- Email: [email protected]
Ultra-thin, multi-layered polyamide membranes: Synthesis and characterization
Author’s Accepted Manuscript Ultra-thin, multi-layered polyamide membranes: synthesis and characterization Xiaoxiao Song, Saren Qi, Chuyang. Y. Tang, Congjie Gao www.elsevier.com/locate/memsci
PII: DOI: Reference:
S0376-7388(17)30554-9 http://dx.doi.org/10.1016/j.memsci.2017.06.016 MEMSCI15328
To appear in: Journal of Membrane Science Received date: 22 February 2017 Revised date: 1 June 2017 Accepted date: 6 June 2017 Cite this article as: Xiaoxiao Song, Saren Qi, Chuyang. Y. Tang and Congjie Gao, Ultra-thin, multi-layered polyamide membranes: synthesis and c h a r a c t e r i z a t i o n , Journal of Membrane Science, http://dx.doi.org/10.1016/j.memsci.2017.06.016 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Ultra-thin, multi-layered polyamide membranes: synthesis and characterization Xiaoxiao Songa,*1, Saren Qib1, Chuyang. Y. Tangb,c*, and Congjie Gaoa a
Center for Membrane and Water Science & Technology, Zhejiang University of Technology,
Hangzhou, 310014, China b
Singapore Membrane Technology Centre, Nanyang Environment and Water Research Institute,
Nanyang Technological University, Singapore 639798. c
The University of Hong Kong, Department of Civil Engineering, Pokfulam, Hong Kong.
[email protected] [email protected] *
Corresponding authors.
Abstract Customization of thickness and roughness of thin film composite reverse osmosis (TFC-RO) membranes provides opportunities to optimize the membrane permeability and fouling resistance. We propose a novel strategy to synthesize ultrathin multi-layered polyamide (ML-PA) membranes with the versatile maneuverability of the salt rejecting layer thickness and roughness. We have employed advanced quartz crystal microbalance with dissipation (QCMD) techniques to study the deposition rate of the ultrathin PA nanolayers with a resolution of approximately 8, 15 and 25 nm per deposition cycle. At brackish water desalination condition, the ML-PA membrane exhibited ~60% flux increase and higher salt rejection compared with the home-made
1
The authors contribute equally 1
TFC membrane fabricated by the conventional method. Benefit from the low roughness, the MLPA membrane shows much better fouling resistance to Bovine serum albumin (BSA).
Graphical
abstract
Keywords: thin film composite reverse osmosis membrane (TFC-RO); quartz crystal microbalance with dissipation (QCMD); polyamide nanolayer 1. Introduction TFC-RO membranes have been researched and developed significantly for the last 30 years and nowadays are widely used in the fields of desalination, wastewater treatment, water softening, etc. [1-4]. TFC RO membranes usually comprise of a salt-rejecting polyamide layers formed by interfacial polymerization (IP) on top of porous polymeric support layers and nonwoven backing 2
layers. Among the layers, the polyamide layer is the most critical selective barrier to determine the overall membrane performance. Conventionally, the TFC membranes are fabricated using a rapid polymerization reaction between a high concentration of aqueous amine monomer solution and an organic acyl chloride monomer solution [5]. Through the process, a relatively thick (a few hundred nanometers), defect-free, and salt-rejecting polyamide (PA) layer. Such PA layers exhibit low water permeabilities (1~5 L/m2.h.bar) [6, 7] and rough surfaces featured with “leaflike” or “ridge-and-valley” micro-structures (average roughness of 50-180 nm) [8, 9].
Recently, significant research efforts have been focused on reducing the thickness and roughness of the PA layers to achieve enhanced water production rate and better anti-fouling property [6, 10]. For example, a recent publication reported a successful formation and isolation of sub-10 nm PA layer on top of a nanowire network using MPD-TMC chemistry [6]. However, the handling and transfer of such sub-10 nm polyamide layer are extremely delicate, and the process is hardly scalable when it comes to realistic production. Another research group utilized the layer-by-layer strategy to form PA layer by using MPD-TMC chemistry in toluene medium [11, 12]. The resultant selective layer was relatively thin (~ 25 nm thickness) and ultra-smooth (~ 4 nm roughness). However, the reported molecular deposition rate was quite slow that it cost up to tens of molecular deposition cycles to achieve good salt rejection, which prolonged the production period beyond the chart of scale production. What’s more, the solvent toluene is environmental-unfriendly. Up to date, no literature has disclosed the applicable and controllable IP strategy under which an ultrathin and ultra-smooth PA layers could be produced readily with non-compromised salt rejection performance.
3
In the current study, we design a multi-layer molecular deposition-polymerization strategy using a bottom-up method starting from an initial layer of 10-20 nm. We successfully demonstrate the maneuverability of polyamide (ML-PA) layer roughness by selecting ultra-low concentrations for both MPD and TMC in a controllable deposition-IP process. The initial PA nanolayer was deposited by alternatively contacting membrane surface to ultra-low MPD aqueous solution (step 1) and TMC organic solution (step 2) with intermediate N2 purging and pure hexane rinsing steps. Subsequently, more nanolayers are fabricated by repeating the reaction cycles. The step-wise growth of IP thickness is governed by controlling the number of the reaction cycles and the duration in each cycle (Figure 1). Such concise and facile process is designed to produce an ultra-thin and ultra-smooth polyamide layer with a controllable rate. The process can be potentially scaled up to create a continuous fabrication process although additional cost may incur as a result of multiple interfacial polymerization steps. According to the best knowledge of the authors, it is the first study to fabricate the ultra-thin polyamide membranes with a scalable, controllable and concise process for water desalination application.
4
Figure 1. Schematic illustration of the fabrication process for ML-PA membranes via a multistep interfacial polymerization.
2. Material and methods 2.1 Chemicals and materials DI water was used to prepare all aqueous solutions unless specified otherwise. Analytical grade sodium chloride (NaCl) with purity over 99% was purchased from Merck (Germany). BSA was purchased from Sigma-Aldrich without further purification. A commercial polysulfone ultrafiltration membrane (UF, 50 kDa, ANDE membrane separation technology and engineering co., Ltd, China) was used as the membrane substrate. The m-phenylenediamine (MPD, SigmaAldrich) and trimesoyl chloride (TMC, Sinopharm Chemical Reagent Co. Ltd.) were used as monomers for interfacial polymerization. The n-hexane (Fisher Scientific) was used as the 5
solvent to dissolve the TMC. Two Dow FilmTec® commercial RO membrane, SW30 and BW30, were evaluated in parallel to facilitate the comparison.
2.2 Preparation of the membrane selective layer The multi-layered polyamide nanolayers were formed on the top of the substrates via the deposition-interfacial polymerization method (Figure 1). For the initial nanolayer deposition, the substrate membrane surface was soaked up with MPD solution with an ultra-low concentration (step 1). Subsequently, excess MPD is removed by N2 purging at 2 bar followed with hexane rinsing. The MPD impregnated surface was then reacted with ultra-low concentration TMC dissolved in hexane for 60 s (step 2). Removal of excess TMC stops the reaction and marks the end of cycle 1. Finally, the cycle was repeated for another n-1 (n=1, 3) times to result in final product ML-PA membranes. Different low MPD and TMC concentration pairs (High, Medium, and Low, while fixing MPD:TMC=10) were examined. The detailed reaction conditions are listed in Table 1. To simplify the discussion, short forms of the resultant membranes are used in the article: PA refers to the polyamide layer, L refers to the low concentration of MPD and TMC, M refers to the medium concentration of MPD and TMC, and H refers to the high concentration of MPD and TMC. The following number affixed to the concentration symbol refers to the number of reaction cycles. For comparison, conventional IP reaction conditions are adopted and the resultant membranes are labeled as TFC. After fabrication, all membranes were kept in DI water at 4 ºC for further characterization.
6
Table 1. The reaction conditions for multi-layered polyamide based membranes Membrane
MPD concentration
TMC concentration
No. of cycles
(wt.%)
(wt.%)
TFC
1
0.1
1
PA_H1
0.1
0.01
1
PA_H3
0.1
0.01
3
PA_M1
0.032
0.0032
1
PA_M3
0.032
0.0032
3
PA_L1
0.01
0.001
1
PA_L3
0.01
0.001
3
2.3 Membrane characterization All membranes were dried in vacuum desiccators for 24 hours before characterization. The membrane surface and cross-section morphologies were characterized by FESEM (Fieldemission scanning electron microscopy, JSM-7600F, JEOL, Japan) following similar procedures with previous studies [13, 14]. The chemical composition of the membrane selective layer was characterized by ATR-FTIR (Attenuated Total Reflection-Fourier Transform Infrared, IRPrestige-21, Shimadzu, Columbia, MD, USA) following procedures described previously [15]. The surface roughness of the membrane was measured by atomic force microscope (AFM, Park XE-100) [16-18]. The contact angle measurements were performed by sessile drop method [8]. Quartz crystal microbalance with dissipation (QCMD) technique is a powerful analytical tool that has been widely used to investigate the deposition thickness on the solid surface [19-21]. 7
The operation process was according to the literature [22]. Briefly, the SiO2 sensors (Q-Sense) were cleaned in a 2% SDS solution followed by a water rinse. The sensor was then dried with nitrogen gas and further cleaned in a UV/Ozone ProCleaner (Bioforce) for 15 min. All the measurements were taken at 22 ˚C. The sensor was firstly stabilized with DI water and then was soaked with MPD at corresponding concentration. After rinsed by n-hexane, an n-hexane containing TMC was applied to the sensor surface. Therefore, the adsorbed mass ∆ms (in the ng.cm-2 range) was recorded, which can be determined from the equation 1.
