Luminescent colloidal carbon dots: optical properties and effects of doping [Invited]

1. Introduction

Carbon is a unique chemical element, whose chemical diversity is unmatched by any other element in the periodic table. Aside from being the foundation of traditional scientific disciplines such as organic chemistry and biochemistry, pure carbon is an inorganic material, which exists in multiple allotropes with a large variety of materials properties [1]. Based on different degrees of hybridization (sp² vs. sp³), graphite, diamond and amorphous carbon, are the main forms of macroscopic carbon. A large number of carbon’s nanoscale allotropes have been discovered in the last decades [2], such as fullerenes [3], carbon nanotubes [4], graphene [5], and carbon nanoparticles [6]. At present, the family of carbon nanoparticles includes graphene quantum dots [7], polymer dots [8], and carbon dots (CDs) [9], although clear classification of these forms can sometimes be difficult [10–12]. Generally, graphene quantum dots consist of a few layers or single-layer graphene sheets, often functionalized with molecules and/or functional groups at the surface [13]. Polymer dots can be viewed as amorphous agglomerates of conjugated and cross-linked polymers, whose properties are mainly determined by the respective constituting monomers [8, 14, 15]. CDs are probably the most diverse class within the family of carbon nanoparticles. They neither inherit the well-defined structure of graphene quantum dots, nor do their properties entirely depend on a monomer unit of the particle. At the same time, they possess a set of characteristic optical properties, in combination with the related structural features, which are both largely determined by the specific synthetic approach. A detailed discussion on the optical and structural properties of CDs, largely related to effects of doping, is in the focus of this review. We mostly consider colloidal carbon dots fabricated by solution-based syntheses (Fig. 1), which offer the greatest flexibility in introducing heteroatoms and other dopants [16–19]. Due to their low toxicity [20], and promising optical properties, such as high fluorescence quantum yields and stability against photo-bleaching [9, 21], these materials are suitable for applications in bio-imaging and solid state lighting [22–26], which we shortly summarize at the end.

 

Fig. 1 Timeline of selected contributions to the synthesis of CDs, with an emphasis on the introduction of dopant atoms in recent years.

Download Full Size | PPT Slide | PDF

2. Synthesis of CDs

Numerous approaches towards the synthesis of CDs have been reported in recent years, with a large variety of starting materials and techniques used. In general, all synthetic methods can be divided into top-down and bottom-up approaches, both with inherent positive and negative aspects for the controlled synthesis of customized CDs. Top-down methods include arc discharge [6], laser ablation [27–35], chemical oxidation in strong acid [36–38], or electrochemical synthesis [39–41], starting from a range of carbon allotropes, such as carbon nanotubes, nanodiamonds or graphite. Bottom-up syntheses, viz. microwave [21, 42–46], ultrasonic [47–50], template-supported [51, 52], and solvothermal treatment [9, 53–68], offer fast and simple access to the production of a wide variety of CDs, with a large number of surface functionalities and tunable properties [16–18]. Importantly, these approaches allow the easy introduction of heteroatoms and other dopants, which significantly affects the properties of the resulting CD materials and will be subject to further discussion in this article. Thus, this chapter introduces synthetic methods towards CDs with a focus on solution-based wet chemical techniques.

2.1. Top-down methods

Within the top-down approach, starting from micrometer- or even larger sized structures, materials are dispersed / cut into smaller fragments to finally reach nanometer size. Fluorescent carbon nanoparticles were discovered by Xu et al. in 2004 [6], as a by-product of carbon nanotubes, produced in a common top-down approach by arc-discharge. After oxidizing the soot with 3.3 N HNO₃ and separating the fractions of their crude product by electrophoresis, they were able to identify green-blue, yellow and orange fluorescent carbon species, which made up in total 10% of the mass of the carbon nanotube suspension. The first attempt to directly synthesize fluorescent carbon nanoparticles was done by Sun et al. in 2006 [27], who performed laser ablation on graphite, yielding a dark-colored soot of aggregated, non-luminescent carbon particles. After further HNO₃ treatment, the surface of the particles was passivated with diamine-terminated oligomeric polyethylene glycol, resulting in fluorescent particles with photoluminescence quantum yields (PL QY) of up to 10%. In their further studies, the same group used CDs for in vivo fluorescence imaging and demonstrated multiphoton imaging [30, 31]. As a follow-up, Hu et al. developed a simplified technique of irradiating graphite with a 1064 nm Nd:YAG pulsed laser in ultrasonic conditions in diethanolamine, diaminehydrate and polyethylene glycol, which was directly attached to the surface of the resulting CDs, exhibiting PL QY of up to 8% [32]. They further showed that the size of as produced CDs could be tuned from 3, 8 to more than 10 nm by changing the laser pulse width from 0.3, 0.9 to 1.5 ms. The corresponding PL QYs of these CDs were found to be 12, 6 and 1%, respectively [33]. More recent, related studies reported the use of different, sometimes exotic carbon sources such as sugarcane bagasse [34], or studied the irradiation parameters, such as the radiation fluence, spot size and irradiation time [35]. Aside from using laser irradiation to cut graphitic carbon into smaller fragments, soot has also been collected from burning candles or natural gas [36, 37]. In both approaches, combustion soot was further processed by oxidative treatment in harsh acidic environment, in order to break down the agglomerates and obtain small fluorescent carbon nanoparticles.

Another top-down synthetic route was initiated by Zhou et al. in 2007, who reported the fabrication of fluorescent carbon nanoparticles by electrochemical etching [39]. The synthesis was carried out in a solution of 0.1 M tetrabutylammonium perchlorate in acetonitrile. By employing a carbon nanotubes covered paper as working electrode and a Pt wire counter electrode, CDs with an emission peak at 410 nm and a PL QY of 6% were synthesized. In a related approach, Li et al. fabricated fluorescent CDs with sizes ranging from 1.2 to 3.8 nm by using graphite rods as anode and cathode material in an alkaline NaOH/EtOH electrolyte [40]. Depending on the current densities applied during the synthesis, different sizes of carbon nanoparticles were achieved, apparently exhibiting size-dependent fluorescence. Similar observations were also reported by Bao et al., who exfoliated fluorescent CDs from carbon fibers, with a Pt counter electrode [41]. In their work, it was further shown that the optical properties of CDs change significantly after further electrochemical surface oxidation, indicating a strong relation between the PL emission and the surface states. So far, top-down methods have led to the discovery of fluorescent carbon nanoparticles, paving the way for further developments in this area, which are mostly related to bottom-up synthetic methods resulting in colloidal nanoparticles.

