Poly(lactic acid) Composites Containing Carbon-Based Nanomaterials: A Review

Poly(lactic acid) (PLA) is a green alternative to petrochemical commodity plastics, used in packaging, agricultural products, disposable materials, textiles, and automotive composites. It is also approved by regulatory authorities for several biomedical applications. However, for some uses it is required that some of its properties be improved, namely in terms of thermo-mechanical and electrical performance. The incorporation of nanofillers is a common approach to attain this goal. The outstanding properties of carbon-based nanomaterials (CBN) have caused a surge in research works dealing with PLA/CBN composites. The available information is compiled and reviewed, focusing on PLA/CNT (carbon nanotubes) and PLA/GBM (graphene-based materials) composites. The production methods, and the effects of CBN loading on PLA properties, namely mechanical, thermal, electrical, and biological, are discussed.


Introduction
The growing environmental awareness and new rules and regulations are forcing the industries to seek more ecologically friendly materials for their products [1]. In the last two decades, industrial and academic research on polymer composites was pursued to provide added value properties to the neat polymer without sacrificing its processability or adding excessive weight [2].
Poly(lactic acid) (PLA), which is derived from natural sources, biodegradable, and bioabsorbable, has had significant demand due to presenting versatile applications in packaging, pharmaceutical, textiles, engineering, chemical industries, automotive composites, biomedical and tissue engineering fields [3]. Its biodegradation time can be tuned, depending on the molecular weight, crystallinity, and material geometry [4]. However, the relatively low glass transition temperature, low thermal dimensional stability, and mechanical ductility limit the number of its applications. A significant body of research has dealt with the use of fillers for improving the properties of PLA [5][6][7]. In this context, carbon based nanomaterials (CBN), offer the potential to combine PLA properties with several of their unique features, such as high mechanical strength, electrical conductivity, thermal stability and bioactivity [8][9][10][11][12][13][14][15][16]. Carbon nanotubes (CNT) and graphene-based materials (GBM) are state of the art and very promising representatives of these materials. CNT have exceptional mechanical properties, aspect ratio, electrical and thermal conductivities, and chemical stability. However, their production methods are usually more complex and expensive, often leaving toxic metal residues [17][18][19][20]. Hence, CBN have been extensively used in polymer composites. In order to take advantage of their large surface area maximizing its effectiveness as filler, dispersion must be efficient, so as to maximize the amount of deagglomerated primary units. Functionalization is often used to improve compatibility with the polymer matrix. However, this can disrupt the sp 2 hybridization of CBN carbon structure and subsequently hinder their properties [131]. Some examples of CBN modifications used on the research works reported in this review are compiled in Figure 1. Some of these involve simple chemical oxidation, prior to surface modification with isocyanates, polymers (ethylene glycol, CBN have been extensively used in polymer composites. In order to take advantage of their large surface area maximizing its effectiveness as filler, dispersion must be efficient, so as to maximize the amount of deagglomerated primary units. Functionalization is often used to improve compatibility with the polymer matrix. However, this can disrupt the sp 2 hybridization of CBN carbon structure and subsequently hinder their properties [131]. Some examples of CBN modifications used on the research works reported in this review are compiled in Figure 1. Some of these involve simple chemical oxidation, prior to surface modification with isocyanates, polymers (ethylene glycol, poly(caprolactone), methyl methacrylate, poly(vinyl pyrrolidone), and PLA), polyols or silanes. The impact of these on the composite properties is discussed in Section 5.

Production of PLA/CBN Composites
Three methods are most frequently used to obtain a dispersion of CBN into a polymer matrix: solution mixing, melt blending, and in situ polymerization [22,104]. Mechanical milling, also called ball milling, has been gaining recognition as an alternative technique with specific advantages, but it has not yet been reported for PLA/CBM composites. High impact milling is performed at room temperature on dry powders, prior to melt processing. Its effectiveness and benefits in relation to other methods have been shown for different polymer/filler systems [132].

