Antistatic Structural Color and Photoluminescent Membranes from Co-assembling Cellulose Nanocrystals and Carbon Nanomaterials for Anti-counterfeiting

Static charges on optical anti-counterfeiting membranes may lead to materials structural changes, dust stain aggravation, and misreading of optical information. Incorporating conductive particles is a common way to transfer accumulative charges, but the key issue is how to achieve high dispersion and effective distribution of particles. According to the strategy of assembly-induced structural colors, cellulose nanocrystals (CNCs) were employed as a solid emulsifier to stabilize hydrophobic carbon nanoparticles (CNPs) in aqueous media; subsequently, by solvent-evaporation-modulated co-assembly under a condition of 30 °C and 20 RH%, the binary suspensions containing 2 wt% CNC and CNPs with the equivalent concentration relative to CNC ranged from 1:40 to 1:10 were used to prepare antistatic composite membranes. Surface chemistry regulation of CNCs was applied to optimize the dispersibility of CNPs and the orientation of assembled CNC arrays, and the hydrophilic CNCs were more favorable for dispersion and assembly of binary suspension systems. Meanwhile, one-dimension carbon nanotube (CNT) and zero-dimension carbon black (CB) were found to show better dispersibility than two-dimension graphene, which was verified by a semiquantitative theoretical study. Moreover, the stable binary systems of CNT/CNC and CB/CNC were chosen for co-assembly as membranes, and the uniaxial orientation could be optimized as the full-width of 9.8° at half-maximum deviation angle while the surface resistivity could also drop down to 3.42 × 102 Ω·cm·cm−1. The structural color character of such paper-homology and antistatic-integrated membranes contributes to optical information hiding-and-reading, and shows great potential as optical mark recognition materials for electrostatic discharge protective packaging and anti-counterfeiting applications.


INTRODUCTION
Materials with structural colors have been widely applied in the fields of optical anti-counterfeiting, [1,2] sensing, [3−5] and so on. Among them, the structural colors induced by assembly have a broad prospect in information encryption because they are free of photo-bleaching and have a long service life, which is different from the common photoluminescent materials based on fluorescence. [6,7] Besides, the monochromaticity of structural colors is usually an important factor for anti-counterfeiting materials. [6] Cellulose nanocrystals (CNCs), a one-dimensional nano-crystal extracted from natural bio-based resources, [8,9] were recently focused due to their self-assembling ability to obtain structural colors. [10,11] The structural colors could be further regulated by controlling self-assembly conditions and the properties of CNCs, such as length, [11] aspect ratio, [12] and surface chemistry structure. [13] Besides, uniaxial-assembled CNCs can even induce an emission based on virtual transitions of electrons, which makes CNCs an excellent candidate material for information security.
However, static charges on anti-counterfeiting CNC membranes with structural colors and uniaxial-assembly-induced emission may destroy the chemical structure of the optical materials, lead to a short service life, and adsorb dust, which affects reading optical information. Coating conductive polymers, like polyaniline, on CNCs is a feasible method for solving the static problem, whereas those polymers can also adsorb emission of CNCs and weaken the structural colors. [14] Another way to increase the conductivity is to add a small number of conductive fillers into the assembly membranes of CNCs. [15,16] Gold nanorods and silver nanowires were reported to co-assemble with CNC, [17] whereas the obtained assembly membranes owned a chiral structure, which was not suitable for assembly-induced emission of CNCs.
Carbon nanomaterials, such as carbon nanotube (CNT), carbon black (CB), and graphite, were also reported to disperse uniformly in CNC suspension. [18−21] Researchers have proved that CNCs could act as solid emulsifier to make hydrophobic carbon nanoparticles (CNPs) dispersed well in aqueous solutions. [22−24] However, we found that CNC might lose the assembling ability after CNPs were introduced in. The reason should be related to the complicated interactions between hydrophilic CNC and hydrophobic CNPs in the aqueous condition. Thus, we further modified CNCs to study the effect of chemical structure on its co-assembling ability with hydrophobic CNPs. The co-assembling mechanism was investigated by studying the co-assembling process of CNCs and CNPs with different dimensions. The mass ratio between CNC and CNPs was also controlled to optimize the photoluminescent and conductive properties of the co-assembling membranes. In general, the co-assembly membranes based on CNPs and CNCs combined both photoluminescent and antistatic properties, and it had potential to be used in electrostatic discharge (ESD) protective packaging materials with optical encryption.

