Design of experiments for microencapsulation applications: A review

1. Introduction

The increased demand for added-value products has been substantially affecting trends in the global market. Therefore, new and innovative technological techniques as microencapsulation [62], [82] and nanoencapsulation [111] have been developed.

Microencapsulation and nanoencapsulation are generally defined as a set of technologies that allows to entrap active ingredients also known as core materials using a surrounding material namely as encapsulating or shell material [65]. A clear distinction between nanoencapsulation and microencapsulation is not consensual among authors, especially in terms of size: some authors consider that nanoparticles size varies between 1 and 1000 nm, however other researchers claim that nanoparticles size should range between 1 and 100 nm [60], [120]. Nevertheless, both technologies aim to create a physical barrier to protect the active ingredient from the external environment, allowing a possible controlled release of it. Usually, micro- and nano-technologies are technically similar: some operational conditions are adapted in order to obtain microparticles or nanoparticles [73].

Nanotechnology is considered a promising technology to effectively entrap compounds [10], [68], [111] for a wide range of industrial sectors such as electronics, engineering, energy storage and biotechnology [18], [33], [34], [79], nevertheless microencapsulation is the focus of the present review because countless brand-new and reinvented microencapsulated products have been available on the retail market.

2. Microencapsulation

The microencapsulation technology was first presented by Green and Schleicher in 1950s with a patent registration for the preparation of capsules containing dyes, which were developed to be incorporated into paper for copying purposes [44], [45]. Nowadays, microencapsulation as it was above described, allows to protect sensitive micro-sized substances from the external environment allowing a controlled release of these micro-sized substances ([11], [44], [64]. The active ingredient also termed as core material can be temporary or permanently protected within a shell of a second material, designated as encapsulating or wall material [23], [24].

The resulting products of microencapsulation techniques are designated microparticles (Fig. 1). Microparticles can be distinguished in microspheres or microcapsules [51] by their internal structure and morphology [62] even though, the terms are often used synonymously. Microspheres and microcapsules are differentiated in reservoir systems and matrix systems, respectively [123].

This technological approach has been explored by pharmaceutical (68%), food (13%), cosmetic (8%), textile (5%), biomedical (3%), agricultural (2%) and electronic (1%) industries [25], [44], [69], [117].

Microencapsulation aims to increase the effectiveness of selected substances in industry [36]. Several authors have been discussing the main advantages of applying microencapsulation techniques in different industry sectors [26], [30], [31], [96], [102]. Nevertheless, pharmaceutical and food industries are the main driving forces in microencapsulation advances.

2.1. Microencapsulation in pharmaceutical industry

Currently, research on microencapsulation for pharmaceutical purposes is focused on finding new drug delivery systems (DDS) to obtain products to reach the market, reducing the adverse reactions and side effects, being suitable for the required administration mode, allowing site-specific delivery, increasing shelf-life, improving patient compliance and allowing a possible controlled and sustained release of compounds [2]. Hence, microencapsulation arises as a potential technological strategy to achieve the above-mentioned goals. Microparticles may be constituted by combinations of the active pharmaceutical ingredients (APIs) and biomaterials. Regarding the microencapsulated APIs, these therapeutic agents can have a short half-life, can be quickly hydrolysed or degraded enzymatically in vivo, which is associated with a more strictly therapeutic regimen (multiple administrations). Therefore, microencapsulation techniques protect the API from degradation, allowing these compounds being appropriately released to obtain the required treatment concentration of API over time [81]. Depending on the biomaterial properties, particularly if they are erodible or non-erodible, they can disappear from where they were administrated or remain there throughout the patient lifetime, respectively. Some examples of microencapsulated APIs are presented in Table 1.

Table 1. Examples of microencapsulated ingredients for pharmaceutical industry.