2.2. Bottom-up methods

Bottom-up syntheses offer the opportunity to synthesize colloidal carbon nanoparticles from a large variety of molecular precursors with different chemical functionalities. Most synthetic approaches include high temperature pyrolysis, by using mircrowaves [42, 44], solvothermal technique [9, 53, 54], or simple thermal combustion of the organic precursor molecules [51, 69, 70]. That is why the syntheses of CDs are not as limited by the reactivity of the precursor materials, as it is the case in a classical organic synthesis. However, possible intermediate reaction products should not be neglected, as different precursor combinations have a strong impact on the properties of the resulting carbon nanoparticles [9, 71, 72]. The first attempt to synthesize fluorescent CDs from molecular organic precursors was reported by Bourlinos et al. in 2008, and included calcination of the salt of octadecylamine and citric acid in ambient atmosphere at 300 °C, yielding an organophilic fluorescent solid. In the same work, hydrophilic CDs were also obtained by treatment of citric acid and N-(2-hydroxyethyl)ethylenediamine in hydrothermal conditions at 300 °C [53]. Since this initial report, solvothermal synthesis has become the most popular method to synthesize CDs. In recent years, various groups were able to conduct CD synthesis from a vast number of natural resources, such as orange juice [73], flour [74], coffee powder [75], soy milk [76], oat meal [77], waste paper [78], lime [79], ginger [80], rice [81], grass [10], or even human urine [82]. This emphasizes the basic concept of CD synthesis, highlighting that it only requires the thermal treatment of substances, which contain carbon, oxygen and hydrogen. However, early observations also suggested the favorable effect of nitrogen on the CD optical properties, which was ascribed either to surface passivation, or to incorporation of this element into the core [83, 84]. Therefore, the focus of the current research switched to the controlled synthesis from selected molecular precursors, which often contain other heteroatoms such as nitrogen or sulfur. In general, the synthesis of doped CDs involves one molecular precursor material, which provides the major carbon framework, whereas another molecular precursor introduces other elements into the structure. The most prominent combination in this respect has been citric acid as carbon source, combined with nitrogen containing molecules, for example ethylenediamine or urea [9, 13, 69]. Zhu et al. reported the synthesis of highly fluorescent CDs (PLQY up to 82%) by reacting citric acid with ethylenediamine at temperatures ranging from 150 – 300 °C under hydrothermal conditions (Fig. 2(a)). Aside from the temperature, optical properties of the resultant nanoparticles were dependent on the ratio between the precursors, and as recently shown, also on the amount of solvent in the synthesis [55]. In terms of the suggested reaction mechanism, which is presented in Fig. 2(a), ethylenediamine and citric acid form a salt at elevated temperatures, before reacting to a large network, due to the multiple functional groups of both precursor molecules. At temperatures around 200 °C, the polymerized network carbonizes, resulting in amorphous CDs with partial sp² crystallinity. Recently, the same group around Song et al. found, that aside from the polymerization reaction, citric acid and ethylenediamine form the fluorescent molecule 5-oxo-1,2,3,5-tetrahydroimidazo[1,2-α]pyridine-7-carboxylic acid (IPCA) in a second simultaneous intramolecular reaction [67, 68]. After separating the reaction products (140 °C) by column chromatography, they collected the highly fluorescent phase and identified the structure of this fluorophore. IPCA can be seen as a derivative of fluorescent citrazinic acid, which is formed by the reaction of citric acid and ammonia, as reported by Sell and Easterfield in 1893 [85]. Several other studies mentioned this reaction in the past [86, 87], before recently Kasprzyk et al. reported that the heating of citric acid in the presence of α,β-diamines, β-amino thiols and β-amino alcohols, (e.g. ethylenediamine, cysteine or ethanolamine) generally results in the formation of this class of fluorophores [88, 89].

 

Fig. 2 (a) Reaction of citric acid and ethylenediamine to form CDs under hydrothermal conditions at 200 °C. Reproduced with permission from [9], Copyright 2013 Wiley-VCH. (b) CDs with tunable emission color by solvothermal treatment of (1,4)/(1,3)/(1,2)-benzenediamine at 180 °C. Reproduced with permission from [56], Copyright 2015 Wiley-VCH. (c) Metal-doped CDs (Cu(II)-N) by pyrolisis of a Cu(II)-EDTA complex at 250 °C W. Reproduced with permission from [90], Copyright 2015 Wiley-VCH.

Download Full Size | PPT Slide | PDF

Another approach towards N-doped CDs was recently reported by Jiang et al., who synthesized color-tunable CDs by varying the precursor materials [56]. Instead of having two precursor molecules, the synthesis contained only phenyldiamines, which were solvothermally heated in ethanol at 180 °C for 12 h (Fig. 2(b)). The variation of the functional group configuration of this precursor from ortho, to meta and para changed the emission color of the resulting CDs from green, to blue and red.

The second prominent heteroatom, which is included into the structure of CDs is sulfur. Aside from limited examples, where only sulfur was reported to be doped into the CDs [91, 92], most studies focus on its co-doping with nitrogen, in order to improve the optical properties of the CDs, in particular their PL QY [57, 93]. Dong et al. synthesized nitrogen-sulfur co-doped CDs via a one-pot hydrothermal treatment of citric acid and L-cysteine at 200 °C (PL QY of 73%) [57]. The presence of nitrogen and sulfur in the structure was confirmed by X-ray photoelectron spectroscopy (XPS), showing pyridinic (N_1s: 399.7 eV) and pyrrolic (N_1s: 400.6 eV) nitrogen species, as well as C-S-C units (S_2p: 163.4 and 164.8) [94]. For the same reaction, Kasprzyk et al. pointed out the formation of highly fluorescent 5-oxo-2,3-dihydro-5H- [1,3]thiazolo[3,2-a]pyridine-3,7-dicarboxylic acid (TPA), which belongs to the same class of fluorophores, mentioned above [88].

Following classical semiconductor doping, the Feng group recently introduced boron and phosphorus containing molecular precursors to the synthesis of fluorescent CDs [58, 59]. Following a similar synthetic protocol, hydroquinone was combined with either phosphorous tribromide or boron tribromide in hydrothermal conditions at 200°C, resulting in two CD species with varying optical properties. Barman et al. reported P-N and B-N co-doped CDs, made from ortho-phosphoric acid and boric acid in combination with diethylenetriamine and citric acid heated at 170°C in a Teflon-lined autoclave [60]. Besides heteroatoms, there have been recent examples of metal doping of CDs. Wu et al. prepared copper-nitrogen co-doped CDs, by thermal pyrolysis of Na₂[Cu(EDTA)) at 250°C in air (Fig. 2(c)). Cu(II)-ions were claimed to be covalently chelated by nitrogen atoms in the graphitic core of the CDs, which was confirmed by Fourier transformed infrared spectroscopy (FT-IR) and electron spin resonance (ESR) measurements [90]. Bourlinos et al. mixed Gadolinium containing gadopentetic acid with Tris-base, and pyrolysed the mixture with betaine hydrochloride as a surface modifier at 250 °C in air. The presence of Gd(III) in the CDs was confirmed by applying these nanoparticles as magnetic resonance imaging positive contrast agent [95].

An important question for most of the doped CD species remains, whether and to which extend, those heteroatoms / dopants are being incorporated into the CD core during the synthesis. Unfortunately, convincing proofs are lacking, as most characterization techniques do not provide sufficient spatial resolution to distinguish between the surface and the core of such small nanoparticles. In addition, it has to be taken into account, that the presence of heteroatoms may favor the formation of fluorophores, which can be incorporated or attached to the CDs. Therefore, traditional organic chemistry, as well as doping of related materials, such as graphene quantum dots, graphene or also graphitic carbon nitride may provide some related information [96–98].

3. Structure and optical properties of CDs

As the term “carbon dots” is used to denote a rather diverse class of materials, their structure may differ a lot within different synthetic techniques, which is particularly true for the case of doped nanoparticles. CDs are generally considered to consist of an amorphous carbon framework, which contains crystalline domains with mostly sp² hybridization, as illustrated by Fig. 3(a)-3(c). The presence of both crystalline sp² and sp³ domains, with corresponding peaks at 1340-1348 (D-band) and 1551-1560 cm⁻¹ (G-band) has been shown by Raman spectroscopy [99, 100]. Such picture is supported by HRTEM images and their electron diffraction analysis, showing lattice fringes and distinct diffraction patterns in most CDs, with distances in the range of 2.1 – 2.4 Å for the (001) lattice plane, and/or 3.2 – 3.8 Å for the (002) lattice plane of graphite, as shown in Fig. 3(a) [51, 101, 102]. The large variation in these distances, especially in the case of the (002) lattice planes is commonly ascribed to turbostratic disorder in multiple stacked sheets [103, 104]. The existence of the (002) graphite lattice planes has been also confirmed from X-ray diffraction (XRD) pattern as evidenced by a broad peak centred at 2θ equal 25°, which corresponds to a lattice distance of 3.4 Å (Fig. 3(d)) [9]. The broad character of the XRD signal reflects the existence of the amorphous framework in CDs at the same time.

 

Fig. 3 (a) HRTEM image of CDs, showing lattice planes with distances of 0.32 and 0.24 nm (3.2 and 2.4 Å). Reproduced with permission from [102], Copyright 2015 American Chemical Society. (b-c) Schematic representation of CDs highlighting the existence of a carbogenic core with sp² domains and different surface functionalizations (OH, COOH, polymers). (d) XRD pattern of overall amorphous CDs indicating the presence of lattice spacing of 0.34 nm belonging to the (002) plane of graphite. (e) FT-IR spectrum of CDs, exhibiting characteristic IR active modes of O-H, N-H, C-H, and C = O. (f) XPS spectrum of CDs confirming the presence of oxygen, carbon and nitrogen. The inset shows the high resolution XPS spectrum of the C_1s, revealing the presence of three carbon species, namely aliphatic, oxygenated and nitrous carbon. ((d)-(f) Reproduced with permission from [9], Copyright 2013 Wiley-VCH.

