Thermoplastic Starch: Current Development and Future Trends

Starch is a biopolymer that is widely available from agriculture/nature and thus is renewable and cheap. In addition, starch-based materials are biodegradable, offering a great advantage over traditional nonbiodegradable synthetic polymers. However, the poor processibility and product performance of starch have greatly impeded the wide application of starch in real applications. This paper reviews the current developments of the production of thermoplastic starch as a commodity renewable material. The biological origins of the feedstock, formulation development, processing requirements, as well as the important aspects that need to be addressed when designing a product from starch are discussed. It is hoped that this paper will provide insights into thermoplastic starch-based materials trends and and inspiration for future research.


INTRODUCTION
Starch is a globally available, low-cost renewable commodity that when processed into a thermoplastic has the potential to replace conventional petroleum plastics. This interest has led to decades of investigation into the large-scale transformation of starch into a thermoplastic material [1,2].
The most common sources of starches produced globally include maize (corn), wheat, rice, potato and tapioca. Starch is produced within plant cells as an energy storage mechanism in the form of spherical granules of size from <1 μm up to 100 μm [3]. The size, amount, and composition of the granules vary between plant species [4].
Starch can be processed by traditional polymer processing techniques, such as extrusion, without signifi cant modifi cation [1]. However, since native starch exists in granules (as a form of energy storage in nature), melt processing of native starch requires thermomechanical treatment, accompanied by a plasticiser (most commonly water-even though some researchers do not consider water as a plasticiser for starch), to produce starch-based materials. During this process, the phase transition ("gelatinisation" or "melting") of starch occurs, which plays a critical role for the plasticisation of starch-indeed this complexity makes the concept of plasticisation of starch different from that of standard synthetic polymers. By plasticisation, raw starch, with structural degradation at multiple levels (molecular, crystalline, and granular) [5], can be transformed into a homogeneous polymeric state, which is known as "thermoplastic starch (TPS)" (in spite of the term, it is noteworthy that plasticised starch is not truly a thermoplastic but more accurately described as a thermomechanoplastic [6]). Figure 1 shows a typical TPS material produced by extrusion. The degree of phase transition can largely determine the processibility and fi nal product properties [2].
Different conditions may also be used for different subsequent processing techniques. For example, starch-polyester blends may be processed in melt state under high shear, 10 2 -10 3 s -1 in extrusion, and 10 3 -10 4 s -1 in injection moulding [8].
To understand starch processing in depth, it is necessary to discuss the biological origins of the feedstock,

Structural Features of Starch
The extensive varieties of starch sources have significant differences in composition and structure. Starch macromolecules can be classifi ed into two main groups, i.e., amylose and amylopectin (even though there are also some intermediate molecules in between the two main types). The amylose is a linear structure of α-(1→4) linkage of glucose units while amylopectin is a highly branched structure of short α-(1→4) chains linked by α-(1→6) bonds [3]. The ratio of the amylose and amylopectin depends on the source and age of the starch as well as the extraction method from the starch granules [3]. Regular starch cultivars (wheat, maize, potato, etc.) have amylose contents ranging between 20-30% [9], while high-amylose starches (with the amylose content being up to 83% [10]) have long been cultivated and used. Besides this, very small amounts of proteins, lipids and phosphorus are also found in starch granules depending on the botanical resource [11,12]. These components can interact with the carbohydrate chains during processing (e.g., Maillard reaction) and then modify the behaviour of the starchbased materials. Moreover, starch contains about 10% bound water [13].
Starch occurs in four different polymorphs, the A, B, C and V crystalline forms. Both A and B forms are left-handed double-helices with six glucose units per turn [3]. Their only difference is the packing of the helices, where A-form has a monoclinic unit cell while B has a hexagonal unit cell [3]. The C-type is simply a mixture of A and B. V-type starches consist of a single left-handed helix with a complexing agent (typically lipids) found inside the helix channel [3]. The V-type polymorph exists more in high-amylose starches [10] and is also prone to be formed by thermal processing like extrusion [14,15].

