd(II). In these experiments, 25 mg of the modified fibers was placed in a series of flasks containing 25 mL of metal ion solutions at the desired concentrations and pH values. The contents of the flasks were stirred at a speed of 100 rpm during the experiments. The flasks were also maintained at the desired temperature during the course of the experiments. After the desired absorption time, the residual concentrations (C_e) of the palladium ions were measured by atomic absorption spectrometer (AAS).

Absorption experiments were also conducted in organic solutions. For these experiments, 25 mg of the modified fibers was added to 25 mL of an organic solution containing Pd(II). In addition 0.5 mL of water was added to wet the modified fiber. After stirring for the desired time, the mixture was filtered and the filtrate was evaporated to dryness. The residue was dissolved in 25 mL aqueous HNO₃ (4 mol L⁻¹) and then the remaining concentration (C_e) of Pd(II) was measured by AAS.

The absorption capacity (q_e, mg g⁻¹) was calculated using:

where C₀ is the initial concentration of Pd(II) (mg L⁻¹), C_e is the equilibrium concentration of Pd(II) (mg L⁻¹), V is the solution volume (L), and W is the weight (g) of the modified fiber. The effects of pH, contact time, initial concentration and temperature on the Pd(II) absorption by the modified PANFs in aqueous or organic solutions were investigated.

The reusability of the modified PANFs was also investigated. A 25 mg sample of used fiber was mixed with 0.1 M thiourea in aqueous HCl or HNO₃ (5 mL, 0.1–2 M) and stirred for 24 h at 298 K. The concentration of desorbed Pd ions was then determined by AAS. This treated fiber was then washed with water, dried and reused in other cycles.

2.3. Application to simulated pharmaceutical products

In pharmaceutical samples, the Pd scavenging process can be inhibited by various binding groups that are used in the compounds.^7,8 So products from two simulated pharmaceutical reactions were prepared. First, a reaction mixture containing a dichloromethane solution of posaconazole with 310 ppm palladium was obtained after a hydrogenation reaction (Scheme 2).^37,38 Second, an intermediate (compound 6) of tedizolid phosphate was synthesized by a Suzuki coupling reaction (Scheme 2).^39,40 The dichloromethane solution of the crude compound 6 contained a large amount (more than 1400 ppm) of palladium after a normal workup.

Scheme 2 Synthetic routes of posaconazole via a hydrogenation reaction and tedizolid phosphate via Suzuki coupling.

A simple setup was then applied to remove the palladium from the products using wetted AETU-PANFs. Solutions with about 5.0 g of the synthesized compounds in 25 mL dichloro-methane were treated with the functionalized fibers. The typical procedures are performed in Sections 4 and 5 of the ESI.†

3 Results and discussion

3.1. Synthesis of Pd(II)-scavenging fibers

Three functionalized fibers (ATU-PANF, AETU-PANF and AEBTU-PANF) were prepared according to the simple procedure shown in Scheme 1. ATU, AETU and AEBTU contain a metal ions binding moiety and an active amino group which can easily be grafted onto PANF. The extent of modification was measured directly by determining the weight gain according to the equation: weight gain = [(W₂ − W₁)/W₁] × 100% where W₁ and W₂ are the weights of PANF and the modified PANF, respectively.

It is easier to aminate the fiber with AETU than with ATU or AEBTU because AETU has higher activity and less steric hindrance. For all three types of fibers, the weight gain increased with the prolonged reaction times. However, too much weight gain reduces the mechanical strength of the modified fiber. For example, AETU-PANF with a 60% weight gain had a breaking strength of 2.3 cN compared to 9.7 cN for the unmodified PANF. In order to find the optimum balance between the extent of modification and mechanical strength, the amination step was optimized for two different feed ratios (Fig. S1†). The best results were obtained for a feed ratio of 1 : 27 (weight of PANF : weight of AETU·HCl) with a reaction time of 4 h. Under these conditions the weight gain was 23.9% representing a functionalization of 1.61 mmol g⁻¹ and the sample has a breaking strength of 9.5 cN which is close to that of the unmodified PANF (9.7 cN).

Similar optimizations were performed for ATU and AEBTU (Fig. S1†). The animation activities of ATU and AEBTU were lower than those for AETU due to their higher polarities. For AEBTU-PANF, the extent of functionalization was only 0.56 mmol g⁻¹ (weight gain: 11.4%). The steric hindrance of AEBTU also contributed to this lower result. The amino groups of ATU have lower reactivities than those in AETU which leads to a longer amination time (6 h) and a lower mechanical strength of the modified fiber (6.3 cN). For ATU-PANF, a weight gain of 11.9% (functional extent: 1.16 mmol g⁻¹) was obtained under the optimized conditions.