Methods
CNC that was extracted by H 2 SO 4 hydrolysis with cotton was coded as SCNC and prepared according to our previous report. The obtained SCNCs were 156.2 ± 40.2 nm in length and 9.0 ± 1.2 nm in diameter. [25] Via replacing H 2 SO 4 solution with 4 mol·L −1 aqueous HCl, CNC powders were also obtained and coded as HCNC, according to a previous report, [26] and the length and diameter were 182.6 ± 59.9 nm and 14.2 ± 4.5 nm, respectively. SCNC was further acetylated according to our previous study, [27] and the modified CNC was coded as ACN with 142.3 ± 33.1 nm in length and 9.0 ± 0.9 nm in diameter. The surface-carboxylated CNC was coded as OCN and was prepared according to a previous report with some changes. [28] Briefly, 2.0 g of SCNC was dispersed in distilled water for 3 min with an ultrasonic cell breaker. TEMPO and NaBr were then added into the SCNC dispersion with magnetic stirring at 20 °C. NaClO solution (7 or 10 mL) was added dropwise into the dispersion at a pH of 10 with a 0.5 mol·L −1 aqueous NaOH solution, and these two kinds of OCNs were coded as OCN-7 and OCN-10, respectively. [29] When the pH was not changing, the reaction was finished. Methanol (6 mL) was then added to react with the excess of NaClO and the pH was adjusted to 7 with 0.5 mol·L −1 HCl. The oxidation degree was controlled by the volumes of the NaClO solution, and the length and diameter of OCN-7 were 143.0 ± 28.9 nm and 8.8 ± 0.9 nm, respectively, while those of OCN-10 were 140.0 ± 26.3 nm and 8.9 ± 1.1 nm, respectively. All the particle sizes were statistically obtained by collecting more than 200 particles to calculate their average values and the corresponding standard deviations.
To prepare the CNP/CNC uniaxial co-assembled membranes, firstly, 0.2 g of CNC (or OCN-10) powders and a certain amount of CNPs (CBs or CNTs) were added into 10 mL water and treated with ultrasound (300 W) for 10 min. Then, a hydrophilic glass plate was vertically inserted into the dispersion. Black membranes were obtained by solvent-evaporation-induced self-assembly at a condition of 30 °C and 20 RH%. The membranes were coded as CNT/SCNC(0.05), CNT/SCNC(0.1), and CNT/SCNC(0.2) when the CNT concentration in 2 wt% SCNC suspension was 0.05 wt%, 0.1 wt%, and 0.

