2.1. Wastewater

The synthetic wastewater was composed of ammonium chloride in distilled water. The targeted NH₄⁺-N concentration was 1250 mg-N/L. The initial pH was adjusted to 9 using 10 N NaOH for the synthetic wastewater. The synthetic wastewater was used to verify the effects of temperature on water transport and ammonia transfer, i.e., the SAT value. The SAT value of synthetic wastewater was compared to those of ADS and HU. The ADS was obtained from Seonam sewage treatment plant in the Republic of Korea. Source-separated HU was collected from Water Quality and Treatment Lab, School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), Republic of Korea. For ADS and HU, suspended solids (SS) in large size were settled down by gravity for 1 hr and removed. For the further removal of SS, filtration with filter paper of 1.2 um pore size (GF/C, Whatman, UK) was conducted as pretreatment to mitigate the fouling on the MD membrane. The pH was adjusted with 35-37 % HCl for the ADS and HU. The original pH of ADS was 6.5 and it was acidified to pH 4.0. The HU showed higher pH of 8.8 and the pH was adjusted to 6.0 and 4.0 in two steps. The initial and adjusted pH, NH₄⁺-N, TN, TP and conductivity before and after the acidification are shown in Tables 3 and 5 for the ADS and HU, respectively. Importantly, the ammonia concentration of acidified ADS and HU at pH 4 were 571 and 4248 mg-N/L, respectively, which may play the major control factor in differentiating the SAT value during the DCMD operation. The pretreated wastewater were stored at 4 °C before use.

2.2. Membrane distillation operation

DCMD experiments were carried out as shown in Fig. 1. The flat sheet hydrophobic membrane was located in the horizontal module that was connected with feed and permeate tanks. The feed and permeate streams were flowing counter-currently by gear pumps. The streams were circulated until the experiments were finished as a batch process. The effective membrane area was 0.003 m². PTFE/PP (PTF045LD0A) with a pore size of 0.45 μm and a thickness of 89 μm was used as a hydrophobic barrier. Only the transportation of water vapor and gas was allowed and the liquid penetration through the membrane pore was prevented by the hydrophobicity of PTFE/PP. Both solutions were circulated to their reservoir tanks. The feed and permeate temperatures were controlled by heating and cooling water baths (WCR-P22, Daihan Scientific, Republic of Korea). The flow rates were adjusted at 2 liters per minute in both hot and cold streams. To verify the effect of temperature, various temperature levels of 40, 50, 60 and 70 °C for feed wastewater was tested while the temperature of the permeate side was fixed at 20 °C using synthetic wastewater containing 1250 mg-N/L of NH₄⁺. The DCMD was operated for 2.5 hr for the synthetic wastewater. ADS and HU were treated at 60 °C and 20 °C for feed and permeate sides, respectively, for 24 hr. Flux and conductivity were measured for both sides every 4 hr. The pH, NH₄⁺-N, TN and TP were analyzed for initial and final samples. After the acidification, the NH₄⁺-N and TN of ADS were mostly the same (Table 3). However, HU showed lowered NH₄⁺-N and TN values due the large dilution by HCl and volatilization of ammonia during the storage (Table 5).

Fig. 1. Schematic diagram of the experimental set-up for direct contact membrane distillation system.

2.3. Analysis

The performance of MD is evaluated by the specific ammonia transfer and permeation flux. The specific ammonia transfer (SAT) can be expressed as follow:

where TAN_transfer is the transfer of total ammoniacal nitrogen and H₂O_transfer is the transfer of water (^{Tun et al., 2016}). To verify the effects of free ammonia on the water flux and ammonia transport, free ammonia was calculated using Eq. 2.