(
⁄ )
(1)
Where ∆f is the consonant frequency, CQCM (=17.7 ng.cm-2Hz-1 at f=5 MHZ) is the mass sensitivity constant and n (=1,3,…) is the overtone number. We assumed the density of the polyamide layer, ρ (=1 g/cm3) is independent of MPD concentration. Then, the thickness of the polyamide layer can be estimated by equation 2. ⁄
(2)
The commercial software package Q-Tools (Q-Sense AB, Gothenburg, Sweden) was used to perform these calculations. 2.4 Membrane performance evaluation The performance (i.e., water flux, (JW) and NaCl rejection (R)) of TFC, ML-PA and commercial RO membranes were evaluated in a cross flow RO setup described elsewhere [14]. Briefly, membrane coupons are cut and mounted into cross-flow membrane cells. A 500 ppm NaCl feed solution was pumped with a constant cross flow velocity of ~ 0.2 m/s under an applied pressure 8
of 10 bar. JW was determined by the gravimetric method. R was determined via equation 3. The conductivity of the permeate (Cp) and the conductivity of feed solution (Cf) was measured by conductivity meter. ⁄
(3)
The method to determine salt permeability, B, can be found in our previous work [13]. For comparison, the commercial RO membranes were evaluated under the same conditions. The fouling experiment was utilized BSA as a model foulant with a concentration of 100 ppm [12, 23]. The PA membrane and PA_H1 membrane were compacted with the same initial water flux of 20 L/m2.h. When the water flux was stable, BSA was dissolved into the feed water to obtain a concentration of 100 ppm. The water flux results were automatically logged into the system connected to the computer. The water flux was normalized by the initial water flux for both TFC and PA_H1.
3. Results and Discussions 3.1 Characterization of PA membranes The ML-PA and home-made conventional TFC membrane surfaces are characterized by AFM, FESEM, and FTIR-ATR techniques. As shown in the AFM study, the conventional TFC membrane surface shows a Ra of ~ 100 nm (Figure 2a and Table 2), which is in line with the range reported in the literature [8, 9]. By contrast, the PA_H1 membrane shows a much lower Ra of ~ 30 nm and a distinctively smoother surface (Figure 2b and Table 2). Accordingly, it can be 9
observed in the surface FESEM images that the rugged surface of conventional TFC membrane is characterized by (Figure 2c) the “leaf-like” and “ridge-and-valley” structures, which have been caused by the uncontrollable and fast reaction rate between high concentration MPD and TMC [24, 25]. By adopting an ultra-low concentration for both MPD and TMC, the typical “leaf-like” and “ridge-and-valley” micro-structures has been totally eliminated on the surface of the PA_H1 membrane (Figure 2d). In the cross-sectional FESEM images, the PA_H1 membrane (Figure 2f) shows a much thinner and more homogeneous skin layer compared with the conventional TFC membrane (Figure 2e). Similarly, smoother surfaces (Figure 3) and thinner skin layers (Figure 4) can be observed for all other ML-PA membranes. Such smooth surface led to a much lower Ra (Table 2), which is one of the most critical properties to lower the fouling tendency of the membrane [9, 10, 26, 27]. The ATR-FTIR results of PA membranes and TFC membranes (Figure 5) show typical absorption peaks at 1663 cm-1 and 1609 cm-1, which are associated with amide I bond and N-H deformation vibration, respectively [28]. Interestingly, the amide II bond at 1541 cm-1 is not observed for ML-PA membranes, probably due to a difference in chemical environment/structure of the amide II bond in ML-PA membranes [15].
10
Figure 2. AFM height images of (a) TFC surface and (b) ML-PA membrane surface (PA_H1). FESEM top surface morphologies of (c) TFC membranes and (d) PA_H1 membrane. FESEM cross-section morphologies of (e) TFC membrane and (f) PA_H1.
11
Figure 3. FESEM surface images of (a) UF substrate membrane, (b) PA_H3, (c) PA_M1, (d) PA_M3, (e) PA_L1, and (f) PA_L3.
12
Figure 4. FESEM cross-section images of (a) PA_H3, (b), PA_M1 (c) PA_M3, (d), PA_L1, and (e) PA_L3. The scale bar is 100 nm in all images.
13
a
100
Trans (%)
90 80 70 60 50
TFC PA_H1 PA_H3
1800 1700 1600 1500 1400 1300 1200 1100 1000 900
800
-1
Wavenumber (cm )
b
100
Trans (%)
90
80
70
60
TFC PA_M1 PA_M3
50 1800
1700
1600
1500
1400
1300
1200
1100
1000
900
800
Wavenumber (cm-1)
c
100
Trans (%)
90
80
70
60
TFC PA_L1 PA_L3
50 1800
1700
1600
1500
1400
1300
1200
1100
1000
900
800
Wavenumber (cm-1)
Figure 5. FTIR results of (a) PA_H, (b) PA_M, (c) PA_L membranes as a comparison to the conventional TFC membrane.
14
Table 2. Surface roughness and contact angle results of PA series membranes Membrane type
Ra (nm)
Contact angle (º)
Substrate
5.51±0.72
46.67±3.84
TFC
101.72±4.88
59.22±7.042
PA_H1
28.75±3.22
59.35±2.041
PA_H3
43.69±2.32
59.67±6.044
PA_M1
10.76±6.17
54.03±2.092
PA_M3
11.65±0.26
60.64±2.67
PA_L1
7.70±2.054
46.73±4.48
PA_L3
13.80±9.86
59.31±4.95
3.2 QCMD characterization To study the in-situ formation of polyamide layer, we have adopted the QCMD technique to monitor the micro-mass changes of ultrathin layers [20, 21]. The detected real time thickness change of the polyamide layer is plotted as a function of the deposition duration in a ThicknessTime plot (Figure 6 and Figure 7). To simulate the membrane formation process at a given concentration, we soaked the sensor of the QCMD instrument with dilute MPD aqueous solution, rinsed it with pure hexane, and soaked it in dilute TMC/hexane solution. The cycle was repeated for another n-1 (n=1, 3) times to examine the deposition scenarios in the PA_Hn, PA-Mn and PA-Ln membranes. In the formation of the PA_H1 membrane (Figure 6a), the initial MPD 15
uptake was low (subtle increase of thickness after MPD soaking). At the point of contacting TMC, the thickness of the PA nanolayer increased quickly and reached a plateau region of approximately 20 nm (δ1) within 10 s (Figure 6a), showing a fast reaction rate between MPD and TMC on immediate contact. In the case of the conventional TFC membrane (Figure 6b), the initial uptake of MPD from solution was much more pronounced than the PA_H1 membrane. When brought into contact with TMC solution, the PA layer thickness quickly increased to ~120 nm at the plateau region. The difference in PA layer deposition thickness of the PA_H1 and conventional TFC membranes is consistant with the observation in the FESEM images (Figure 2e and Figure 2f).
Figure 6. QCMD Thickness-Time plots of (a) PA_H1 membrane and (b) conventional TFC membrane.
In subsequent cycles of PA_H membrane, the PA layer gains roughly 25 nm increase per cycle (Figure 7a). Interestingly, the average PA growth rate in later cycle decreases, as reflected by a gradually decreased slope (during TMC contact) and an increased Δtn (time needed to finish 16
reaction). PA_M and PA_L membranes have similar Thickness-Time profile with the PA_H membranes, but the thickness of each nanolayer is only ~ 15nm (Figure 7b) and ~ 8 nm (Figure 7c) owing to the much lower MPD and TMC concentration. Apart from the thickness, the growth rate of PA_M and PA_L membranes are also much lower than that of PA_H membrane. For example, the Δt1(~ 600 s) of PA_M and PA_L membrane is much longer than that of the PA_H membrane (<10 s).