3.2. Characterization of the modified fibers

3.2.1. Elemental analysis (EA). The elemental analysis of PANF, AETU-PANF, ATU-PANF, AEBTU-PANF and AETU-PANF-Pd(II), are shown in Table 1. The amount of functionalized groups in the modified fibers can be calculated from the sulfur content using the following equation:where F_e (mmol g⁻¹) is the functionalized content of the modified fibers, n_l is the number of sulfur atoms in the thiourea derivatives, and S_c is the sulfur content (%) of the synthesized fiber. The weight gain was also used to calculate the extent of functionalization F_w (mmol g⁻¹) and these results are shown in Table 1.

Table 1 EA data of PANF, AETU-PANF, ATU-PANF, AEBTU-PANF, AETU-PANF-Pd(II)

Sample	C (%)	H (%)	N (%)	S (%)	Fe (mmol g^−1)	Fw (mmol g^−1)
PANF	70.75	5.94	26.12	0.093	N/A	N/A
AETU-PANF	61.68	6.98	22.44	5.11	1.59	1.61
ATU-PANF	56.12	6.36	21.44	1.85	0.58	1.16
AEBTU-PANF	60.12	6.36	24.44	2.81	0.44	0.56
AETU-PANF-Pd(II)	46.65	5.875	17.17	3.43	N/A	N/A

---

For AETU-PANF and AEBTU-PANF, the F_e value is nearly equal to F_w. For ATU-PANF, the F_e value is much lower than the F_w. Some of the –CN groups may be converted to –CONH₂ groups during the prolonged refluxing of the amination step (Fig. S1†). After the modification of PANF with the thiourea groups, the amount of carbon and nitrogen decrease significantly in all the samples.

3.2.2. Fourier transform infrared (FTIR) spectroscopy. The FTIR spectra of PANF and the modified fibers are shown in Fig. 1. The FTIR spectrum of PANF (Fig. 1a) has a characteristic C≡N peak at 2230 cm⁻¹ and a C=O peak from the ester moiety at 1730 cm⁻¹.^28,30 This confirms that PANF is a copolymer of acrylonitrile and methylacrylate. Comparing the FTIR spectrum of AETU-PANF (Fig. 1b) with that of PANF, the most striking changes are the disappearance of the 1730 cm⁻¹ peak, the weakening of the 2230 cm⁻¹ peak and the appearance of a broad absorption peak at 3270–3140 cm⁻¹ which is due to the amide N–H stretching vibration.³⁰ These changes suggest that most of the ester groups and some of the C≡N groups were transformed to amide bonds when the N-(2-aminoethyl)thiourea was introduced. In the spectrum of AETU-PANF the peak at 1290 cm⁻¹ (due to C=S vibrations) indicates the insertion of thiourea units into the PANF.¹² The FTIR of AETU-PANF-Pd(II) exhibits a new absorption peak at 1083 cm⁻¹ which can be assigned to a Pd(II) and thiourea complex.

Fig. 1 FTIR spectra of (a) PANF, (b) AETU-PANF, (c) AEBTU-PANF, (d) ATU-PANF, (e) AETU-PANF-Pd(II).

3.2.3. Scanning electron microscopy (SEM). Fig. 2 shows the SEM images of PANF, AETU-PANF, AETU-PANF-Pd(II), ATU-PANF and AEBTU-PANF. After modification, AETU-PANF became thicker and the surface became coarser (Fig. 2b). The SEM image of AETU-PANF-Pd(II) (Fig. 2e) shows that the material has a rougher and more porous surface with granular flakes. This is the result of the absorption of Pd(II).

Fig. 2 SEM images of (a) PANF, (b) AETU-PANF, (c) ATU-PANF, (d) AEBTU-PANF (e) AETU-PANF-Pd(II).

Compared to AETU-PANF, the fibers in ATU-PANF and AEBTU-PANF (Fig. 2c and d) were thinner and their surfaces are smoother. That illustrates that ATU-PANF and AEBTU-PANF have lower functionalities than AETU-PANF.

3.2.4. EDS (energy dispersive spectroscopy). In order to confirm that the Pd(II) was absorbed by AETU-PANF, EDS (Energy Dispersive Spectroscopy) was performed on the AETU-PANF-Pd(II) sample. As shown in Fig. 3, Pd(II) is present in the sample indicating that Pd(II) is loaded on the surface of AETU-PANF. This is consistent with the IR, elemental analyses and XPS results.

Fig. 3 EDS spectrum of AETU-PANF-Pd(II).

3.2.5. X-ray diffraction (XRD). The crystalline structures of the modified and original fibers were also investigated by XRD. The XRD spectrum of PANF (Fig. 4a) shows an intense reflection peak at 2θ = 17° which corresponds to the (100) diffraction of the hexagonal lattice formed by the parallel close packing of the molecule rods.⁴¹ This indicates that PANF adopts a stiff rod-like conformation due to the intermolecular repulsions between the nitrile groups. The XRD peak at 2θ = 17° is weaker in the modified fibers (Fig. 4b–d), which indicates that the crystalline phase of the fibers does not change much after the modifications. When the AETU-PANF sample is coordinated with Pd(II), the peak at 2θ = 17° becomes very small and a new peak appears at 2θ = 32.8° (Fig. 4e).