Characterization
Fourier-transformed infrared (FTIR) spectra of CNC, OCNs, and ACN were recorded with a Nicolet 6700 FTIR spectrometer (Nicolet Instruments) in the range of 4000−400 cm −1 , using the method of KBr platelets. Ultraviolet and visual adsorption (UV-Vis) tests were carried out on a Cary Lambda 750 S spectrophotometer. Photoluminescence (PL) spectra of CNP/CNC membranes were tested on the 5JI-004 (Hitachi Hightechnologies Corporation). The conductivity of CNP dispersion was measured at the concentration of 0.01 wt%, 0.05 wt%, and 0.1 wt% for CNPs after the dispersion had been ultrasonic treated for 5 min. Zeta potential of CNCs and CNPs aqueous suspension (1 mg·mL −1 ) was measured by a NanoBrook Omni (Brookhaven). The atomic force microscopy (AFM) measurements were carried out on the Dimension icon from BRUKER, whose surface scans were performed in a Scanasyst mode. The full width at half-maximum (FWHM) of the included-angle of the CNC arrays was counted by collecting angle deviations of more than 200 CNCs in the AFM image. The surface resistivity of the co-assembled membranes was obtained by measuring the cube resistance, and all the data were repeatedly tested 5 times to calculate the average values.
The measurement of the oxidation degree of OCNs was carried out by a previous method, and the calculation of oxidation degree (DO) and carboxyl content was carried out by Eq. (1) and Eq. (2), respectively: [30] where V 1 and V 2 are the initial and terminal volumes of NaOH (in mL) during titration, c is the concentration of NaOH (in mol·L −1 ), ω is the mass of OCN samples (in mg), and m OCN is also the mass of OCN (in g). The contents (%) of carbon (C), hydrogen (H), and sulfur (S) in different CNCs were measured from elemental analysis (Elemental Vario EL Cube, Germany), and the degree of acetyl substitution (DS surface-acetyl ) for ACN was calculated according to our previous report with the Eq. (3): [27,31] DS surface-acetyl = n surface-acetyl n surface-OH = where DS surface-acetyl is the amount of surface acetyl groups on ACN, n surface-OH is the amount of surface hydroxyl groups on CNCs, and ΔC is the increment of carbon content after surface acetylation (ΔC = C ACN − C CNC ).
To make quantitative measurement of the stability of CNP dispersion, the mixed dispersion of CNPs and CNCs was prepared at a mass ratio of 1:2 with ultrasonical treatment of a cell crusher (300 W, 10 min). Centrifugal sedimentation was carried out at different rotating speeds with a Bench Centrifuge (3-30 KS, Sigma). Characterization of UV-Vis spectrophotometry was obtained after the centrifugation. Samples were loaded with a 1 cm quartz dish and under the conditions of scanning interval in 800−200 nm and the scanning rate at an intermediate speed. All the UV-Vis data were tested twice, and the error was less than 0.1%.
According to the previous report, the absorbance values of the CNP supernatants at 500 nm were collected as a function of relative centrifugal force (RCF). [32] The stability constant of CNPs could be obtained with Eq. (4): where y is the absorbance values at 500 nm, A 0 is the initial absorbance of CNP/CNC suspension without centrifugation, x is the RCF (kg), and k s is the rate constant of centrifugal settling (kg −1 ) with a large k s indicating poor dispersion stability. [32] Eq. (4) can be written in a linear form by taking the natural logarithm of both sides of the equation as Eq. (5): The absorbance values at 500 nm reflected the CNP con-centration in aqueous solution.

Semi-quantitative Theoretical Calculation of Phase Separation
The dispersion of CNPs in the assembly structure of cellulose nanocrystal (CNC) was studied thermally with the change of Gibbs free energy of phase separation. [33] In general, where ΔG a,b is the Gibbs free energy change from uniform to phase separation. G a-a , G b-b , and G a-b are the Gibbs free energy of two connected dispersing nanoparticles, two connected assembly nanoparticles, and connected dispersing nanoparticle and assembly nanoparticle, respectively. n is a constant.
Here, we labelled CB, CNT, graphene, and SCNC with subscript 0, 1, 2, and 1', respectively, to discuss ΔG of each system. Researches have proved that the interaction between graphene nanoparticles should be strong due to their π-π stacking, [34] leading to a low G 2-2 . The high lipophilicity of graphene and high hydrophilicity of SCNC should lead to a high G 2-1' . We thus believed that ΔG 2-1' should semi-quantitatively be a negative one and should induce spontaneous phase separation of CNCs and graphene. The AFM image of the co-assembly membrane of graphene and SCNC also confirmed such phase separation (Fig. 1c). Although CB and CNT had lower zeta potentials (ZPs) than that of graphene (seen in Table 1), the lower dimension of CB and CNT should result in a fact that the π-π stacking of CB and CNT should be much less than that of graphene. Thus, ΔG 0-1' and ΔG 1-1' should be larger than ΔG 2-1' . However, co-assembling with SCNC, CB also aggregated severely (Fig. 1b) while CNT did not (Fig. 1a). Thus, the compatibility between CNPs and CNC should be further modified. spectrum of OCN, the -C＝O stretching vibration band appeared at 1616 cm −1 , indicating that carboxyl groups were introduced on the OCN surfaces. [36] The band located at 1033 cm −1 was attributed to the stretching vibration (-SO 3 ) of SCNC. [31] Compared with SCNC, HCNC lacked -SO 3 group derived from the H 2 SO 4 hydrolysis, so the stretching vibrations of the -SO 3 group located at 1033 cm −1 was absent in the spectrum of HCNC. The results of elemental analysis in Table 2 also proved an increase in the content of carbon element from SCNC to ACN, which is consistent with the results of the infrared analysis above. The contents of elements in HCNC, SCNC, and ACN samples are displayed in Table 2. Oxidation degree (DO) of OCN-7 and OCN-10 was calculated to be 3.9% and 4.7%, respectively, with Eq. (1). Also, the carboxyl content was calculated to be 0.24 and 0.3 for OCN-7 and OCN-10, respectively, with Eq. (2). Besides, DS surface-acetyl of ACN was measured and calculated to be 37.1% with Eq. (3).