Solution Mixing
Solution mixing is a simple procedure, requiring no special equipment, and allowing for straightforward scale-up. This method typically consists of three steps: (i) dispersion of the nanomaterial in a suitable solvent using sonication or mechanical stirring; (ii) dissolution of the polymer in the previous dispersion, under appropriate stirring; and (iii) removal of the solvent by distillation or lyophilization. Often the dispersion is cast into a flat mold, and then the solvent is evaporated. Flat composite slabs are therefore obtained. For this reason, the procedure is often called "solvent casting". As an alternative, the dispersion may be cast onto a low surface energy material (e.g., PTFE coated surface) using a blade applicator (doctor blading). After solvent evaporation, thin composite films are obtained. The viscosity of the dispersion needs to be adjusted for this procedure, which can be done by changing the concentration of polymer [133]. If production of fibers is desired, the third step can be replaced by electrospinning. This technique allows obtaining fibers that are much smaller in diameter (ranging from micrometers to nanometers) than those produced by conventional techniques. The basis of electrospinning is to charge the polymer solution in the spinneret tip with a high voltage, so that the electrostatic repulsion overcomes the surface tension of the solution, causing its ejection. The solvent vaporizes while the jet is in the air, producing a continuous fiber which deposits on the ground collector [27].
Complete solvent removal is a critical issue when using solution mixing to prepare composites, since toxicity concerns may arise when organic solvents are used. In addition, presence of residual solvent induces plasticization of the polymer matrix, which may alter significantly its mechanical properties [134][135][136].
PLA is soluble in organic solvents such as chlorinated solvents, benzene, tetrahydrofuran (THF), dimethyl formamide (DMF) and dioxane, but insoluble in ethanol, methanol, and aliphatic hydrocarbons. CBN are hydrophobic, therefore cannot be easily dispersed in polar solvents. However, they can be oxidized or modified with hydrophilic groups in order to allow dispersion in such solvents. Solubility limitations can also be overcome to a certain point by using ultrasonication to produce short-time metastable dispersions of CBN in organic solvents, which can then be mixed with polymer solutions [137].

Melt Blending
Melt blending is an economically attractive, environmentally friendly and highly scalable method for preparing nanocomposites. This strategy involves direct addition of the nanomaterial into the molten polymer, allowing optimization of the state of dispersion by adjusting operating parameters such as mixing speed, time and temperature. Due to the absence of solvent, the only compatibility issue is placed in terms of the nanofiller towards the polymer matrix [27,48]. The drawbacks of this procedure are the low bulk density of CBN, that makes the feeding of the melt-mixer a troublesome task and the lower degree of dispersion that is usually attained when compared to solvent mixing [137,159].
In addition to melt blending not being as effective as the solution mixing method or in situ polymerization in terms of the ability to achieve good filler dispersion, damage to the nanofillers or polymer may occur under severe conditions. Some studies have shown that processing conditions can have an impact on the molecular weight of PLA [171]. This can be mainly attributed to the presence of impurities such as acidic species, peroxide groups, metallic ions or other residual products that can increase the degradation of PLA during melt mixing [172].

In Situ Polymerization
In situ polymerization for production of polymer composites generally involves mixing the filler in neat monomer, or a solution of monomer, in the presence of catalysts and under proper reaction conditions [173]. The polymer chains grow on the filler surface, being covalently bonded. In situ polymerization generally results in more homogeneous particle dispersion than melt blending [174]. Use of this approach for polymerizing lactide in the presence of CNT has been reviewed by Brzeziński and Biela [175]. Contrary to CNT, that usually are post-treated, GBM already present some chemical groups that can be used in further functionalization, such as grafting polymer chains via atom transfer radical polymerization. Examples of in situ polymerization on GBM include polymers such as polyaniline (PANI), polyurethane (PU), polystyrene (PS), poly(methyl methacrylate) (PMMA) and polydimethylsiloxane (PDMS) [24].
Concerning the particular case of PLA/CBN, only a few examples of in situ polymerization can be found in the literature. Ring opening polymerization of L-lactide in presence of GBM has been reported by Yang et al. [176] and Promoda et al. [177]. Carboxyl-functionalized CNT have been grafted with PLA by Li and co-workers [178].
The above-mentioned composite production methods can be used both with GBM and CNT, and are congregated in Figure 2.

Properties of PLA/CBN Composites
Numerous researchers have studied the properties of PLA combined with other materials, in order to tune key properties regarding specific applications [48]. The current review is focused on the effect of incorporating two carbon-based nanomaterials, CNT and GBM, in PLA. CNT are known for two decades and have well established large-scale production methods. GBM, which have been raising a growing interest from the scientific community, are cheaper and, in principle, comparable in properties to CNT [177].

Mechanical Properties
Physico-chemical interactions between fillers and polymer phase contribute to load transfer and distribution along the CBN network. Table 1 shows that solution mixing is the most commonly

Properties of PLA/CBN Composites
Numerous researchers have studied the properties of PLA combined with other materials, in order to tune key properties regarding specific applications [48]. The current review is focused on the effect of incorporating two carbon-based nanomaterials, CNT and GBM, in PLA. CNT are known for two decades and have well established large-scale production methods. GBM, which have been raising a growing interest from the scientific community, are cheaper and, in principle, comparable in properties to CNT [177].