Fig. 4 XRD spectra of (a) PANF, (b) AETU-PANF, (c) AEBTU-PANF, (d) ATU-PANF, (e) AETU-PANF-Pd(II).

3.2.6. X-ray photoelectron spectrometry (XPS). In order to gain further information about the chemical changes that occurred in AETU-PANF after Pd absorption, the samples were evaluated using XPS. The XPS spectra of AETU-PANF and AETU-PANF-Pd(II) are presented in Fig. 5a. For AETU-PANF, there are five peaks were observed at 533.7, 399.6, 287.8, 228.3 and 161.4 eV which correspond to C, N, O and S, respectively. The S 2p peak at 161.4 eV is evidence for the introduction of the thiourea moieties into the PANF. After absorption, the XPS spectrum of AETU-PANF-Pd(II) also has two Pd(II) absorption peaks at 337.7 and 342.9 eV.⁴²

Fig. 5 XPS spectra of (a) total survey scan before and after absorption of Pd, (b) Pd 3d for AETU-PANF-Pd(II), (c) S 2p before absorption, (d) S 2p after absorption, (e) N 1s before absorption, (f) N 1s after absorption.

The high resolution XPS spectra of the S 2p peak (C=S) (Fig. 5c and d), show that the peak for AETU-PANF-Pd(II) is at a higher binding energy (163.1 eV) than that of AETU-PANF (161.4 eV). This indicates that the C=S group coordinates with the Pd(II).⁴³ The high resolution N 1s spectra for AETU-PANF (Fig. 5e) can be deconvoluted into two peaks which correspond to O=C–N (400.4 eV) and –NH₂ (399.3 eV).^43,44 After the absorption of Pd(II), the S 2p binding energy for EATU-PANF increases due to the donation of electron to Pd and the binding energy of N 1s remained almost the same.

3.3. Influence of functional group structure on Pd absorption

Next the influence of the functional groups on the modified PANF samples was investigated. The three thiourea derivatives differ in their spacer lengths and their coordination environments. The ATU-PANF derivative has a short distance between the thiourea groups and polymer backbone. As seen in Fig. 6, this short distance inhibited the Pd(II) absorption. AETU-PANF contains an ethylene group which improves the flexibility and minimizes the steric hindrance. This longer spacer improved the absorption properties of AETU-PANF giving it a higher absorption rate and a lower absorption limit.

Fig. 6 Scavenging Pd(II) by the modified fibers (15 mL aqueous solution, initial concentration 300 mg L⁻¹, 4 mol equiv. of absorbent per Pd(II), pH 7.0, 25 °C).

As mentioned in Section 3.1., a high extent of functionalization was not achieved for AEBTU-PANF due to the steric and polar inhibition. The Pd absorption by AEBTU-PANF was equally as good as that for AETU-PANF (Fig. 6). This result implies that the two thiourea moieties of AEBTU complex independently with Pd(II), i.e. a seven-membered complexation ring is not formed with the two sulfur atoms. Considering the preparation and adsorption behavior, AETU-PANF was selected for further investigation.

3.4. Influence of pH

The initial pH of a solution can seriously influence absorption because pH can affect the existence forms of metal ions and change the coordination environment of the absorbents. The effect of pH on the absorption of Pd(II) by AETU-PANF is presented in Fig. 7. The maximum Pd(II) absorption capacity (165 mg g⁻¹) was reached at pH = 7.0. At lower pH values, the active sites tend to be protonated, which depresses the ability of those sites to coordinate with Pd(II). Basic pH values were obtained using aq. NH₄OH. At pH > 7.0, competitive coordination of Pd(II) with HO⁻ and NH₃ may inhibit the coordination and absorption of Pd(II).

Fig. 7 Effect of pH (initial concentration 300 mg L⁻¹, absorbent 1.0 g L⁻¹, 25 °C, contact time 24 h).

3.5. Kinetics of absorption

Quantitative kinetics analysis is essential for designing absorption systems and understanding the absorption mechanism. Fig. 8 presents the absorption kinetics of AETU-PANF for Pd(II) at 25 and 50 °C from 0 to 46.5 h (initial concentration of 300 mg L⁻¹). The amount of absorption increased with time until reaching a maximum at about 24 h. At 50 °C, the absorption rate was slightly faster but the maximum absorption capacity was slightly lower.

Fig. 8 Absorption kinetics of AETU-PANF for Pd(II) (initial concentration 300 mg L⁻¹, absorbent 1.0 g L⁻¹, pH 7.0, 25 °C and 50 °C).

