The RO process was firstly proposed by Reid of Florida University in the beginning of 1953 [23,36]. After then, the workers at Florida University announced a cellulose diacetate (CDA) film with outstanding semipermeable performance in which salt rejection was 96% or more, but its water permeability was still very small and not applicable for commercial application [37,38]. However, their pioneering work clearly showed a possibility of conducting a RO process using a semipermeable membrane for seawater desalination. After that, a remarkable progress was achieved by Loeb and Sourirajan of the University of California, Los Angeles (UCLA); they prepared an asymmetric-type CDA membrane that had quite a thinner salt rejection layer and a loose support sublayer using the same material of CDA [39,40]. After this invention, accelerated research on cellulose-acetate (CA)-based materials for practical applications was conducted to improve the membrane performance in the industry [41]. The details in the early history of RO membrane development are available elsewhere [37,41,42,43,44,45]. As CA-based polymers, cellulose triacetate (CTA) [46], cellulose acetate butylate [47], CA blends such as CDA with CTA [45,48], the cross-linked CA [49], and the composite membrane (a thin CA film on a porous support consisting of the other polymer) [50] were demonstrated in addition to the CDA. Towards its commercialization, among them, CTA membranes were well developed by some companies because CTA had a better stability in a wider range of temperatures and pH, higher resistance to chemicals and biological attack, and better membrane performance compared to the initial CDA, even though CTA was more difficult to handle in the membrane preparation than CDA [41]. Then, the preparation conditions such as dope solution composition and post-treatment condition were also further developed and optimized for the applicable industrial production. For example, the dope solution contained CA polymer, acetone, and magnesium perchlorate at −10 °C casted on a glass plate. The casted solution was set for a certain time as an evaporation procedure and then immersed in cold water [39,40]. Since the first method was difficult, an easier preparation procedure with high membrane performance was required in the industry [41]. Although various types of solvent for the dope and coagulation bath were then studied (such as acetone, acetic acid, formamide, chloroform, pyridine, dioxane and its mixtures), these solutions were difficult to handle in the industry. Therefore, polar aprotic solvents such as N-methyl pyrrolidone (NMP), dimethylacetamide (DMAc) and dimethylsulfoxide (DMSO) were preferably used as the alternative solvents because of their easy handling in the industrial production.

In addition to the membrane development, subsequent module configurations were also developed, such as tubular [42,51], plate-and-flame [52], spiral-wound [53,54], and HF [55]. Among them, the HF configuration had superior advantages to the spiral-wound in the total membrane area, the packing density, and the module performance if the original membrane performance was the same. The spiral-wound type CA membrane was commercialized by many companies and used as the industry standard through the 1960s to mid-1970s [37,56]. However, CA-based membranes were still not widely used for the large-scale desalination purpose at that time [57]. The CA-based spiral one was mainly used for pure water production for industrial processes and ultrapure water production in the semiconductor industry [57]. After that, in the case of the spiral-wound type, an alternative membrane was found via the pioneering work by Cadotte from 1972 [56,58], which was based on a situ condensation method using monomeric amine and monomeric acid halide. This innovative invention made a rapid shift from the CA-based spiral-wound to the PA-based one because of the preferable features of the PA-based membrane, such as higher flux and higher pH resistance.

With the purpose of achieving seawater desalination in mind, a significant portion of research and development was carried out using a CTA-HF membrane and module by Dow Chemical based on a research contract with the Office of Saline Water in the U.S. Department of the Interior [37]. Their development results of the RO module for brackish water desalination and for seawater desalination were published in 1970 and 1974, respectively [59,60], and the CTA-based HF RO module for low-pressure brackish water desalination was subsequently marketed in 1974 [37]. The research developments for the CA-based HF were also conducted by other companies such as Monsanto, Toyobo, and others, in addition to Dow Chemical. After that, following the early work by Dow Chemical, Toyobo successfully developed a CTA-HF RO module for high-pressure one-pass seawater desalination and subsequently marketed it in 1979 [37,45,61]. The CTA-HF RO module was often used in the fishing industry as a small-scale desalination apparatus on a shipping boat [62].

After these efforts, because of the rapid increasing demand for drinking water in the world, RO membranes have been acceptably and widely used for the large-scale seawater desalination purpose since the 1990s; the total global cumulative contracted capacities since 1965 were about 14, 25, and 60 million m³/day in 1990, 2000, and 2010, respectively [1]. In the case of the HF type, only CTA is still used for the RO membrane for seawater desalination, unlike the spiral-wound type, because of its competitive total module performance of CTA-based HF membrane module compared with the PA-based spiral-wound one [29].