From Cellulose Micro- to Nanocrystals

The O-H bending peak around 1630 cm^-1 of adsorbed water is particularly interesting to study the amount of adsorbed water inside the cellulosic structure.

This peak will receive special attention in Chapter 5 Section 5.2.2.2, where the amount of water can be traced after the infiltration of specific ions into HM-CNC films.

Figure 2.10c shows the Thermogravimetric analysis (TGA) (Simultaneous Ther-mal Analyser STA 449 F3 Jupiter) of Avicel powder between RT and 500 °C (step of 0.05 ºC). The assessment of the thermal stability of the used materials is impor-tant for final device production, where sometimes processing temperatures above 150 °C are needed for thin-film deposition or annealing purposes. As Avicel will merely serve as the starting material for HM-CNC production TGA serves solely for comparison reasons with the obtained HM-CNCs and the as-received C-CNCs (see Section 2.4.3.2). Due to strong inter- and intramolecular hydrogen bond-ing cellulose does not melt but rather undergoes thermal degradation. Thermal degradation of cellulose proceeds in three main steps:[25]

1. Dehydration of the cellulose to produce dehydrocellulose.[26] Characterized through an initial decrease of mass associated to the loss of absorbed and adsorbed water from within the cellulosic structure (until around 270 °C) 2. Depolymerization by an endothermic process into tar (from 270 °C

on-wards).

3. Decomposition via exothermic reactions into gaseous products and char residue.

The main mass loss is associated to the last two steps and can be verified in Figure 2.10c with a decrease of mass from around 95% to 12%. The temperature range for the onset of degradation of Avicel (or native cellulose for that matter) is high enough for usual post-deposition annealing treatments of amorphous semi-conducting devices produced on paper substrates.[27] However, the transition from microcrystalline cellulose to cellulose nanocrystals induces some significant thermal property changes that will impair usual post-deposition annealing, as will be explored more in detail in Chapter 5.

most prominently during mechanical, chemical or enzymatic treatments. CNMs present unique characteristics connected to their nanoscale size, resulting in large surface areas and also their intrinsic fibril morphology. Consequently, CNMs have huge potential applications, including adhesives, cosmetics, barrier mem-branes, nanocomposites, batteries and supercapacitors, catalysis, textiles, health care, biomedical, cement, in drilling fluids, drug delivery and many more.[28]

CNMs can be further divided into CNCs, Cellulose Nanofibrils (CNFs), tunicate CNCs, algal cellulose and BC. Each subgroup naturally presents very distinct degrees of crystallinity, allomorph ratios and particle morphologies. Figure 2.11 depicts, apart from the appearance of native lignocellulosic fibers (a), three dis-tinct types of CNMs, namely CNFs (b), CNCs (c) and bacterial cellulose (d). Be-low the macro photographs (membranes, dish-cast from the starting materials in the petri-dishes), SEM images show the appearance of the membranes on the nanoscale.

Figure 2.11 – Cellulosic Materials on various scales with SEM images. a) Native lignocellulosic fibres, b) CNFs, c) CNCs, d) BCs. The membranes in c) show irides-cence owing to the chiral nematic arrangement of the CNCs inside the membranes (more on this topic in Chapter 3); this arrangement does however not show in the corresponding SEM image, as it is a surface image. In the SEM image in d) dead bacteria are indicated by red circles.

From the SEM images and the corresponding length scales it becomes evident that when moving from micro- to nanocellulose there is a considerable reduction in fibre diameter for CNFs, CNCs and bacterial cellulose. This provides a drastic increase in the volume-to-surface area for CNMs. Moreover, regarding CNCs, no individual fibres are visible anymore. This hints on a reduction not only in fibre diameter but also in length, as the individual CNCs form an apparently uniform surface. On closer inspection however, individual CNCs can be observed.