Mechanical Properties
Physico-chemical interactions between fillers and polymer phase contribute to load transfer and distribution along the CBN network. Table 1 shows that solution mixing is the most commonly reported method for incorporation of CBN in PLA. The most frequently used solvents are chloroform, DMF and THF. The filler concentrations most often tested are between 0.1-2 wt %. Maximum improvements in Young's modulus (E), storage modulus (E'), and tensile strength (σ max ) are found for concentrations between 0.25-5 wt % for CNT, and between 0.1-1 wt % for GBM. The larger improvement in E, relative to unfilled PLA, is of 372%, for 0.25 wt % MWCNT sonicated in a PLA/chloroform dispersion, followed by compression molding of the dried mixture [138]. For GBM, the best performance is an increase of 156% with incorporation of 0.4 wt % GNP-M, also by sonication, but followed by film casting using doctor blading. In this study, comparison is made with GO, which yields a maximum E increase at 0.3 wt % loading. Figure 3 presents microscopy images demonstrating good dispersion of the fillers in the PLA matrix [135].
reported method for incorporation of CBN in PLA. The most frequently used solvents are chloroform, DMF and THF. The filler concentrations most often tested are between 0.1-2 wt %. Maximum improvements in Young's modulus (E), storage modulus (E'), and tensile strength (σmax) are found for concentrations between 0.25-5 wt % for CNT, and between 0.1-1 wt % for GBM. The larger improvement in E, relative to unfilled PLA, is of 372%, for 0.25 wt % MWCNT sonicated in a PLA/chloroform dispersion, followed by compression molding of the dried mixture [138]. For GBM, the best performance is an increase of 156% with incorporation of 0.4 wt % GNP-M, also by sonication, but followed by film casting using doctor blading. In this study, comparison is made with GO, which yields a maximum E increase at 0.3 wt % loading. Figure 3 presents microscopy images demonstrating good dispersion of the fillers in the PLA matrix [135].  The maximum increase on E' is of 1500%, achieved with incorporation of 0.5 wt % rGO-KH792 in PLLA, by simple stirring, casting on PTFE mold, and vacuum drying the resultant films at 120 • C for 48 h [157]. However, this increase only occurs around PLA transition temperature (60-65 • C). At ambient temperature, the best result is an increase of 67% with incorporation of 3 wt % A-SWCNT-Si (acid treated and grafted with 3-isocyanatoporpyl triethoxysilane) in PLA by sonication, followed by drying and compression molding at 190 • C [144]. The maximum increase in σ max is of 129 wt %, obtained with incorporation of 0.4 wt % GNP-M in PLA by sonication and film casting by doctor blading [135]. For CNT the best result is an increase of 47% obtained with MWCNT grafted with PLA, and then incorporated at a loading of 1 wt % in PLA by sonication in chloroform, separation, drying and compression molding at 180 • C [141]. When considering CNT without modification, the best result reported is an increase of 9% for 1.2 wt % MWCNT incorporated in PLA by solution mixing, followed by drying and compression molding at 180 • C with a pressure of 1000 Kg [142].
Melt-blending is less frequently reported than solution mixing for production of PLA/CBN composites, probably due to the lower availability of the necessary equipment. Results show that it tends to be not as effective in improvement of mechanical properties, as solution mixing. The best performance in terms of E (↑88%) and E' (↑76%) is reported by Lin et al. [160] for an incorporation of 3 wt % MWCNT grafted with stearyl alcohol (MWCNT-C 18 OH) in PLA by melt blending (180 • C, 5 min, 50 rpm), using Ti(OBu) 4 for transesterification, followed by compression molding at the same temperature. When PLA is not transesterified, E and E' increases were of 74% and 44%, respectively. The maximum increase in σ max (40%) is obtained incorporating 0.08 wt % rGO using a twin-screw mixer (175 • C, 8 min, 60 rpm), followed by compression molding at 180 • C [168]. The incorporation by melt blending (180 • C, 20 min, 50 rpm) of 0.25 wt % GNP-M5 and C in PLA followed by compression molding at 190 • C, prevented its mechanical properties decay after 6 months degradation in phosphate-buffered saline at 37 • C [180].
In situ polymerization is the least used technique. It has been reported by Pramoda et al. [177], who performed PLA ring-opening polymerization in presence of 1 wt % of GO functionalized with butanediol and GO modified with POSS silsesquioxane. In the first case, improvements of 1% and 14% in E and hardness are obtained, respectively. In the second, the performance is increased by 33% and 45%, in the same order.
Comparing the results for CNT and GBM, we can conclude that both can effectively improve PLA mechanical properties, whether by solution mixing and melt blending. However, use of GBM usually implies lower amounts of GBM than of CNT. Several chemical modifications have been tried to improve compatibility with the polymer matrix, with ineffective results is some cases. Functionalization with carboxyl groups is the most common and effective procedure to improve CNT compatibility with PLA matrix [146]. On the other hand, no relation has been observed between CBN morphological properties (size, length, and diameter) and the mechanical performance of the composites. PLA was modified with benzoyl chloride and pyridine (PLAm), then acid chloride groups were added by reaction with thionyl chloride and triethylamine, then fMWCNT were added and the mixture centrifuged and filtered to remove excess filler and salts. Finally, sonication in chloroform and film casting was performed MWCNT functionalized with COOH using Fenton reactant and then reacted with SOCl 2 and ethylene glycol (fMWCNT) d = 9.5 nm l = 1.5 µm 95% purity Not clear ∆E↑17%, ∆σ max ↑8% (comparing to PLAm) [140] Sonication in chloroform, coagulation with methanol, filtration, vacuum drying, and compression molding

