Diffusive transport through surface functionalized nanoporous alumina membranes by atomic layer deposition of metal oxides

Diffusive transport through surface functionalized nanoporous alumina membranes by atomic layer deposition of metal oxides

Accepted Manuscript Title: Diffusive transport through surface functionalized nanoporous alumina membranes by atomic layer deposition of metal oxides ...

633KB Sizes 1 Downloads 8 Views

Accepted Manuscript Title: Diffusive transport through surface functionalized nanoporous alumina membranes by atomic layer deposition of metal oxides Authors: V. Vega, L. Gelde, A.S. Gonz´alez, V.M. Prida, B. Hernando, J. Benavente PII: DOI: Reference:

S1226-086X(17)30142-9 http://dx.doi.org/doi:10.1016/j.jiec.2017.03.025 JIEC 3338

To appear in: Received date: Revised date: Accepted date:

21-11-2016 13-2-2017 16-3-2017

Please cite this article as: V.Vega, L.Gelde, A.S.Gonz´alez, V.M.Prida, B.Hernando, J.Benavente, Diffusive transport through surface functionalized nanoporous alumina membranes by atomic layer deposition of metal oxides, Journal of Industrial and Engineering Chemistryhttp://dx.doi.org/10.1016/j.jiec.2017.03.025 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 proof before it is published in its final 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.

Diffusive transport through surface functionalized nanoporous alumina membranes by atomic layer deposition of metal oxides V. Vega1,2,*, L. Gelde3, A. S. González2, V.M. Prida2, B. Hernando2, J. Benavente3 1

Laboratorio de Membranas Nanoporosas. Servicios Científico-Técnicos, Universidad de

Oviedo, 33006 Oviedo, Spain. 2

Departamento de Física, Facultad de Ciencias, Universidad de Oviedo, 33007 Oviedo,

Spain. 3

Departamento de Física Aplicada I, Facultad de Ciencias, Universidad de Málaga, 29071

Málaga, Spain. *

[email protected]

1

Graphical abstract

2

Highlights  Alumina membrane surface and pore wall modification with metal oxides by ALD.  Functionalization by single and double oxide layers is achieved.  A wide range of metal oxide coatings are investigated in detail.  The impact of surface modification is revealed by membrane potential measurements.

3

Abstract Changes associated to surface functionalization of nanoporous alumina membranes by atomic layer deposition (ALD) of metal oxides (Al2O3, SiO2, TiO2, Fe2O3, ZnO) are presented. ALD modification of the alumina membranes reveals a reduction up to 25-35 % in porosity, and confirms the presence of the metal oxide layer coating the pores. Its effect on the membrane permselectivity and other characteristic transport parameters was determined from membrane potential measurements, being correlated with changes in morphology and physic-chemical characteristics of the alumina membranes. According to our results, ALD provides a straight-forward and efficient method to adjust membrane performance for specific applications.

Keywords: Nanoporous alumina membranes; ALD surface coating; SEM analysis; membrane potentials; permselectivity.

1. Introduction

Nanoporous structures consisting of parallel straight channels having well-defined lattice parameters such as the nanopore size diameter, length and pore density (or porosity), which can be varied in a broad range of selected values, have been extensively described lately due to their increased applications in drug delivery, nanofiltration, biosensors and nanofluidics [1-6]. Nanoporous alumina membranes (NPAMs), having spatially self-ordered and vertically aligned one-dimensional (1D) channel structure, attract great interest in the preparation of monodisperse and self-ordered 1D nanostructures within their pores [7] as the growth-limiting template for matrix materials engineering, such as solid state acid catalyst packing, which improves the efficiency of mass transfer and the even distribution of active components on the matrix, therefore reducing byproducts due to excessive local reactions [8]. Efficient separation technology of neutral molecules needs nanoporous membranes with a controlled pore size, but the surface charge can also play a significant effect on transport or 4