17
18
Figure 7. QCMD Thickness-Time plots of PA_H, PA_M, and PA_L membranes. In Figure 7a, δn (n=1,2,3) represents the thickness increment in the cycle of n; Δtn (n=1,2,3) represents the time required for polyamide layer growth in the cycle of n.
3.3 RO performance of ML-PA membranes The RO performance of the home made TFC and ML-PA membranes are listed in Figure 8. The water flux of all ML-PA membranes follows the order of PA_H1 < PA_M1 < PA_L1 and PA_H3 < PA_M3 < PA_L3 while the salt rejection follows the reverse order (refer to Table 3 and Figure 8). In the high concentration scenario, the PA_H1 membrane showed ~ 60% water flux increase compared with the conventional TFC membrane and a slightly higher NaCl rejection. The improved water permeability can be explained by lower hydraulic resistance of PA_H1 membrane due to its reduced active layer thickness [6]. Considering that NaCl rejection is given by JW/(JW+B) for RO membranes [29], the higher water flux JW of PA_H1 may benefit the rejection by diluting the salt flux through the membrane, provided the membrane salt permeability B is comparable to the TFC membrane. Further studies are needed to systematically investigate the underlining mechanisms for the simultaneous enhancement in water permeability and salt rejection, particularly in the context of ultrathin rejection layers. The PA_H3 membrane exhibits a lower water flux than the PA_H1 membrane due to the thicker selective layer, as confirmed by the QCMD and FESEM results. However, the similar B/A values (~ 3 KPa) of the two membranes imply that the PA_H3 membrane is as dense as the PA_H1 membrane, implying that the multilayers in PA_H3 membrane have a similar homogeneous structure. In the medium concentration scenario, the PA_M1 membrane shows comparatively larger B/A value compared 19
with the PA_H1 membrane, while the B/A value could be reduced by depositing more cycles. For PA_M3 membrane, the B/A value is lowered to ~ 5 KPa, and the respective salt rejection is 95%. This suggests that PA_M1 may have some defects on the initial PA layer, resulting in salt leakage and large B/A value. After deposition of 3 layers, the defects are repaired, resulting in a lower B/A value and an improved salt rejection. The PA_L series follows a similar trend with the PA_M series, yet starting with an even higher value of B/A (PA_L1 membrane). Even after the deposition of 3 layers, the rejection could only be elevated to 65%, suggesting the PA_L3 membrane may still have some defects. Comparing PA_H, PA_M and PA_L membranes, concentrated MPD and TMC form a ML_PA membrane with higher roughness, lower water flux, and higher NaCl rejection; while dilute MPD and TMC solutions form a ML_PA membrane with lower roughness, higher water flux and lower NaCl rejection. More deposition cycles are needed to eliminate potential defects and achieve satisfactory NaCl rejection if ultra-low concentration of reactants is chosen. The ML-PA and TFC membranes are benchmarked against commercial RO SW30 and BW30 membranes (Table 3). To ensure fair comparison, both commercial and home-made membranes were tested under identical experimental conditions. In the current study, SW30 had a water permeability of 1 L m-2h-1bar-1 and NaCl rejection of 96%, and BW30 had a water permeability of 2.5 L m-2h-1bar-1 and NaCl rejection of 95%. The rejection values reported for SW30 and BW30 in the current study are in good agreement with several existing studies using the same membranes [30-32]. Nevertheless, these values are lower than the ones provided by the manufacturer [33, 34] (99.4% for SW30 and 99.5% for BW30), which is likely due to different hydrodynamic testing conditions (see [30] and the footnotes of Table 3). Since the rejection of a membrane is strongly dependent on the testing pressure [35], an identical applied pressure of 10 20
bar was used throughout the current study. When compared with the commercial SW30 membrane, the water flux of all the ML-PA membranes was higher except for the PA_H3 membrane. On the other hand, PA_H1, PA_H3, and PA_M3 membranes achieved better salt rejections than SW30. Noticeably, PA-H1 and PA-H3 membranes achieved higher salt rejection but showed lower water permeability as compared to the commercial BW30 membrane (Table 3). It should be noted that the recipe to optimize SW30 and BW30 membrane properties as well as substrate properties, which has been optimized for more than twenty years, is still kept as a secret. There is a great possibility that improved performance (i.e., water flux or rejection) of ML-PA membranes will be obtained if these parameters are optimized.
1000
100
100 60
40 10 JW home-made SW30 BW30
1 PA
PA_H1
20
R home-made SW30 BW30 PA_H3
PA_M1
R (NaCl rejection, %)
JW (Water flux, L/m2.h)
80
0 PA_M3
PA_L1
PA_L3
Name
Figure 8. Water flux (JW, solid black symbols) and rejection (R, empty red symbols) of the MLPA membranes with different multi-polyamide layers and concentration. The horizontal black line and red line are the water flux and rejection of commercial SW30 and BW30 membranes, 21
respectively, which measured under the same conditions as the ML-PA membranes in the current study.
Table 3 RO performance of PA series membranes Membrane type
A L m-2h-1bar-1
Ra
B 10-8 m/s
B/A KPa
TFC
1.06±0.11
0.97±0.0052
1.12±0.024
3.6378833
PA_H1
1.66±0.20
0.98±0.0020
1.07±0.0035
2.6976223
PA_H3
0.66±0.068
0.97±0.0058
0.53±0.14
2.9170649
PA_M1
11.61±1.70
0.37±0.070
599±292
185.64726
PA_M3
1.51±0.15
0.95±0.015
2.08±0.89
4.9828752
PA_L1
75.91±13.28
0.034±0.034
106422±84674
5047.6839
PA_L3
5.07±0.16
0.65±0.094
85.58±32.08
60.492957
SW30
1.02±0.08
0.96±0.002 (0.994b)
2.83±0.05
4.166667
BW30
2.50±0.16
0.95±0.005 (0.995b )
3.70±0.18
5.263158
Note: a. In the current study, the testing pressure was 10 bar using 500 ppm NaCl as feed solution. b. The rejection values in the parenthesis were obtained from the membrane manufacturer. Specifically, SW30 [33] was tested at 55 bar using 32,000 ppm NaCl as feed solution, and BW30 [34] was tested at 15.5 bar using 2000 ppm NaCl as feed solution.
3.4 The anti-fouling performance of PA membranes
22
The standard bovine serum albumin (BSA) fouling filtration tests were carried out for PA_H1 membranes and TFC membrane to study the fouling behavior of both membranes. To make a valid comparison, the same initial water flux and cross flow velocity were kept (20 L/m2 h) for all the membranes [36]. The trends of normalized water flux on the initial water flux are presented in Figure 9. For both TFC and PA_H1 membranes, water flux rapidly reduced in the initial five hours and reached a stable region where the water flux underwent no further significant decrease. At steady state, a much lower water flux reduction rate (~ 12%) was observed for PA_H1 as compared to that of TFC (~ 25%). The lower flux decline reveals the lower tendency of BSA deposition on membrane surface. Previous studies have shown that smoother membrane surfaces have lower fouling tendency when other membrane properties are similar (i.e., surface charge and hydrophobicity [10, 26]). In this work, the zeta potential (~-50 mV at pH=5.6) and contact angle (~60˚, refer to Table 2) properties are similar for PA_H1 and TFC membranes. Hence, it could be reasoned that the smoother surface of PA_H1 membrane play a crucial role in fouling resistance.
1.00
PA_H1 TFC
Normalized water flux
0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0
5
10
15
20
25
Time (h)
23
Figure 9. The normalized water flux over time for the home-made conventional TFC membrane and the PA_H1 membrane. The normalized water flux is obtained from the water flux normalized by the initial water flux as the function of filtration time. Both of the initial water flux of TFC and PA_H1 were kept at 20 L/m2 h. 100 ppm BSA was used as model foulant. 4. Conclusions In summary, we have demonstrated a novel approach to synthesize a high-permeability, highrejection, and low-fouling RO membrane via a multi-layer IP strategy. We discovered that the roughness, structure, and thickness of the PA layer could be regulated by manipulating monomer concentration and the number of fabrication cycles. The novel ultra-thin PA-H1 membrane showed ~60% permeability increase compared with the home-made conventional TFC membrane and achieved better salt rejection at the same time. The key contribution of this method opens a new avenue to fabricate an ultrathin, defect-free, and low roughness selective layer with customizable thickness, roughness, water permeability, and salt rejection for the first time. In future study, the monomer concentration, deposition durations, and monomer materials can be examined to further optimize the structure and functionality of the membrane. Such potential for manipulation of nano-scale structures is a key driving force to a wide spectrum of potential applications such as water/solvent filtration membranes, biomedical devices, coatings, and gas separation membranes. Despite the good potential, further investigations of other contaminants that could foul the membrane more severely (e.g., sodium alginate) are needed to analyse the cost and benefit trade-off relationship for ML-PA membranes comprehensively.