Download Full Size | PPT Slide | PDF

Besides the presence of a carbogenic core with different kinds of carbon hybridization, CDs also contain different surface functional groups. Most common surface functionalities include carboxylic acids, alcohols (Fig. 3(b)), or in the presence of an amino precursor, amine groups. Post-preparative surface passivation of CDs with PEG or other polymer-like surface ligands is also possible, which is schematically presented in Fig. 3(c) [30]. This passivation not only prevents agglomeration of nanoparticles, but is also beneficial for the optical properties of CDs [105, 106]. Fourier transform infrared spectroscopy (FT-IR) is widely employed to identify the characteristic bonds at the surface of CDs, which are either present after the synthesis or as a result of post-preparative functionalization steps. Figure 3(e) shows a FT-IR spectrum of CDs which reveals stretching vibrations from O-H, N-H, and C-H bonds at wavenumbers between 3700 and 2700 cm⁻¹ [9]. At lower wavenumbers, there are two other distinct features, which could be assigned to a C = O stretching mode at 1635 cm⁻¹, and the N-H bending mode at 1570 cm⁻¹. At the wavenumbers lower than 1500 cm⁻¹, specific peak assignment to particular modes is generally difficult, as both C-C and C-O bonds are active in this region.

For elemental analysis of CDs, X-ray photoelectron spectroscopy (XPS) is often employed. It not only allows estimation of the absolute ratio between the different elements present in the structure (Fig. 3(f)), but also provides information on different chemical states of the elements. The XPS C_1s peak located at 284-289 eV belongs to three different carbon species, namely carbon bound to carbon at 284.7 eV, carbon bound to oxygen at 286.2 eV and carbon bound to nitrogen at 287.7 eV [9]. In case of the nitrogen doping, the presence of pyrrolic nitrogen, pyridinic nitrogen or nitrogen bound to oxygen can be revealed in a similar way [105].

Same as for structural features, optical properties of CDs also depend on the chemical composition of the precursors used during bottom-up synthesis. Reaction parameters such as temperature and pressure further influence the polymerization and carbonization steps. CDs, which are synthesized from precursors containing only carbon and oxygen atoms are generally considered as “undoped CDs” but may also be denoted as oxygen-CDs (O-CDs). A large variety of CDs contain additional elements, most commonly nitrogen (N-CDs), but also sulfur (S-CDs) or chlorine (Cl-CDs). Yet another kind of CDs contain two or more doping atoms at the same time in addition to carbon and oxygen, for example nitrogen-sulfur co-doped CDs (N,S-CDs).

Raman spectra of CDs provide information on crystallinity of CDs. The intensity ratio I_D/I_G of the Raman D (disorder) and G (crystalline) carbon bands is often used to determine the quality of graphene sheets or carbon nanotubes [100]. These characteristic carbon Raman bands are also found in CDs, where the G band confirms the existence of the sp² hybridized (crystalline) core, and the D band accounts for defects and disordered carbon associated with amorphous carbon. Small I_D/I_G ratios (~0.5) show that the carbonization process during the synthesis lead to a high crystalline CD core, while larger ratios indicate a growing disorder and/or amount of amorphous carbon within CDs. Figure 4 shows Raman spectra for O-, N-, and N,S-CDs [57]. O-CDs show the largest I_D/I_G ratio of ~0.8 (Fig. 4(a)), N-CDs of ~0.6 (Fig. 4(a)) and N,S-CDs of ~0.5 (Fig. 4(b)). Apparently, O-CDs are characterized by the largest disorder. As discussed later, this is in line with the absence of well pronounced absorption peaks (refer to Fig. 6(c)) because these are related to crystalline, sp² hybridized carbon in the CD core.

 

Fig. 4 Raman spectra of (a) O-CDs (blue), N-CDs (green) and (b) N,S-CDs (black). The Raman D to G band peak intensity ratio I_D/I_G changes depending on the dopants. Reproduced with permission from [57], Copyright 2013 Wiley-VCH.

Download Full Size | PPT Slide | PDF

Sudolská et al. modeled absorption of O-CDs using a simple multilayer model consisting of oxygen functionalized pyrene and coronene building blocks [107]. The aromatic carbon rings form a conjugated sp² system that overlaps when several building blocks are stacked up. They identified a high energy UV absorption band of π→π* nature at 260 nm, and a lower energy band at 300 nm related to the interlayer π→π* charge transfer on its high energy side and to n→π* transitions on its lower energy side. As shown in Fig. 5(a), for the high energy side at 295 nm the charge depletion in the middle layer and its accumulation in the outer layer illustrate interlayer charge transfer, while for the low energy side at 310 nm, the charge accumulation can be found rather around the functional groups. π→π* transitions originate from the sp²-hybridized carbon core and n→π* are due to non-binding electron orbitals, introduced by the oxygen functionalization at the CD edge.

 

Fig. 5 (a) Calculated charge depletion (blue) and charge accumulation (red) for ground and excited states of O-CDs for different excitation wavelengths. At 295 nm excitation, charge depletion in the middle layer and charge accumulation in the outer layer illustrates an interlayer charge transfer. Reproduced with permission from [107], Copyright 2015 American Chemical Society. (b) Simulated absorption spectrum of N-CDs. The inset shows the single layer CD model with edge functionalization (grey: carbon; red: oxygen; blue: nitrogen; white: hydrogen). (c) Simulated emission spectrum of N-CDs based on the lowest excited state. ((b)-(c) Adapted with permission from [44], Copyright 2014 American Chemical Society [44].

Download Full Size | PPT Slide | PDF

Strauss et al. modelled absorption and fluorescence spectra of N-CDs using two differently sized, single layer, amide capped graphene sheets (Fig. 5(b),5(c)) and a bilayer model consisting of two stacked graphene sheets [44]. The sheets were all non-planar due to the out-of-plane distortions of the sp² network caused by functional groups. The employed bilayer model was found to have a larger interlayer spacing than general π-stacked dimers but still allowed to consider electronic interactions. A larger sheet size, stacking and hydroxylation all caused a redshift of the spectral features, while pyridinic nitrogen and epoxy-oxygen atoms caused a blueshift. The model of Strauss et al. was successful in describing the π→π* character of all considered states, but could not be used to address the processes at longer excitation and emission wavelengths. Both, simulated absorption (Fig. 5(b)) and emission spectra (Fig. 5(c)) were found to be qualitatively similar to experimental data. We note that experimentally measured absorption spectra of O-CDs are typically unstructured without any particular peaks (Fig. 6(c)) [57, 108], while N-CDs show well resolved π→π* and n→π* peaks (Fig. 6(a),6(c)) [9].

 

Fig. 6 (a) Absorption, PLE and emission spectrum of N-CDs. (b) Corresponding excitation dependent emission of N-CDs. (a)-(b) Reproduced with permission from [9], Copyright 2013 Wiley-VCH. (c) Absorption spectra appear unstructured for O-CDs while N- and Cl-CDs show pronounced absorption features. (d) Emission spectra of O-, N- and Cl-CDs at 360nm excitation. Only N-CDs show strong emission. (c)-(d) Reproduced with permission from [108], Copyright 2013 The Royal Society of Chemistry.

Download Full Size | PPT Slide | PDF

Due to their structural similarity to the crystalline CD core, stacked polyaromatic hydrocarbons (PAHs) can mimic absorption features that are observed in CDs, as outlined in the theoretical model of Sudolská et al. [107], Fu et al. experimentally compared the optical properties of N-CDs with those of PAHs [102], and found out that a mixture of anthracene, pyrene and perylene in a polymer matrix can indeed closely reproduce, both the absorption and emission of CDs (Fig. 7(a),7(b)), as well as their excitation dependent emission (Fig. 7(c)). Therefore, Fu et al. concluded that CDs can be considered as molecular nanocrystals with PAHs embedded in a sp³-hybridized carbon matrix. The large emission Stokes shift was assigned to exciton self-trapping in stacked PAHs, while the width and excitation dependence of emission was related to different band gaps of PAHs (Fig. 7(d)). The exciton energy decrease by self-trapping is related to the exciton-phonon coupling constant, which is dependent on the intermolecular distance between stacked PAHs. These findings highlight that structural differences on a molecular level in CDs alter their optical properties.