Electrical Properties
Neat PLA is electrically insulating with a low electrical conductivity (σ ≈ 1 × 10 −16 S m −1 ), and high sheet resistance (ρ ≈ 5 × 10 12 Ω sq −1 ) [144,160]. Since CNT and reduced forms of GBM present high electrical conductivity, they can be incorporated in PLA to improve its conductivity. This sort of composites have potential to be used as electrical stimulating implants, since PLA is used as a biodegradable matrix in orthopedic material. Other advantages of increasing PLA conductivity are the possibility of using it as antistatic coating/material or for electromagnetic shielding [104]. The minimum amount of filler required to form a conductive network within the polymer is called percolation threshold, and should be as low as possible in order to keep processing simple (relatively low viscosity of the melt) and low costs. Table 2 shows that, once again, the most used method to incorporate CBN on PLA for electrical properties evaluation is solution mixing. The amount of fillers ranges from 0.01 to 10 wt %. The best result, considering electrical conductivity (σ) with CNT is 3.5 × 10 −3 S m −1 , obtained incorporating 10 wt % MWCNT in PLA by sonication in chloroform, followed by drying and compression molding at 200 • C during 15 min [138]. Results are also often presented in terms of sheet resistance (ρ ), being the lowest value reported by Shao et al. [183], of 1 × 10 2 Ω sq −1 achieved incorporating 5 wt % MWCNT previously oxidized (treated with HCl and HNO 3 ) in PLA by solution mixing, followed by electrospinning of aligned nanofibers (d ≈ 250 nm). The alignment of the fibers slightly improved sheet resistance, comparing with random meshes. Interestingly, Yoon et al. [143] observe a considerable sheet resistance of 1 × 10 5 Ω sq −1 , with incorporation of 1 wt % MWCNT-COOH, also oxidized by treatment with strong acids (H 2 SO 4 and HNO 3 ). For GBM, the maximum conductivity reported is 2.2 S m −1 , higher than for CNT, obtained incorporating 1.25 wt % rGO-g (reduced with ammonia) in PLA by sonication in DMF. Interestingly, the solvent used for dispersion of CNT in PLA is always chloroform and for GBM is always DMF.
Melt-blending is the second most used approach to disperse CBN in PLA in order to improve its electrical properties, being most often performed by twin-screw extrusion, followed by compression molding. The highest σ considering CNT is 50 S m −1 , which is reported by Pötschke et al. [184]. These authors prepare MWCNT mixtures by twin-screw extrusion, followed by piston spinning at different speeds. They conclude that non-spun mixtures with 5 wt % MWCNT in PLA present the same conductivity as 3 wt % mixtures after piston spinning at a speed of 20 m min −1 . Microscopy images in Figure 4 allow to observe good MWCNT dispersion and orientation due to spinning process.