Quantitative analysis towards dispersion stability of CNP/CNC suspension
By controlling the surface chemistry structure of CNCs to adjust their hydrophilicity, we could obtain appropriate CNCs for coassembling with hydrophobic CNPs. As shown in Figs. 3(a) and 3(b), the dispersion of different CNPs could be stable in water when appropriate CNCs were introduced. However, some sediment could be observed in the systems of CNT/HCNC and CNT/ACN after 24 h. Moreover, CB aggregated completely after 24 h in water with SCNC, which could be seen in Fig. 3(d). By contrast, CNT/SCNC, CNT/OCN-7, CNT/OCN-10, and all CB/CNC (except CB/SCNC) were still stable after several weeks. The combination of CNCs and graphite showed little effect, which is in accordance with the result of semi-quantitative theoretical calculation. Thus, this work only discussed CNPs of CNTs and CBs, and the properties of graphite/CNC systems are displayed in the electronic supplementary information (ESI).
A dispersion constant, k s , was used to measure the stability of CNP and CNP/CNC dispersion quantitatively. Figs. 4(a)−4(c) show that the CNT dispersion with SCNC or OCNs adding owned a lower k s value than neat one at different concentrations, which means these kinds of CNCs could improve the CNT dispersion stability in water. However, introducing ACN and HCNC even increased the k s value of CNT dispersion. These results imply that CNCs with higher surface charge    could increase the dispersion stability of CNTs in water. In details, SCNC and OCNs could be negatively charged due to their abundant sulfonic groups and carboxyl groups, respectively, while HCNC and ACN could not. The surface charge of CNT and various CNCs could be compared through the ZP data displayed in Tables 1 and 3, and are further analyzed in the following discussion. By contrast, the k s values of CB/HCNC and CB/OCN systems were similar to that of neat CB, as shown in Figs. 4(d) and 4(e). Although the k s value of CB/ACN sample was about twice as that of neat CB at the CB concentration of 0.05 wt%, it was not high enough compared with CNT/SCNC, which means the CB/ACN system was also stable at that concentration. Besides, according to the results of element analysis, the sulfonic acid groups and carboxyl groups could improve the dispersion of CNT (1D carbon particle), and excessive sulfonic acid groups could even prevent the dispersion of CB (0D carbon particle). However, the k s value of CB/SCNC sample was much higher than that of neat CB, which means that SCNC limited rather than improved the dispersion stability of CB in water. Since the neat CB dispersion owned a much lower k s value than that of CNT dispersion, the higher ionic strength from the negative charge of SCNC might account for the main effect on CB dispersion. We thus used NaCl to control the ionic strength of all CB/CNC (except SCNC) system at the same level of the pure CB/SCNC sample (seen in ESI). Then the k s value of CB/CNC (except SCNC) systems increased to similar values of CB/SCNC system, as shown in Fig. 4(f), which proves the effect of ionic strength on CB dispersion.