Impact of Amylose/Amylopectin Ratios
In spite of the fact that starches with a higher amylopectin content (lower amylose content) have a more ordered supramolecular structure and thus display a higher enthalpy of gelatinisation, as shown from differential scanning calorimetry (DSC) studies [16], the granules of high-amylose content starches are more compactly packed [17,18], and usually require more energy to be destroyed for the production of TPS [5,19,20]. Therefore, poor processibility of high-amylose content starches is encountered. However, compared to those from low-amylose content starches, TPS materials made from high-amylose starches often show more fl exibility and superior strength and stiffness, and display less deterioration of the mechanical properties due to ageing [9,21,22] (even though their postprocessing shrinkage has been reported [9]). On the other hand, the use of starch with a lower amylose content (higher amylopectin content) could reduce the melt viscosity (improve the processibility); but it should be noted that in this case the feed rate into the extruder needs to be increased to keep the torque rate constant and to avoid unstable processing [9]. It has also been reported that a higher amylopectin content could lead to fi lms with a greater elongationat-break value due to the higher molecular weight of amylopectin [9]. The amylose/amylopectin ratio of starch could largely infl uence the produced materials. Generally, starch with a higher amylose content has better mechanical properties [7]. Nevertheless, an increase in the amylose content can reduce the transparency of fi lms, due to the materials containing more granules that are not destroyed, suggesting less homogeneity [9]. In addition, examination of 100% amylose or amylopectin TPS substructures has also been carried out. Rindlav-Westling et al. [21] reported that pure amylose fi lms had a relative crystallinity of around 30%, while pure amylopectin fi lms were entirely amorphous. Blending of the two polysaccharides would result in variation in crystallinity due to amylose inducing crystallisation of amylopectin [21]. Purifi ed amylose has a higher crystallisation tendency than natural amylose due to the change in molecular weights-a decrease in the molecular weight leads to a decrease in the chain size, allowing for a greater tendency to form continuous networks [21].

Impact of Chemical Modifi cation
Chemical modifi cation (e.g., hydroxylation [23][24][25] and acetylation [26][27][28][29][30]) of starch by substituting ester or ether groups for the hydroxyls is an effective way to improve the processing and product properties (mechanical properties, water resistance, etc.). As a typical example, hydroxypropylated starch shows much improved processibility (lower viscosity, specifi c mechanical energy input (SME), and die pressure) and more fl exible mechanical nature of fi nal fi lms [23,31]. Besides, Thunwall et al. [32] observed a more Newtonian behaviour of hydroxypropylated and oxidised potato starch compared to normal potato starch. It is proposed that the modifi cation of groups may weaken the starch intra-and inter-molecular interactions and facilitate the formation of "ellipsoids". Meanwhile, it should be noted that chemical modifi cation often decreases the polysaccharide molecular weight, potentially leading to materials with less mechanical strength, and also that the toxic chemical residues may modify the biodegradability and negatively impact the life-cycle assessment (LCA) of the fi nal products.

Plasticisers
The addition of a plasticiser can be used to enable and regulate the degree of melting and disruption of starch granules and change a number of fi lm properties to be more desirable for end use [9]. For a plasticiser to be effective for the production of TPS it needs to be compatible with starch, which means it needs to be hydrophilic [33]. Having a boiling point higher than the processing and drying conditions is also desirable to prevent the evaporation of the plasticiser out of the material both during processing and post processing. A plasticiser can reduce the internal hydrogen bonding while increasing intermolecular spacing, and usually decreases the crystalline-to-amorphous ratio [34]. As a result of the increased mobility of polymer chains, it is common to see a reduction in the glass transition temperature by the addition of plasticiser [9].
The most common plasticisers used in TPS fi lm production include water. However, when water is used as the sole plasticiser, unstable processing could be observed due to the water evaporation. Further, the fi nal products based on starch containing only water usually have poor mechanical properties, especially due to the brittleness, since its fi nal temperature is lower than its glass transition temperature (T g ). Regarding this, polyols such as glycerol, sorbitol and glycol have also been used for the production of TPS [35][36][37][38][39][40][41][42][43][44][45]. It has been found, however, that polyols are less effective plasticisers than water [9], as the gelatinisation of starch becomes less easy with larger molecules of polyols [46]. Nevertheless, polyols are non-volatile plasticisers, overcoming the corresponding drawback of water. More non-volatile plasticisers used for TPS include nitrogen-based compounds (urea, ammonium-derived, amines, etc.) [47][48][49][50][51][52][53][54][55][56][57][58][59][60][61] and citric acid [62][63][64]. In particular, citric acid has been regarded as an effective plasticiser for starch as the acidity of citric acid can promote the fragmentation and dissolution of starch granules, as well as the effective interactions with the C-O groups of starch (more than glycerol) and thus the restriction of hydrogen bonds between the hydroxyl groups of TPS. Therefore, citric acid can result in a greater degree of the amorphous state and a low degree of recrystallisation (or "retrogradation", a term specifi cally for starch) [65].
Along with promoting plasticisation, the addition of a plasticiser like glycerol improves the fi lm's extensibility and reduces elasticity but greatly reduces the tensile strength [34]. However, it is noteworthy that at low concentrations (below around 10-15%) glycerol may exhibit an anti-plasticisation effect due to its higher affi nity with water than with starch [13]. This can increase the gelatinisation temperature of native starch as well as increase the glass transition temperature and brittleness of the corresponding TPS materials [13]. A further consequence of the addition of a plasticiser is the signifi cant increase in water or oxygen permeability of the material [34,66] (and the permeability depends on water more than on polyols [67]). Also, plasticisers like glycerol and sorbitol were found to allow fi lms to have higher water solubility than other less hygroscopic plasticisers due to their strong affi nity with water [34].