Electrical Properties
Neat PLA is electrically insulating with a low electrical conductivity (σ ≈ 1 × 10 −16 S m −1 ), and high sheet resistance (ρ□ ≈ 5 × 10 12 Ω sq −1 ) [144,160]. Since CNT and reduced forms of GBM present high electrical conductivity, they can be incorporated in PLA to improve its conductivity. This sort of composites have potential to be used as electrical stimulating implants, since PLA is used as a biodegradable matrix in orthopedic material. Other advantages of increasing PLA conductivity are the possibility of using it as antistatic coating/material or for electromagnetic shielding [104]. The minimum amount of filler required to form a conductive network within the polymer is called percolation threshold, and should be as low as possible in order to keep processing simple (relatively low viscosity of the melt) and low costs. Table 2 shows that, once again, the most used method to incorporate CBN on PLA for electrical properties evaluation is solution mixing. The amount of fillers ranges from 0.01 to 10 wt %. The best result, considering electrical conductivity (σ) with CNT is 3.5 × 10 −3 S m −1 , obtained incorporating 10 wt % MWCNT in PLA by sonication in chloroform, followed by drying and compression molding at 200 °C during 15 min [138]. Results are also often presented in terms of sheet resistance (ρ□), being the lowest value reported by Shao et al. [183], of 1 × 10 2 Ω sq −1 achieved incorporating 5 wt % MWCNT previously oxidized (treated with HCl and HNO3) in PLA by solution mixing, followed by electrospinning of aligned nanofibers (d ≈ 250 nm). The alignment of the fibers slightly improved sheet resistance, comparing with random meshes. Interestingly, Yoon et al. [143] observe a considerable sheet resistance of 1 × 10 5 Ω sq −1 , with incorporation of 1 wt % MWCNT-COOH, also oxidized by treatment with strong acids (H2SO4 and HNO3). For GBM, the maximum conductivity reported is 2.2 S m −1 , higher than for CNT, obtained incorporating 1.25 wt % rGO-g (reduced with ammonia) in PLA by sonication in DMF. Interestingly, the solvent used for dispersion of CNT in PLA is always chloroform and for GBM is always DMF.
Melt-blending is the second most used approach to disperse CBN in PLA in order to improve its electrical properties, being most often performed by twin-screw extrusion, followed by compression molding. The highest σ considering CNT is 50 S m −1 , which is reported by Pötschke et al. [184]. These authors prepare MWCNT mixtures by twin-screw extrusion, followed by piston spinning at different speeds. They conclude that non-spun mixtures with 5 wt % MWCNT in PLA present the same conductivity as 3 wt % mixtures after piston spinning at a speed of 20 m min −1 . Microscopy images in Figure 4 allow to observe good MWCNT dispersion and orientation due to spinning process.        Considering ρ , the best performance is obtained incorporating 3 wt % MWCNT-C 18 OH (MWCNT modified with DCC and stearyl alcohol) using and external mixer, followed by compression molding at 180 • C during 5 min, resulting in a ρ of 1 × 10 −1 Ω sq −1 [160]. This is the most effective modification performed, considering the sheet resistance values obtained with incorporation of the same amount of non-modified MWCNT, which was 3 × 10 5 Ω sq −1 . For GBM, the higher σ is 2.6 × 10 −4 S m −1 , resultant from dispersion using an internal mixer at 180 • C, of 5 wt % PFG (graphene nanoparticles functionalized with methylmethacrylate) [164]. For rGO, a non-functionalized GBM, the best conductivity value is obtained for 2 wt % incorporation in PLA using a twin-screw extruder and compression molding. The value obtained is of 1 × 10 −9 S m −1 , being higher than for the other concentrations tested. It can be compared, for example, with a σ of 1 × 10 −13 S m −1 for 0.2 wt % [168]. In most works evaluated, electrical properties improve with the increase of filler amount.
In situ polymerization is the least explored technique, despite interesting results being obtained by Yang et al. [176], which incorporate 0.01-2 wt % trGO (thermally reduced) in PLA by ring-opening melt polymerization of L-lactide in presence of the filler. As example, σ obtained is 5 × 10 −6 and 1.6 × 10 −2 S m −1 for 1.5 and 2 wt %, respectively.
An interesting study by Chiu et al. [88], shows that purification of MWCNT by sonication with strong acids improved fillers compatibility and dispersibility in PLA, resulting in better electrical conductivity. The values of σ for incorporations of 7 wt % are 5 × 10 −8 and 2 × 10 −6 S m −1 , respectively for non-purified and purified MWCNT. Purification introduced polar functional groups on the CNT surface, allowing better dispersion, which resulted in more deagglomerated particles that formed a wider conductive network on PLA matrix.

