**1. Introduction**

The human immunodeficiency virus (HIV) is a global health burden. At the end of June 2019, approximately 24.5 million people were accessing antiretroviral therapy [1]. In efforts to improve access to HIV drugs, simplification of process chemistry, reformulation, dose reduction, inclusion of new drug classes and new therapeutic strategies have been developed and contributed to the reduction of the HIV burden [2]. Efavirenz (EFV) is a non-nucleoside reverse-transcriptase inhibitor (NNRTI) used, in combination, for first line treatment of HIV. EFV is a BCS Class II or poorly soluble and highly permeable compound with an aqueous solubility of <10 mg/mL and low bioavailability of approximately 40% [3]. Caco-2 and intestinal permeability studies suggest EFV is highly permeable [4] and the clinical efficacy may be limited by low solubility. The log P of 4.6 for EFV suggests it may be a good candidate for formulation into a lipid based drug delivery system [5,6].

Lipid-based drug delivery systems (LBDDS) specifically, self-nanoemulsifying drug delivery systems (SNEDDS) are a useful approach for overcoming poor solubility and low oral bioavailability of some drugs [7–9]. The different classes of LBDDS include micro and nanoemulsions, of which the latter are single optically isotropic and metastable liquid solutions with droplet sizes ranging

between 20 nm and 600 nm with some published reports suggesting a maximum particle size of 300 nm and 1000 nm [10–12]. Microemulsions are mixtures of oil, water and surfactant which are single optically isotropic, transparent thermodynamically stable solutions which form spontaneously when thermodynamic variables such as temperature and composition are met with droplet sizes of up to 100 nm [13].

Crude edible vegetable oils are readily available, economic and green chemistry products with functional food properties that exhibit numerous health benefits for patients and have been used in LBDDS [14]. Grapeseed, flaxseed and soybean oil are rich in poly unsaturated fatty acid, such as α-linoleic acid, content which is positively associated with cardiovascular health due to down-regulation of low-density lipoprotein cholesterol production. Diets rich in α-linoleic acid inhibit lymphocyte proliferation and the immune response in healthy humans and thus may be beneficial to individuals that present with autoimmune disorders [15]. Nanoemulsions that require low-energy input are relatively simple and inexpensive to manufacture if an appropriate surfactant mixture for a specific oil phase is used however, research relating to the formation, phase behavior and microstructure of nanoemulsions produced using food grade materials has not been conducted. Crude edible oils are difficult to solubilize in o/w nanoemulsions [16] and low-energy production methods take advantage of the intrinsic physicochemical properties of the components in order to generate sub-micron droplets. The use of a co-surfactant and/or co-solvent may be required to solubilize multicomponent oil or crude cold pressed materials sufficiently, to ensure that single or isotropic phases are formed. Ethanol used in concentrations ranging between 10% and 20% *v*/*v* is acceptable for food applications and can be used as a co-solvent to produce nanoemulsions. Legislation about the use of ethanol as an excipient, particularly for pediatric medicines differs from country to country, consequently formulations should comply with the regulatory requirements in the country of origin [17]. Co-solvents may improve the solubility of the oil, surfactant, co-surfactant and active pharmaceutical compounds while also functioning as penetration enhancers or altering bulk properties such as viscosity, density, refractive index and interfacial tension of aqueous solutions [18].

Nanoemulsions have small droplet sizes that ensure a large interfacial surface area is available to facilitate drug absorption and nanoemulsions or lipid based nanoformulations can enhance intestinal lymphatic transport of drugs and therefore circumvent the hepatic first pass effect [19]. Mechanisms of transport include an increase in membrane fluidity that facilitates transcellular absorption and opening of tight junctions that facilitate paracellular transport [20].

Critical quality attributes (CQA) of nanoemulsions include droplet size (PS), polydispersity index (PDI), zeta potential (ZP) and drug loading capacity (DLC). In addition, there is a need to understand physiological processes such as hepatic uptake and accumulation, tissue diffusion, tissue extravasation and renal excretion as a function of droplet size. Nano carriers of diameter between 100 and 150 nm circulating in blood vessels do not easily leave the capillaries that perfuse tissues such as the kidney, lung, heart and brain. In contrast smaller droplets between 20 and 100 nm may distribute into the bone marrow, spleen and liver sinusoids and may leave the vasculature via fenestrated capillaries in the perfused organs [21,22]. In the case of efavirenz and associated central nervous system adverse effects, small droplet sizes < 100 nm may lead to more pronounced side effects due to the associated increase in blood brain barrier permeability and the pathophysiology of acute and/or chronic CNS disease [23,24]. The PDI describes the uniformity or lack thereof of the size distribution of particles and PDI values < 0.05 are only observed for monodisperse standards whereas PDI values > 0.7 indicate that the sample exhibits a broad size distribution. The ZP can influence the stability of products, cellular uptake and intracellular trafficking of emulsions [25]. Generally, nanoemulsions with a high positive or negative ZP are electrically stabilized while emulsions with a low ZP tend to coagulate or flocculate leading to poor physical stability. High positive values for ZP > 30 mV or negative values < −30 mV tend to exhibit enhanced physical stability. In contrast low values for ZP of <5 mV can lead to droplet agglomeration [26]. The optimization of mixtures to produce a product of predefined specifications through estimation of the effects of formulation components on the mixture can be

achieved using design of experiments (DoE) such as box-Behnken or D-optimal mixture designs with the aid of response surface methodology (RSM). The D-optimal mixed design has been applied to product formulation in the food, pharmaceutical and cosmeceutical industries as a reduced number of experiments are used to generate data for which interactions between variables can be identified using statistical tools thereby avoiding the shortcomings of traditional "one factor at a time" experimental and/or manufacturing approaches [27–29].