Additives (Including Nanofi llers)
In addition to plasticisers, other additives can be introduced to the extrusion feed for various purposes. For example, to reduce the material's tendency to stick to the die and clogging it, lubricants can be used such as magnesium stearate [68], calcium stearate [69] and fl uoro-elastomers [70].
Various nanofi llers have been used to improve the properties (typically as mechanical properties) of TPS and/or to add functional properties. These include phyllosilicates (e.g., montmorillonite (MMT)), polysaccharide nanofi llers (e.g., cellulose nanowhiskers, starch nanocrystals); carbonaceous nanofi llers (e.g., carbon nanotubes), and various metalloid oxides, metal oxides, and metal chalcogenides, as having been comprehensively reviewed elsewhere [71][72][73]. As the most frequently used nanofi ller, MMT, as a layered nanofi ller, when dispersed homogeneous in TPS, can increase the barrier properties as well as reinforce the material [74]. It is worth noting that unmodifi ed MMT can be dispersed in TPS more evenly than any other modifi ed MMT counterparts, due to the former's stronger hydrophilicity which results in stronger polar interactions, especially hydrogen bonds formed between the -OH groups of the MMT and of the starch molecules [75][76][77][78]. This example hints at the importance of considering the nanofi ller's hydrophilicity for choosing effective nanofi llers for TPS materials.

Characterisation of Native Starch for Design of Starch Processing
As starch undergoes phase transition during thermal processing, it is important to understand the thermal phase transition of starch before practical processing work. This can be done by various techniques such as microscopy with hot-stage, DSC, X-ray diffraction (XRD), and nuclear magnetic resonance (NMR) [1,104]. Among these, DSC has gained the most popularity due to its convenience of use and accurate results. However, as the thermal phase transition (gelatinisation) of starch occurs only with water (and/or other plasticisers), it is signifi cant to avoid the loss of water during measurement; otherwise the endotherm of water evaporation would largely interfere with the signal of starch phase transition (which is much weaker than that of the former). Regarding this, a stainless steel high-pressure pan was used in DSC for understanding the phase transitions of different starches in pure water [16] or in water-glycerol mixtures [46]. Processing of starch usually involves shear treatment, which is expected to greatly facilitate the phase transition of starch as it can disrupt the multilevel structures of starch granules. This cannot be understood solely by DSC. Regarding this, an attempt was made by Xie et al. [105] to study the phase transition of starch with a low moisture content (45%) and under shear stress by using dynamic mechanical analysis (DMA). The results showed that the variations in the physical parameter, damp (tanδ) in DMA, occurred both before thermal transition (as indicated by the heat fl ow in DSC) started and after thermal transition completed. However, the shear treatment in DMA was quite small. By using a twin-rotor mixer (Haake Rheomix ® ), Xue et al. [106] and Wang et al. [20] investigated the phase transition of starch under a strong shear treatment, just like in an extruder. The torque curve could be used to refl ect the viscosity change and to estimate the time needed to achieve the steady molten starch phase, which could provide useful information for designing practical extrusion processing. Nevertheless, more studies are urgently needed to understand the changes of starch under a combined thermomechanical treatment.

Processing Techniques for Starch
There are a number of techniques used to produce TPS materials, including the most frequently used extrusion, as well as injection moulding, compression moulding, internal mixing, etc. Extrusion can be considered as a core processing technique and is usually coupled with, or prior to, other processes such as injection moulding, compression moulding, and fi lm casting. While solution fi lm casting could be the simplest method for the preparation of starch-based fi lms, it is considered to be much less effi cient for industrialscale production and thus is not discussed in the current paper. More details of thermal processing techniques for the production of TPS can be found in a review paper [1].