Acknowledgements: 24
We thank the Singapore Ministry of Education (Grant #MOE2011-T2-2-035, ARC 3/12) for the financial support of the work. We are also grateful to Dow Company for supplying membrane samples. The authors thank the Singapore membrane technology centre (SMTC) for helping with the QCMD measurements.
Appendix: Table 3 RO performance of PA series membranes (2000 ppm feed) Membrane type
A
R
L m-2h-1bar-1
B
B/A
10-8 m/s
KPa
TFC
1.06
0.96
1.24
4.38
PA_H1
1.64
0.98
1.17
2.56
PA_H3
0.63
0.97
0.46
2.66
PA_M1
10.95
0.35
646
212
PA_M3
1.49
0.94
2.64
6.38
PA_L1
74.50
0.028
71839
3471
PA_L3
4.89
0.64
78.4
57.2
(Feed: 2000 ppm NaCl. Applied pressure: 15 bar)
25
References [1] R.J. Petersen, Composite reverse osmosis and nanofiltration membranes, Journal of Membrane Science, 83 (1993) 81-150. [2] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse osmosis desalination, Desalination, 216 (2007) 1-76. [3] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis desalination: Water sources, technology, and today's challenges, Water Research, 43 (2009) 2317-2348. [4] A.G. Fane, C.Y.Tang, R.Wang, Membrane technology for water: microfiltration, ultrafiltration, nanofiltration, and reverse osmosis, Treatise on Water Science, Academic Press, Oxford, 2011. [5] V. Freger, Nanoscale Heterogeneity of Polyamide Membranes Formed by Interfacial Polymerization, Langmuir, 19 (2003) 4791-4797. [6] S. Karan, Z. Jiang, A.G. Livingston, Sub-10 nm polyamide nanofilms with ultrafast solvent transport for molecular separation, Science, 348 (2015) 1347-1351. [7] D. Cohen-Tanugi, R.K. McGovern, S.H. Dave, J.H. Lienhard, J.C. Grossman, Quantifying the potential of ultra-permeable membranes for water desalination, Energy and Environmental Science, 7 (2014) 1134-1141. [8] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes, Desalination, 242 (2009) 149-167. [9] E.M. Vrijenhoek, S. Hong, M. Elimelech, Influence of membrane surface properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes, Journal of Membrane Science, 188 (2001) 115-128. [10] M. Elimelech, X. Zhu, A.E. Childress, S. Hong, Role of membrane surface morphology in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis membranes, Journal of Membrane Science, 127 (1997) 101-109. [11] W. Choi, J.-E. Gu, S.-H. Park, S. Kim, J. Bang, K.-Y. Baek, B. Park, J.S. Lee, E.P. Chan, J.-H. Lee, TailorMade Polyamide Membranes for Water Desalination, ACS Nano, 9 (2015) 345-355. [12] J.E. Gu, S. Lee, C.M. Stafford, J.S. Lee, W. Choi, B.Y. Kim, K.Y. Baek, E.P. Chan, J.Y. Chung, J. Bang, J.H. Lee, Molecular layer-by-layer assembled thin-film composite membranes for water desalination, Advanced Materials, 25 (2013) 4778-4782. [13] Q. Saren, C.Q. Qiu, C.Y. Tang, Synthesis and characterization of novel forward osmosis membranes based on layer-by-layer assembly, Environmental Science and Technology, 45 (2011) 5201-5208. [14] S. Qi, R. Wang, G.K.M. Chaitra, J. Torres, X. Hu, A.G. Fane, Aquaporin-based biomimetic reverse osmosis membranes: Stability and long term performance, Journal of Membrane Science, 508 (2016) 94103. [15] V.T. Do, C.Y. Tang, M. Reinhard, J.O. Leckie, Degradation of Polyamide Nanofiltration and Reverse Osmosis Membranes by Hypochlorite, Environmental Science & Technology, 46 (2012) 852-859. [16] S. Qi, W. Li, Y. Zhao, N. Ma, J. Wei, T.W. Chin, C.Y. Tang, Influence of the properties of layer-by-layer active layers on forward osmosis performance, Journal of Membrane Science, 423-424 (2012) 536-542. [17] S. Qi, C.Q. Qiu, Y. Zhao, C.Y. Tang, Double-skinned forward osmosis membranes based on layer-bylayer assembly-FO performance and fouling behavior, Journal of Membrane Science, 405-406 (2012) 2029. [18] C.Y. Tang, Y.N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes. II. Membrane physiochemical properties and their dependence on polyamide and coating layers, Desalination, 242 (2009) 168-182. 26
[19] K. Ariga, Y. Lvov, T. Kunitake, Assembling alternate dye-polyion molecular films by electrostatic layer-by-layer adsorption, Journal of the American Chemical Society, 119 (1997) 2224-2231. [20] F. Caruso, K. Niikura, D.N. Furlong, Y. Okahata, 1. Ultrathin multilayer polyelectrolyte films on gold: Construction and thickness determination, Langmuir, 13 (1997) 3422-3426. [21] F. Höök, B. Kasemo, T. Nylander, C. Fant, K. Sott, H. Elwing, Variations in coupled water, viscoelastic properties, and film thickness of a Mefp-1 protein film during adsorption and cross-linking: A quartz crystal microbalance with dissipation monitoring, ellipsometry, and surface plasmon resonance study, Analytical Chemistry, 73 (2001) 5796-5804. [22] X. Li, R. Wang, F. Wicaksana, Y. Zhao, C. Tang, J. Torres, A.G. Fane, Fusion behaviour of aquaporin Z incorporated proteoliposomes investigated by quartz crystal microbalance with dissipation (QCM-D), Colloids and Surfaces B: Biointerfaces, 111 (2013) 446-452. [23] Q.L. Li, Z.H. Xu, I. Pinnau, Fouling of reverse osmosis membranes by biopolymers in wastewater secondary effluent: Role of membrane surface properties and initial permeate flux, Journal of Membrane Science, 290 (2007) 173-181. [24] Y. Song, P. Sun, L.L. Henry, B. Sun, Mechanisms of structure and performance controlled thin film composite membrane formation via interfacial polymerization process, Journal of Membrane Science, 251 (2005) 67-79. [25] G.-Y. Chai, W.B. Krantz, Formation and characterization of polyamide membranes via interfacial polymerization, Journal of Membrane Science, 93 (1994) 175-192. [26] X. Zhu, M. Elimelech, Colloidal fouling of reverse osmosis membranes: Measurements and fouling mechanisms, Environmental Science and Technology, 31 (1997) 3654-3662. [27] D. Rana, T. Matsuura, Surface Modifications for Antifouling Membranes, Chemical Reviews, 110 (2010) 2448-2471. [28] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: I. FTIR and XPS characterization of polyamide and coating layer chemistry, Desalination, 242 (2009) 149-167. [29] A.C. Sagle, E.M. Van Wagner, H. Ju, B.D. McCloskey, B.D. Freeman, M.M. Sharma, PEG-coated reverse osmosis membranes: Desalination properties and fouling resistance, Journal of Membrane Science, 340 (2009) 92-108. [30] E.M. Van Wagner, A.C. Sagle, M.M. Sharma, B.D. Freeman, Effect of crossflow testing conditions, including feed pH and continuous feed filtration, on commercial reverse osmosis membrane performance, Journal of Membrane Science, 345 (2009) 97-109. [31] D. Mukherjee, A. Kulkarni, W.N. Gill, Flux enhancement of reverse osmosis membranes by chemical surface modification, Journal of Membrane Science, 97 (1994) 231-249. [32] J.T. Arena, B. McCloskey, B.D. Freeman, J.R. McCutcheon, Surface modification of thin film composite membrane support layers with polydopamine: Enabling use of reverse osmosis membranes in pressure retarded osmosis, Journal of Membrane Science, 375 (2011) 55-62. [33] S. Qi, C.Q. Qiu, Y. Zhao, C.Y. Tang, Double-skinned forward osmosis membranes based on layer-bylayer assembly—FO performance and fouling behavior, Journal of Membrane Science, 405–406 (2012) 20-29. [34] A.C. Ferrari, J.C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K.S. Novoselov, S. Roth, A.K. Geim, Raman Spectrum of Graphene and Graphene Layers, Physical Review Letters, 97 (2006) 187401. [35] J.E. Cadotte, R.J. Petersen, R.E. Larson, E.E. Erickson, A new thin-film composite seawater reverse osmosis membrane, Desalination, 32 (1980) 25-31.
27
[36] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Fouling of reverse osmosis and nanofiltration membranes by humic acid—Effects of solution composition and hydrodynamic conditions, Journal of Membrane Science, 290 (2007) 86-94.
