Introduction

Sulfur-containing polymers represent a compelling class of modern materials recognized for their diverse properties, including metal coordination, high refractive index, self-healing capacity, and semiconducting behavior1,2,3,4. Among these materials, polydithiocarbamates are distinctive macromolecules exhibiting excellent mechanical properties, a desirable photorefractive index, and dynamic reversibility5,6. However, the structural and functional exploration of these polymers has been limited by synthetic constraints. Early synthetic routes to polydithiocarbamates involved multicomponent polymerization of isocyan, dithiol, and sulfur or the addition of diisocyanates to dithiols, the polymerization of five-membered cyclic trithiocarbonates with amine monomers, or multicomponent polymerization utilizing carbon disulfide, diamines, and activated alkenes such as diacrylates or divinyl sulfones (Fig. 1a–c)5,6,7,8,9. While feasible, these methods often necessitate the use of diisocyanates, cyclic sulfur-containing monomers, or costly and limited monomers such as diacrylates/divinyl sulfones, leading to extended reaction times. Furthermore, these approaches frequently yield polydithiocarbamates incorporating secondary amides within the polymer backbone. Recent advances have enabled the synthesis of polydithiocarbamates devoid of NH groups through the use of activated internal alkynes, aliphatic secondary amines, and CS2 (Fig. 1d)10. Nevertheless, this method suffers from substantial substrate limitations. A general approach to the synthesis of N-alkylated polydithiocarbamates remains an outstanding challenge.

Fig. 1: Overview of established synthetic routes to polydithiocarbamates.
figure 1

a Multicomponent polymerization of isocyanides, thiols, and elemental sulfur. b Polyaddition reaction between isocyanates and thiols. c Polymerization involving primary amines, acrylates, and carbon disulfide (CS2). d Synthesis of structurally specific N-alkylated polydithiocarbamates from activated internal alkynes, aliphatic secondary amines, and CS2. e A general synthesis of N-alkylated polydithiocarbamates using thiocarbonyl fluoride, secondary amines, and thiols.

Full size image

Dithiocarbamates, possessing a unique thiocarbonylthio group (Z−C(=S)S−R), have emerged as effective photoiniferters capable of mediating radical polymerization through an initiation, propagation, reversible combination, and chain transfer mechanism11,12. The Z-group plays a crucial role in the chain transfer process by governing the reactivity of the C = S bond towards radicals and influencing the stability of the intermediate radical species13. Notably, among dithiocarbamates, only those incorporating N-alkylated dithiocarbamate functionalities have been demonstrated to function as photoiniferters in polymerization. Incorporation of dithiocarbamate iniferter units into polyurethane networks has enabled the construction of light-responsive dynamic networks exhibiting photoactivated strengthening and healing capabilities14,15,16. Recently developed macro-photoiniferters with dithiocarbamate chain ends allow for the creation and reconfiguration of 3D microstructures via direct laser writing, significantly mitigating migration issues during fabrication and within the resulting structures17,18. These advances highlight the substantial potential of polydithiocarbamates in the fabrication of advanced, transformable materials for diverse applications.

Building upon our group’s previously reported versatile approach to synthesizing polythioamides from diboronic acids, secondary diamines, and thiocarbonyl fluoride (CSF2) as a central linking unit19, we present herein a facile and general method for synthesizing novel N-alkylated polydithiocarbamates, employing CSF2 to bridge secondary diamines and dithiols (Fig. 1e). This approach enabled the synthesis of diverse structures, including previously inaccessible substrates. The resulting polydithiocarbamates, incorporating conjugated groups, function as superior macro-photoiniferters for radical polymerization, exhibiting enhanced photoactivity compared to their small-molecule analogs. This enhanced photoactivity enables direct photolysis of the macro-photoiniferters for fully open-to-air 3D printing. Furthermore, the polydithiocarbamates mediate light absorption and chain transfer reactions, significantly enhancing the resolution of the 3D-printed materials. Critically, the dormant dithiocarbamate functionalities embedded within the structural framework can be reactivated, allowing for iterative modifications and demonstrating the potential of our approach for living 3D printing.