The data of permeation flux and the conductivity measured by an analytical balance (CUX6200H, CAS corporation, the republic of Korea) and a HQ40d portable conductivity meter (HQ40d, Hach, USA) were deposited in a data logging system. The pH of feed and permeate were measured by a portable pH meter (Accumet research AR15, Fisher Scientific, USA). The initial and final ammonia concentrations were analyzed by using Kjectec Auto 2300 system (Auto 2300 system, FOSS, Denmark). Total nitrogen (TN) and total phosphate (TP) were quantified by using the Hach system (26722-45 and 27426-45, respectively, Hach, USA) and UV-VIS Spectro-photometer (DR 5000-02, Hach, USA). The morphology and composition of the fouling layer were studied using scanning electron microscopy coupled with the energy dispersing spectrometry (FE-SEM-EDS) (S-4200, HITACHI, USA).

3.1. Temperature effect on water and ammonia transfer

The hydrophobic PTFE flat sheet membrane was located in the horizontal module that was connected to the feed and permeate tanks in the DCMD system. The water vapor pressure difference between the MD membrane is the main driving force for the water flux in the MD system. The initial pH was set to 9 for all experiments. To determine the effect of temperature on the water flux, various temperatures of 40, 50, 60 and 70 °C were applied while the permeate temperature was fixed at 20 °C in all experiments. The increase in the feed temperature resulted in the enhanced water flux (Fig. 2). The initial fluxes were 18, 31, 51 and 75 L/m².hr for 40, 50, 60 and 70 °C, respectively. From 1 hr of operation, the water flux was stabilized and became steady.

Fig. 2. Flux variation with respect to temperature of feed solution.

The total masses of produced water were 74, 174, 317 and 518 g for 40, 50, 60 and 70 °C of the feed temperature, respectively. The transferred ammoniacal nitrogen to the produced is shown in Table 1. The ammonia in the feed at a concentration of 1250 mg-N/L of NH₄⁺ transferred to the permeate with the maximum amount of 0.89 g at 70 °C. High feed temperature caused the efficient volatilization of ammonia. Consequently, the partial pressure difference motivation of the ammonia transfer through the hydrophobic pores of the MD membrane. Although the transferred ammoniacal nitrogen was increased with the feed temperature, the SAT value was reduced due to the faster water transfer than the ammonia transfer. The smallest SAT value was found as 1.71 × 10^-3 for 70 °C. A high feed temperature provides additional advantages. The liquid viscosity reduces with high temperature. Also, the improved fluidity of feed enhances turbulent movement to mitigate the temperature polarization near the MD membrane. Based on these results, the high feed temperature was applied in this study to obtain low SAT values for real wastewaters. However, instead of 70 °C, a moderate 60 °C was applied as the feed temperature to avoid rigorous crystallization of the inorganic components on the MD membraned due to the reinforced polarization of inorganic compounds. Actually, severe fouling of CaCO₃ scaling occurs due to the feed temperature (^{He et al., 2008}).

3.2. Effects of acidification on the water flux and permeate quality

To mitigate the ammonia transfer, ADS and HU were acidified before the MD process. The water flux and permeate conductivity correlated with the initial pH of strong nitrogenous feed were investigated for ADS (Fig. 3). The ADS showed near-neutral acidity and the pH was lowered from 6.5 to 4.0. The temperature difference between 60 °C and 20 °C for feed and permeate sides, respectively generated effective water flux (Fig. 3a). The water initial flux of the original ADS feed solution was round about 40 L/m².hr which is similar to the flux for the synthetic wastewater in Fig. 2 for the 60 ~ 20 °C temperature difference. After 24 hr, the flux for the original ADS was significantly declined to 70.0 % (27.7 L/m².hr) of the initial water flux. In the acidified condition of pH 4, the initial water flux was stabilized after 8 hr of operation. The average water flux was steady as 37.6±1.1 L/m².hr which implies reduced fouling at the acidic condition. However, the conductivity was worsen when the ADS was acidified (Fig. 3b). The increase of permeate conductivity according to acclimated permeate volume for original and acidified anaerobic digestion supernatant is shown in Fig. 4. The conductivities were almost saturated as 83.2 and 136.7 μS/cm when the acclimated permeate volume reached 0.98 and 0.86 L for the original and acidified ADS, respectively.

Fig. 3. Water flux (a) and conductivity (b) for original and acidified anaerobic digestion supernatant during the membrane distillation operation.