Highlights
A scalable novel method for ultrathin, defect-free and multi-layered polyamide membrane; QCMD technique is employed to study in-situ growth of the ML-PA membrane; The ultrathin ML-PA membrane outperforms conventional TFC by ～60% production rate; The roughness of PA membrane is significantly reduced in this approach, result in less fouling;
28
PII: DOI: Reference:
S0376-7388(17)30554-9 http://dx.doi.org/10.1016/j.memsci.2017.06.016 MEMSCI15328
To appear in: Journal of Membrane Science Received date: 22 February 2017 Revised date: 1 June 2017 Accepted date: 6 June 2017 Cite this article as: Xiaoxiao Song, Saren Qi, Chuyang. Y. Tang and Congjie Gao, Ultra-thin, multi-layered polyamide membranes: synthesis and c h a r a c t e r i z a t i o n , Journal of Membrane Science, http://dx.doi.org/10.1016/j.memsci.2017.06.016 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Ultra-thin, multi-layered polyamide membranes: synthesis and characterization Xiaoxiao Songa,*1, Saren Qib1, Chuyang. Y. Tangb,c*, and Congjie Gaoa a
Center for Membrane and Water Science & Technology, Zhejiang University of Technology,
Hangzhou, 310014, China b
Singapore Membrane Technology Centre, Nanyang Environment and Water Research Institute,
Nanyang Technological University, Singapore 639798. c
The University of Hong Kong, Department of Civil Engineering, Pokfulam, Hong Kong.
[email protected] [email protected] *
Corresponding authors.
Abstract Customization of thickness and roughness of thin film composite reverse osmosis (TFC-RO) membranes provides opportunities to optimize the membrane permeability and fouling resistance. We propose a novel strategy to synthesize ultrathin multi-layered polyamide (ML-PA) membranes with the versatile maneuverability of the salt rejecting layer thickness and roughness. We have employed advanced quartz crystal microbalance with dissipation (QCMD) techniques to study the deposition rate of the ultrathin PA nanolayers with a resolution of approximately 8, 15 and 25 nm per deposition cycle. At brackish water desalination condition, the ML-PA membrane exhibited ~60% flux increase and higher salt rejection compared with the home-made
1
The authors contribute equally 1
TFC membrane fabricated by the conventional method. Benefit from the low roughness, the MLPA membrane shows much better fouling resistance to Bovine serum albumin (BSA).
Graphical
abstract
Keywords: thin film composite reverse osmosis membrane (TFC-RO); quartz crystal microbalance with dissipation (QCMD); polyamide nanolayer 1. Introduction TFC-RO membranes have been researched and developed significantly for the last 30 years and nowadays are widely used in the fields of desalination, wastewater treatment, water softening, etc. [1-4]. TFC RO membranes usually comprise of a salt-rejecting polyamide layers formed by interfacial polymerization (IP) on top of porous polymeric support layers and nonwoven backing 2
layers. Among the layers, the polyamide layer is the most critical selective barrier to determine the overall membrane performance. Conventionally, the TFC membranes are fabricated using a rapid polymerization reaction between a high concentration of aqueous amine monomer solution and an organic acyl chloride monomer solution [5]. Through the process, a relatively thick (a few hundred nanometers), defect-free, and salt-rejecting polyamide (PA) layer. Such PA layers exhibit low water permeabilities (1~5 L/m2.h.bar) [6, 7] and rough surfaces featured with “leaflike” or “ridge-and-valley” micro-structures (average roughness of 50-180 nm) [8, 9].
Recently, significant research efforts have been focused on reducing the thickness and roughness of the PA layers to achieve enhanced water production rate and better anti-fouling property [6, 10]. For example, a recent publication reported a successful formation and isolation of sub-10 nm PA layer on top of a nanowire network using MPD-TMC chemistry [6]. However, the handling and transfer of such sub-10 nm polyamide layer are extremely delicate, and the process is hardly scalable when it comes to realistic production. Another research group utilized the layer-by-layer strategy to form PA layer by using MPD-TMC chemistry in toluene medium [11, 12]. The resultant selective layer was relatively thin (~ 25 nm thickness) and ultra-smooth (~ 4 nm roughness). However, the reported molecular deposition rate was quite slow that it cost up to tens of molecular deposition cycles to achieve good salt rejection, which prolonged the production period beyond the chart of scale production. What’s more, the solvent toluene is environmental-unfriendly. Up to date, no literature has disclosed the applicable and controllable IP strategy under which an ultrathin and ultra-smooth PA layers could be produced readily with non-compromised salt rejection performance.
3
In the current study, we design a multi-layer molecular deposition-polymerization strategy using a bottom-up method starting from an initial layer of 10-20 nm. We successfully demonstrate the maneuverability of polyamide (ML-PA) layer roughness by selecting ultra-low concentrations for both MPD and TMC in a controllable deposition-IP process. The initial PA nanolayer was deposited by alternatively contacting membrane surface to ultra-low MPD aqueous solution (step 1) and TMC organic solution (step 2) with intermediate N2 purging and pure hexane rinsing steps. Subsequently, more nanolayers are fabricated by repeating the reaction cycles. The step-wise growth of IP thickness is governed by controlling the number of the reaction cycles and the duration in each cycle (Figure 1). Such concise and facile process is designed to produce an ultra-thin and ultra-smooth polyamide layer with a controllable rate. The process can be potentially scaled up to create a continuous fabrication process although additional cost may incur as a result of multiple interfacial polymerization steps. According to the best knowledge of the authors, it is the first study to fabricate the ultra-thin polyamide membranes with a scalable, controllable and concise process for water desalination application.
4
Figure 1. Schematic illustration of the fabrication process for ML-PA membranes via a multistep interfacial polymerization.
2. Material and methods 2.1 Chemicals and materials DI water was used to prepare all aqueous solutions unless specified otherwise. Analytical grade sodium chloride (NaCl) with purity over 99% was purchased from Merck (Germany). BSA was purchased from Sigma-Aldrich without further purification. A commercial polysulfone ultrafiltration membrane (UF, 50 kDa, ANDE membrane separation technology and engineering co., Ltd, China) was used as the membrane substrate. The m-phenylenediamine (MPD, SigmaAldrich) and trimesoyl chloride (TMC, Sinopharm Chemical Reagent Co. Ltd.) were used as monomers for interfacial polymerization. The n-hexane (Fisher Scientific) was used as the 5
solvent to dissolve the TMC. Two Dow FilmTec® commercial RO membrane, SW30 and BW30, were evaluated in parallel to facilitate the comparison.
2.2 Preparation of the membrane selective layer The multi-layered polyamide nanolayers were formed on the top of the substrates via the deposition-interfacial polymerization method (Figure 1). For the initial nanolayer deposition, the substrate membrane surface was soaked up with MPD solution with an ultra-low concentration (step 1). Subsequently, excess MPD is removed by N2 purging at 2 bar followed with hexane rinsing. The MPD impregnated surface was then reacted with ultra-low concentration TMC dissolved in hexane for 60 s (step 2). Removal of excess TMC stops the reaction and marks the end of cycle 1. Finally, the cycle was repeated for another n-1 (n=1, 3) times to result in final product ML-PA membranes. Different low MPD and TMC concentration pairs (High, Medium, and Low, while fixing MPD:TMC=10) were examined. The detailed reaction conditions are listed in Table 1. To simplify the discussion, short forms of the resultant membranes are used in the article: PA refers to the polyamide layer, L refers to the low concentration of MPD and TMC, M refers to the medium concentration of MPD and TMC, and H refers to the high concentration of MPD and TMC. The following number affixed to the concentration symbol refers to the number of reaction cycles. For comparison, conventional IP reaction conditions are adopted and the resultant membranes are labeled as TFC. After fabrication, all membranes were kept in DI water at 4 ºC for further characterization.
6
Table 1. The reaction conditions for multi-layered polyamide based membranes Membrane
MPD concentration
TMC concentration
No. of cycles
(wt.%)
(wt.%)
TFC
1
0.1
1
PA_H1
0.1
0.01
1
PA_H3
0.1
0.01
3
PA_M1
0.032
0.0032
1
PA_M3
0.032
0.0032
3
PA_L1
0.01
0.001
1
PA_L3
0.01
0.001
3
2.3 Membrane characterization All membranes were dried in vacuum desiccators for 24 hours before characterization. The membrane surface and cross-section morphologies were characterized by FESEM (Fieldemission scanning electron microscopy, JSM-7600F, JEOL, Japan) following similar procedures with previous studies [13, 14]. The chemical composition of the membrane selective layer was characterized by ATR-FTIR (Attenuated Total Reflection-Fourier Transform Infrared, IRPrestige-21, Shimadzu, Columbia, MD, USA) following procedures described previously [15]. The surface roughness of the membrane was measured by atomic force microscope (AFM, Park XE-100) [16-18]. The contact angle measurements were performed by sessile drop method [8]. Quartz crystal microbalance with dissipation (QCMD) technique is a powerful analytical tool that has been widely used to investigate the deposition thickness on the solid surface [19-21]. 7
The operation process was according to the literature [22]. Briefly, the SiO2 sensors (Q-Sense) were cleaned in a 2% SDS solution followed by a water rinse. The sensor was then dried with nitrogen gas and further cleaned in a UV/Ozone ProCleaner (Bioforce) for 15 min. All the measurements were taken at 22 ˚C. The sensor was firstly stabilized with DI water and then was soaked with MPD at corresponding concentration. After rinsed by n-hexane, an n-hexane containing TMC was applied to the sensor surface. Therefore, the adsorbed mass ∆ms (in the ng.cm-2 range) was recorded, which can be determined from the equation 1.