Results and discussion

Synthesis and characterization of polydithiocarbamates

Monomers 1a-5a were synthesized according to our previously reported methods (Fig. 2a)19. To optimize reaction conditions, we investigated the polymerization of 2,3-dihydroquinoxaline-1,4-bis(carbothioyl)difluoride (1a) with p-benzyl dithiol (1b) Under initial conditions (room temperature, THF solvent, 2.0 equiv. K2CO3), a polydithiocarbamate with a moderate molecular weight (Mw = 5.6 kDa) was obtained in 75% yield after 12 h (Supplementary Table 1, Entry 1).

Fig. 2: Synthesis of N-alkylated polydithiocarbamates using thiocarbonyl fluoride, secondary amines and thiols.
figure 2

a The synthetic route to thiocarbamoyl fluoride monomers and the scope of the corresponding thiocarbamoyl fluoride and thiol reactants investigated in this study. b General reaction scheme illustrating the base-catalyzed polycondensation of thiocarbamoyl fluorides with thiols to form the target N-alkylated polydithiocarbamates.

Full size image

We subsequently investigated the influence of various bases on dithiol activation (Supplementary Table 1, Entries 2–10). While inorganic bases (CsF, KF) promoted the reaction to some extent, their limited solubility in THF resulted in low molecular weight polydithiocarbamates. Sodium and potassium alkoxides yielded similar results to the inorganic bases. In contrast, organic bases led to higher molecular weight products. Notably, 4-(dialkylamino)pyridines (DMAP) and 4-pyrrolidinopyridine (4-Ppy) significantly enhanced polymerization, affording polydithiocarbamates with Mw of 8.8 kDa and 85% yield, potentially due to unique interactions with the thiocarbonyl group20.

Solvent effects were explored next (Supplementary Table 2). High-polarity solvents such as N-methyl-2-pyrrolidone (NMP) favored polymerization, yielding the polydithiocarbamate with Mw of 13.1 kDa and 88% yield. Further optimization of monomer ratio, catalyst loading, temperature, and reaction time led to the optimal conditions (Supplementary Table 3 and 4). Under these conditions, polydithiocarbamate P1 with Mw of 14.5 kDa and 90% yield was obtained as a yellow solid.

We next explored the capacity of this method to access structurally diverse polymers (Fig. 2b). Reactions of 1a with aliphatic and aromatic dithiols (1b-5b) afforded polymers P1-P5 in 72-90% yields. Expanding the monomer scope, we synthesized long-chain aliphatic thiocarbamoyl fluorides (2a-3a) to produce polymers P6-P8. Notably, compared with previous work8,9, P6 was obtained without the need for an additional ester group in the main chain, offering increased synthetic efficiency compared to previous methods. (Sulfonylbis(4,1-phenylene))bis(ethylaminothiofluoride) (4a) reacted with various dithiols (7b,1b,6b,4b) to yield high molecular weight polydithiocarbamates (P9-P11) in good yields. Remarkably, benzylcarbamoyl fluoride (5a), a substrate known for its propensity to undergo free radical reactions and side reactions with thiols (3b,1b,5b) under similar conditions21, participated smoothly in this polymerization, highlighting the method’s selectivity (P13-P15).

The structures of the synthesized polymers were confirmed by various spectroscopic techniques. Comparative analysis with a structurally characterized small molecule model compound (M1) supported the proposed polydithiocarbamate structures (Supplementary Table 5). Fourier-transform infrared (FT-IR) spectroscopy provided clear evidence for successful polymerization. The absence of the characteristic S-H stretching band from the starting dithiol (1b) in the P1 spectrum confirmed its consumption. Additionally, the C=S stretching vibration peak originating from the monomer (1a) shifted from 1683 cm−1 to 1517 cm−1 in P1 (Supplementary Fig. 1), indicating the formation of thiocarbamate groups. 1H and 13C nuclear magnetic resonance (NMR) spectroscopy further corroborated these findings. In the ¹H NMR spectrum of P1, the singlet peaks corresponding to the piperazine group remained unchanged compared to M1 at δ 4.60 and δ 4.66. Notably, the 13C NMR spectra of both M1 and P1 displayed a characteristic peak at δ 197 for the C=S carbon, confirming the presence of thiocarbamate moieties (Supplementary Fig. 1). Similar spectral features were observed for polymers P2-P15. Their FT-IR spectra displayed prominent C=S stretching vibration peaks, a characteristic fingerprint of polydithiocarbamates (Supplementary Fig. 1). Additionally, the 13C NMR spectra of these polymers all exhibited the characteristic carbon resonance associated with the C=S group, providing conclusive evidence for their polydithiocarbamate structures. Furthermore, these amorphous polydithiocarbamates (Supplementary Fig. 3) exhibited notable thermal stability, with decomposition temperatures at 5 % weight loss (Td) under nitrogen spanning the range of 225 to 305 °C (except P12) (Supplementary Fig. 4).