The difference in the dispersion of CNP caused by CNC dispersity
Since CNCs could affect the stability of CNP dispersion, they might be able to prevent CNPs from aggregation during coassembling, and we thus used AFM to analyze CNP/CNC systems. In the drop-coating membrane of CNT/OCN dispersion, the OCNs were found to adhere on the CNTs (Fig. 5b), making CNTs separate from each other rather than entangling. By contrast, in the CNT/HCNC system, HCNCs tended to aggregate instead of adhering on the CNTs, so CNTs entangled and aggregated (Fig. 5d). Considering that the dispersion stability of HCNC itself in water was poor (the HCNC aggregated seriously, as shown in Fig. 5c), those results indicate that CNCs with a higher ZP value could well disperse in water, as seen in Table 3. [37,38] It is reported that the stability of the colloid system could be evaluated by the corresponding regions of the absolute ZP values, i.e. lower than 10 mV indicates highly unstable, 10-20 mV indicates relatively stable, 20-30 mV indicates moderately stable, and higher than 30 mV indicates as highly stable. [39] Thus, the aqueous suspension of OCNs and SCNCs with the absolute ZP value of higher than 30 mV  Table S1 in ESI) with 0.01 wt% CB. should be very stable, while the aqueous suspensions of HCNCs and ACNs were relatively stable due to their ZP values around -20 mV, and tended to aggregation. Those well dispersed CNCs could adsorb on positively charged CNTs and separate them from aggregation, resulting in the potential of CNTs to be well distributed in the co-assembled membranes, as shown in Scheme 1(a). No apparent adhesion between CB and HCNC was observed, as shown in Fig. 5(e) and schematically displayed in Scheme 1(b), which might be attributed to a much lower specific surface of CB than CNT. [40,41] However, when SCNCs were introduced into the CB dispersion, CBs aggregated seriously (Fig. 1b), which is consistent with the result that the CB/SCNC system owned the highest k s value among all CB/CNC systems.

The orientation of the CNP/ CNC co-assembled membranes
According to the quantitative analysis and AFM tests, we chose CNT/SCNC, CNT/OCN, and CB/OCN systems to co-assemble and fabricate anti-static photoluminescent membranes. The obtained membranes were characterized by the distribution of the included angle between CNC long-axis and vertical direction in the CNP/CNC co-assembly by AFM tests as shown in Fig. 6. FWHM of the included angle was used to evaluate the orientation of CNC assembly. From Fig. 7 and Table 4, we found that CNT/OCN and CB/OCN membranes had a large FWHM, while the FWHM of CNT/SCNC membranes was much smaller. OCNs had a high ZP and they could easily adhere on the surface of CNTs, which caused a random assembly of OCNs. By contrast, CNTs distributed very well in CNT/OCN membranes, so CNT/OCN membranes displayed much higher conductivity than the others but lower photoluminescence. Meanwhile, there were weak interactions between CBs and OCNs, so that CBs were separated by OCNs and finally aggregated in the CB/OCN membranes, which led to a high resistivity and bad solid-state luminescence. CNT/SCNC membranes had lower FWHM and could luminesce. Besides, a high concentration of CNTs could bring out good conductivity. These results would be further discussed in the following sections.