(
⁄ )
(1)
Where ∆f is the consonant frequency, CQCM (=17.7 ng.cm-2Hz-1 at f=5 MHZ) is the mass sensitivity constant and n (=1,3,…) is the overtone number. We assumed the density of the polyamide layer, ρ (=1 g/cm3) is independent of MPD concentration. Then, the thickness of the polyamide layer can be estimated by equation 2. ⁄
(2)
The commercial software package Q-Tools (Q-Sense AB, Gothenburg, Sweden) was used to perform these calculations. 2.4 Membrane performance evaluation The performance (i.e., water flux, (JW) and NaCl rejection (R)) of TFC, ML-PA and commercial RO membranes were evaluated in a cross flow RO setup described elsewhere [14]. Briefly, membrane coupons are cut and mounted into cross-flow membrane cells. A 500 ppm NaCl feed solution was pumped with a constant cross flow velocity of ~ 0.2 m/s under an applied pressure 8
of 10 bar. JW was determined by the gravimetric method. R was determined via equation 3. The conductivity of the permeate (Cp) and the conductivity of feed solution (Cf) was measured by conductivity meter. ⁄
(3)
The method to determine salt permeability, B, can be found in our previous work [13]. For comparison, the commercial RO membranes were evaluated under the same conditions. The fouling experiment was utilized BSA as a model foulant with a concentration of 100 ppm [12, 23]. The PA membrane and PA_H1 membrane were compacted with the same initial water flux of 20 L/m2.h. When the water flux was stable, BSA was dissolved into the feed water to obtain a concentration of 100 ppm. The water flux results were automatically logged into the system connected to the computer. The water flux was normalized by the initial water flux for both TFC and PA_H1.
3. Results and Discussions 3.1 Characterization of PA membranes The ML-PA and home-made conventional TFC membrane surfaces are characterized by AFM, FESEM, and FTIR-ATR techniques. As shown in the AFM study, the conventional TFC membrane surface shows a Ra of ~ 100 nm (Figure 2a and Table 2), which is in line with the range reported in the literature [8, 9]. By contrast, the PA_H1 membrane shows a much lower Ra of ~ 30 nm and a distinctively smoother surface (Figure 2b and Table 2). Accordingly, it can be 9
observed in the surface FESEM images that the rugged surface of conventional TFC membrane is characterized by (Figure 2c) the “leaf-like” and “ridge-and-valley” structures, which have been caused by the uncontrollable and fast reaction rate between high concentration MPD and TMC [24, 25]. By adopting an ultra-low concentration for both MPD and TMC, the typical “leaf-like” and “ridge-and-valley” micro-structures has been totally eliminated on the surface of the PA_H1 membrane (Figure 2d). In the cross-sectional FESEM images, the PA_H1 membrane (Figure 2f) shows a much thinner and more homogeneous skin layer compared with the conventional TFC membrane (Figure 2e). Similarly, smoother surfaces (Figure 3) and thinner skin layers (Figure 4) can be observed for all other ML-PA membranes. Such smooth surface led to a much lower Ra (Table 2), which is one of the most critical properties to lower the fouling tendency of the membrane [9, 10, 26, 27]. The ATR-FTIR results of PA membranes and TFC membranes (Figure 5) show typical absorption peaks at 1663 cm-1 and 1609 cm-1, which are associated with amide I bond and N-H deformation vibration, respectively [28]. Interestingly, the amide II bond at 1541 cm-1 is not observed for ML-PA membranes, probably due to a difference in chemical environment/structure of the amide II bond in ML-PA membranes [15].
10
Figure 2. AFM height images of (a) TFC surface and (b) ML-PA membrane surface (PA_H1). FESEM top surface morphologies of (c) TFC membranes and (d) PA_H1 membrane. FESEM cross-section morphologies of (e) TFC membrane and (f) PA_H1.
11
Figure 3. FESEM surface images of (a) UF substrate membrane, (b) PA_H3, (c) PA_M1, (d) PA_M3, (e) PA_L1, and (f) PA_L3.
12
Figure 4. FESEM cross-section images of (a) PA_H3, (b), PA_M1 (c) PA_M3, (d), PA_L1, and (e) PA_L3. The scale bar is 100 nm in all images.
13
a
100
Trans (%)
90 80 70 60 50
TFC PA_H1 PA_H3
1800 1700 1600 1500 1400 1300 1200 1100 1000 900
800
-1
Wavenumber (cm )
b
100
Trans (%)
90
80
70
60
TFC PA_M1 PA_M3
50 1800
1700
1600
1500
1400
1300
1200
1100
1000
900
800
Wavenumber (cm-1)
c
100
Trans (%)
90
80
70
60
TFC PA_L1 PA_L3
50 1800
1700
1600
1500
1400
1300
1200
1100
1000
900
800
Wavenumber (cm-1)
Figure 5. FTIR results of (a) PA_H, (b) PA_M, (c) PA_L membranes as a comparison to the conventional TFC membrane.
14
Table 2. Surface roughness and contact angle results of PA series membranes Membrane type
Ra (nm)
Contact angle (º)
Substrate
5.51±0.72
46.67±3.84
TFC
101.72±4.88
59.22±7.042
PA_H1
28.75±3.22
59.35±2.041
PA_H3
43.69±2.32
59.67±6.044
PA_M1
10.76±6.17
54.03±2.092
PA_M3
11.65±0.26
60.64±2.67
PA_L1
7.70±2.054
46.73±4.48
PA_L3
13.80±9.86
59.31±4.95
3.2 QCMD characterization To study the in-situ formation of polyamide layer, we have adopted the QCMD technique to monitor the micro-mass changes of ultrathin layers [20, 21]. The detected real time thickness change of the polyamide layer is plotted as a function of the deposition duration in a ThicknessTime plot (Figure 6 and Figure 7). To simulate the membrane formation process at a given concentration, we soaked the sensor of the QCMD instrument with dilute MPD aqueous solution, rinsed it with pure hexane, and soaked it in dilute TMC/hexane solution. The cycle was repeated for another n-1 (n=1, 3) times to examine the deposition scenarios in the PA_Hn, PA-Mn and PA-Ln membranes. In the formation of the PA_H1 membrane (Figure 6a), the initial MPD 15
uptake was low (subtle increase of thickness after MPD soaking). At the point of contacting TMC, the thickness of the PA nanolayer increased quickly and reached a plateau region of approximately 20 nm (δ1) within 10 s (Figure 6a), showing a fast reaction rate between MPD and TMC on immediate contact. In the case of the conventional TFC membrane (Figure 6b), the initial uptake of MPD from solution was much more pronounced than the PA_H1 membrane. When brought into contact with TMC solution, the PA layer thickness quickly increased to ~120 nm at the plateau region. The difference in PA layer deposition thickness of the PA_H1 and conventional TFC membranes is consistant with the observation in the FESEM images (Figure 2e and Figure 2f).
Figure 6. QCMD Thickness-Time plots of (a) PA_H1 membrane and (b) conventional TFC membrane.
In subsequent cycles of PA_H membrane, the PA layer gains roughly 25 nm increase per cycle (Figure 7a). Interestingly, the average PA growth rate in later cycle decreases, as reflected by a gradually decreased slope (during TMC contact) and an increased Δtn (time needed to finish 16
reaction). PA_M and PA_L membranes have similar Thickness-Time profile with the PA_H membranes, but the thickness of each nanolayer is only ~ 15nm (Figure 7b) and ~ 8 nm (Figure 7c) owing to the much lower MPD and TMC concentration. Apart from the thickness, the growth rate of PA_M and PA_L membranes are also much lower than that of PA_H membrane. For example, the Δt1(~ 600 s) of PA_M and PA_L membrane is much longer than that of the PA_H membrane (<10 s).
17
18
Figure 7. QCMD Thickness-Time plots of PA_H, PA_M, and PA_L membranes. In Figure 7a, δn (n=1,2,3) represents the thickness increment in the cycle of n; Δtn (n=1,2,3) represents the time required for polyamide layer growth in the cycle of n.