Pseudo-first-order equations (eqn (3)) and pseudo-second-order equations (eqn (4)) are frequently used in the kinetic analysis of absorption:^45,46

where q_t (mg g⁻¹) is the absorption capacity at a given time t, q₁ and q₂ (mg g⁻¹) are the saturated absorption capacities calculated according to the pseudo-first-order and the pseudo-second-order models (mg g⁻¹), respectively and k₁ and k₂ are the rate constants of the pseudo-first-order model (h⁻¹) and the pseudo-second-order model (g mg⁻¹ h⁻¹) respectively. The two equations were used to fit the experimental data at different temperature and the kinetic parameters are listed in Table 2. The absorption kinetics of the fiber for Pd(II) are better described by the pseudo-second-order equation (Fig. S2†). This suggests that the rate-determining step for Pd(II) is chemisorption, where valence forces are involved in electron sharing or exchange between the modified fiber and the Pd(II).

Table 2 Kinetic parameters for Pd(II) absorption by AETU-PANF

	T (K)	k1 (h^−1)	q1 (mg g^−1)	R^2
Pseudo-first-order	298	0.275 ± 0.003	153.5 ± 3.2	0.4302 ± 0.0008
	323	0.667 ± 0.006	143.1 ± 4.1	0.4003 ± 0.001

---

	T (K)	k2 (g mg^−1 h^−1)	q2 (mg g^−1)	R^2
Pseudo-second-order	298	0.00204 ± 0.00004	174.8 ± 2.5	0.9921 ± 0.0006
	323	0.00412 ± 0.00005	164.5 ± 2.8	0.9982 ± 0.0004

---

3.6. Isotherm of absorption

Absorption isotherm studies are very important to explain the interactions between metal ions and absorbents. The absorption isotherm for Pd(II) by AETU-PANF was performed at 25 °C and the result is shown in Fig. 9.

Fig. 9 Absorption isotherm of Pd(II) by AETU-PANF (initial concentration 30–1160 mg L⁻¹, absorbent 1.0 g L⁻¹, pH 7.0, contact time 24 h).

The absorption data was fit with two common absorption models: the Langmuir and Freundlich isotherm models. The Langmuir isotherm model was developed to describe absorption processes occurring on absorbents with homogeneous and flat surfaces. The model assumes that each absorptive site can only be occupied once in a one-on-one manner.⁴⁷ The Freundlich isotherm model, on the other hand, is used to describe absorption processes occurring on absorbents with heterogeneous surfaces.⁴⁸ The linear forms of the Langmuir and Freundlich equations are expressed as follows:^47,48

where q_e (mg g⁻¹) is the equilibrium absorption capacity, C_e is the metal ions concentration in solution (mg L⁻¹) at equilibrium, q_max (mg g⁻¹) is the maximum Langmuir absorption capacity, K_L (L mg⁻¹) is the Langmuir constant related to the energy of absorption. K_L increases with the strength of the absorption bond. In eqn (6), K_F ((mg g⁻¹) (mg L⁻¹)^−1/n) is the Freundlich constant and 1/n is the heterogeneity factor for the Freundlich model.

The fit of the absorption data with the Langmuir and the Freundlich models are shown in Fig. S3† and the parameters are summarized in Table 3. The Langmuir model is a better fit with a fitting correlation coefficient of 0.9982. This implies that each absorptive site is occupied by only one Pd(II) atom, and that interactions between the absorbed molecules are negligible. The calculated maximum absorption capacity for AETU-PANF using the Langmuir isotherm model is 169.2 mg g⁻¹ which is almost the same as the actual maximum absorption capacity (166.4 mg g⁻¹).

Table 3 Isotherm parameters for Pd(II) absorption by AETU-PANF

Langmuir	qmax (mg g^−1)	KL (L mg^−1)	R^2
	169.2 ± 2.3	67.7 ± 2.2	0.9982 ± 0.0005
Freundlich	1/n	KF ((mg g^−1) (mg L^−1)^−1/n)	R^2
	0.14 ± 0.01	63.3 ± 0.6	0.9251 ± 0.0008

---

The absorption process can be evaluated by calculating the separation factor (R_L) using the equation:⁴⁹

where C₀ (mg L⁻¹) is the initial concentration of Pd(II) and K_L (L mg⁻¹) is the Langmuir constant. R_L indicates whether the monolayer absorption is irreversible (R_L = 0), favorable (0 < R_L < 1), linear (R_L = 1) or unfavorable (R_L > 1). For the absorption of Pd(II) on AETU-PANF, the calculated values of R_L were between 0.3485 and 0.0126, indicating that the absorption is favorable.

3.7. Selective absorption

Some lightweight metals, such as Ca, K, Mg and Na are frequently encountered in the synthesis of APIs. Therefore the removal selectivity of the AETU-PANF was investigated using a multicomponent aqueous mixture containing C, K, Mg, Na and Pd ions. The initial concentration of each metal ion was 60 mg L⁻¹. The selectivity was evaluated using distribution coefficients (D, L g⁻¹) which were calculated using:⁵⁰where C₀ is the initial concentration of metal ions (mg L⁻¹), C_e is the residual concentration of metal ions in the treated solution (mg L⁻¹), V is the solution volume (L), and W is the dry fiber weight (g).