Class of active pharmaceutical ingredient	Active pharmaceutical ingredient	Main goal of the study	References
Antibiotic	Gentamicin	Description of process-related issues of proteins microencapsulation.	[54]
	Erythromycin	Characterization of erythromycin-loaded microspheres by double emulsion solvent evaporation technique.	[89]
	Doxycycline	Development of a controlled release system for doxycycline microencapsulation for human periodontal pocket treatment.	[85]
Enzymes	Lysozyme (group of glycoside hydrolases)	Analysis of stability parameters of the microencapsulated compound (lysozyme).	[90]
Vaccine	SPf66	Evaluation of Ƴ-irradiation on pharmaceutical properties of PLGA based microspheres loaded with SPf66, a malaria preventing vaccine.	[58]
Vaccine	Tatanus toxoid	Evaluation of process-related issues of proteins microencapsulation.	[15]
Anti-cancer agent	Pactitaxel	Preparation and characterization of microparticles loaded with pactitaxel intended for a controlled release.	[97]
	Disodium norcantharidate	Preparation and characterization of poly-caprolactone based microparticles loaded with this anti-cancer agent.	[119]
	Anti-CD40 antibody	Preparation and characterization of anti-CD40 anti-body modified magnetic into poly-caprolactone-polyethylene glycol-poly-caprolactone microspheres.	[42]
Protein	Insulin	Evaluation of process-related issues of proteins microencapsulation.	[15]
	Rhi (recombinant human insulin)	Evaluation the influence of process parameters on Rhi-loaded microparticles size distribution.	[80]
	Recombinant human epidermal growth factor (rhEGF)	Encapsulation and evaluation of rhEGF for chronic gastric ulcer healing.	[47]
	Clonidine	Investigation of a possible sustained release of a hydrophilic molecule for intra-articular administrations.	[41]
	Ovalbumin	Study the mechanism of ovalbumin-loaded microparticles formation by a double emulsion solvent evaporation technique.	[27]
Nucleotide	Deoxyribonucleic acid (DNA)	Preparation of DNA loaded and porous microspheres by leaching of fore formation for antisense therapy.	[4]

An efficient DDS is the one that allows the API to reach the target site, in the required time and for the desired time. Four major factors are considered to achieve an efficient DDS: administration route, pattern of API release, method of delivery and production process also known as formulation process [108]. Many of the non-microencapsulated API are administered repeatedly which makes the therapeutic regimen more frequent and always under medical supervision and as so, microencapsulation arises as a potential drug delivery strategy to overcome multiple issues associated to multiple administrations. Formulated microparticles must be biocompatible, stable, safe and demonstrate predictable degradation kinetics. However, other factors such as chemical modifications on the particle surface can optimize the system and thus be possible to use microencapsulation for drug delivery systems [51].

Nevertheless, there are few microencapsulated pharmaceutical products available on the market [110]. This can be explained with regards to the size control and size distribution is difficult, resulting in low reproducibility of the production process, especially on a large scale. Thus, DDSs are difficult to be approved. Additionally, it is considered in the case of microencapsulation of APIs is difficult to maintain the bioactivity of the therapeutic agent during all processing steps (preparation, storage and release). The APIs can even lose their therapeutic capacity and even increase the unwanted side effects due to deactivation of the therapeutic agent [118]. Despite the difficulties that have been encountered in the implementation of microencapsulation for DDSs, traditional therapies have been progressively replaced by more advanced technologies such as microencapsulation.

2.2. Microencapsulation in food industry

The food industry is the second main driving force for microencapsulation progress. Increasingly demanding consumers and product requirements are the major motivations for microencapsulation research intended to food industry. In fact, demanding consumers have been required the addition of functional ingredients in the final product. Usually, these ingredients are environmental and/or processing instable and as so, microencapsulation arises as technological approach to overcome the above-mentioned problems and therefore obtain an effective protection of these instable ingredients. Additionally, these compounds may be prone to degradation in gastrointestinal conditions and consequently, an effective protection of these ingredients may be required. Functional ingredients can be used to regulate color, flavour or texture of the final product. Additionally, they can be used as preservatives, being possible to extend their shelf life. As well as functional ingredients, bioactive ingredients have been encapsulated to preserve their stability during food processing and storage and additionally, to avoid undesired interactions between other ingredients present in the food matrix which could lead to a faster product degradation and loss of some proprieties. Thus, the unstable bioactive microencapsulated ingredients are protected and kept totally functional. Moreover, microencapsulation allows to potentiate specific flavours and aromas, to mask undesirable odours and tastes or even to increase ingredient bioactivities. Furthermore, microencapsulation can enhance physico-chemical properties of food ingredients in order to allow an easier handling, provide a desired and adequate concentration of the active ingredient, to promote a uniform dispersion of the active ingredient in the food matrix and to avoid undesired reactions [30], [31], [76]).