 

Fig. 7 Comparison between N-CDs in aqueous solution (black dotted line) and a PMMA film containing a mixture of PAH molecules anthracene/pyrene/perylene in molar ratio of 10:10:1 (blue solid line). (a) Normalized absorption spectra. (b) Normalized emission spectra at different excitation wavelengths: 340nm (blue), 420nm (cyan), 440nm (orange), 480nm (red). (c) Excitation wavelength (x-axis) dependent emission tracking the position of the emission maximum (y-axis). (d) Proposed model for exciton-self trapping in CDs and stacked PAHs which is used to explain the large Stokes shift, while the emission bandgap is determined by each single PAH unit. Reproduced with permission from [102], Copyright 2015 American Chemical Society.

Download Full Size | PPT Slide | PDF

Most CDs emit strong blue luminescence upon the excitation within their low energy UV absorption band (Fig. 6(a),6(d)). This emissive band is located around 350 nm and is commonly ascribed to n→π* transitions in CD literature [109], while Sudolská et al. also mentioned the contribution of the interlayer π→π* charge transfer [107]. The n→π* transitions are of special importance in CDs, as they are related to both doping and surface functionalization at the carbon core edge. In contrast, upon excitation within the high energy π→π* UV absorption band (conjugated carbon rings within the core), a weak luminescence is observed. Fluorescence intensity also strongly depends on the doping atoms used. As shown in Fig. 6(d), N-CDs show high emission intensity, while Cl-CDs and O-CDs only emit weakly. With increasing excitation wavelength, CDs commonly show an excitation dependent emission: The emission peak redshifts and its emission intensity drops in accord with the decreasing absorption (Fig. 6(b)). Figure 7(c) tracks the position of the emission peak with an increasing excitation wavelength. It maintains the same position until an excitation wavelength of about 380 nm. This corresponds to the tail of the n→π* absorption peak (Fig. 6(a)). Once the excitation wavelength increases further, the surface state region of a low absorption strength starting from about 400 nm is excited and emission strongly redshifts. This kind of low intensity, redshifted emission is often associated with surface- or defect states in the amorphous carbon shell of CDs [109–113]. The defect states of CDs are also related to functional groups at the CD surface. We note that CDs are often labeled as ‘excitation independent’ in literature, if no emission redshifts occur upon excitation within the n→π* UV absorption band. However, excitation of low energy surface states will still lead to an emission redshift. For example, Dong et al. report ‘excitation independent’ N,S-CDs, whose emission still redshifts at wavelengths longer than 400 nm with a significantly decreased emission intensity [57]. Recently, citric acid derived CDs were found to contain a significant amount of organic fluorophores [67, 68, 88, 89], which show no excitation dependent emission, and thus contribute to the claimed “excitation independence”, such was the case for N,S-CDs [57].

Xu et al. studied the interplay of the amorphous sp³ surface region of N-CDs with the crystalline sp² core at different oxidation levels of CDs [84]. They found that the oxidation increases the sp³ shell region and reduces the sp² core structures, with most oxygen groups located in the shell. Exemplary TEM images of these CDs are shown in Fig. 8(h)-8(k), with core and amorphous surface regions marked separately. The authors also found a correlation between the emission intensity and increasing nitrogen content in N-CDs, and proposed that nitrogen bound to the crystalline sp² core disorders the hexagonal ring structure, introducing emissive trap states within the core and thus enhancing the emission. The oxidized nitrogen in the sp³ shell did not contribute to emission enhancement. Furthermore, they assigned CD core and shell as separate emission centers while the dominating emission originated from the core structure.

 

Fig. 8 (a) Ensemble PL spectrum of O-CDs in water (1) and single CD emission spectra (2-5). The inset shows O-CDs in aqueous solution in ambient light and greenish ensemble emission upon excitation with 488nm laser. (b-e) Single O-CDs excited with an azimuthally polarized laser beam. This is used to determine the characteristics of the excitation transition dipole moment. (f) Calculated excitation pattern of a single CD assuming a linear horizontal dipole. The double arrow illustrates orientation of the dipole. (g) Reference image without any CD. The excitation patterns suggest the presence of a single, fixed dipole. Adapted with permission from [114], Copyright 2014 American Chemical Society. (h-k) HRTEM images of N-CDs (h,i) and oxidized N-CDs (j,k). Crystalline core and amorphous shell are marked separately. Oxidized N-CDs show a smaller core and a larger shell. ((h)-(k) Reproduced with permission from [84], Copyright 2013 Wiley-VCH.

Download Full Size | PPT Slide | PDF

Ghosh et al. have conducted single particle luminescence measurements on O-CDs [114], synthesized using sucrose and PEG₁₅₀ as precursors with a short heating time in a microwave oven, which resulted in two CD species with different crystallization patterns and sizes (7-20 nm and 2-5 nm). They excited primarily low energy surface states of CDs (488 nm excitation wavelength) emitting from green to orange, and have found that the broad ensemble emission spectrum was formed by the superposition of single CDs emission (Fig. 8(a)). The excitation and emission transition dipole moments of single CDs were found to be parallel, by tracking the excitation pattern upon excitation with an azimuthally polarized laser beam (Fig. 8(b)-8(g)). They suggested that the emission of CDs originated from a single, fixed emissive dipole center, and claimed strong optical phonon coupling with the emission center. In contrast, Yu et al. reported a weak electron-phonon coupling in green emitting O-CDs [115]. Ghosh et al. assumingly excited only low energy surface states within the amorphous sp³ shell of CDs, while Yu et al. studied the emission originating from the core or edge states.

Depending on the synthesis method and precursors, N-doping of CDs can influence not only the amount of dopant atoms incorporated into the CDs, but also the size of the CDs as well as their structural properties. Zhang et al. synthesized N-CDs using CCl₄ as carbon source and NaNH₂ as both nitrogen source and the dechlorination reagent by a solvothermal method in toluene at 200°C [54]. The resulting N-CDs showed an increasing amount of incorporated N, as well as larger sizes and emission redshifted with the reaction time. Specifically, a reaction time of one hour resulted in 1.5 nm average sized CDs with 3.1% N-content and blue emission, while after 8 hours, 3.3 nm sized CDs with 7.2% N-content and green emission were obtained, and the PL QY increased from 3.5% to 13.4%. Longer reaction times of up to 16 hours led to CDs with yellow emission, which was accompanied by a slight decrease in PL QY. While a longer reaction time yielded CDs of increasing size, the authors excluded a quantum confinement effect as the reason behind the observed emission redshift, as a control sample of nitrogen-free CDs had a larger size than the comparable N-CDs but showed a blue-shifted emission. Chen et al. synthesized N-CDs using a mild solvothermal treatment in methanol at 70°C for 24 hours, with 2-azidoimidazole as precursor [116]. They proposed a reaction mechanism including self-polymerization of 2-azidoimidazole followed by formation of an interconnected polymer network that finally undergoes carbonization to form CDs. Size separation of the resulting N-CDs by chromatography yielded three distinctive CDs species: Blue emitting N-CDs with 2.1 nm average size and 19% N-content, cyan emitting N-CDs (4.6 nm in size, 31% N) and cyan-green emitting N-CDs (5.4 nm in size, 34% N). Similar to the findings of Zhang et al. [54], the increasing N content and size of these CDs led to an emission redshift. Furthermore, a reaction temperature of 50 °C yielded green emitting N-CDs, while its increase to 100 °C resulted in the blue emissive nanoparticles. Chen et al. concluded that both polymerization and nitrogen incorporation are less efficient at high reaction temperatures.