AETU-PANF (15 mg) was immersed in a solution (15 mL) containing the above metal ions and the residual ions concentrations were measured after stirring. The distribution coefficients (D) for the different ions are presented in Table 4. Clearly AETU-PANF selectively adsorbs Pd(II) over lightweight metal ions. These results can be explained by the theory of hard and soft acids and bases. The thioamide group (NH–C=S) can undergo a thion–thiol tautomerization in solution.⁵¹ The –C=N group contributes to the high Pd(II) absorption capacity, and the –SH group contributes to the excellent Pd(II) selectivity, since it is a very soft base with a higher affinity for Pd(II) than for the lightweight metal ions.

Table 4 Absorption selectivity of AETU-PANF for metal ions absorption (initial concentration 60 mg L⁻¹ for each ion; AETU-PANF 1 g L⁻¹; contact time 12 h; stirring rate 100 rpm, 25 °C)

	Pd(II)	Ca(II)	K(I)	Mg(II)	Na(I)
Distribution coefficient (L g^−1)	285	0.05	0.03	0.07	0.02

---

3.8. Absorption in organic solutions

Generally, most palladium-catalyzed reactions are carried out in organic solutions. The removal of Pd(II) in organic solutions is desirable. Pd(II) did not readily chelate with dried AETU in organic solvents. However when AETU-PANF was wetted with 2% (V_water/V_{organic solvent}), the fiber did absorb Pd(II) as shown in Fig. S4.† The absorption rates increased in wetted organic solvents compared to that in aqueous solution and all the absorption in all the organic solvents reached equilibrium within 3 h. The saturated absorption capacities (q_e) of AETU-PANF for Pd(II) in the tested solvents are shown in Fig. 10. The dielectric constants of the solvents are also shown. The q_e values were much higher in solvents with lower dielectric constants.

Fig. 10 Absorption of Pd(II) by AETU-PANF in organic solutions (initial concentration 300 mg L⁻¹; absorbent 1.0 g L⁻¹; contact time 24 h; stirring rate 100 rpm, 25 °C).

The modified PANF fibers are different from other traditional supporting materials.^31,52 The PANF fibers not only have modified surfaces but the bulk of the material is also modified to a depth of several hundred nanometres. Thus, the fibers contain a hydrophilic polar microenvironment consisting of the functional groups (ligands) and the polymeric segments. Less polar organic solvents cannot easily enter into this microenvironment, so does not the Pd(II) which is surrounded by organic molecules. However, when a small amount of water was added, the water easily enters into the microenvironment. As a result, the modified fiber is wetted and swollen which then allows the hydrophilic Pd(II) to more easily enter into the microenvironment. In this proposed model, the Pd(II) are forced from the hydrophobic organic phase to this hydrophilic “special microenvironment”. The coordination absorption can then take place efficiently in this microenvironment where there are higher concentrations of both the ligands from the functional groups and the Pd(II). Therefore, the absorption rates in wetted organic solvents are higher compared to that in aqueous solution.

As shown in Fig. 10, the saturated absorption capacities (q_e) are different in the various organic solvents. In solvents with high dielectric constants, the water tends to dissolve in the organic phase rather than to penetrate into the microenvironment. So the modified fiber is harder to be wetted and does not swell as much in solutions with high dielectric constants. This makes it difficult for the hydrophilic Pd(II) to get into the microenvironment, which results in the lower q_e.

3.9. Application to simulated pharmaceutical products

Palladium-catalyzed organic reactions often result in contamination of the final products with palladium. The AETU-PANF was used to scavenge Pd(II) from two APIs in dichloromethane solutions and the results are shown in Fig. 11 and 12. The concentration of Pd(II) in both solutions declined rapidly in the presence of AETU-PANF. During the absorption process, the Pd-contaminated solutions became colourless and the fibers turned brown. When oxygen is not purged from the posaconazole hydrogenation workup solution, any residual palladium is expected to be found as Pd(II) (Scheme 2). Since dichloromethane was employed as the extraction solvent, it was selected as the solvent for these scavenging processes.

Fig. 11 Scavenging palladium from posaconazole in dichloromethane with different amounts of AETU-PANF (initial Pd concentration 310 ppm, stirring rate 100 rpm, 25 °C).

Fig. 12 Scavenging palladium from compound 6 in dichloromethane with different amounts of AETU-PANF (initial Pd concentration 1430 ppm, stirring rate 100 rpm, 25 °C).

The effective removal of Pd(II) from actual process solutions critically depends on the nature of the metal complex as well as on the API. The competitive absorption process can be summarized as follows (Scheme 3): first the Pd(II) coordinates with the API or the intermediate. Then the API–Pd complex diffuses from the bulk solution to the absorbent surface. Finally the absorption of metal ion occurs on in the absorbent's active sites.

Scheme 3 Competitive absorption process with APIs.