The microencapsulated active ingredients may be bioactive molecules (e.g. flavouring agents, sweeteners, colorants and vitamins) or living cells as probiotics [37]. Some examples of microencapsulated ingredients for food industry are presented in Table 2. Clearly, bioactive molecules are the most commonly microencapsulated compounds. Many of the microencapsulated bioactive molecules present antioxidant capacity. These molecules are capable to oxidizing themselves instead or before others and consequently, protecting them. They may be used in pharmaceutical and food industries as active ingredients or supplements (preservatives) [88], [91].

Table 2. Examples of microencapsulated ingredients for food industry.

Class of food ingredient	Application field	Examples	References
Bioactive molecule	Flavouring agent	Lemon oil	[13]
		Peppermint oil	[70]
		Vanilla oil	[122]
	Sweetener	Aspartame	[93], [94]
		Sucralose	[93], [94]
		Xylitol	[100]
	Colorant	Annatto	[29]
		β-carotene	[35]
		Turmeric oleoresin	[35]
	Fat-soluble vitamins	Vitamin A	[61]
		Vitamin D2	[106]
		Vitamin E	[104]
	Water soluble vitamins	Vitamin C	[30], [31]
Living cells	Probiotics	Bifidobacterium spp.	[49]
		Bifidobacterium longum	[5]
		Lactobacillus acidophilus	[38]

It should be pointed out that microencapsulation of bioactive molecules, functional ingredients or living cells for food industry must consider several factors such as technological concerns (manufacturing and storage properties), economic feasibility and consumers satisfaction [3].

2.3. Challenges in microencapsulation in pharmaceutical and food industries

Much of inherent challenges of microencapsulation of ingredients are the same for both pharmaceutical and food industries and as so, this topic is going to be addressed for both industries.

Several authors have been discussing the main advantages of applying microencapsulation techniques in pharmaceutical and food industry sectors. Numerous difficulties and disadvantages to industrial application of microencapsulation techniques have been pointed out for both application fields such as the poor entrapment of the active compound, the impossibility to scale-up some processes [46], [66], [103], the requirement of multi-step processes [17], [125] the downstream high concentration of undesired sub-products or residues [32], [99], [101], [114], [121], [124], [126], the long-time requirement of some processes (time-consuming processes) [50], [109], the requirement of high energy inputs and the demand of complex equipment [53], [63], [105], [107].

Many parameters can affect microencapsulated products and their final characteristics. These parameters may be divided into three classes: properties of materials, formulation parameters and operating conditions (Table 3). The main properties affected are microparticles mean size, particle size distribution, microparticles surface morphology, product yield and encapsulation efficiency. The final properties of microparticles may affect the active compound release rate [77].

Table 3. Parameters affecting microparticles final properties.

Type of factor	Factor	Studies	References
Properties of materials	Wall material	Evaluation the effect of wall materials on fish oil microencapsulation by spray-drying.	[92]
		Assessment the effect of wall material type and oil load on flaxseed oil microencapsulation by spray-drying.	[87]
		Investigation of the effect of different wall materials on red-fleshed pitaya (Hylocereus polyrhizus) seed oil microencapsulation by spray-drying.	[78]
		Evaluation the use of gum arabic, maltodextrin and a modified starch as wall materials on cardamom oleoresin microencapsulation by spray-drying.	[71]
		Assessment the effect of wall material on extra-virgin olive oil microencapsulation by spray-drying.	[22]
	Core material	Analysis of the solubility of core materials in aqueous polymeric solutions on curcumin microencapsulation by coacervation.	[7]
Formulation parameters	Viscosity	Interpretation of the influence of viscosity and other physicochemical properties of acai (Euterpe oleracea Mart.) microencapsulation by spray-drying.	[115]
	Type of solvent	Analysis of the influence of using ethyl acetate as a dispersed solvent on microspheres properties.	[98]
	Use of additives	Investigation of the effect of salt addition on proteins microencapsulation.	[127]
Operation conditions	Stirring speed	Evaluation of the influence of stirring speed on drug release from microcapsules.	[59]
	Temperature	Consideration of the influence of spray-dryer air temperatures on mandarin oil microencapsulation.	[20]
	Temperature	Assessment of the influence of air inlet temperature on the microencapsulation of flaxseed oil by spray drying.	[116]