Jiang et al. synthesized full-colour tuneable CDs with varying nitrogen content by using phenylenediamine isomers as precursors in a solvothermal method in ethanol (180 °C, 12 h reaction time) (Fig. 2(b) and Fig. 9) [56]. Having exactly the same amount of atoms, the difference within the phenylenediamine isomers was in the substitution pattern of the amine groups. Meta-substitution resulted in blue emitting CDs (m-CDs, Fig. 9(a)), ortho-substitution in green emitting CDs (o-CDs, Fig. 9(b)) and para-substitution in red emitting CDs (p-CDs, Fig. 9(c)). Both the size and the nitrogen content of the CDs varied: blue m-CDs were the smallest with an average size of 6 nm and 4% of incorporated nitrogen, green o-CDs were 8 nm in size with 7% of nitrogen, and red p-CDs were the largest with 10 nm size and 16% nitrogen. Also in this work, an emission redshift with increasing size and nitrogen content was observed. PL QYs were found to be 5% for m-CDs, 10% for o-CDs and 20% for p-CDs, and emission lifetime of m-CDs was bi-exponential and 1 ns on average, while o- and p-CDs showed mono-exponential emission decays with 4 ns and 9 ns lifetime, respectively.

 

Fig. 9 (a)-(c) UV/Vis absorption spectra of phenylenediamine isomers m-PD (a), o-PD (b) and p-PD (c) (red lines) and corresponding m-CDs, o-CDs and p-CDs (black line) and PL emission spectra of m-CDs, o-CDs and p-CDs (blue line for 365nm excitation, green line for 420nm and 510nm, respectively). (d) Photograph of m-CDs, o-CDs and p-CDs emission under 365nm UV illumination in ethanol. Reproduced with permission from [56], Copyright 2015 Wiley-VCH.

Download Full Size | PPT Slide | PDF

In another report, the nitrogen content was varied by adjusting the amount of water in a hydrothermal synthesis with commonly used precursors citric acid and ethylenediamine [55]. The size, emission peak positions and emission lifetimes were found to be the similar for each CD sample, while the nitrogen content decreased with the amount of water used during synthesis. The PL QY peaked at a medium amount of water used, attributed to a balance between the amidation and the carboxylation.

Qian et al. synthesized CDs by a solvothermal method with CCl₄ as a carbon source and diamines of different chain lengths as the nitrogen source [72]. They used diamines separated by two (1,2-ethylenediamine), three (1,3-propanediamine) or four carbon atoms (1,4-butanediamine), such that the corresponding CDs are abbreviated here as e-CDs, p-CDs, and b-CDs. e-CDs (the shortest diamine chain) had the lowest oxygen and the highest nitrogen content. The highest oxygen and the lowest nitrogen content was found for p-CDs (mid-length chain), while b-CDs (the longest diamine chain) were in an intermediate state. e-CDs (the highest N-content) showed the lowest PL QY of 20% with a maximum emission at 406 nm, followed by b-CDs with 25% PL QY and emission at 398 nm and p-CDs (the lowest N-content) with 36% PL QY and emission at 390 nm. These results show that while different precursors result in different amount of dopant atoms incorporated in CDs, the emissive properties of CDs may not be necessarily related to that, which can be due to the different degree of carbonization of the resulting CDs depending on the precursors used. Qian et al. assigned the observed fluorescence enhancement to incorporation of pyridinic and pyrrolic nitrogen atoms, which introduced defect states in the conjugated carbon lattice of the CD. In addition, pyrrolic nitrogen atoms were found to undergo protonation. The resulting hydrogen bonds and the altered electron transfer were claimed to enhance emission properties of CDs.

In 2013, Zhu et al. published one of the most commonly used protocols for the synthesis of N-CDs in a hydrothermal reaction using citric acid and ethylenediamine (Fig. 2(a)) [9]. Under optimal proportion of all starting materials, these N-CDs reached a PL QY of 80%. The authors distinguished between the crystalline, carbonized carbon core, which is produced as a result of high temperature synthesis, and a surface or molecule-like state, which dominates at lower reaction temperatures, when CDs are polymerized but not fully carbonized. They have claimed that both structural regions of CDs play an important role for their emission, and that the excitation dependent emission of CDs originates from molecular states on the surface of CDs.

Whether trap- or molecular states on the surface, or size effects (quantum confinement) of CDs play a dominant role for their excitation dependent emission remains controversial. As discussed before, most publications suggest that the size plays little if any role for the emission of CDs. Contrary to that point of view, Bhattacharya et al. reported on size dependent quenching of N-CDs in ferritin nanocages [117]. N-CDs were synthesized following the common hydrothermal procedure using citric acid and ethylenediamine. They ascribed the excitation dependent emission of CDs to inhomogeneous size distribution of CDs, and proposed that the quantum confinement of surface trap states is responsible for CD emission.

Strauss et al. addressed the fluorescence mechanism of N-CDs using citric acid and urea as the carbon and nitrogen source, respectively. The carbonization of CDs was induced in a microwave reactor under rapid heating for 5 minutes [44]. One sample of CDs was kept under pressure of 15 bar, labelled pCDs, while another sample was kept under atmospheric pressure, labelled aCDs. Both, pCDs and aCDs showed the common absorption peak at 350 nm with a corresponding emission at 450 nm. Interestingly, aCDs also showed a second absorption peak at 410 nm with a related emission at 519 nm, and a long absorption tail throughout the whole visible spectrum. The second absorption peak was related to the presence of another species of CDs formed during the synthesis under atmospheric pressure. Further spectral studies in this work focused on the pCDs, which, despite being assigned to a single species, showed multiexponential fluorescence decay in the nanosecond range (6 ns). Fluorescence lifetimes were found to decrease at longer wavelengths, indicating different deactivation pathways in the red part of the spectrum. Computational modelling of these CDs (see Fig. 5(b) and the related discussion above) succeeded to mimic experimental absorption spectra, as well as the blue emission of CDs, but failed to accurately predict emission properties at longer excitation and emission wavelengths. Importantly, Strauss et al. were able to evaluate the influence of sp² lattice defects by both nitrogen doping and functional groups on the emission of CDs. Their model revealed a blueshift of spectral properties by pyridinic nitrogen at the CD edge and to a lesser extent by epoxy-oxygen atoms, while hydroxylation caused a spectral redshift.

Wang et al. have compared the standard hydrothermal synthesis approach using citric acid and ethylenediamine (EDA) with the case where the nitrogen precursor EDA was exchanged for diethylenetriamine (DETA) and triethylenetetramine (TEPA), which increased the amount of nitrogen and the chain length of the precursor [71]. Neither the nitrogen content (~16%) nor the average size (~4 nm) of the resulting CDs changed significantly upon using those different nitrogen precursors, while the PL QY did. Nitrogen atoms were incorporated into the heterocyclic carbon ring system during the synthesis, specifically around the core termination or core edge in pyrrolic and pyridinic nitrogen form or as an amide group. The oxidation state of N-CDs (oxygen/carbon ratio) did not influence PL QY, while the amount of nitrogen – carbon double bonds correlated with PL QY leading to the conclusion that an increasing amount of cyclic imines enhanced the conjugated system in the carbon core and improved the PL QY. PL lifetimes slightly decreased with the chain length of amines used, being 14 ns, 13 ns and 10 ns for CD-EDA, CD-DETA and CD-TEPA, respectively. Notably, the PL decays were monoexponential, which is in contrast to the previously mentioned study by Strauss et al. [44], where CDs synthesized with the longest chain nitrogen precursor TEPA showed the best PL stability (but low PL QY), while CDs synthesised with employment of shorter chain molecules EDA and DEPA had a photobleaching behaviour similar to organic fluorophores (but higher PL QY). Wang et al. concluded that these results indicate possible incorporation of organic fluorophores into the carbon domains.