Fig. 4. The increase conductivity of permeate according to the acclimated permeate volume for original and acidified anaerobic digestion supernatants.

The relationship between the acidification and conductivity of permeate is controversial. Actually, the positive effect of low pH on the permeate quality of a DCMD system for ADS was found by monitoring conductivity (^{Yan, Liu et al., 2019}). In this study, it was suggested that the difference of the final conductivity of permeate attributed to the initial conductivity of ADS depending on the acidification. The higher initial conductivity of ADS may lead to the higher conductivity in the final permeate. For ADS in this study, the initial conductivity before and after the acidification were 20410 and 23000 μS/cm, respectively (Table 3). The addition of 35-37 % HCl for acidification resulted in the higher initial conductivity of ADS. For this reason, the final conductivity of the acidified ADS was 125 μS/cm which is larger than that of the original ADS (89 μS/cm). However, the conductivity rejection efficiency of the DCMD system, i.e., comparison of the final conductivity of the permeate and the initial conductivity, were similar as 99.6 % and 99.5 % for the original and acidified ADS, respectively.

As in Table 2, the amounts of produced water were similar. However, the SAT value was much lower at the acidic condition (2.33 × 10^-6) than that of the original ADS (1.87 × 10^-5) due to the low ammonia partial pressure in the permeate side. Also, the SAT value was much lower than the SAT value of synthetic wastewater at the same temperature condition (2.48 × 10^-3) due to lower ammonia concentration (571 mg-N/L) and acidified condition (pH 4.0). Consistently, the acidification offered high quality of permeate in terms of NH₄⁺-N (Table 3). The final ammonia concentration in the permeate was reduced from 19 to 2.3 mg-N/L by the acidification. Therefore, it can be concluded that acidification is an effective means to obtain a better water flux and controlled ammonia transfer for ADS. However, the TN and TP showed an insignificant difference. Besides, the final conductivity of the acidified ADS (125 μS/cm) was larger than that of the original ADS (89 μS/cm). This implies that the acidification of ADS can facilitate the transfer of the impurities rather than ammonia.

The acidification was applied to the weakly basic HU (pH 8.8) in two steps of pH 6.0 and 4.0 (HU6 and HU4, respectively). For the water flux, the acidification showed a clear negative effect on the water flux (Fig. 5a) which is distinguished from the acidified ADS (Fig. 3a). The final water fluxes for HU6 and HU4 were 18.4 % and 40.9 % lower than the original HU, respectively. The reduced water flux can attribute to the serious fouling by organic and inorganic impurities on the MD membrane deteriorating the water flux in the acidic condition. Interestingly, the low pH prevented the rise in the conductivity of the permeate (Fig. 5b). The conductivity of permeate for the original HU continuously increased up to 6230 μS/cm according to the acclimated permeate volume (Fig. 6) while the conductivity was saturated for HU6. The conductivities became steady as 1418 μS/cm when the acclimated permeate volume reached 1.33 L for HU6. The conductivity of HU4 increased to 191.7 μS/cm, but the conductivity was extremely low in comparison to the original HU and HU6.

Fig. 5. Water flux (a) and conductivity (b) for original and acidified human urine during the membrane distillation operation.

Fig. 6. The increase of permeate conductivity according to the acclimated permeate volume for original and acidified human urine during the membrane distillation operation.

Based on these results, it is considered that the characteristics of the fouling and conductivity leakage are entirely different between ADS and HU. ^{Yan, Liu et al. (2019)} described the low pH prevented the conductivity leakage by lowering the ammonia transfer through the MD membrane. In the case of ADS, the final ammonia concentration of the permeates were in the low level as 19 and 2.3 mg-N/L for original and acidified ADS, respectively. In contrast, HU showed extremely high ammonia concentration and a significant reduction in the ammonia concentration in the permeates as 3000, 402 and 16 mg-N/L for HU, HU6 and HU4, respectively. In the case of HU, the residual ammonia may exhibit a predominant effect on the conductivity in the permeate. However, the permeate composition for ADS and HU is still unclear. To prove the effect of acidification on the permeate quality and conductivity, there is need for the further investigation of the permeate composition depending on the feed pH.