3. Design of experiments for microencapsulation applications

Traditionally, a common approach to analyse which is the main influencing factors in microencapsulation area is studying the influence of one variable while the others remain constant. This methodology known as one-variable-at-time (OVAT) or interchangeably one-factor-at-time (OFAT) leads to non-optimized final products or processes [40]. This may be explained by factor interactions that OVAT methodology does not count: different variables can interact and be responsible for a specific system behaviour [67]. Another disadvantage of OVAT methodology is the relative huge number of experiments required to be performed which makes OVAT methodology a material, reagent and time consuming approach [19].

Recently, multivariate statistical approaches as design of experiments (DOE) have been used to overcome the main disadvantages of OVAT procedures [28]. Among the most relevant multivariate statistical methods used in microencapsulation field, the surface response methodology (RSM) is the most used one. The application of RSM for microencapsulation optimization processes have been described essentially in pharmaceutical and food research sectors.

3.1. Response surface methodology implementation

The response surface methodology is defined as a set of mathematical and statistical approaches to express relationships between factors and responses [9] . Factors also known as independent variables are independent parameters that can be independently changed. On the other hand, responses or dependent variables are the measured values from the experiments performed. The experiments belong to an experimental domain which is the experimental area under study. The maximum and minimum levels of experimental variables define the limits of the domain. Additionally, levels of a variable must be defined. Levels of variables are the values at which the experiment should be performed [12]. Unlike OVAT methodology, RSM outlines the effect of factors alone or in combination generating a mathematical model. The mathematical relationship between factors and responses is given by Eq. (1)[9].(1) $η = f (x_{1} \dots x_{n}) + ε$ where η stands for the response, x1 , … , xn, represents the factors being n the number of independent variables, f stands for the unknown function of η and εaccounts for the statistical errors not considered in f. Generally, the sources of εare measurement errors considering that ε follows a normal distribution with mean and variance equal to zero.

The RSM optimization processes performed in four stages as depicted in Fig. 2.

3.1.1. Screening experiments: determination of factors and their levels

As was previously described many parameters can affect microparticles production and their properties. In many systems are arduous to determine the effects of all parameters; therefore, screening experiments may be a useful tool to identify factors and their interactions with relevant effects on the process. Generally, full or fractional designs are used in screening experiments, essentially due to their efficiency: in the early stages of the optimization processes, a limited number of experiments may be representative [48]. After the selection of the most influencing parameters follows the determination of factors levels. Incidentally, this is a key point in the optimization process: levels incorrectly chosen can lead to a non-optimized response. The selected factors during the screening process usually have different units or ranges. Many examples can be given for each type of factor as presented in Table 2. If the factors under study are the wall material: core material ratio (% w/w), viscosity (mPa·s) and operation temperature (°C), then a normalization process has to be performed because the three parameters have different units and therefore they are not directly comparable. Consequently, a normalization process shall be performed before a regression analysis. Factors are coded (normalized) according to the following equation (Eq. 2):(2) $X = \frac{x - (x_{\max} + x_{\min}) / 2}{(x_{\max} - x_{\min}) / 2}$ where, X stands for the normalized factor, x, the natural variable and xmax and xmin, the maximum and minimum values for x, respectively. Each factor varies between − 1 and + 1. The units of the selected factors are not more significant and therefore they are normalized.