Recently, as previously mentioned in chapter 2.2, Song et al. reported on the identification of a strong blue emitting fluorophore 5-oxo-1,2,3,5-tetrahydroimidazo[1,2-α]pyridine-7-carboxylic acid (IPCA), generated during their synthesis of CDs [67]. Synthesis was conducted in a similar way as in the previously discussed approach by Wang et al. [71], using citric acid and EDA as precursors in a hydrothermal reaction. IPCA showed the largest contribution to fluorescence of the resulting product, with overall QY of almost 86%, and a spectral profile showing a stunning similarity to CDs. Comparing Fig. 10(c) with Fig. 10(d) reveals that absorption and emission features are located at almost the same positions. Song et al. thus considered IPCA as an important emissive centre of CDs, which is either attached to the surface of CDs or integrated into the carbogenic core. Nitrogen plays an important role in the emission of this molecular centre, which supported the previously discussed observation by Wang et al. on the correlation of the amount of nitrogen – carbon double bonds with PL QY of CDs [71]. Different to IPCA, which showed a monoexponential PL decay, CDs cores had a multiexponential PL decay and their emission varied with the excitation wavelength. Therefore, Song et al. concluded, that the CD cores possess their own intrinsic emission, different to the molecular emission state of IPCA. As shown in Fig. 10(a), these different emissive states of CDs could be controlled by the reaction temperature. At low temperatures, CDs with a dominant light emissive molecular state are formed, while CDs with PL contributions from carbon core states can be obtained at high temperatures. It was furthermore shown, that IPCA can be used as a precursor on its own in a hydrothermal reaction leading to CDs, and can thus be considered one of the reaction intermediates on the way to full carbonization of the starting precursor material. It could therefore also improve integration of highly active imines into the carbon core. Song et al. introduced a scheme shown in Fig. 10(b) that involves different pathways during the synthesis of CDs. The precursor materials citric acid and ethylenediamine can either form IPCA in a (reversible) condensation reaction or polymerise under participation of ethylene diamine as a linking agent. Irreversible carbonization can take place from either of the intermediate states to form carbon cores with possible integration of fluorophores. Carbonization reactions slightly vary for each intermediate state and thus could lead to structural differences in the resulting CDs.

 

Fig. 10 Formation of organic fluorophores and CDs during the synthesis and their optical properties. (a) Low temperature hydrothermal reaction yields mainly molecular fluorophores, while high temperature results in crystalline carbonized cores. (b) Schematics of the reaction pathways involved into the citric acid based CD synthesis showing different intermediate states that lead to carbonized CDs with integrated fluorophores. (c) Exemplary absorption and emission spectra of N-CDs and (d) those of IPCA fluorophores. Spectral features of IPCA are very similar to N-CDs, making it difficult to distinguish between this molecular fluorophore and the carbonized state of CDs. ((a), (b), (d) Adopted with permission from [67], Copyright 2015 The Royal Society of Chemistry; (c) Reproduced with permission from [46], Copyright 20012 The Royal Society of Chemistry.

Download Full Size | PPT Slide | PDF

In a recently published article, the Yang group generalized their results on the fluorophore formation during the CD synthesis [68]. They have stressed that citric acid can react with a whole family of amines via several condensation reactions forming strong emissive fluorophores with heterocyclic ring structures. As already mentioned above, Kasprzyk et al. reported the formation of highly luminescent fluorophores (up to 79% PL QY) by a condensation of citric acid with specific β-amines [89], and assigned them to ring-fused 2-pyridones, which can also incorporate other doping atoms such as sulfur. The incorporation of nitrogen within the 2-pyridone ring structure is in agreement with the positive correlation of the amount of cyclic imine and nitrogen – carbon double bonds with PL QY of the resulting CDs, as discussed by Wang et al. [71]. The above mentioned fluorophores feature absorption and emission bands very similar to those commonly observed in CDs, which makes it difficult to distinguish these components without suitable separation methods. A high energy UV absorption band at 260 nm was assigned to π→π* transition of sp² carbons and another band at 350 nm was due to n→π* transitions by non-bonding orbitals such as C = O. At the same time, the organic fluorophores do not show any excitation dependence of their emission profiles, which is commonly found in CDs. It can therefore be speculated that for those cases in literature where the excitation independence was claimed for CDs, the emission originated from fluorophores rather than from CDs. At the same time, fluorophores can be integrated into the amorphous shell of CDs or become attached to their surface.

Kwon et al. conducted surface functionalization of N-CDs with para-substituted anilines [105], which featured the same amine head group but had different functional groups containing, oxygen, nitrogen and sulfur, namely methoxy-, methylamino- and methylthio- groups. N-CDs without functionalization showed the π→π* absorption band at 250 nm and the n→π* band at 360 nm with an emission peak at 420 nm for either 250 nm or 360 nm excitation wavelengths (Fig. 11(b)). Upon functionalization, the intrinsic n→π* emission at 420 nm was quenched and a new emission peak appeared at 565 nm for anilines with the methoxy- group (Fig. 11(c)), 590 nm for the methylthio- and 600 nm for the methylamino- group. These sharp, extrinsic emission peaks originated from the para-substituted anilines, while the respective emission redshifts were related to the π-electron donating strength of the functional groups, with oxygen possessing the lowest contribution, followed by sulfur and nitrogen. PL QY dropped from 40% for unmodified N-CDs to 31% for oxygen, 21% for nitrogen and 33% for sulfur containing anilines, which was ascribed to a loss of photoexcited electrons in the process of internal conversion based on the comparison of transient absorption spectra at short and long excitation wavelengths. This also confirmed the existence of competitive intrinsic and extrinsic PL centers with internal conversion pathways from high-energy intrinsic states to low energy extrinsic state as illustrated in Fig. 11(a). The data of Kwon et al. suggest a strong interaction between the carbogenic core of the CDs and the functional groups at their surface via conjugated n→π* orbitals.

 

Fig. 11 (a) Energy level diagram of surface functionalized CDs. Excited states in the π→π* band can either emit or be internally converted to lower lying n→π* band states and emit intrinsically from there. If surface functionalized anilines are present, energy is transferred and extrinsic emission from functionalized surface groups occur. The molecular orbitals are assumed. (b) Corresponding excitation-emission map for bare N-CDs and (c) for attached para-substituted anilines with methoxy group. Almost all n→π* band emission is quenched and emission only originates from the surface functionalized para-substituted anilines. Reproduced with permission from [105] licensed by CC BY 4.0.

Download Full Size | PPT Slide | PDF

Hu et al. compare chlorine doped Cl-CDs with N- and O-CDs (Fig. 6(c),6(d)) [108]. While nitrogen doping was found to enhance emission properties, chlorine doping improved photocatalytic activity. Chlorine radicals were thought to enhance charge carrier separation and migration. By introducing a simple energy band-bending model, Hu et al. could separate individual influences of doping atoms on the charge carrier distribution. Enhanced charge carrier separation of Cl-CDs was explained by creation of an internal electrical field within the CD due to opposing influences of surface groups.

Apart from nitrogen, other elements have been used for doping of CDs. A useful method to enhance emission properties of CDs reported by Dong et al. has been realised by nitrogen and sulfur co-doping [57]. The resulting N,S-CDs were synthesized by hydrothermal treatment using citric acid as carbon source and L-cysteine as both nitrogen and sulfur source. They showed a low Raman I_D/I_G ratio (~0.5, Fig. 4(b)) and bright blue luminescence at 420 nm, with PL QY of up to 78%. The co-doping with sulfur enhanced the high energy luminescent states introduced by nitrogen, while at the same time reduced broad, low energy states associated with oxygen. The absorption strength of surface states was found to be very low resulting in a weak long wavelength emission. Thus, N,S-CDs are a good example of CDs with high QY and a weak excitation dependence. However, Kasprzyk et al. reported the formation of the fluorophore TPA from the reaction of citric acid and L-cysteine [88], so that without further studies, a fluorophore contribution to the excitation independence of the N,S-CDs reported by Dong et al. cannot be excluded.

Following a report on solely sulfur doped CDs (S-CDs) which showed a high PL QY of up to 67% [118], Xu et al. prepared N,S-CDs using an ultra-low doping precursor ratio [119]. In a one-pot hydrothermal approach, sodium citrate was used as a carbon source and sulfamide as a nitrogen and sulfur source in an optimal ratio of 10:1, with a maximum PL QY of up to 55%. The PL QY was significantly altered when any of the reaction parameters such as precursor ratio, reaction temperature and reaction time varied. While nitrogen and sulfur were found to be incorporated into the CD core mainly by C-N and C-S bonds, the PL QY correlated with the sulphate content and the amount of oxidized carbon atoms at the CD surface. Sulphate was thought to push connected carbon atoms into a high oxidation state thus increasing the electron density and introducing new radiative recombination pathways. Xu et al. also calculated the theoretical density of states for a 268-atom CD model, and found that impurities caused by solely S or N doping shifted the related density-of-state peaks towards the LUMO, while impurities by N and S co-doping shifted them towards the HOMO. This was ascribed to the dangling S- and N- bonds serving as an energy gradient, which enhanced electron transfer from the surface, while the dangling S-N bonds increased hole transfer from the surface. These results fit well to the general observation that CDs are good electron and hole acceptors as well as donors and stresses once more the importance of dopant atoms for the electronic properties of CDs [60, 120–122].