Optical properties are also very distinct between the native cellulose mem-branes and the memmem-branes obtained from CNMs. Whereas native cellulose yields an opaque, translucent appearance, membranes derived from CNMs are usually

transparent (membranes from CNCs may show iridescence – see Chapter 3). This optical behaviour is connected to the scattering of the incoming light by fibres and voids in between them. The native cellulosic membrane of Figure 2.11a presents strong surface light scattering and surface Fresnel reflections, effectively hinder-ing the transmission of light in the visible range.[29] Transmittance spectra of the four membranes from Figure 2.11 are shown in Figure 2.12a.

Figure 2.12 – a) Transmittance spectra of the membranes in Figure 2.11. Light scattering mechanisms in b) native cellulosic paper and in c) paper made from CNMs. ©IOP Publishing. Reproduced with permission [30]. All rights reserved.

CNMs, on the other hand, pack more tightly and leave less voids in between the fibres, where light can be scattered. This concept is schematized in Figure 2.12b and c, where the incident light is either scattered (b) or transmitted (c) through the membrane. In the group of CNMs, when moving from NFCs over Bacterial Cellulose to CNCs the transparency increases steadily, which can be connected to increasingly tighter packing. In this regard nanocrystals pack more tightly than nanofibers, leaving less voids for possible scattering. The superior transparency of CNMs make them particularly interesting for optoelectronic or coating and packaging applications.

2.4.2.1 CNCs

Obtaining CNCs through acid hydrolysis by strong mineral acids is nowadays the main route for the production of CNCs not only in laboratories but also on larger scales, as successfully demonstrated by, for instance, CelluForce©.[31] Since the discovery of CNC liberation by sulfuric acid, other acids, including hydrochloric or phosphoric, have been shown.[32–35] Depending on the used hydrolysis acid, the functionalization of the most exposed C6 carbon can either present hydroxyl, sulfate or phosphate groups (see Figure 2.13).

Figure 2.13 – Three examples of oxygen functionalization on the C6 carbon atom exemplified on a cellobiose unit at the surface of the CNC obtained through a) hydrochloric, b) sulfuric and c) phosphoric acid hydrolysis.

This in turn has a tremendous impact on colloidal stability, ionic strength and most importantly the liquid crystalline behaviour of the CNCs in aqueous sus-pension. It was shown that colloidal stability suffers considerably when CNCs are obtained through hydrochloric hydrolysis.[36] This can be connected to the low surface charge of the hydroxyl groups and consequent inter-particle aggregation.

To improve colloidal stability the surface charge of the HCl hydrolysed CNCs can be increased by including a post-synthesis treatment with TEMPO (2,2,6,6,-tetramethylpiperidine-1-oxyl), resulting in carboxyl groups on the surface or through polymer grafting.[37, 38] On the other hand, sulfuric and phosphoric acid treatments have shown excellent colloidal stability due to the high surface charge and size of the bulky sulfate and phosphate groups.[1] In fact, the ini-tial approach followed by Rånby turned out to be the most promising one for the production of CNCs with good colloidal stability and most favourable liquid crystalline properties.

2.4.2.2 CNCs from Sulfuric Acid Hydrolysis

Nowadays the sulfuric acid hydrolysis is the most commonly used approach to obtain CNCs. Especially when aiming for liquid crystalline properties to produce

chiral nematic films or membranes. Colloidally stable CNCs have been success-fully prepared by this method with a great variety of sources, including BC, MCCs (Avicel), cotton, soft or hardwood pulps.[36, 39–42]