The most common approaches in screening experiments are the two-level full-factorial designs, also known as 2k-designs where the 2 stands for the number of factor levels and k for the number of factors each with a high and low value. In fractional designs, also known as 2k − p, the number of experiments is reduced by p [9], [12], [48].

The choice of the screening design may affect the final response, actually, the screening design should be chosen regarding the goals and limitations of the optimization process.

3.1.2. Choice of the experimental design

The simplest mathematical model that can be used in RSM is a linear function (Eq. 3).(3) $y = β_{0} \sum_{i = 1}^{k} β_{i} x_{i} + ε$ where, β0 is a constant term, βi are the coefficients of the linear parameters, xistands for factors, ε represents the residues and k stands for the number of variables. For an accurate application, the responses should be fitted in the linear model. However, a linear RSM model does not account for any curvature of the response. For factor interactions analysis, a second order model must be used (Eq. 4).(4) $y = β_{0} \sum_{i = 1}^{k} β_{i} x_{i} + \sum_{1 \leq i \leq j}^{k} β_{ij} x_{i} x_{j} + ε$ Where βij stands for the coefficient of interactions between factors.

If a maximum or a minimum has to be determined, a quadratic term ought to be added into second order model, giving rise to the following model (Eq. 5):(5) $y = β_{0} \sum_{i = 1}^{k} β_{i} x_{i} + \sum_{1 \leq i \leq j}^{k} β_{ij} x_{i} x_{j} + \sum_{i = 1}^{k} β_{ii} x_{i}^{2} + ε$ where, βii stands for the coefficient of the quadratic factor. According to Eq. 5, the estimation of factors and consequently, the optimization process has to be carried out in at least three factor levels for all factors.

The model presented in Eq. 5 can be presented in a matrix notation (Eq. 6)(6) $y = Xβ + ε \Leftrightarrow [\begin{matrix} y_{1} \\ ⋮ \\ y_{n} \end{matrix}] = [\begin{matrix} 1 \\ ⋮ \\ 1 \end{matrix} \begin{matrix} x_{11} \\ ⋮ \\ x_{n 1} \end{matrix} \dots \begin{matrix} x_{1 k} \\ ⋮ \\ x_{nk} \end{matrix}] [\begin{matrix} β_{0} \\ ⋮ \\ β_{k} \end{matrix}] + [\begin{matrix} ε_{1} \\ ⋮ \\ ε_{n} \end{matrix}]$

The least squares method (MLS) allows to solve Eq. 6 matrix system. The description of this method is not within the scope of this paper.

Nowadays, a huge variety of computer packages are available to solve RSM problems, so much so that computer packages may differ on specific inputs like the number of runs, blocks or even how experimental points are selected [9].

The most known second-order models used in RSM are the central composite design (CCD), the face-centred composite design (FCCD) and the central rotatable composite design (CRCD). These systems are differentiated according to number of experiments required to run.

3.1.3. Assessment of the predicted model

A mandatory condition to use RSM is the predicted model has to describe the experimental domain. The reliability of the fitted model is usually tested through the analysis of variance (ANOVA). The ANOVA tests compare the variation due to changes in combinations of factors levels with the variation caused by measurement errors (random errors) of responses. Hereby, this evaluation is essential to determine if the model fits and describes the experimental domain [21], [43]. Variation sources can be compared using a Fisher distribution (F-test) or alternatively, a Student's distribution (t-test). Variations associated to the model and not to random errors, arise if the F-probability is < 0.05 (correspondent to a 95% confidence level). A t-test, explores which factors and/or interactions have statistical meaning. If t-probability is smaller than 0.05, the factor or the interaction between factors are considered to be significant [52].

3.1.4. Determination of the optimal conditions

Linear models (represented generically on Eq. 3) allow to specify in which direction the original design should be studied in order to obtain optimal conditions. Nevertheless, if the experimental region cannot be changed, the optimization study should find out the optimal operational conditions by visual inspection [12].