General Concept of Starch Extrusion
The main extrusion facility for starch extrusion processing can be a single-screw extruder (SSE) or a twin-screw extruder (TSE). The differences of these two types of extruders have been discussed in detail elsewhere [107]. It is noteworthy that while a SSE can handle the high viscosity of starch and provide a high processing pressure for continuous metering of starch through die shapes, a TSE is much better for the fi rst stage of TPS production from raw starch powder, as its twin screws have a self-wiping ability. Moreover, a TSE has a large operational fl exibility (the control of individual barrel zone temperatures, multiple locations for feeding solids and liquids, and versatile screw confi gurations for different degrees of mixing/ kneading) and is useful for intensive mixing and compounding of components into TPS.
A schematic of the twin-screw extrusion process for the production of TPS is shown in Figure 2 [2]. Here, the native starch granules are introduced, preferably by a powder gravimetric feeder, into the extruder in a hungry way (i.e., the feeding capacity is less than the conveying capacity of the screws at the feeding port). Liquids (typically plasticiser) could be introduced in a following location, preferably using a pressureinjection way to ensure uniform mixing between the solid(s) and the liquid(s). The granules undergo exposure to high temperatures and pressures in the extruder, resulting in the disruption, and gelatinisation/melting, of starch granules [16]. After gelatinisation, the TPS will undergo further melting and compression, which may be increased with increased use of high mixing (kneading or reversing) elements in the screw. It is important to note that degradation of starch molecules also occurs with the thermal mechanical treatment in the extruder, with amylopectin molecules suffering more than amylose [5,108]. And the mechanism for scission of the polymer chains is believed to preferentially take place close to the centre of the molecule, causing the size distribution to narrow and converge toward a maximum stable size [108]. The TPS is forced through the extrusion line and fi nally out of a die at the end of the extruder at a controlled rate (possibly with the aid of a melt pump to further control the fl ow rate).
It is interesting to note that some special techniques have been used for the production of TPS. For example, Stute et al. [109] claimed that starch can be gelatinised by a high pressure (upwards of 800 MPa) alone. However, this high pressure cannot be easily achieved without special equipment, and to date this method of processing has only been applied to starch slurries.

Characteristics of Starch Extrusion Process
The primary aim of extrusion is to ensure complete melting and suffi cient mixing, while avoiding degradation (to maintain good mechanical properties), to obtain a homogenous TPS melt. This can only be achieved by careful investigation of the key processing parameters, including raw material feed rate, temperature, screw speed, and screw profi le.
The temperature profi le in extrusion is a vital processing parameter that needs to be established mostly depending on the feed material. A typical temperature profi le for all starches will start from a low temperature (e.g., room temperature), ramp up to a highest value at the middle of the barrel, and then is reduced to some extent at the sections near the die and at the die. The use of a low temperature at the sections near the feed port is to prevent the loss of water immediately after the raw starch is fed in, since water is the most frequently used plasticiser. Then higher temperatures at the middle sections, especially coupled with kneading elements there, could provide a strong thermomechanical treatment to destroy starch granules and transform starch into TPS. A reduced, appropriate, temperature at the die could assist stable extrusion of the material from the die (without undesirable foaming if water and/ or other volatile substances are used in the formulation) and facilitate the orientation and crystallisation of the material for better properties (in particular mechanical).
It is noteworthy that the different abovementioned parameters can further determine three more important parameters, namely pressure, residence time, and SME. For example, a change in the screw speed will also modify the pressure, SME, and even temperature. Thus, it is often diffi cult to isolate the effects of the latter three parameters.
According to Gropper et al. [110], SME can be calculated from the following equation: where N is screw speed (rpm), T is torque (N·m), Q is feed rate (kg/h), and SME is in unit of kJ/kg. In general, an increased ratio of N/Q can result in a higher SME value and greater macromolecular degeneration, leading to decreased viscosity [2]. Also, it has been found that higher screw speeds can lead to lower gelatinisation of the starch because they decrease the residence time of the starch material in the extruder [111]. However, increasing the screw speed may also lead to a higher mixing torque [112].

Reactive Extrusion of Starch
Reactive extrusion (REX) means concurrent reaction and extrusion. It can be used to produce either chemically modifi ed starch or starch-based blends with high conversion effi ciency and a rapid rate of production, which has already been extensively reviewed in several papers [113][114][115]. This technique has recently been used more innovatively. For example, Frost et al. [116] created plasticised starch-silica (SiO 2 )-poly(vinyl alcohol) (PVOH) composite fi lms by REX using tetraethyl orthosilicate (TEOS) as a precursor. The TEOS-to-SiO 2 conversion effi ciencies of up to 41.3% were achieved, and the resulting fi lms had well dispersed SiO 2 particles of size ranging from 20 μm to < 1 μm, which were responsible for the increased tensile strength and Young's modulus.
In addition, some of the studies of starch-PLA blends by Xiong et al. [102,103,117], as mentioned in Section 3.3, could also be considered as involving REX, as the compatibiliser, starch, and PLA reactively interacted with each other during melt processing.