During hydrolysis, the acid diffuses into the pulp fibres and removes polysac-charide material that is closely bonded to the microfibril surface. This initial reaction removes most of the amorphous regions throughout the cellulose start-ing material. This is followed by a breakdown of the cellulose polymer chains (transverse glycosidic bond cleavage) into slender individual crystallites.[see 10, p. 36] This remaining part corresponds to the highly crystalline regions of the cellulose, as already depicted in Figure 2.4 in Section 2.3. After reaction quench-ing, several centrifugation steps and dialysis to increase the pH, a cloudy aqueous CNC suspension is obtained. In this suspension the CNCs are naturally connected in a tight network through Van-der-Waals interactions. To break this network, in-crease colloidal stability and promote liquid crystalline properties, mechanical energy, usually in the form of tip sonication is applied. This procedure is very rele-vant for the CelluForce© CNCs in Chapter 3 Section 3.7.4, where certain amounts of tip sonication energies will be investigated regarding colloidal stability, phase separation, liquid crystalline and photonic properties of dried films. Tip sonica-tion consequently breaks up agglomerated CNCs and provides a clear suspension with a slight Tyndall effect (for isotropic suspensions only), where blue light is scattered by the suspended particles giving the suspension a blue appearance.[43]

Figure 2.14 compares HM-CNC suspensions before and after tip sonication.

Figure 2.14 – Macroscopic images taken of HM-CNC isotropic suspensions after (left) and before (right) tip sonication broke up agglomerated CNC particles. The slight Tyndall effect becomes evident in the vial on the left. Scale bar indicates 0.5 cm.

Depending on the used source, the cellulose content can be quite different. For example, whereas cotton-seed fluffcontains ≤94%, flax or hemp contain ≤80%

wood contains only≤55%.[30, 44, 45] Consequently, the chosen starting material will in part also determine the obtainable synthesis yield. Yield and reproducibil-ity optimization regarding hydrolysis conditions is a complex topic and not sub-ject to this work. For further insight, the reader is referred to “Cellulose Nanocrys-tals: Properties, Production and Applications” by Wadood Y. Hamad.[10]

The synthesis conditions for the home-made CNCs (HM-CNCs) followed in this work is derived from the works of Revol and Gray with minor adaptions.[41, 46] In short, the synthesis proceeds as follows:[30, 47, 48]

“Avicel from Sigma-Aldrich (Avicel PH-101, derived from cotton as indicated by the supplier, Sigma-Aldrich, particle size≈50µm) was used without any further pu-rification. A total mass of 10 g is hydrolyzed by sulfuric acid (Sigma-Aldrich, 95 – 97%) diluted to 64 wt%, with an acid/solid ratio of 8.5:1, during 130 min at 45 °C, under vigorous stirring, and quenched with ultrapure water (Millipore Elix Advantage 3 system). The resultant material was centrifuged with ultrapure water with consecu-tive cycles and the CNCs suspension collected for pH values between 1.9 and 3.9. The product was dialyzed (Spectra/Por 4 membrane, molecular cut-off12–14 kDa) against ultrapure water, for at least a month until a constant pH was reached. The obtained suspension was treated by tip sonication in a 500 W ultrasonic processor (Vibra-Cell from Sonics) at 60% power with cycles of 1 s on 1 s offfor 45 min.”

CNCs provided by CelluForce© (C-CNCs) were obtained through the same sul-furic acid hydrolysis from bleached Kraft pulp (as pioneered by Gray’s group and scaled up at FPInnovations Pointe Claire, QC, Canada).[41, 49–52] The hydrolysis is followed by dilution, separation from residual acid and neutralization through sodium hydroxide solution and finally spray-dried into powder form.[53] In this regard, spray-drying refers to a process where the CNC suspension is sprayed in a high temperature environment, which results in a complete removal of the solvent and agglomeration of the CNCs into a dry powder.

The final CNC dimensions are influenced not only by the source material, but also by the overall acid hydrolysis conditions and ionic strength.[54] As the acid hydrolysis is a diffusion-controlled process the final CNC dimensions will inevitably present a certain heterogeneity, where the average dimensions also de-pend on the chosen starting material. In general, regardless of the preparation method, CNCs are disperse in width and, in particular, in length (where typical values range from 30% to 50% dispersity).[14] Table 2.3 shows a comparison of

three distinct sources with their respective CNC dimensions for the same hydrol-ysis conditions.

Table 2.3 – Comparison of three different cellulose source materials and the re-sulting CNC dimensions for similar acid hydrolysis conditions.