For quadratic models, the optimal conditions correspond to a maximum, minimum or a saddle point (critical points). The coordinates of a critical point are calculated deriving the quadratic form of Eq. 5 (Eq. 7) in order to x1and x2and then equating them to zero. The coordinates of the critical point are obtained solving Eqs. (8), (9) in order to find the values of x1 and x2.(7) $y = b_{0} + b_{1} x_{1} + b_{2} x_{2} + b_{11} x_{1}^{2} + b_{22} x_{2}^{2} + b_{12} x_{1} x_{2}$ (8) $\frac{\partial y}{\partial x_{1}} = b_{1} + 2 b_{11} x_{1} + b_{12} x_{2} = 0$ (9) $\frac{\partial y}{\partial x_{2}} = b_{2} + 2 b_{22} x_{2} + b_{12} x_{1} = 0$

3.1.5. Graphical representation of the model equation

The model equation predicted using a RSM can be visualized on response surface plots and their related contour plots. The response surface plots are three-dimensional graphs that represent the relationship between factors and the response. Surface response plots are n-dimensional surfaces in the (n + 1)-dimensional space. If the system is represented by three or more variables, the graphical representation is only possible if one or more variables are considered constant.

The two-dimensional depiction of the response surface plot is the related contour plot. The shape of lines in contour plots may predict the response type. A maximum or minimum response corresponds to ellipses or circles around the plot centre. Further interpretations can be performed comparing both plots regarding the predicted model within the experimental domain. Concerning Fig. 3, the surface plot (A) and the contour plot (B) are represented in a general two variables optimization of a maximum within the experimental region. Regarding Fig. 4, the surface (A) and the contour (B) plots of a maximum point included in the experimental domain with a graphical plateau in relation to x1are depicted. The variation of x1 levels does not affect the response. On the other hand, Fig. 5 is presented a maximum response outside the experimental domain. The initial design should be displaced in order to obtain the maximum response inside the experimental area. A minimum response within the experimental domain is displayed on Fig. 6, whereas a saddle point is depicted in Fig. 7. Despite the result of Eqs. (8), (9) are found in a saddle point, this is not an optimal response. The optimum response can be found through visual inspection. Graphical representations presented in this article were made using the JMP 13 statistical software.

4. Applications of response surface methodologies in microencapsulation techniques

Innovate experimental designs and optimization methodologies have been applied in microencapsulation techniques to overcome the main barriers for the microencapsulation industrial application and moreover to improve and/or reformulated micro-enclosed products. Through representative examples, the application of RSM optimization studies in pharmaceutical and food industries are reviewed and listed in Table 4, Table 5. Likewise, studies about the application of design of experiments are comprehensively and critically reviewed and discussed.

Table 4. Examples of DOE applications within microencapsulation field for pharmaceutical industry.