The five-membered triazole ring of posaconazole contains three nitrogen atoms and has been found to be a good ligand for Pd(II).⁵³ Posaconazole also has a piperazine moiety which can chelate with Pd(II).⁵⁴ Scavenging Pd from posaconazole solutions would be inhibited by these moieties. Various reagents, such as L-tartaric acid, citric acid, activated charcoal and SiliaMetS®Thiol, have been employed to scavenge Pd(II) from posaconazole in dichloromethane (Table S5†).^6–8,23 All of these gave poorer results than AETU-PANF. SiliaMetS®Thiol and SiliaMetS®DMT are excellent scavengers for Pd(II) but the results with AETU-PANF were still better. As explained in Section 3.8., it is the special microenvironment of AETU-PANF which is responsible for the better Pd scavenging abilities.

The efficiency of a chelating fiber depends on the amount used and the treatment time. Thus the Pd(II) scavenging Pd(II) ability of AETU-PANF was tested with different equivalents of AETU-PANF (Fig. 11). After 4 h, 4 molar equivalents of the fiber reduced the Pd content in the posaconazole solution to 0.14 ppm. The Pd content of the isolated API solid was 0.8 ppm which meets the oral limit set by FDA.⁵ Using 8 molar equivalents of AETU-PANF reduced the palladium content in the solution to 0.07 ppm and in the isolated API solid to 0.4 ppm. This meets the parenteral requirements.⁵

Products derived from Suzuki coupling reactions usually contain much more Pd than hydrogenation products.¹⁸ When a soluble Pd(dppf)Cl₂ catalyst is used to prepare an tedizolid phosphate intermediate, the Pd(II) remains entirely in the crude products (Scheme 2). AETU-PANF was used to remove the Pd(II) from a dichloromethane solution of the Pd-contaminated compound 6, and the result is shown in Fig. 12. After 4 h, 4 molar equivalents of AETU-PANF lowered the Pd-load of the solution from 1430 ppm to 2 ppm. Since compound 6 is only an intermediate of tedizolid phosphate, the subsequent synthesis process would result in a final product with a lower and acceptable level of Pd.

3.10. Absorption limit of AETU-PANF

In addition to the absorption capacity, the absorption limit is also an important parameter. Both residual Pd(II) concentration and removal percentage (%) were employed to evaluate the absorption limit of Pd(II), respectively.where C₀ is the initial concentration of Pd(II) (mg L⁻¹) and, C_e is the equilibrium concentration of Pd(II) (mg L⁻¹). The amount of absorbent, absorption time and pH value were optimized to achieve the lowest absorption limits and the results are shown in Table 5. Absorption limits of 0.03 ppm and 0.07 ppm were obtained for aqueous and toluene solutions, respectively. When 10 molar equivalents of AETU-PANF were employed, the residual Pd content did not change.

Table 5 AETU-PANF Absorption limit (initial concentration 20 mg L⁻¹, stirring rate 100 rpm, 25 °C)

Solvent	Absorbent (equiv.)	pH	t (h)	Residual Pd (ppm)	Removal percentage (%)
Water	4	7	1	0.48	97.6
Water	4	7	24	0.03	99.9
Water	10	7	24	0.03	99.9
Water	4	2	24	0.11	99.5
Toluene	4	7	24	0.07	99.7

---

Some absorption limits and capacities for Pd(II) using different absorbents are listed in Table 6. Compared with other materials, AETU-PANF has a superior absorption limit and absorption capacity proving that AETU-PANF is an excellent scavenger for palladium ions.

Table 6 Absorption properties of different absorbents

Absorbent	Absorption limit	Capacity (mmol g^−1)	Lit.
Trimercaptotriazine	<1 ppm	N/A	8
Maleate acid	0.5 ppm	N/A	6
Silica gel-bound trimercaptotriazine	<1 ppm	0.67	18
Polystyrene-bound ethylenediamine	<0.1 ppm	>0.4	9
Polystyrene-bound trimercaptotriazine	2 ppm	0.32	10
L-Cysteine methyl ester physisorbed on carbon black	<2.2 ppm	0.48	11
AETU-PANF	0.03 ppm	1.61	This work

---

3.11. Reusability of AETU-PANF

Reusability is a key factor for the development of a commercially valuable Pd-scavenging material. Therefore the Pd(II) saturated AETU-PANF were soaked in 5 mL of 0.1 M thiourea in 0.1–2.0 M aqueous HCl or HNO₃. The adsorbed Pd(II) was completely desorbed by 0.1 M thiourea in aqueous HCl (1 M). After cleaning, the fiber was then reused for four more absorption–desorption cycles and the results were presented in Table 7. After five cycles the adsorption efficiency of the AETU-PANF remained at about 96% of its original efficiency.