Using N-CDs as a starting material, co-doping with heteroatoms other than sulfur has as well been studied. Barman et al. used phosphor (P) and boron (B) as co-doping atoms [60], following the common hydrothermal treatment using citric acid as a carbon source and diethylenetriamine as nitrogen source and boric acid as boron source or orthophosphoric acid as a phosphor source. XPS analysis confirmed incorporation of B or P together with nitrogen into the carbon core. The n→π* absorption peak located at 350 nm for N-CDs was ~15 nm blue-shifted for N,B-CDs, and was slightly red-shifted for N,P-CDs. The emission peak for all three kinds of CDs was located at 452 nm, with N,P-CDs showing the brightest emission (PL QY 70%), followed by N-CDs (64%) and N,B-CDs (39%). Both nitrogen and phosphor are thought to donate additional electrons into the n→π* system, while boron induces slight structural relaxations to change its electronic structure and create holes in CDs, thus enhancing non-radiative recombination. Attaching nickel (II) phthalocyanine as a fluorophore to the surface of these CDs allowed to study differences in the charge transfer processes. N,B-CDs showed the strongest PL quenching and the longest decay times attributed to simultaneous electron (due to the nitrogen n-type doping) and hole (due to the boron p-type doping) transfer processes. N,P-CDs and N-CDs showed less PL quenching, respectively, attributed to dominating electron transfer processes (n-doping, only).

Yang et al. reported selenium (Se) doped CDs with a yellow emission centered around 563 nm [123]. High resolution XPS confirmed Se incorporation in the form of C-Se-Se and C-Se bonds with a doping content of about 11%. Wu et al. introduced copper doping into the CD matrix using chelating Cu-N complexes (Fig. 2(c)) [90], which significantly increased the electron accepting and donating ability of Cu-N-CDs as well as their conductivity. The amount of copper doped into CDs varied from 2.1% to about 0.6% by increasing the reaction temperature from 250 °C to 350 °C, respectively. Cu-N-CDs had a Raman I_D/I_G ratio of 0.8 and increased absorption strength compared to undoped CDs, which was credited to possible Cu-graphite charge transfer states. Bright blue emission around 450 nm has been detected upon excitation of the n→π* band at 365 nm, and showed a strong excitation wavelength dependent character. It therefore appears that, in contrast to sulfur, Cu doping enhances the absorption and emission of both core and surface states with an absence of citric acid derived fluorophores.

Doped CDs can also be synthesized from much more complex precursors. As already mentioned before, green synthetic routes can involve food-based precursors, such as coffee or fruit juice [75, 79]. Very recently, the Baker group collected the fragrant urine of an asparagus eating student to synthesize sulfur doped “Pee-CDs” (though with only 3% PL QY) [82]. A proper purification of the precursor materials is a common challenge for most green synthetic routes. Bhunia et al. used vitamin B₁ (Thiamine) as carbon, nitrogen and sulfur source and a phosphate salt containing either one or three phosphate atoms as an additional co-doping source [93], to produce P,N,S-CDs with up to 76% PL QY. Using low reaction temperature (90 °C) in water resulted in the formation of blue emitting CDs (emission peak at 440 nm), while using a high reaction temperature (130 °C) in ethylene glycole resulted in green emitting CDs (470 nm). It was suggested that the phosphor incorporation into the CD core introduces additional sp² defect states and thus enhances the luminescence.

4. Applications of CDs

The ease of synthesis, bright fluorescence and advancing methods to modify the emission of CDs by doping, functionalization or other means, opens the door for a broad range of applications of these materials, as has been summarized in a number of reviews [2, 17–19, 26, 124]. In this section, we briefly discuss the use of CDs in analytical ion sensing, bioimaging, and optoelectronics.

The large surface area of solution dispersed CDs with amine- and carboxyl functionalities, allows their use as sensors for different ionic species. Quenching of the CD emission caused by the presence of specific ions has been widely used for analytical detection of different metal cations, in particular mercury ions in the range of several nanomoles [63, 125, 126], or even down to 0.05 nanomoles [127]. Such detection sensitivity is reaching the detection limit of inductively coupled plasma mass spectrometry and the electrochemical methods [128, 129]. Other studies reported detection of Cu²⁺ [130], Ag⁺ [72], and Be²⁺ [131]. Not only metal cations but also anions, such as S^2- [132], and CN^- [133], and DNA, labeled with ethidium bromide [134], have been detected using CDs. Besides the direct case when PL emission of CDs drops in the presence of metal ions, their emission might also be affected when a complex of CDs with some metal cations is sensitive to other molecules. Liu et al. illustrated a sensitive probe for glutathione, based on the CDs–Cu²⁺ system which showed a linear range of 0.5–80 μmol/L and a detection limit of 0.48 μmol/L [10]. CDs synthesized from citric acid / hyperbranched polyethyleneimine [135], and citric acid / ethylenediamine [9], showed varying sensitivities for Na⁺ (20% difference), Fe²⁺ (30%) and Zn²⁺ ions (2.5 times). Mohapatra et al. attributed the possible detection mechanism of nitrogen and sulphur co-doped CDs to the non-radiative electron transfer from the excited state to the d-orbital of the metal ion. The soft–soft acid–base interaction between the sulphur part of the CDs and Hg²⁺ makes this fluorescence probe selective towards Hg²⁺ in contrast to other metal ions [127].

Regarding bioimaging applications, there are a number of claimed advantages of CDs, namely high PL QY, solubility in water, a broad excitation band, the presence of functional surface groups facilitating their conjugation with various biomolecules, and last but not least low toxicity. Semiconductor quantum dots (QDs) for example have been reported to exhibit cytotoxic effects in concentrations varying from 62.5 µg mL⁻¹ to 400 µg mL⁻¹ depending on the core material, surface coating and the type of tested cell lines [136], while CDs exhibited very low or no cytotoxicity, even in much higher concentrations [16, 25]. It was reported that even with a concentration of CDs equal to 5 mg mL⁻¹ after 24 h of incubation, LC₅₀ of HeLa cells has not been reached [137]. For the in vivo studies, LD₅₀ was reported for the injected dose of 16.2 mg of CdSe/CdS QDs per kg of body weight of Kunming mice within 24 h [138]. While the death ratio for the injected dose of 26 mg kg⁻¹ of these QDs was 60% and reached 100% for 40 mg kg⁻¹, no signs of mortality were detected for CDs, even for higher doses of 51 mg kg⁻¹ [139].

Liu et al. used PEGylated-CDs (PEG: polyethylene glycol) to fluorescently label human breast cancer MCF-7 and human colon adenocarcinoma HT-29 cells [140]. Excitation in the visible (458 nm) or in the near-infrared (NIR) with two-photons (femtosecond pulsed laser at 800 nm) was used to activate green CD luminescence inside cells, as illustrated by Fig. 12(a), 12(b)). The upconversion ability of CDs is yet another useful optical property of these materials [30], which makes it possible to excite them within the so-called biological window of transparency in NIR and is also related to deeper tissue penetration by light. Green fluorescence from PEG-CDs in both cell lines could readily be detected, with the results suggesting, that the CDs resided mostly in the cell membrane and cytoplasm. Cao et al. estimated two-photon absorption cross-section of CDs: The typical average value at 800 nm was 39,000 ± 5000 GM (Goeppert-Mayer unit, with 1 GM = 10−50cm⁴s/photon) [30], which is comparable to other two-photon luminescent nanomaterials, e.g. 780 - 10,300 GM for CdSe QDs at 800 nm [141], and 47,000 GM for CdSe/ZnS core/shell QDs at 605 nm [142].

 

Fig. 12 Fluorescence images obtained from MCF-7 (a) and HT-29 (b) cells labelled with PEGylated CDs under two-photon excitation at 800 nm. Reproduced with permission from [140], Copyright 2015 American Chemical Society. Intravenous injection of CDs: (c) bright field, (d) as detected fluorescence (Bl, bladder; Ur, urine), and (e) color-coded images. The same order is used for the images of the dissected kidneys (c’–e’) and liver (c”–e”). Reproduced with permission from [31], Copyright 2012 American Chemical Society.