Source Material Lengths (nm) Widths (nm) Aspect ratio Ref

Cotton 100 - 300 7 14 - 43 [40]

Softwood kraft pulp 105 - 255 3-5 21 - 85 [36]

Valonia Algae 100 - 2000 20 5 - 100 [55]

As will be seen later (Section 3.6 Table 3.2) the CNC’s dimensions and cor-responding aspect ratios play an important role during slef-assembly and were shown to have an impact on the resulting pitch and consequently photonic prop-erties of dry CNC films.[56] Additionally, phase separation behavior following Onsager’s hard-rod model is affected by the CNCs’ aspect ratios, as demonstrated before in Chapter 1 Section 1.3.2.6.

With the described sulfuric acid hydrolysis conditions, a sulfation (derivatiza-tion) of the cellulosic hydroxyl groups (predominantly on the C6 carbon) occurs.

This promotes an addition of surface sulfate ester groups and transforms cellulose-OH to cellulose-OSO₃H (as shown in Figure 2.13b), where the hydrogen counter ion may later be exchanged with other alkali ions, such as Na, as is the case for the C-CNCs. The pH of aqueous CNC colloids depends on the type of this counter ion, where the sodium salt form has been shown to be chemically and thermally more stable, especially as a powder.[39, 57] This is why commercially available CNCs are often sold as spray or freeze-dried with Na as the counter-ion.

The obtained HM-CNCs from the above-described reaction are in colloidal suspension and called never-dried CNCs. Naturally when dealing with colloidal systems, where high colloidal stability is wanted, it is much desired to tune the interaction potential between individual particles to the maximum of repulsive.

Due to the surface sulfate groups the CNCs are equally electrostatically charged and repulsion occurs naturally. Thus, after sufficient sonication treatment the electrostatic interaction between individual particles is great enough to keep the particles farther apart than the action of Van-der-Waals forces, stabilizing them.

At this point the suspension is kinetically stable, where temporarily even minor thermal fluctuations (on the order of k_BT) are not enough to induce aggrega-tion.[14]

Nevertheless, over long periods of time desulfation occurs (the main cause of chemical CNC degradation in stable suspension), promoting changes in colloidal

stability, dispersion uniformity, sulfur content, pH and conductivity. More im-portantly however, even with very high zeta potential (ζ), prolonged storage at room temperature leads to particle aggregation, which has shown to have con-siderable impact on the self-assembly properties and consequently the photonic band-gap.[58, 59] It is thus advised to always store CNC suspensions in the fridge at low temperatures (4 °C). Even when stored in the fridge some tip-sonication is usually employed before drop- or dish casting the CNC suspension to break up potential agglomerates.

In the case of the used spray-dried C-CNCs the interparticle separation dis-tance was reduced to less than the Debye screening length. In this case interparti-cle separation reaches distances, where Van-der-Waals forces start to take effect.

From this point on it is in fact possible to separate the particles again, but only through the application of considerable amounts of energy through, for instance, sonication. Therefore, emphasis is given on this specific topic of tip sonication, playing a central role in redispersing sodium neutralized spray-dried C-CNCs in Chapter 3 Section 3.7.4.

Counter-ion exchange either through membranes or direct electrolyte addition has a considerable impact on self-assembly behaviour and may alter the pitch (see Chapter 6). Furthermore, increasing the pH through the addition of, for instance NaOH into HM-CNC suspensions (as will be explored in Chapter 5 Section 5.2.2.1) also strongly impacts the liquid crystalline and self-assembly behaviour. However, the addition of ions into CNC films is important for this work as they improve electrochemical properties of dried HM-CNC films, which will alter their behaviour as solid-state electrolytes in field-effect transistors in Chapter 5 Section 5.2.2.1.

It is important to note that after obtaining CNCs in a stable suspension they can be deposited through various techniques to obtain solid films or membranes (as explained more in detail in Chapter 3). The following section on the character-ization of the CNCs, used in this work, studies CNCs in both forms: suspensions and films.