3.3 RO performance of ML-PA membranes The RO performance of the home made TFC and ML-PA membranes are listed in Figure 8. The water flux of all ML-PA membranes follows the order of PA_H1 < PA_M1 < PA_L1 and PA_H3 < PA_M3 < PA_L3 while the salt rejection follows the reverse order (refer to Table 3 and Figure 8). In the high concentration scenario, the PA_H1 membrane showed ~ 60% water flux increase compared with the conventional TFC membrane and a slightly higher NaCl rejection. The improved water permeability can be explained by lower hydraulic resistance of PA_H1 membrane due to its reduced active layer thickness [6]. Considering that NaCl rejection is given by JW/(JW+B) for RO membranes [29], the higher water flux JW of PA_H1 may benefit the rejection by diluting the salt flux through the membrane, provided the membrane salt permeability B is comparable to the TFC membrane. Further studies are needed to systematically investigate the underlining mechanisms for the simultaneous enhancement in water permeability and salt rejection, particularly in the context of ultrathin rejection layers. The PA_H3 membrane exhibits a lower water flux than the PA_H1 membrane due to the thicker selective layer, as confirmed by the QCMD and FESEM results. However, the similar B/A values (~ 3 KPa) of the two membranes imply that the PA_H3 membrane is as dense as the PA_H1 membrane, implying that the multilayers in PA_H3 membrane have a similar homogeneous structure. In the medium concentration scenario, the PA_M1 membrane shows comparatively larger B/A value compared 19
with the PA_H1 membrane, while the B/A value could be reduced by depositing more cycles. For PA_M3 membrane, the B/A value is lowered to ~ 5 KPa, and the respective salt rejection is 95%. This suggests that PA_M1 may have some defects on the initial PA layer, resulting in salt leakage and large B/A value. After deposition of 3 layers, the defects are repaired, resulting in a lower B/A value and an improved salt rejection. The PA_L series follows a similar trend with the PA_M series, yet starting with an even higher value of B/A (PA_L1 membrane). Even after the deposition of 3 layers, the rejection could only be elevated to 65%, suggesting the PA_L3 membrane may still have some defects. Comparing PA_H, PA_M and PA_L membranes, concentrated MPD and TMC form a ML_PA membrane with higher roughness, lower water flux, and higher NaCl rejection; while dilute MPD and TMC solutions form a ML_PA membrane with lower roughness, higher water flux and lower NaCl rejection. More deposition cycles are needed to eliminate potential defects and achieve satisfactory NaCl rejection if ultra-low concentration of reactants is chosen. The ML-PA and TFC membranes are benchmarked against commercial RO SW30 and BW30 membranes (Table 3). To ensure fair comparison, both commercial and home-made membranes were tested under identical experimental conditions. In the current study, SW30 had a water permeability of 1 L m-2h-1bar-1 and NaCl rejection of 96%, and BW30 had a water permeability of 2.5 L m-2h-1bar-1 and NaCl rejection of 95%. The rejection values reported for SW30 and BW30 in the current study are in good agreement with several existing studies using the same membranes [30-32]. Nevertheless, these values are lower than the ones provided by the manufacturer [33, 34] (99.4% for SW30 and 99.5% for BW30), which is likely due to different hydrodynamic testing conditions (see [30] and the footnotes of Table 3). Since the rejection of a membrane is strongly dependent on the testing pressure [35], an identical applied pressure of 10 20
bar was used throughout the current study. When compared with the commercial SW30 membrane, the water flux of all the ML-PA membranes was higher except for the PA_H3 membrane. On the other hand, PA_H1, PA_H3, and PA_M3 membranes achieved better salt rejections than SW30. Noticeably, PA-H1 and PA-H3 membranes achieved higher salt rejection but showed lower water permeability as compared to the commercial BW30 membrane (Table 3). It should be noted that the recipe to optimize SW30 and BW30 membrane properties as well as substrate properties, which has been optimized for more than twenty years, is still kept as a secret. There is a great possibility that improved performance (i.e., water flux or rejection) of ML-PA membranes will be obtained if these parameters are optimized.
1000
100
100 60
40 10 JW home-made SW30 BW30
1 PA
PA_H1
20
R home-made SW30 BW30 PA_H3
PA_M1
R (NaCl rejection, %)
JW (Water flux, L/m2.h)
80
0 PA_M3
PA_L1
PA_L3
Name
Figure 8. Water flux (JW, solid black symbols) and rejection (R, empty red symbols) of the MLPA membranes with different multi-polyamide layers and concentration. The horizontal black line and red line are the water flux and rejection of commercial SW30 and BW30 membranes, 21
respectively, which measured under the same conditions as the ML-PA membranes in the current study.
Table 3 RO performance of PA series membranes Membrane type
A L m-2h-1bar-1
Ra
B 10-8 m/s
B/A KPa
TFC
1.06±0.11
0.97±0.0052
1.12±0.024
3.6378833
PA_H1
1.66±0.20
0.98±0.0020
1.07±0.0035
2.6976223
PA_H3
0.66±0.068
0.97±0.0058
0.53±0.14
2.9170649
PA_M1
11.61±1.70
0.37±0.070
599±292
185.64726
PA_M3
1.51±0.15
0.95±0.015
2.08±0.89
4.9828752
PA_L1
75.91±13.28
0.034±0.034
106422±84674
5047.6839
PA_L3
5.07±0.16
0.65±0.094
85.58±32.08
60.492957
SW30
1.02±0.08
0.96±0.002 (0.994b)
2.83±0.05
4.166667
BW30
2.50±0.16
0.95±0.005 (0.995b )
3.70±0.18
5.263158
Note: a. In the current study, the testing pressure was 10 bar using 500 ppm NaCl as feed solution. b. The rejection values in the parenthesis were obtained from the membrane manufacturer. Specifically, SW30 [33] was tested at 55 bar using 32,000 ppm NaCl as feed solution, and BW30 [34] was tested at 15.5 bar using 2000 ppm NaCl as feed solution.
3.4 The anti-fouling performance of PA membranes
22
The standard bovine serum albumin (BSA) fouling filtration tests were carried out for PA_H1 membranes and TFC membrane to study the fouling behavior of both membranes. To make a valid comparison, the same initial water flux and cross flow velocity were kept (20 L/m2 h) for all the membranes [36]. The trends of normalized water flux on the initial water flux are presented in Figure 9. For both TFC and PA_H1 membranes, water flux rapidly reduced in the initial five hours and reached a stable region where the water flux underwent no further significant decrease. At steady state, a much lower water flux reduction rate (~ 12%) was observed for PA_H1 as compared to that of TFC (~ 25%). The lower flux decline reveals the lower tendency of BSA deposition on membrane surface. Previous studies have shown that smoother membrane surfaces have lower fouling tendency when other membrane properties are similar (i.e., surface charge and hydrophobicity [10, 26]). In this work, the zeta potential (~-50 mV at pH=5.6) and contact angle (~60˚, refer to Table 2) properties are similar for PA_H1 and TFC membranes. Hence, it could be reasoned that the smoother surface of PA_H1 membrane play a crucial role in fouling resistance.
1.00
PA_H1 TFC
Normalized water flux
0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0
5
10
15
20
25
Time (h)
23
Figure 9. The normalized water flux over time for the home-made conventional TFC membrane and the PA_H1 membrane. The normalized water flux is obtained from the water flux normalized by the initial water flux as the function of filtration time. Both of the initial water flux of TFC and PA_H1 were kept at 20 L/m2 h. 100 ppm BSA was used as model foulant. 4. Conclusions In summary, we have demonstrated a novel approach to synthesize a high-permeability, highrejection, and low-fouling RO membrane via a multi-layer IP strategy. We discovered that the roughness, structure, and thickness of the PA layer could be regulated by manipulating monomer concentration and the number of fabrication cycles. The novel ultra-thin PA-H1 membrane showed ~60% permeability increase compared with the home-made conventional TFC membrane and achieved better salt rejection at the same time. The key contribution of this method opens a new avenue to fabricate an ultrathin, defect-free, and low roughness selective layer with customizable thickness, roughness, water permeability, and salt rejection for the first time. In future study, the monomer concentration, deposition durations, and monomer materials can be examined to further optimize the structure and functionality of the membrane. Such potential for manipulation of nano-scale structures is a key driving force to a wide spectrum of potential applications such as water/solvent filtration membranes, biomedical devices, coatings, and gas separation membranes. Despite the good potential, further investigations of other contaminants that could foul the membrane more severely (e.g., sodium alginate) are needed to analyse the cost and benefit trade-off relationship for ML-PA membranes comprehensively.
Acknowledgements: 24
We thank the Singapore Ministry of Education (Grant #MOE2011-T2-2-035, ARC 3/12) for the financial support of the work. We are also grateful to Dow Company for supplying membrane samples. The authors thank the Singapore membrane technology centre (SMTC) for helping with the QCMD measurements.