Table 7 Reuse of the AETU-PANF for the absorption of Pd(II) (initial concentration 300 mg L⁻¹, absorbent 1.0 g L⁻¹, pH 7.0, contact time 24 h; stirring rate 100 rpm, 25 °C)

Cycle number	1	2	3	4	5
Absorption capacity (mg g^−1)	165.1	164.1	159.1	158.7	158.5

---

4 Conclusions

A novel chelating fiber (AETU-PANF) for Pd(II) scavenging was prepared by immobilizing N-(2-aminoethyl)thiourea on polyacrylonitrile fibers. The modified fiber can be used in aqueous solution with a high absorption capacity of 166.4 mg g⁻¹ and a low absorption limit of 0.03 ppm. The absorption process follows a pseudo-second-order model, suggesting that chemisorption is the rate controlling step. The absorption isotherm fits with a Langmuir isotherm which indicates that the dominant absorption is one Pd(II) per active site. The modified fiber has excellent Pd(II) absorption selectivity in the presence of Ca, K, Mg and Na ions. In addition the fiber can also effectively remove Pd(II) from organic solutions. These absorptions are faster than those in aqueous solutions which is due to existence of a special microenvironment. The modified PANF fiber was also used to effectively scavenge palladium from dichloromethane solutions of posaconazole and the intermediate of tedizolid phosphate. In conclusion, AETU-PANF is a highly effective material for removing Pd(II) from both aqueous and organic solutions.

Acknowledgements

The authors are grateful for the financial support from the National Natural Science Foundation of China. (No. 21306133 and No. 21572156) and the Tianjin Research Program of Application Foundation and Advanced Technology (No. 14JCYBJC22600).