Thermal Properties
Several works studied thermal properties of PLA containing CBN. CNT incorporations range from 0.01 to 15 wt %, while for GBM lower amounts are needed 0.01-2 wt % (Table 3). However, for both CBN, slight or no changes are observed in the composites' thermal properties, especially when low fillers amounts are used [135,146,156,157,[160][161][162]167]. The most frequently used techniques to evaluate thermal properties in polymer composites are thermogravimetric analysis (TGA), differential scanning calorimetry (DSC), and dynamic mechanical analysis (DMA). TGA allows determination of thermal degradation temperatures (T d ) and DSC and DMA phase transition temperatures (T g -glass transition temperature, T m -melting temperature, and T c -cold crystallization temperature).
A positive deviation in T d is expected when there is good compatibility between CBN and the polymer matrix, combined with good dispersion of the fillers. This leads to restriction of PLA's chains motions, delaying thermal decomposition. Also, CBN can induce the formation of a crystallization region on their surfaces, which absorbs some heat as temperature of the composite increases. However, the incorporation of too high amounts of CBN can lead to the formation of agglomerates, which represent structural defects in the matrix, decreasing thermal stability [145]. Some works also attribute improvements in thermal stability to the barrier effect caused by the CBN, which creates a "tortuous path" delaying permeation of oxygen and the escape of volatile degradation products, and also to char formation [146,150,167]. Increases in T g are usually also associated with good interaction between CBN and polymer matrix, leading to constraint of PLA's molecular mobility by hydrogen bonding and electrostatic attraction [139,140,146,150]. T m increases are usually attributed to a nucleation effect caused by the CBN, which increases the degree of crystallinity [146,150,176]. For the same reason, T c usually decreases with CBN incorporation [141,146,153,162,170,176].
When using solution mixing, the highest variation in terms of T g is an increase of 10 • C, obtained using 1 wt % MWCNT purified by treatment with strong acids. Comparing with non-purified filler at the same loading, the increase is 5 • C higher. This is explained by purified MWCNT having stronger interfacial interactions with PLA matrix, imposing increased restriction to the mobility of macromolecular chains, and therefore rising T g . Also, T d (decomposition temperature) presents an increase of 10 • C for purified materials [88]. For T m , the higher increase is of 16 • C for 0.3 and 1 wt % MWCNT-PCL (functionalized with poly(caprolactone)) incorporated in PLA aligned fibers by sonication in dichloromethane and electrospinning. Also, T c decreases more than 10 • C, due to MWCNT inducing heterogeneous crystallization [145]. However, the higher decrease in T c (<20 • C), is obtained by Moon et al. [138], with the incorporation of 3-10 wt % MWCNT, with a length of about 2000 µm. In literature, the degradation temperatures of the polymeric materials determined by TGA are presented in different terms. For example, as T di (beginning of thermal degradation), T d5 (decomposition temperature for 5 wt % loss), and T d50 (decomposition temperature for 50% weight loss). For T di , the highest increase is of 20 • C, obtained incorporating 2.5 wt % MWCNT-COOH (carboxylated with strong acids) by sonication in PLA dispersed in dichloromethane and THF, followed by vacuum drying and compression molding [146]. Considering T d50 , the best result is an increase of 1-3 • C, in a work above described [145].
GBM incorporation also induces changes on thermal properties of PLA. For T g , an increase of 7 • C was obtained sonicating 0.4 wt % GNP in PLA films prepared by solvent casting [135]. The highest increases in T m have been of 5 • C, for samples obtained by compression molding of PLA with 0.5 wt % GO grafted with PLA, produced by vacuum drying a dispersion in chloroform [150]. Significant decrease in T c , of 20 • C, is observed for PLA with 2 wt % GO, obtained by solvent mixing [153]. Thermal stability of PLA has been shown to improve with addition of GBM. 2 wt % GONSs (graphene oxide nanosheets) increases T di by 16 • C in samples produced by solvent mixing [156]. Also, T d5 is increased by 11 • C sonication of 0.2 wt % GNSs (graphene nanosheets) in PLA dispersed in DMF, dried under vacuum to produce composites [152]. Finally, T d max (T of maximum degradation rate) increases 33 • C for PLA filled with TRG, produced by solution mixing [154]. Chemical modifications of MWCNT are reported to increase thermal properties of the composites. For example, directly comparing with PLA/MWCNT(non-modified), the incorporation of 1 wt % MWCNT grafted with PLA in the same PLA matrix, results in increases of about 3 • C in T g and decreases of 9 • C in T c [141]. Treatment with strong acids followed by silanization of SWCNT [144], which are incorporated in PLA at loading ranging from 0.1 and 3 wt %, results in increases of about 5 • C in T g .
Concerning composites produced by melt-blending, the highest increases in T g are of 5-6 • C, for PLA micro-fibers with 3 wt % MWCNT to PLA [184]. Also, T c is observed to decrease at most 12 • C with incorporation of 0.5 and 2 wt % MWCNT [170]. Chieng et al. [167], study on the thermal properties of PLA/PEG (9:1) blends with addition of 0.1-1 wt % GNP, reveals no variations on T g , T m , and T c . However, T di , T max , and T 50 , increase by 56, 53, and 44 • C, respectively, for 0.5 wt % loadings.
In situ polymerization of L-lactide in presence of TRG in amounts from 0.01 to 2 wt % result in considerable increases on T g , T m , and T dmax . For example, at 2 wt % loading, increases of 5, 14, and 18 • C are obtained, respectively [176]. In a different work reporting in situ polymerization of L-lactide, covalent functionalization of GO with both 1,4-butanediol, and polyhedral silsesquioxane results in increases in T g (18, 20 • C), T c (15, 8 • C), T m (7, 5 • C), and T d5 (23, 11 • C) comparing with PLA/GO composites at 1 wt % loadings [177]. Table 3. Thermal properties of PLA/CBN composites in comparison with non-modified PLA. Production methods and CBN characteristics. PLA was modified with benzoyl chloride and pyridine (PLAm), then acid chloride groups were added by reaction with thionyl chloride and triethylamine, then fMWCNT were added and the mixture centrifuged and filtered to remove excess filler and salts. Finally, sonication in chloroform and film casting was performed MWCNT functionalized with COOH using Fenton reactant and then reacted with SOCl 2 and ethylene glycol (fMWCNT). d = 9.5 nm l = 1.5 µm 95% purity