Living 3D printing

Upon photolysis, dithiocarbamates can generate both carbon-centered and sulfur-centered radicals. While the former readily initiates free radical polymerization with vinyl monomers such as poly(ethylene glycol) diacrylate (PEGDA), the latter exhibits lower reactivity and primarily deactivates transient radical species, which can subsequently undergo photochemical dissociation22. Electron paramagnetic resonance (EPR) spectroscopy confirmed the generation of alkyl radicals from P1 upon photolysis in 2-methyl-2-nitrosopropane dimer (MNP) solvent (Supplementary Fig. 5a). These radicals were effectively trapped by MNP, producing characteristic EPR signals. The addition of PEGDA introduced a new set of peaks (Supplementary Fig. 5b), indicating preferential reaction of the alkyl radicals with PEGDA double bonds over MNP.

The photoactivity of dithiocarbamates is strongly influenced by adjacent functional groups. Introduction of conjugated groups, such as phenyl, benzyl, and ester moieties, can enhance their photoreactivity23,24,25,26. Highly active polydithiocarbamates function as macro-photoiniferters, initiating the polymerization of acrylate monomers upon light exposure. However, some polydithiocarbamates (P3, P7, P8, P9, P11 and P13) with R-groups exhibiting weak leaving abilities do not function as photoiniferters (Supplementary Fig. 6).

Photoresins were then formulated by dissolving P1 (or M1) and PEGDA in NMP at varying molar ratios (Supplementary Table 6). For M1, increasing its molar ratio led to decreases in both double bond conversion and storage modulus (G’) (Fig. 3). This behavior is attributed to M1’s role as a chain transfer agent, which reduces double bond conversion, crosslinked chain length, and increases chain ends, consequently lowering the material’s stiffness27,28. The polymerization rate, as indicated by the slope (ΔG’/Δt), also decreased due to deactivation (Supplementary Table 7). Concurrently, delay and gelation times were extended with increasing M1 content (Supplementary Fig. 8 and Supplementary Table 7). This further proves that excessive M1 concentrations promote chain deactivation, ultimately delaying the gelation process. Under identical dithiocarbamate group concentrations, P1 exhibited higher double bond conversion and gelation speed than M1. This disparity may be attributed to P1’s lower migration rate, resulting in decreased chain deactivation compared to M1, thereby accelerating its polymerization. Regarding double bond conversion rate, P1 demonstrated an initial increase followed by a decrease with increasing molar ratio, indicating a shift from predominantly chain incorporation to chain transfer reactions. Collectively, P1 exhibits higher double bond conversion and storage modulus relative to M1, suggesting enhanced initiating efficiency. While traditional photoiniferter or RFAT 3D printing methods typically require stringent deoxygenation or supplementary initiators29,30,31,32, the high initiating efficiency inherent to P1 facilitates open-to-air 3D printing, obviating the need for these additions.

Fig. 3: Kinetic analysis of the photopolymerization process.
figure 3

a Real-time monitoring of double bond conversion in the photoresins, as measured by ATR-FTIR spectroscopy. b Photo-rheological analysis showing the evolution of storage and loss moduli upon UV exposure. Note: The specific formulations of the photoresins are provided in Supplementary Table 6.