Notes and references

1. V. Farina, J. T. Reeves, C. H. Senanayake and J. J. Song, Chem. Rev., 2006, 106, 2734–2793.
2. N. Miyaura and A. Suzuki, Chem. Rev., 1995, 95, 2457–2483.
3. J. Hassan, M. Sévignon, C. Gozzi, E. Schulz and M. Lemaire, Chem. Rev., 2002, 102, 1359–1470.
4. I. P. Beletskaya and A. V. Cheprakov, Chem. Rev., 2000, 100, 3009–3066.
5. The United States Pharmacopoeial Convention, United States Pharmacopeia, United Press, Inc., Baltimore, 36 edn, 2012, p. 232.
6. P. W. Manley, M. Acemoglu, W. Marterer and W. Pachinger, Org. Process Res. Dev., 2003, 7, 436–445.
7. C. Y. Chen, P. Dagneau, E. J. Grabowski, R. Oballa, P. O'Shea, P. Prasit, J. Robichaud, R. Tillyer and X. Wang, J. Org. Chem., 2003, 68, 2633–2638.
8. V. W. Rosso, D. A. Lust, P. J. Bernot, J. A. Grosso, S. P. Modi, A. Rusowicz, T. C. Sedergran, J. H. Simpson, S. K. Srivastava, M. J. Humora and N. G. Anderson, Org. Process Res. Dev., 1997, 1, 311–314.
9. Y. Urawa, M. Miyazawa, N. Ozeki and K. Ogura, Org. Process Res. Dev., 2003, 7, 191–195.
10. K. Ishihara, M. Nakayama, H. Kurihara, A. Itoh and H. Haraguchi, Chem. Lett., 2000, 29, 1218–1219.
11. J. Panchompoo, L. Aldous and R. G. Compton, J. Mater. Chem., 2011, 21, 9513–9522.
12. M. Monier, I. M. Kenawy and M. A. Hashem, Carbohydr. Polym., 2014, 106, 49–59.
13. R. Zhao, X. Li, B. Sun, M. Shen, X. Tan, Y. Ding, Z. Jiang and C. Wang, Chem. Eng. J., 2015, 268, 290–299.
14. P. K. Neghlani, M. Rafizadeh and F. A. Taromi, J. Hazard. Mater., 2011, 186, 182–189.
15. M. Trautmann, S. Lubahn and H. J. Holdt, React. Funct. Polym., 2014, 83, 84–97.
16. M. Yamada, M. R. Gandhi, Y. Kondo, K. Haga, A. Shibayama and F. Hamada, RSC Adv., 2015, 5, 60506.
17. C. Xiong, Y. Li, G. Wang, L. Fang, S. Zhou, C. Yao, Q. Chen, X. Zheng, D. Qi, Y. Fu and Y. Zhu, Chem. Eng. J., 2015, 259, 257–265.
18. D. Barbaras, J. Brozio, I. Johannsen and T. Allmendinger, Org. Process Res. Dev., 2009, 13, 1068–1079.
19. J. Brown, A. Chighine, M. A. Colucci, N. Galaffu, S. C. Hirst, H. M. Seymour, J. J. Shiers, R. D. Wilkes, J. G. Williams and J. R. H. Wilson, Tetrahedron Lett., 2008, 49, 4968–4971.
20. Q. Zou, L. Zou and H. Tian, J. Mater. Chem., 2011, 21, 14441–14447.
21. Y. Wang, Y. Qi, Y. Li, J. Wu, X. Ma, C. Yu and L. Ji, J. Hazard. Mater., 2013, 260, 9–15.
22. S. Nishiyama, Y. Sudo and Y. Hara, JP 2011041919 A, 2011.
23. SiliaMrts@Metal Scarvegers and Siliabond@Metal Scarvagers, see: http://www.silicycle.com; PhosphonicS@Metal Scarvegers, see: http://www.phosphonics.com.
24. Smopex@Metal Scarvegers, see: http://www.jmcatalysts.com; CHEMMATRIX@Metal Scarvegers, see: http://www.matrix-innovation.com.
25. L. Zhang, X. Zhang, P. Li and W. Zhang, React. Funct. Polym., 2009, 69, 48–54.
26. G. Li, L. Zhang, Z. Li and W. Zhang, J. Hazard. Mater., 2010, 177, 983–989.
27. Y. Fan, H. J. Liu, Y. Zhang and Y. Chen, J. Hazard. Mater., 2015, 283, 321–328.
28. S. Deng, R. Bai and J. P. Chen, Langmuir, 2003, 19, 5058–5064.
29. X. Chang, Y. Li, G. Zhan, X. Luo and W. Gao, Talanta, 1996, 43, 407–413.
30. X. Chang, Z. Su, G. Zhan, X. Luo and W. Gao, Analyst, 1994, 119, 1445–1449.
31. X. Liu, X. Liu, M. Tao and W. Zhang, J. Mater. Chem. A, 2015, 3, 13254–13262.
32. G. C. Yang, Z. X. Chen and Z. J. Zhang, React. Funct. Polym., 2002, 51, 1–6.
33. A. W. Trochimczuk and B. N. Kolarz, Eur. Polym. J., 2000, 36, 2359–2363.
34. A. A. Atia, Hydrometallurgy, 2005, 80, 98–106.
35. C. Ni, C. Yi and Z. Feng, J. Appl. Polym. Sci., 2001, 82, 3127–3132.
36. P. Liu, Z. Su, X. Wu and Q. Pu, J. Anal. At. Spectrom., 2002, 17, 125–130.
37. D. R. Andrews, D. Gala, J. Gosteli, F. Guenter, W. Leong, I. Mergelsberg and A. Sudhakar, WO 9633178 A1, 1996.
38. K. Shalini, N. Kumar, S. Drabu and P. K. Sharma, Beilstein J. Org. Chem., 2011, 7, 668–677.
39. W. B. Im, S. H. Choi, J. Y. Park, S. H. Choi, J. Finn and S. H. Yoon, Eur. J. Med. Chem., 2011, 46, 1027–1039.
40. O. Urbina, O. Ferrández, M. Espona, E. Salas, I. Ferrández and S. Grau, Drug Des., Dev. Ther., 2013, 7, 243–265.
41. S. M. Badawy and A. M. Dessouki, J. Phys. Chem. B, 2003, 107, 11273–11279.
42. S. Shrestha, M. Nagib and E. J. Biddinger, J. Electrochem. Soc., 2016, 163, D74–D82.
43. M. Zhang, Y. Zhang and R. Helleur, Chem. Eng. J., 2014, 264, 56–65.
44. Y. Yuan, G. Zhang, Y. Li, G. Zhang, F. Zhang and X. Fan, Polym. Chem., 2013, 4, 2164–2167.
45. Y. S. Ho and G. McKay, Process Biochem., 1999, 34, 451–465.
46. Y. S. Ho and G. McKay, Process Saf. Environ. Prot., 1998, 76, 332–340.
47. I. Langmuir, J. Am. Chem. Soc., 1918, 40, 1361–1403.
48. H. M. F. Freundlich, J. Phys. Chem., 1906, 57, 385–470.
49. L. Zhou, C. Shang, Z. Liu, G. Huang and A. A. Adesina, J. Colloid Interface Sci., 2012, 366, 165–172.
50. E. Birlik, A. Ersöz, E. Açıkkalp, A. Denizli and R. Say, J. Hazard. Mater., 2007, 140, 110–116.
51. T. A. Yousef, G. M. Abu El-Reash, O. A. El-Gammal and R. A. Bedier, J. Mol. Struct., 2013, 1035, 307–317.
52. X. L. Shi, M. Zhang, Y. Li and W. Zhang, Green Chem., 2013, 15, 3438–3445.
53. Z. D. Bugarcic, S. T. Nandibewoor, M. S. A. Hamza, F. Heinemann and R. van Eldik, Dalton Trans., 2006, 2984–2990.
54. M. M. Shoukry, A. A. Shoukry and M. N. Hafez, J. Coord. Chem., 2010, 63, 652–664.

---

Footnote

† Electronic supplementary information (ESI) available: Preparation of thiourea derivatives; typical procedure for removal of pharmaceutical contaminant; supplementary tables and figures are provided. See DOI: 10.1039/c6ra09689a

---