Biological Properties
Most nanomaterials may present toxicity at concentrations above a certain threshold when in isolated form, i.e., when not incorporated in a polymer matrix [40,186]. Biocompatibility of the composites must be tested when considering uses as biomaterials. Table 4 shows that PLA/CBN composites (films and nanofibers) do not tend to decrease in vitro metabolic activity of several cell types, or cause increases up to 40% until 72 h incubations. Also, the selection of production method used (melt blending or solvent mixing followed by casting, doctor blading, spin coating or electrospinning), does not seem to influence cell proliferation. For long term incubations, McCullen et al. [187] shows that scaffolds of PLA with 1 wt % MWNTs do not to influence metabolic activity of adipose-derived human mesenchymal stem cells (hMSCs) at 7 days. At 14 days, cells present increased metabolic activity and longitudinal alignment induced by the scaffolds. Sherrell et al. [188] reports PLGA (1:1) with a surface layer of graphene applied by CVD to increase PC-12 cells average length of neurites by 2.5 fold when electrical stimulated. Also, hemocompatibility improvements are reported with both incorporation of 0.4 wt % GNP by solvent mixing followed by doctor blading [149] and 4 wt % MWCNT by extrusion followed by injection molding [189] in PLA. In the last case, MWCNT alignment is associated with decreased platelet adhesion and activation. Thus, alignment seems to be generally benefit for biocompatibility. The bioeffectiveness of electrical stimulation together with nanofibers and its fillers alignment is confirmed by Shao et al. [183], which cultures osteoblasts at the surface of PLA/MWCNT-ox (3 wt %) produced by solution mixing followed by electrospinning. They observe improvements in cell elongation (190%) and metabolic activity (20%) for random nanofibers (d ≈ 250 nm) under DC 100 µA, comparing to unstimulated controls. For aligned fibers the previous values increase by 90 and 40%, respectively. The aspect ratio is higher for the latter, comparing with random stimulated fibers ( Figure 5). Finally, An et al. [190] find that PLA composite films and nanofibers with 3 wt % PU and 5 wt % GO almost completely suppress Escherichia coli and Staphylococcus aureus growth after 24 h, not affecting MC3T3-E1 cells metabolic activity. This effect is attributed to GO potentially inducing oxidative stress or physical disruption on bacteria.

Biological Properties
Most nanomaterials may present toxicity at concentrations above a certain threshold when in isolated form, i.e., when not incorporated in a polymer matrix [40,186]. Biocompatibility of the composites must be tested when considering uses as biomaterials. Table 4 shows that PLA/CBN composites (films and nanofibers) do not tend to decrease in vitro metabolic activity of several cell types, or cause increases up to 40% until 72 h incubations. Also, the selection of production method used (melt blending or solvent mixing followed by casting, doctor blading, spin coating or electrospinning), does not seem to influence cell proliferation. For long term incubations, McCullen et al. [187] shows that scaffolds of PLA with 1 wt % MWNTs do not to influence metabolic activity of adipose-derived human mesenchymal stem cells (hMSCs) at 7 days. At 14 days, cells present increased metabolic activity and longitudinal alignment induced by the scaffolds. Sherrell et al. [188] reports PLGA (1:1) with a surface layer of graphene applied by CVD to increase PC-12 cells average length of neurites by 2.5 fold when electrical stimulated. Also, hemocompatibility improvements are reported with both incorporation of 0.4 wt % GNP by solvent mixing followed by doctor blading [149] and 4 wt % MWCNT by extrusion followed by injection molding [189] in PLA. In the last case, MWCNT alignment is associated with decreased platelet adhesion and activation. Thus, alignment seems to be generally benefit for biocompatibility. The bioeffectiveness of electrical stimulation together with nanofibers and its fillers alignment is confirmed by Shao et al. [183], which cultures osteoblasts at the surface of PLA/MWCNT-ox (3 wt %) produced by solution mixing followed by electrospinning. They observe improvements in cell elongation (190%) and metabolic activity (20%) for random nanofibers (d ≈ 250 nm) under DC 100 µA, comparing to unstimulated controls. For aligned fibers the previous values increase by 90 and 40%, respectively. The aspect ratio is higher for the latter, comparing with random stimulated fibers ( Figure 5). Finally, An et al. [190] find that PLA composite films and nanofibers with 3 wt % PU and 5 wt % GO almost completely suppress Escherichia coli and Staphylococcus aureus growth after 24 h, not affecting MC3T3-E1 cells metabolic activity. This effect is attributed to GO potentially inducing oxidative stress or physical disruption on bacteria.  Osteoblast elongation is presented as the aspect ratio (b,c). Scale bars represent 30 µm [191].
In an in vivo study, Kanczler et al. [192] observe that PLA-CB 0.1 wt % scaffolds seeded or not with fetal femur-derived cells, when implanted in a murine critical-size femur segmental defect model aid the regeneration of bone defect. Pinto et al. [193] report both PLA/GNP-M5 (2 wt %) and CNT-COOH (0.3 and 0.7 wt %) to be biocompatible, both in vitro and in vivo (2 weeks subcutaneous implantation in C57Bl/6 mice). Also, PLA/GNP-M5 and C 0.25 wt % composites have not release toxic products after 6 months degradation in phosphate-buffered saline at 37 °C [180]. This is relevant considering that long-term biocompatibility must be assured for safe PLA/CBN composites implantation. In an in vivo study, Kanczler et al. [192] observe that PLA-CB 0.1 wt % scaffolds seeded or not with fetal femur-derived cells, when implanted in a murine critical-size femur segmental defect model aid the regeneration of bone defect. Pinto et al. [193] report both PLA/GNP-M5 (2 wt %) and CNT-COOH (0.3 and 0.7 wt %) to be biocompatible, both in vitro and in vivo (2 weeks subcutaneous implantation in C57Bl/6 mice). Also, PLA/GNP-M5 and C 0.25 wt % composites have not release toxic products after 6 months degradation in phosphate-buffered saline at 37 • C [180]. This is relevant considering that long-term biocompatibility must be assured for safe PLA/CBN composites implantation.    Biocompatible, both in vitro (human fibroblasts, HFF-1) and in vivo (2 weeks subcutaneous implantation in C57Bl/6 mice) [193] Laser sintering CB (carbon black)-not found Scaffolds (several shapes)-surface selective laser sintering (CB) Carbon black (d = 360 nm, surface area = 100 m 2 g −1 ) SSLS-PLA/CB 0.1 wt % scaffolds SSLS-PLA/CB 0.1 wt % scaffolds seeded or not with fetal femur-derived cells aided regeneration of murine bone defect [192]