Full size image

We subsequently conducted open-to-air 3D printing using photoresins incorporating P1 (Supplementary Table 6). Given P1’s UV light absorption and subsequent generation of sulfur-centered radicals, which can combine with growing radicals to form dormant species and potentially enhance resolution33, we investigated the influence of P1 concentration on print quality. A spoke pattern test (Fig. 4a) was employed to evaluate resolution across various resin formulations34. At low P1 concentrations, increasing the concentration accelerates the gelation rate, initially leading to reduced resolution (e.g., suboptimal resolution observed for P1 0.50, which exhibited the fastest gelation). However, as the P1 concentration increases further, chain transfer activity becomes more dominant, resulting in improved resolution at higher concentrations (Fig. 4b–d)33. Correspondingly, for samples printed with M1 as the photoiniferter, the resolution gradually improved as the concentration increased (Fig. 4e–g). This observation supports M1’s role as a chain transfer agent.

Fig. 4: Printing resolution evaluation using a spoke-like pattern.
figure 4

a Spoke pattern illustration and resolution quantification. This panel illustrates the spoke pattern, where the increasing gap between adjacent spokes from the center to the periphery indicates resolution. Resolution is quantified by the unresolved fraction, defined as the ratio between the unresolved diameter (marked by the red dashed circle) and the outer diameter (marked by the yellow dashed circle). Microscopic images of printed structures. These panels display microscopic images of printed structures for various resin formulations: b P1 0.25, c P1 0.50, d P1 1.00, e M1 0.25, f M1 0.50, and g M1 1.00. Red dashed circles highlight the unresolved areas in each image. All scale bars represent 1 mm.

Full size image

Employing P1 1.00, which offered a balance of curing speed and significantly enhanced resolution, we successfully fabricated intricate structures such as a lattice, a fish, and a hexagonal pattern (Fig. 5). FT-IR analysis confirmed the retention of dithiocarbamate groups whitin the 3D printed objects (Supplementary Fig. 9), suggesting the potential to be reactivated to allow for chain extension. To explore this, a square prism base was initially 3D printed using formulation P1 1.00. Subsequently, a second layer, containing only the blue fluorescent monomer 1,2-bis(4-acryloyloxyphenyl)−1,2-diphenylethylene (A2TPE, Supplementary Table 8) without initiator, was printed onto the surface of the base35,36,37. This enabled visual monitoring of the subsequent 3D printing process directly on the base object (Fig. 6a, b). SEM analysis revealed a seamless interface between the surface-modified layer and the substrate, indicative of a living polymerization mechanism (Fig. 6c).

Fig. 5: A series of 3D printed structures.
figure 5

a Lattice structure. b Fish structure. c Hexagonal structure. All scale bars are 1 mm.

Full size image
Fig. 6: Living 3D printing.
figure 6

a Living 3D printed object. This panel shows a representative 3D printed object fabricated using the living polymerization approach. The scale bars are 1 mm. b Living 3D printed object under UV light (λmax = 365 nm). This panel displays the same living 3D printed object when exposed to UV light, highlighting its photoresponsive properties. The scale bars are 1 mm. c SEM image of the cross-section of the living 3D printed object. The red dashed box outlines the region of surface functionalization, indicating the ability for further chemical modification. d SEM image of the cross-section of the 3D printed object with P1 as photo-iniferter. e SEM image of the cross-section of the 3D printed object when Omnirad 819 (BAPO) was used as the initiator for comparison.

Full size image

To demonstrate the reactivation of the dithiocarbamate group within the gel matrix, further experiments were conducted. A cube 3D printed using P1 as a photoiniferter was post-cured under UV light to maximize vinyl group conversion. The sample was subsequently cut into two pieces, which were then brought into contact under ambient conditions (room temperature and open-air environment) and irradiated with UV light (I0 = 300 mW/cm2, λmax = 365 nm) for 30 minutes. Minimal healing was observed, likely due to the near-complete consumption of vinyl groups preventing new crosslinking. Subsequently, application of a solution of N,N-Dimethylacrylamide (DMAm, 50 wt% in NMP) to the fracture interface resulted in noticeable healing after 10 minutes of UV irradiation (Supplementary Fig. 11c). To further confirm the role of the dithiocarbamate group, a cube was prepared by 3D printing using M1 as a photoiniferter. After cutting this cube into two pieces, a DMAm solution (50 wt% in NMP) was uniformly applied to the fracture surfaces and exposed to UV light. Similar to P1, photoinduced healing was observed (Supplementary Fig. 11d). In contrast, for cubes printed with a traditional photoinitiator Omnirad 819 (BAPO), almost no healing was observed after applying the DMAm solution to their cross-sectional surfaces and subsequent UV exposure (Supplementary Fig. 11e). These results confirm that dithiocarbamate groups can be reactivated under UV light to form new polymer networks, highlighting their essential role in facilitating healing. These findings demonstrate the feasibility of our approach for living 3D printing.