Impact of Extrusion on Structures and Properties of Thermoplastic Starch Products
Since extrusion processing is complex, it is not an easy job to design a perfect extrusion process for the production of TPS materials with mostly desirable properties. This needs repeated practical tests with an extruder and improvements in the extrusion process, taking into consideration different technical parameters (type of feed starch, formulation, extrusion processing conditions, and also subsequent environmental conditions [relative humidity and temperature] of storage or annealing). It is hoped that native starch granules can be completely transformed into a homogenous plasticised state by extrusion, but unfortunately TPS products, especially those from high-amylose starches (which are preferably used for better properties nevertheless), normally still contain remaining starch granules. This can be detected by various techniques such as microscopy and X-ray diffraction (XRD). Using a light microscope, Li et al. [7] showed that granules that were not disrupted still largely exist in extruded fi lms from high-amylose starch. The XRD can also be useful in detecting the granule remnants and its original crystalline structure. As mentioned before about original polymorphs of starch, a strong signal indicating A-or B-type by XRD usually means a large amount of remaining starch granules [5]. Moreover, atomic force microscopy (AFM) can be a good way to image fi lm samples and obtain topographic images of the starch fi lms surface at a micrometre level. It is found that there were granular envelopes with a size of around 10 μm with a surround matrix of amylose, and when fi lms were more gelatinised, the surface contained less granular envelopes [118]. When these envelopes were surrounded by dark regions, it might represent surface tension developed by concentrated amylose solutions in relation to granule remnants as the liquid evaporates during fi lm formation [118].
For reducing the granule remnants in TPS products, a change in parameters such as increased temperature or screw speed may be able to make a more homogenous fi lm. It has been suggested that during extrusion the mechanical shear is more important in disruption of the granules and their crystalline structures than thermal treatment [5]. However, care needs to be taken, as the mechanical shear is also more signifi cant in contributing to the degradation of amylopectin molecules [5], potentially negatively affecting the mechanical properties of the fi nal products.

PROGRESS OF STARCH PROCESSING RESEARCH AT UQ
At The University of Queensland (UQ), we have focused on many fundamental aspects in starch processing, including understanding of fundamentals of native starch [10,[119][120][121], starch degradation during extrusion [5,108], processing rheology of TPS [122], effects of starch type and chemical modifi cation on the processing and properties [23,31]; development of new TPS polymer blend products [123][124][125][126], optimisation of the related processing [127], understanding of performance of TPS by post-processing environmental conditions [128][129][130]; as well as the very recent work on plasticisation of starch using novel plasticisers [131]. Product wise our research has focused on the development of water soluble rigid sheet products and blown fi lms for dry goods packaging and water resistant injection moulding grades for moulding applications [132]. Very recently we have embarked on a new project focusing on thin starch fi lms for barrier packaging. This project, through • the development of new biopolymers, • the understanding of barrier properties for biopolymer fi lms and laminates, and • smart design of fi lms for specifi c packaging applications, will enable the robust development of high-performance thin fi lm products with enhanced barrier properties. This technology will greatly expand the product range of current TPS fi lms into attractive markets, such as meat trays and smart packaging, by diminishing reliance on one raw material feed stock, control of novel processing conditions, and by providing possible environmental advantages. An example of a novel thin fi lm is shown in Figure 3.
There is a lack of understanding of processing starch fi lms at low gauge (under 50 microns), as well as the subsequent relationships of processing-structureproperties of these fi lms. A range of fi lm properties such as pH resistance, bacterial resistance, temperature resistance, barrier resistance and changes in these properties over time, have also not been explored.

CONCLUSIONS
Clearly starch has always offered a very attractive low-cost base for new biodegradable polymers due to its low material cost and ability to be processed on conventional plastic processing equipment. However, poor processibility and product properties narrow the applicability of TPS to niche markets (e.g., edible food packaging, disposable packaging). Thus, the main technological drive in the future will be the engineering of more advanced properties into these kinds of low-cost materials. This development will most probably be in the form of integrating research in parallel from: a. New TPS formulation (like alternative plasticisation methods, blends, nanocomposites, and reactive modifi cation), and b. Better fundamental understanding of TPS processing and its effects on extruded starch structure and properties (gelatinisation, plasticisation, retrogradation, degradation, and fi nal material structure and properties).
It is hoped that this paper has addressed key aspects of both these parallel development streams.