Conclusions
Both CNT and GBM nanofillers are effective at improving PLA thermo-mechanical and electrical properties. However, lower amounts of GBM (0.1-1 wt %) are usually needed when comparing with CNT (0.25-5 wt %). Melt-blending is less reported than solution mixing for production of PLA/CBN composites, maybe because it implies use of specialized equipment. Moreover, results show that melt blending suffers from some drawbacks, since viscous shear is less effective than solvent sonication for promoting exfoliation/deagglomeration of CBN. In situ polymerization is the least reported technique, with further research being needed to demonstrate its advantages over the previous production methods.
Surface modifications of CBN can be used to improve compatibility with a polymer matrix. Functionalization with carboxyls is the most common and effective procedure to improve CNT dispersibility and compatibility with PLA. Some authors refer that purification with strong acids introduces polar groups in the carbon surface, which results in positive interaction with PLA. Besides straightforward chemical oxidation of CBN, other chemical modifications which lead to better performance after incorporation in PLA, comparing with non-modified CBN, include reaction with isocyanates, polyols, or silanes, and grafting with polymers (ethylene glycol, poly(caprolactone), poly(methyl methacrylate), poly(vinyl pyrrolidone), and PLA).
When comparing reduced and oxidized forms of GBM as PLA fillers, like rGO and GO, only in the case of increasing electrical conductivity the reduced forms show clearly better performance.
Based on the available data, no relation can be determined between CBN morphological properties (size, length, and diameter) and the composites performances.
The alignment of PLA/CNT fibers, has been shown to improve electrical conductivity. Electrical properties also improve with the increase of the amount of CBN incorporated.
Concerning biological properties, the composite production process does not influence cell metabolic activity, which does not decrease comparing to non-filled PLA. Furthermore, increases up to 40% in cell viability can be induced by GBM incorporation. Improvements in hemocompatibility are achieved with incorporation of both CNT and GBM. Also, both fiber/filler alignment and electrical stimulation, improve cell metabolic activity and elongation. Short term in vivo studies reveal PLA/CBN composites to be biocompatible, and no release of toxic degradation products is found up to 6 months in vitro degradation of PLA/GBM composites. Incorporation of GO has lead to suppression of Escherichia coli and Staphylococcus aureus growth, without compromising the composite biocompatibility. However, there is still no information on antimicrobial activity of these composites on other types of microorganisms or with other types of GBM. Also, long-term in vivo biocompatibility of PLA/CBN composites needs to be assured prior to their clinical use.
Some other relevant topics for future research include obtaining a better understanding of how the fillers physico-chemical properties, and their alignment inside the polymer matrix, affect the composites properties. In situ polymerization of PLA in presence of CBN is a not well developed topic, being worthwhile of further exploration due to the potential for optimization of the degree of interaction and dispersion of CBN in the polymer matrix. Mechanical milling is an increasingly interesting technique for mixing filler nanoparticles with a polymer matrix, but has not yet been reported for producing PLA/CBN composites. This is expected to change in the near future. Finally, emerging technologies, like 3D printing, will surely contribute to the conception of materials appropriate for the broad potential applications of PLA/CBN composites.