Concurrently, the stair-stepping layer pattern presents a significant challenge in DLP 3D printing. This phenomenon originates from the layer-by-layer curing of the photosensitive resin. UV exposure leads to sufficient curing in the region proximal to the light source, while the distal upper part experiences insufficient curing. This difference arises because the already cured material absorbs UV light, hindering photopolymerization in the subsequent layer. As a result, the resin can exhibit discontinuities at the interlayer interface, manifesting as the characteristic stair-stepping pattern38,39. Existing strategies to mitigate or eliminate layer patterns include modifications to resin formulation40, the use of grayscale printing41, Oscillation-Assisted Digital Light Processing42,43, continuous liquid film confined44 and conducting post-printing thermal treatment45. Interestingly, our utilization of P1 as a photoiniferter for DLP 3D printing effectively eliminates the stair-stepping layer pattern (Fig. 6d), in contrast to formulations using BAPO, which display prominent stair-stepping (Fig. 6e). This improvement is likely a consequence of the repeated activation of dithiocarbamate groups during both 3D printing and post-curing. This continuous activation generates radicals which, through chain initiation and transfer, consume residual vinyl monomers and dynamically heal interlayer interfaces. This mechanism effectively removes interfacial discontinuities, thus preventing the formation of stair-stepping layer patterns. This unique capability of P1 to eliminate stair-stepping layer patterns in DLP 3D printing, without requiring printer modifications or additional pre- or post-processing steps, represents a simple yet powerful approach that could substantially enhance the flexibility and output quality of DLP 3D printing.

In summary, we have developed a facile and versatile synthetic route to previously inaccessible polydithiocarbamates utilizing secondary diamines, dithiols, and thiocarbonyl fluoride as a central building block. These novel polymers function as efficient macro-photoiniferters for direct, open-to-air 3D printing. Compared to small-molecule analogs, our polydithiocarbamates exhibit superior photoactivity, leading to enhanced resolution in the resulting 3D-printed structures. Critically, the inherent reactivity of the dithiocarbamate moieties enables living 3D printing. More interestingly, P1 can eliminate the stair-stepping layer patterns that represent a considerable challenge in DLP 3D printing. This is achieved without requiring printer modifications or additional pre- or post-processing steps, presenting a simple yet powerful approach that could substantially enhance the flexibility and output quality of DLP 3D printing. We demonstrate that thiocarbonyl fluoride is a valuable platform for the synthesis of diverse sulfur-containing polymers, and that our polydithiocarbamate-based macro-photoiniferters will facilitate the creation of advanced materials with broad applications in biomedicine, energy, and materials science.

Methods

Chemicals and materials

Sodium trifluoromethanesulfinate (95%), Triphenylphosphine (PPh3, 99%), Chlorodiphenylphosphine (CDPP, 98%), 1,2,3,4-Tetrahydroquinoxaline (98%), 4,4’-Sulfonyldianiline (99%), N, N’-(1,4-Phenylenebis(Methylene))Dianiline (97%), 1,6-Dimercaptohexane (97%), N-methyl-2-Pyrrolidinone (NMP, 99.5%), Methanol (MeOH, 99.5%) were purchased from Adamas. 1,3-Di(piperidin-4-yl)propane (98%), N, N’-Dimethyl-1,6-diaminohexane (98%), 1,4-Benzenedimethanethiol (95%), 1,3-Benzenedimethanethiol (98%), 2,2′-(Ethylenedioxy)diethanethiol (97%), Bis(4-mercaptophenyl) sulfide (98%), Glycol dimercaptoacetate (97%), N1, N3-Bis(2-mercaptoethyl)isophthalamide (98%) were purchased from Bide Pharmatech Co., Ltd. All of chemical reagents were used as received without further purification.