Download Full Size | PPT Slide | PDF

Yang et al. demonstrated that CDs injected in various ways into mice remain strongly fluorescent in vivo [31]. It was shown that the intravenously injected CDs are primarily excreted via urine. Such an excretion pathway has been widely reported in the literature for other kinds of PEGylated nanoparticles, such as PEGylated QDs [143]. Figure 12 illustrates the in vivo distribution of CDs. When injected intravenously for whole-body circulation studies, CDs emission from the bladder area was observed, and in 3 h after injection their emission could be detected in the urine. At 4 h after injection, mouse organs were harvested and the CDs were found to have accumulated mainly in the kidney (which is consistent with an excretion pathway via the urine) and scantly in the liver.

Yet another application of CDs in the biomedical field is related to their use as nanocarrier for drug delivery. CDs are able to deliver drugs to the targeted tumor locations by conjugation with various compounds for active targeting. Tang et al. reported loading of the anticancer drug doxorubicin, labelled by red fluorescent dye into PEG chains on the surface of green fluorescent CDs [144]. Such a conjugate exhibited time dependent sustained doxorubicin release at acidic pH levels, simulating lysosomal uptake in cells of cancerous tissue. Förster resonance energy transfer (FRET) between CDs and doxorubicin could be used to monitor the drug release. Oxaliplatin was conjugated with CDs by Zheng and associates [145], and the obtained hybrid system was successfully tested as a theranostic agent in vivo. Mukherjee et al. reported on two kinds of polymer functionalized CDs for theranostic application [146]. Multi-arm PEG functionalized CDs (MA-PEG-CD) showed a weak NIR emission within the biological window (Fig. 13(b)), while using poly(N-isopropylacrylamide) (PNIPAM) resulted in thermoresponsive CDs (TR-CDs) enabling the controlled release of a loaded drug (Fig. 13(a)). The CDs were synthesized using Molasse (a mixture of sugars or carbohydrates) and the respective polymers with or without adding the drug (Pentoxifylline) in a one-pot reaction. A sufficient amount of the polymer was preserved resulting in large CDs (Fig. 13(c)-(e)) with a crystalline core and extended amorphous shell (Raman I_D/I_G ratio ~1.2). Their absorption spectra showed a rather uncommon monotonically decreasing absorption over the visible range, which might be caused by overlapping absorption of different constituents smearing out specific absorption features. The emission was centered around 450 nm for both types of CDs without showing any fluorescence signal of the drug. Upon excitation at 785 nm, MA-PEG-CDs showed weak NIR emission, centered around 812 nm, which could be used for fluorescence imaging within the biological window (Fig. 13(b)).

 

Fig. 13 Polymer functionalized CDs for imaging and drug delivery. (a) Synthetic pathway leading to the formation of CDs for two different applications. Left: Use of highly branched polymers (multi-arm PEG, purple) leads to CDs with NIR emission and burst the release of the drug Pentoxyfilline (red). Right: Use of a thermoresponsive polymer PNIPAM (blue) results in CDs which can encapsulate the drug (red) and release it upon heating. (b) NIR emission spectrum of MA-PEG-CDs in cells upon 785nm excitation. Region 1 corresponds to scattering of the excitation beam and thus the entire cell appears luminescent (inset). Region 2 corresponds to the NIR emission of CDs and can be clearly detected in the sample (inset). (c-d) TEM image of MA-PEG-CDs at two different magnifications. (e) AFM image of MA-PEG-CDs. (f) Summary of measured physical parameters for MA-PEG-CDs (top, grey) and TR-CDs (bottom, white). Reproduced with permission from [146], Copyright 2015 Wiley-VCH.

Download Full Size | PPT Slide | PDF

Light emission of CDs has been also utilized in optoelectronic devices, such as in electrically driven CD-based light emitting diodes (LEDs) [147]. The authors used a layered architecture, based on an emissive layer of CDs, which was passivated with 1-hexadecylamine. This emissive layer was sandwiched between a poly(ethylenedioxythiophene):polystyrene sulphonate (PEDOT:PSS) hole injection layer and a 1,3,5-tris(N-phenylbenzimidazol-2-yl) benzene (TPBI) electron transport layer (Fig. 14(a)). For this device, a maximum external quantum efficiency of 0.083%, at a current density of 5 mA/cm², and a maximum brightness of 35 cd/m² was achieved. Zhang et al. reported multi-color electroluminescence from the similar CD-LEDs [23], which were operated at different driving voltages. With this approach, blue, cyan, magenta and white emission could be obtained. The maximum brightness was 24 cd/m² and 90 cd/m² for blue and white light, respectively, with an EQE of > 0.08%. The performance of these CD-based devices has been so far quite modest as compared to much more mature QD-based LED technology [148–150], so that specific ligands which enhance the overall conductivity of the CD active layer, a suitable matrix, and proper choices of energy matched carrier injection layers are crucial for the electrically driven CD LEDs.

 

Fig. 14 (a) Schematic diagrams of the CD-White-LEDs’ cross-section and (b) energy band diagram of CD-White-LED, with (c) molecular structures of PEDOT: PSS and TPBI, and the schematic drawing of the CD Reproduced from http://pubs.rsc.org/en/Content/ArticleLanding/2011/CC/C0CC05391K#!divAbstract [147] with permission from The Royal Society of Chemistry, (d) Emission from CDs and Zn-doped AgInS₂ nanocrystals deposited on a 380nm UV LED chip. Reproduced with permission from [152], Copyright 2014 The Royal Society of Chemistry.

Download Full Size | PPT Slide | PDF

In an alternative approach towards the optically driven LEDs, Xin and associates demonstrated bright blue - chromaticity coordinates at (0.19, 0.28), orange (0.45, 0.44) and warm white (0.34, 0.37) light from 3 different kinds of CDs produced by pyrolysis of epoxy-enriched polystyrene. The devices were made by coating CDs on the solid-state lighting unit comprised of a 370 nm excitation light-emitting chip [151].

Sun et al. combined blue-emitting CDs with green and red emitting ZnCuInS₂/ZnSe/ZnS core/shell QDs in PMMA matrix coated on a commercial UV chip with the emission peak wavelength centered at 385 nm. The resulting optically driven LED had a high colour rendering index (CRI) of 93, and by changing the concentration of each material in a PMMA matrix, it was possible to tune the region of the colour temperature from 3825 to 6452 K along with the black body locus [24]. In yet another combination of phosphors, white LEDs were produced by combining three carbon-based phosphors based on blue-emitting CDs and two different kinds of (green and red emitting) polymer dots in a polyvinyl pyrrolidone matrix [15]. The resulting devices had a high colour rendering index (85–96) and widely variable colour temperatures (2805–7786 K). Chung et al. produced a white LED on the base of a 380 nm UV LED chip, combining blue-emitting CDs and Zn-doped AgInS₂ nanocrystals emitting at 618 nm (Fig. 14(b)) [152]. An excellent CRI of 96 was achieved with a colour temperature of 4411 K for warm white light at 50 mA.

5. Conclusions

In conclusion, we provided an overview on synthetic strategies, optical properties and highlighted a few examples of applications of CDs. A focus on bottom up methods allowed us to track the reactions of CDs from the molecular precursors to the final luminescent nanoparticles, where subtle changes of the reaction conditions and in particular incorporation of doping atoms influence the structure of the resulting CDs and determine their optical properties. Organic fluorophores, polymerized nanostructures and finally carbonized, crystalline material combine to a CD consisting of a crystalline core and amorphous surface region with a large amount of functional surface groups. Especially the citric acid derived CDs can include significant amounts of highly luminescent organic fluorophores with similar spectral properties as the crystalline CDs. Heteroatoms incorporated into the conjugated carbon ring structure of the core or attached to the surface of these nanoparticles influence electronic configuration of CDs. Dopant atoms such as nitrogen and sulfur can be specifically used to enhance light emissive properties of CDs, while other heteroatoms such as chlorine can be used to enhance charge carrier separation. CDs can serve as both electron acceptors and donors, opening applications in photocatalysis or for the detection of ions via emission quenching. Furthermore, high PL QYs, good chemical stability, low cytotoxicity and availability of customized surface functional groups make CDs suitable for fluorescent labelling and targeted drug delivery and release.