Appendix: Table 3 RO performance of PA series membranes (2000 ppm feed) Membrane type
A
R
L m-2h-1bar-1
B
B/A
10-8 m/s
KPa
TFC
1.06
0.96
1.24
4.38
PA_H1
1.64
0.98
1.17
2.56
PA_H3
0.63
0.97
0.46
2.66
PA_M1
10.95
0.35
646
212
PA_M3
1.49
0.94
2.64
6.38
PA_L1
74.50
0.028
71839
3471
PA_L3
4.89
0.64
78.4
57.2
(Feed: 2000 ppm NaCl. Applied pressure: 15 bar)
25
References [1] R.J. Petersen, Composite reverse osmosis and nanofiltration membranes, Journal of Membrane Science, 83 (1993) 81-150. [2] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse osmosis desalination, Desalination, 216 (2007) 1-76. [3] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis desalination: Water sources, technology, and today's challenges, Water Research, 43 (2009) 2317-2348. [4] A.G. Fane, C.Y.Tang, R.Wang, Membrane technology for water: microfiltration, ultrafiltration, nanofiltration, and reverse osmosis, Treatise on Water Science, Academic Press, Oxford, 2011. [5] V. Freger, Nanoscale Heterogeneity of Polyamide Membranes Formed by Interfacial Polymerization, Langmuir, 19 (2003) 4791-4797. [6] S. Karan, Z. Jiang, A.G. Livingston, Sub-10 nm polyamide nanofilms with ultrafast solvent transport for molecular separation, Science, 348 (2015) 1347-1351. [7] D. Cohen-Tanugi, R.K. McGovern, S.H. Dave, J.H. Lienhard, J.C. Grossman, Quantifying the potential of ultra-permeable membranes for water desalination, Energy and Environmental Science, 7 (2014) 1134-1141. [8] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes, Desalination, 242 (2009) 149-167. [9] E.M. Vrijenhoek, S. Hong, M. Elimelech, Influence of membrane surface properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes, Journal of Membrane Science, 188 (2001) 115-128. [10] M. Elimelech, X. Zhu, A.E. Childress, S. Hong, Role of membrane surface morphology in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis membranes, Journal of Membrane Science, 127 (1997) 101-109. [11] W. Choi, J.-E. Gu, S.-H. Park, S. Kim, J. Bang, K.-Y. Baek, B. Park, J.S. Lee, E.P. Chan, J.-H. Lee, TailorMade Polyamide Membranes for Water Desalination, ACS Nano, 9 (2015) 345-355. [12] J.E. Gu, S. Lee, C.M. Stafford, J.S. Lee, W. Choi, B.Y. Kim, K.Y. Baek, E.P. Chan, J.Y. Chung, J. Bang, J.H. Lee, Molecular layer-by-layer assembled thin-film composite membranes for water desalination, Advanced Materials, 25 (2013) 4778-4782. [13] Q. Saren, C.Q. Qiu, C.Y. Tang, Synthesis and characterization of novel forward osmosis membranes based on layer-by-layer assembly, Environmental Science and Technology, 45 (2011) 5201-5208. [14] S. Qi, R. Wang, G.K.M. Chaitra, J. Torres, X. Hu, A.G. Fane, Aquaporin-based biomimetic reverse osmosis membranes: Stability and long term performance, Journal of Membrane Science, 508 (2016) 94103. [15] V.T. Do, C.Y. Tang, M. Reinhard, J.O. Leckie, Degradation of Polyamide Nanofiltration and Reverse Osmosis Membranes by Hypochlorite, Environmental Science & Technology, 46 (2012) 852-859. [16] S. Qi, W. Li, Y. Zhao, N. Ma, J. Wei, T.W. Chin, C.Y. Tang, Influence of the properties of layer-by-layer active layers on forward osmosis performance, Journal of Membrane Science, 423-424 (2012) 536-542. [17] S. Qi, C.Q. Qiu, Y. Zhao, C.Y. Tang, Double-skinned forward osmosis membranes based on layer-bylayer assembly-FO performance and fouling behavior, Journal of Membrane Science, 405-406 (2012) 2029. [18] C.Y. Tang, Y.N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes. II. Membrane physiochemical properties and their dependence on polyamide and coating layers, Desalination, 242 (2009) 168-182. 26
[19] K. Ariga, Y. Lvov, T. Kunitake, Assembling alternate dye-polyion molecular films by electrostatic layer-by-layer adsorption, Journal of the American Chemical Society, 119 (1997) 2224-2231. [20] F. Caruso, K. Niikura, D.N. Furlong, Y. Okahata, 1. Ultrathin multilayer polyelectrolyte films on gold: Construction and thickness determination, Langmuir, 13 (1997) 3422-3426. [21] F. Höök, B. Kasemo, T. Nylander, C. Fant, K. Sott, H. Elwing, Variations in coupled water, viscoelastic properties, and film thickness of a Mefp-1 protein film during adsorption and cross-linking: A quartz crystal microbalance with dissipation monitoring, ellipsometry, and surface plasmon resonance study, Analytical Chemistry, 73 (2001) 5796-5804. [22] X. Li, R. Wang, F. Wicaksana, Y. Zhao, C. Tang, J. Torres, A.G. Fane, Fusion behaviour of aquaporin Z incorporated proteoliposomes investigated by quartz crystal microbalance with dissipation (QCM-D), Colloids and Surfaces B: Biointerfaces, 111 (2013) 446-452. [23] Q.L. Li, Z.H. Xu, I. Pinnau, Fouling of reverse osmosis membranes by biopolymers in wastewater secondary effluent: Role of membrane surface properties and initial permeate flux, Journal of Membrane Science, 290 (2007) 173-181. [24] Y. Song, P. Sun, L.L. Henry, B. Sun, Mechanisms of structure and performance controlled thin film composite membrane formation via interfacial polymerization process, Journal of Membrane Science, 251 (2005) 67-79. [25] G.-Y. Chai, W.B. Krantz, Formation and characterization of polyamide membranes via interfacial polymerization, Journal of Membrane Science, 93 (1994) 175-192. [26] X. Zhu, M. Elimelech, Colloidal fouling of reverse osmosis membranes: Measurements and fouling mechanisms, Environmental Science and Technology, 31 (1997) 3654-3662. [27] D. Rana, T. Matsuura, Surface Modifications for Antifouling Membranes, Chemical Reviews, 110 (2010) 2448-2471. [28] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: I. FTIR and XPS characterization of polyamide and coating layer chemistry, Desalination, 242 (2009) 149-167. [29] A.C. Sagle, E.M. Van Wagner, H. Ju, B.D. McCloskey, B.D. Freeman, M.M. Sharma, PEG-coated reverse osmosis membranes: Desalination properties and fouling resistance, Journal of Membrane Science, 340 (2009) 92-108. [30] E.M. Van Wagner, A.C. Sagle, M.M. Sharma, B.D. Freeman, Effect of crossflow testing conditions, including feed pH and continuous feed filtration, on commercial reverse osmosis membrane performance, Journal of Membrane Science, 345 (2009) 97-109. [31] D. Mukherjee, A. Kulkarni, W.N. Gill, Flux enhancement of reverse osmosis membranes by chemical surface modification, Journal of Membrane Science, 97 (1994) 231-249. [32] J.T. Arena, B. McCloskey, B.D. Freeman, J.R. McCutcheon, Surface modification of thin film composite membrane support layers with polydopamine: Enabling use of reverse osmosis membranes in pressure retarded osmosis, Journal of Membrane Science, 375 (2011) 55-62. [33] S. Qi, C.Q. Qiu, Y. Zhao, C.Y. Tang, Double-skinned forward osmosis membranes based on layer-bylayer assembly—FO performance and fouling behavior, Journal of Membrane Science, 405–406 (2012) 20-29. [34] A.C. Ferrari, J.C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K.S. Novoselov, S. Roth, A.K. Geim, Raman Spectrum of Graphene and Graphene Layers, Physical Review Letters, 97 (2006) 187401. [35] J.E. Cadotte, R.J. Petersen, R.E. Larson, E.E. Erickson, A new thin-film composite seawater reverse osmosis membrane, Desalination, 32 (1980) 25-31.
27
[36] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, Fouling of reverse osmosis and nanofiltration membranes by humic acid—Effects of solution composition and hydrodynamic conditions, Journal of Membrane Science, 290 (2007) 86-94.
Highlights
A scalable novel method for ultrathin, defect-free and multi-layered polyamide membrane; QCMD technique is employed to study in-situ growth of the ML-PA membrane; The ultrathin ML-PA membrane outperforms conventional TFC by ～60% production rate; The roughness of PA membrane is significantly reduced in this approach, result in less fouling;
28