Polymer characterization

1H, 13C, and 19F NMR spectra were recorded on a 500 MHz Bruker DRX 500. Proton magnetic resonance (1H NMR) spectra were recorded at 500 MHz. Carbon magnetic resonance (13C NMR) spectra were recorded at 126 MHz. Fluorine magnetic resonance (19F NMR) spectra were recorded at 470 MHz. Chemical shifts are reported in parts per million (ppm), and the residual solvent peak is used as an internal reference: proton (DMSO δ 2.50) (CDCl3 δ 7.26), carbon (DMSO δ 39.52) (CDCl3 δ 77.0). GC-MS data were recorded on an ISQ LT Single Quadrupole Mass Spectrometer, coupled with a Trace 1300 Gas Chromatograph (Thermo Fisher Scientific). High-resolution mass spectral data were acquired on Thermo Scientific Q Exactive HF Orbitrap-FTMS (electrospray ionization: ESI). GC analyses were performed on an Agilent 7890 A instrument (Column: Agilent 19091J-413: 30 m × 320 μm × 0.25 μm, carrier gas: H2, FID detection. The number-(Mn) and weight-(Mw) average molecular weights and dispersity (Đ = Mw/Mn) of the polymers were estimated by a Waters 1515 gel permeation chromatography system. Samples were eluted in series through a TSKgel Alpha column. DMF/LiBr solution (0.05 M of LiBr) solution was used as an eluent at a flow rate of 1 mL/min. A set of monodispersed polystyrenes, covering the Mw range of 107 g/mol, was utilized as standards for molecular weight calibration. Thermogravimetric analysis was conducted on NETZSCH STA 449 C under a nitrogen atmosphere at a heating rate of 10 °C/min. FT-IR spectra were determined on a Nicolet iS10 FT-IR spectrometer. UV-Vis spectra were recorded on an EVOLUTION220 spectrophotometer (Thermo Fisher Scientific) using 10 mm quartz cells.

In situ EPR measurements

Electron paramagnetic resonance (EPR) spectra were obtained on the Bruker EMX PLUS spectrometer equipped with a 500 W xenon lamp (USHIO Optical Modulex SXS5 U1501XQ) as the illumination source. The EPR spectra were recorded under light irradiation conditions.

Characterization of photosensitive resins

Double bond conversions of photosensitive resins were continuously monitored using real-time ATR-FTIR (Nicolet iS10 series, ThermoFisher), with a light intensity of I0 = 60 mW/cm2 (λmax = 365 nm). The double bond conversion was calculated by monitoring the intensity ratio of the C = C torsional peak at 810 cm-1 to the C=O stretching vibration at 1720 cm-1, using the latter as an internal standard.

The photo-rheology curve was obtained using a NETZSCH Kinexus rotational rheometer with a sample thickness of 0.200 mm, at a frequency of 1 Hz, and under a shear stress of 0.5 Pa with a light intensity of I0 = 60 mW/cm2 (λmax = 365 nm). The test duration was 9 min with a sampling interval of 0.05 s, starting 420 seconds after the light was turned on. The light sources used for the test were all Omnicure S2000.

3D printing

The resin was used to fabricate 3D hydrogel scaffolds using a commercial DLP 3D printer (Asiga Max X27, 385 nm). All printed structures are post-cured for 10 min under UV light (I0 = 300 mW/cm2, λmax = 365 nm) to convert the double bonds as fully as possible.

Photo-healing

Some cuboid structures were printed with P1 (Supplementary Fig. 11c), M1 (Supplementary Fig. 11d), and BAPO (Supplementary Fig. 11e) as photoiniferter or initiators. Subsequently, the sample was cut into two pieces. A solution of DMAm (50 wt% in NMP) was applied to the fracture surfaces. The two pieces were then brought into contact under ambient conditions (room temperature and open-air environment) and irradiated with UV light (I0 = 300 mW/cm2, λmax = 365 nm) for 10 minutes. Then, the sample was soaked in NMP for 5 minutes to remove excess monomers and linear polymers.