Application	ESD EOD (NOE)	EOD			References
Active ingredient Coating material Additives	ESD EOD (NOE)	Factor(s)	Response(s)	Optimal formulation	References
Aciclovir Poly(d,l-lactide-co-glycolide) Gelatine	52 CRCD (30)	-Aciclovir content (mg) -Gelatine content (mg)	-PY (%) -EE (%) -Initial burst release -Cumulative amount of aciclovir released from 1 to 14 days (μg·mg of microspheres− 1) -Cumulative amount of aciclovir released at the end of the assay (μg·mg of microspheres− 1)	-Aciclovir content: 80 mg -Gelatine content: 80 mg -PY: 70.14% -EE: 70.77% -Initial burst release: constant release rate during 63 days -Cumulative amount of aciclovir released from 1 to 14 days: 26.96 μg·mg− 1of microspheres -Cumulative amount of aciclovir released at the end of the assay: 118.83 μg·mg− 1of microspheres	[84]
Ibuprofen PCL–PEG–PCL n.d.	24 LM (16)	-Shaking speed (rpm) -Time of contact (h) -PVA concentration (%) -IBF concentration (%)	-EE (%)	-Shaking speed: 200 rpm -Time of contact: 0.5 h -PVA concentration: 0.1% -IBF concentration: 35% -EE: 88.86%	[8]
Acetaminophen TA-CMC Colloidal silicon dioxide	25–1 CRCD (16)	-Feed rate (mL·min− 1) -Inlet temperature (°C) -Drug concentration in the feed (g·L− 1) -Polymer concentration (g·L− 1) -Additive concentration (g·L− 1) -SiO2concentration (g·L− 1)	-PY (%) -Residual moisture content of spray-dried product (%)	-Feed rate: 20 mL·min− 1 -Inlet temperature: 140 °C -Drug concentration in the feed: 6 g·L− 1 -Polymer concentration: 6 g·L− 1 -Additive concentration: 1.8 g·L− 1 -SiO2concentration: 8 g·L− 1 -PY: 82.5% -Residual moisture content of spray-dried product: 1.03%	[16]
Acetaminophen TA-CMC Colloidal silicon dioxide				-Feed rate: 20 mL·min− 1 -Inlet temperature: 140 °C -Drug concentration in the feed: 6 g·L− 1 -Polymer concentration: 6 g·L− 1 -Additive concentration: 1.8 g·L− 1 -SiO2concentration: 8 g·L− 1 -PY: 82.5% -Residual moisture content of spray-dried product: 1.03%
Acetaminophen OA-CMC Colloidal silicon dioxide				-Feed rate: 20 mL·min− 1 -Inlet temperature: 140 °C -Drug concentration in the feed: 6 g·L− 1 -Polymer concentration: 6 g·L− 1 -Additive concentration: 1.8 g·L− 1 -SiO2concentration: 8 g·L− 1 -PY: 80.8% -Residual moisture content of spray-dried product: 1.02%
Acetaminophen MCC Colloidal silicon dioxide				-Feed rate: 30 mL·min− 1 -Inlet temperature: 160 °C -Drug concentration in the feed: 10 g·L− 1 -Polymer concentration: 10 g·L− 1 -Additive concentration: 0.3 g·L− 1 -SiO2concentration: 18 g·L− 1 -PY: 86.8% -Residual moisture content of spray-dried product: 1.02%

n.d. – not defined; CMC - Carboxymethyl cellulose; CRCD – central rotatable composite design; EE – encapsulation efficiency; EOD – experimental optimization design; ESD – experimental screening design; IBF – ibuprofen; LM – linear model; MCC – Microcrystalline Cellulose; NOE – number of experiments; OA – oxalic acid; PCL – polycaprolactone; PEG – polyethylene glycol; PY – product yield; TA -tartic acid.

Table 5. Examples of DOE applications within microencapsulation field for food industry.