Anti-static Properties of CNP/CNC Co-assembled Membranes
The surface resistivity of CNP/CNC co-assembled membranes with different concentrations is collected in Table 5. CNT/ SCNC membranes had a high resistivity at the initial concentration of 0.05 wt% and 0.1 wt%, which was 3.22 × 10 6 and 1.15 × 10 6 Ω·cm·cm −1 , respectively. In contrast, CNT/SCNC(0.2) displayed a lower surface resistivity as 5.62 × 10 4 Ω·cm·cm −1 . According to "Handbook for Electrostatic Discharge Protective Packaging (GJB/Z86-97)", the membranes with a surface resistivity between 10 5 and 10 12 Ω·cm·cm −1 are the static dissipative materials, while the static conductive materials have the surface resistivity of less than 10 5 Ω·cm·cm −1 , which are the best for electrostatic discharge (ESD) protective packaging. However, when the resistivity of the membrane is too low as less than 10 4 Ω·cm·cm −1 , it cannot be dir-ectly used as ESD protective packaging material because it may produce electrostatic induction and electrostatic discharge in the electrostatic field and damage the product.
The surface resistivity of CB/OCN was very high. This is attributed to the spherical structure of CBs. CBs could be well dispersed in aqueous solution, but the introduction of OCNs separated them in the co-assembled membranes. Thus, CBs could not contact each other, causing a low surface resistivity of the membrane.

Photoluminescent Properties of CNP/CNC Co-assemble Membranes
The results of the photoluminescent properties are shown in

CONCLUSIONS
To enhance the potential of optical anti-counterfeiting application, CNPs were introduced to impart antistatic function to the uniaxial arrays of CNCs with structural color. In this case, it is crucial that hydrophilic CNCs served as a solid emulsifier to promote the dispersion stability of CNPs in blending media and hence achieved uniform distribution of CNPs with uniaxial CNC arrays in the binary-component membranes. The surface chemistry regulation of CNCs determined the hydrophilic/ hydrophobic properties and thus affected the colloidal stability of binary CNP/CNC suspensions. A semi-quantitative theoretical calculation predicted the requirement in Gibbs free energy for two-phase separation of these binary colloid systems containing CNCs with various surfaces structures and CNPs with different dimensions. As expected, 1D CNTs and 0D CBs could co-exist with hydrophilic CNCs in water and showed enough dispersion stability, but 2D graphene consistently tended to aggregation due to a very low requirement of Gibbs free energy from uniform dispersion to individual component aggregation. As a result, CNT and CB together with hydrophilic SCNC and OCN were selected to combine as the CNP/CNC pairs for co-assembly as the binary-component membranes, but an unexpected factor, i.e. high ionic strength of SCNC, aggravated the aggregation of CBs and hence led to the failure in the coassembly of the CB/SCNC pair. The solvent-evaporationmodulated co-assembly process still produced uniaxial CNC arrays with structural color in the binary CNP/CNC membranes integrated with antistatic function, which could emit blue light under ultraviolet irradiation. As usual, the surface resistivity dramatically dropped down with an increase of CNP, and 1D CNT showed a more predominant contribution to conductivity than 0D CB. Moreover, by selecting the CNO/CNC pair and controlling the CNPs loading-level, the co-assembled membranes could be suitable as static dissipative material with the surface resistivity of 1.82 × 10 7 −1.15 × 10 6 Ω·cm·cm −1 and static productive materials with the surface resistivity of 5.62 × 10 4 and 3.42 × 10 2 Ω·cm·cm −1 , respectively. As for luminescence properties, highly uniaxial orientation favored the enhancement of luminous intensity. The CNT/SCNC coassembled systems with a weight equivalent of CNT versus SCNC as 1:10 showed the narrowest full-width as 9.8° at halfmaximum deviation angle from assembly orientation, and showed the highest luminous intensity among all the coassembled specimens; combined with the surface resistivity of 5.62 × 10 4 Ω·cm·cm −1 , it showed a great potential as optical mark recognition materials with ESD protective function. Overall, this kind of composite membranes, which integrate conductive carbon nanoparticles into bio-based structural color arrays, take the advantages of paper homology and are free from photobleaching, and could be considered as a good antistatic anti-counterfeiting candidate by virtue of optical information hiding-and-reading function.

Electronic Supplementary Information
Electronic supplementary information (ESI) is available free of charge in the website at https://doi.org/10.1007/s10118-020-2414-x.