As expected, the acidification for HU showed a significant effect to minimize ammonia transfer to the permeate side. The ammonia transfer is a more critical issue for HU than for ADS due to much higher alkalinity (pH 8.8) and ammonia concentration (5700 mg-N/L). The amounts of transferred ammonia were 7.6, 1.0 and 0.038 g for HU, HU6 and HU4, respectively (Table 4). Thus, the SAT values were exponentially reduced from 3.00 × 10^-3 to 4.02 × 10^-4 and 1.64 × 10^-5 for HU6 and HU4, respectively. Likely, the ammonia concentrations were significantly low as 402 and 16 mg-N/L for HU6 and HU4, respectively in comparison to 3000 mg-N/L in the permeate for the original HU (Table 5). TN was also reduced by 99.6 %, but the effect of acidification was limited for TP control by only 33.3 %. This implies that the acidification of HU is more effective to control ammonia transfer, TN and conductivity than ADS.

3.3. Characterization of fouling on the membrane

SEM observations demonstrated that during 24 hr operation, the fouling of the membrane had occurred on the feed side of the examined module. A comparison of the pristine PTFE membrane (Fig. 7a) with the used membrane (Fig. 7b, c, d, e, and f) showed that a large part of the membrane surface on the feed side was covered with a deposited layer formed during the DCMD operation. The elemental composition normalized to the total weight of elements detected by EDS showed a significant amount of C, O, N, Na, P, S, Cu and trace amounts of Fe, Cl and Ca for original and acidified ADS (Table 6). The crystals were accumulated on the membrane surfaces. As can be seen from SEM images (Fig. 7b and 7c) and the EDS spectrum, the foulants were recognized as NaCl, CaCO₃ and CuSO₄. A hydrophobic characteristic of the MD membrane enhances the adsorption of organic materials onto the membrane surface, and proteins exhibited a very high tendency to deposit on the hydrophobic membranes. The present investigation confirmed that acidification can intensify fouling that completely covers the membrane surfaces (Fig. 7c and 7f), which can reduce the membrane permeability and increased temperature polarization. From the EDS spectrum, it was confirmed that organic and ammonia-induced fouling accompanied by salt crystals on the membrane surface.

Fig. 7. SEM images for (a) pristine PTFE membrane and fouling formed by (b) original anaerobic digestion supernatant (pH 6.5), (c) acidified anaerobic digestion supernatant (pH 4.0), (d) original human urine (pH 8.8), (e) acidified human urine (pH 6.0) and (f) acidified human urine (pH 4.0).

The difference in the C content was clear between ADS and HU. The high C content of the ADS foulant implies the higher organic fouling phenomenon. From ADS, organic carbons including organics such as proteins, humic acids, and polysaccharides can be deposited on the MD membrane (^{Yan, Yang et al, 2019}). It is reasonable that the non-biodegradable organic compounds of ADS or biopolymer of active anaerobic sludge after the anaerobic digestion play a significant role in organic carbon deposition on the MD membrane. In contrast, HU contains relatively low concentration of organic carbon but high concentration of inorganic carbon including aqueous urea, sodium chloride and uric acid (^{Zhao et al., 2013}). It is interesting that acidified ADS and HU at pH 4 showed relatively high N content of 8.18 and 28.03 %, respectively. The first suspectible nitrogen compound of the major foulant is amine group (-NH₂) which was recognized as N-H stretching representing protein-like organics in the Fourier transform infrared spectroscopy (^{Yan, Yang et al., 2019}). The C-N bond representing polypeptide chain was also found from the foulant on the MD membrane for the treatment of undigested manure (^{Zarebska et al., 2014}). The counter anion for the inorganic fouling would be SO₄^2- for acidified ADS considering S in Table 6. For HU4, the suspectible major counter anion of chloride showed dramatically increased concentration to 53.12 %. Besides, the small amounts of detected foulants like Cr, Cu S, and Na might be aggregated on the MD membrane with colloidal particles for acidified ADS and HU at pH 4.