Application	ESD EOD (NOE)	EOD			References
Active ingredient Coating material Additives	ESD EOD (NOE)	Factor(s)	Response(s)	Optimal formulation	References
Fish oil n.d. n.d.	33 FCCD (17)	-Aqueous phase content (%) -Oil concentration in total solids of emulsion (%) -Emulsification time (min)	-ExE (%) -EE (%)	-Aqueous phase content: 87.12% -Oil concentration in total solids of emulsion: 10.82% -Emulsification time: 13.23 min -ExE: 6.17% -EE: 88.67%	[1]
Lemon myrtle oil n.d. n.d.	n.d. FCCD (40)	-Feed concentration (% w/w) -Oil concentration (% w/w of feed concentration) -Outlet drying air temperature (°C)	-Oil retention (%) -Surface oil content (mg/100 g of podwer)	-Feed concentration: 40% w/w -Oil concentration: 18% w/w of feed concentration -Outlet drying air temperature: 65 °C Oil retention: 83.82% -Surface oil content: 23.04 mg/100 g of powder	[56], [57]
Holy basil essential oil n.d. Gelatin	n.d. FCCD (13)	-Gelatine concentration (% w/v) -Holy basil essential oil amount (mL)	-PY (%) -EE (%) -Final oil content (%)	-Gelatine concentration: 11.75% w/v -Holy basil essential oil amount: 31 mL -PY: 98.80% -EE: 95.41% -Final oil content: 66.50%	[112]
Rosemary essential oil n.d. n.d.	52 FCCD (12)	-Wall materials concentration (%) -Oil load % (w/w)	-Moisture content -Wettability (s) -Hygroscopicity (%) -Total oil (%) -Oil retention (%)	-Wall materials concentration: 20.9% -Oil load: 29.4% w/w -Wettability: 1315 s -Hygroscopicity: 11% -Total oil: 11% -Oil retention:39%	[39]
Flaxseed oil n.d. Lecithin (emulsifier) Xanthan gum (emulsifier)	n.d. FCCD (17)	-Concentration of lecithin (% w/w) -Concentration of oil loading (% w/w) -Concentration of xanthan gum (% w/w)	-EE (%) -Flaxseed oil droplet (nm)	-Concentration of lecithin: 1.14% w/w -Concentration of oil loading: 22.78% w/w -Concentration of xanthan gum: 0.10% w/w -EE: 92.3% -Flaxseed Oil Droplet: 446.9 nm	[87]
Peanut sprout extract Triacylglycerol (first coating material) Maltodextrin (secondary coating material) Polyglycerol polyricinoleates (primary emulsifier) Polyoxyethylene sorbitan monolaurate (secondary emulsifier)	24 CCD (31)	-Core material:triacylglycerol ratio -Core material:maltodextrin ratio -Polyglycerol polyricinoleates concentration (% w/v) -Polyoxyethylene sorbitan monolaurate concentration (% w/v)	-PY (%)	-Core material:triacylglycerol ratio: 1:2 -Core material:maltodextrin ratio: 1:1.23 -Polyglycerol polyricinoleates concentration: 1.25% w/v -Polyoxyethylene sorbitan monolaurate concentration: 1.21% w/v -PY: 98.7%	[74], [75]
Starch oleate derivatives from native corn n.d. n.d.	53 CCD (20)	-Oleic acid content (g) -Time of the process (s) -Temperature of the process (°C)	-Degree of substitution	-Oleic acid content 1.2 g -Time of the process: 201 min. -Temperature of the process: 160 °C -Degree of substitution: 0.021	[72]
Starch oleate derivatives from hydrolysed corn starch with longer glucose chain n.d. n.d.				-Oleic acid content 1.53 g -Time of the process: 334 min. -Temperature of the process: 150 °C -Degree of substitution: 0.0353
Starch oleate derivatives from hydrolysed corn starch with shorter glucose chain n.d. n.d.				-Oleic acid content 1.36 g -Time of the process: 300 min. -Temperature of the process: 129 °C -Degree of substitution: 0.0503
Probiotics of raspberry juice n.d. Maltodextrin(carbon source and parameter to assess the prebiotic potential)	n.d. CCD (20)	-Inlet temperature (°C) -Total solids: maltodextrin ratio (product of starch hydrolysis) -Feed rate (mL·min− 1)	-Recovery (%) -Survival percentage (%) -Color (ΔE)	-Inlet temperature: 100 °C -Total solids: maltodextrin ratio: 1:1 -Feed rate: 40 mL·min− 1 -Recovery: 48.7% -Survival percentage: 87.17% -Color: 57.21	[6]

n.d. – not defined; CCD – central composite design; EE – encapsulation efficiency; EOD – experimental optimization design; ESD – experimental screening design; ExE - exergy efficiency; FCCD – face centred composite design; NOE – number of experiments; PY – product yield.

Henceforth, for both research fields reviewed (Table 4, Table 5), the specific application (the active ingredient, the coating material and the use of additives) is firstly presented. Afterwards, parameters related to the experimental optimization design (EOD) are reported namely the type of experimental screening design (ESD), the type of EOD and the number of runs performed. A widely variety of factorial designs can be performed. Notwithstanding, full screening experiments do not have to be performed: a fractional factorial design may describe the system under study. Once a fractional design is applied, a reference for the fractional design is reported in both tables (application column) considering a general representation of xk − p. Therefore, the number of runs can be less than xk if a fractional design is considered or more if replicates are considered in modeling. Posteriorly, the factors (x) and the response(s) are presented. Finally, the optimal settings are described: the desired values for both factors and responses are presented.