concentration of charged biomolecules and ions through membranes with pore radii ≤ 20 nm [9-12]. Among the different techniques used for the fabrication of membranes with a welldefined nanostructure, track-etched polymeric membranes [13-14] and NPAMs obtained by the two-step aluminium anodization method [15] are the most commonly used. In both cases, porosity and pore radius depend on preparation conditions and post-treatments [13, 16-17]. However, narrow nanopores and highly selective membranes can be obtained by different kinds of surface modification and membrane functionalization [18-23]. In particular, the atomic layer deposition (ALD) technique has been used for a suitable tuning of the membranes pore size, as well as to modify surface physicochemical characteristics such as charge or hydrophilicity [11,21], which in turn can be of great interest in the case of filtration/diffusion of aqueous solutions, since they may affect to membranes selectivity. As it is well known, ALD technique is based on the vapor phase deposition method, but with a great potential for producing very thin and conformal coating of oxide layers, achieving an optimal control of their thickness and composition at atomic scale (see ref. 24 for a detailed description of ALD technique). Based on sequential and self-limiting reactions, ALD offers exceptional deposition conformality on high aspect ratio (length-to-diameter) structures and with tunable thin film composition. Consequently, ALD is an outstanding technique for the surface modification with a homogeneous oxide layer on the inner-pore walls of nanoporous membranes. However, due to the high temperature values required for some of the reactions, ALD has been mostly used during the last decade for the functionalization and modification of ceramic membranes surfaces, but lately it has also been employed for tailoring the pore size of polymeric (polycarbonate) membranes [21, 25], thus significantly opening its applicability nowadays. In this work, we describe the conformal deposition of metal oxides by ALD technique on the porous surface of ceramic nanoporous alumina membranes, obtained by the two-step aluminium anodization method using oxalic acid solution (Ox NPAM sample). A wide range of metal oxide single layers (Al2O3, SiO2, TiO2, Fe2O3 and ZnO) were deposited, as well as combinations of double layered structures consisting of a first layer of SiO2 plus a second layer of Al2O3 or ZnO:Al (AZO). Morphological modification and chemical changes in the pores of the Ox NPAM sample associated to ALD functionalization treatment were determined by scanning electron microscopy (SEM) and local energy dispersive X-ray spectroscopy (EDS) analysis, leading to a pore size reduction around 25 % and to the presence of the modifying material along the whole pores length, respectively. Moreover, the effect of both modifications on diffusive transport has also been considered by determining differences in the ionic permselectivity, diffusion coefficients and effective fixed charge from 5

membrane potential measurements carried out with NaCl solutions at different concentrations. A correlation between these two latter parameters has been obtained. 2. Experimental

2.1. Synthesis of nanoporous alumina membranes

NPAMs have been synthesized by the well-established two-step anodization approach [15, 26] in oxalic acid electrolytes. High purity aluminium discs (Al 99.999 %), having 0.5 mm in thickness and 25 mm in diameter, were employed as starting substrates for the fabrication of NPAMs. The aluminium discs were firstly cleaned by sonication in isopropanol and ethanol, and subsequently electropolished in a vigorously stirred mixture of perchloric acid and ethanol (1:3 vol.) at 5ºC. An electropolishing voltage of 20 V was applied between the sample and a Platinum counter-electrode. The first anodization process was performed in 0.3 M oxalic acid aqueous electrolyte, at 1-3 ºC, and under an anodization voltage of 40 V, applied between the sample and the Pt counter electrode during 24 h. Vigorous stirring was employed during the anodization steps in order to ensure the homogeneity of the bulk electrolyte concentration, and its temperature was kept constant by an external recirculating cooler. The aluminium oxide layer grown after the first anodization step was selectively removed by immersing the samples in an acidic solution of CrO3 and H3PO4 at 35ºC for 24 hours. The second anodization step was performed under the same anodization conditions as the first one, but with a longer time duration of 33 h to adjust the final thickness of NPAMs around of ~ 60 µm, providing them robustness enough to ensure their integrity during the cell manipulation in flow measurements. Afterwards, the remaining aluminum substrate was partially removed under chemical etching in an aqueous mixture of HCl and CuCl2, by exposing an area of around 1 cm2 of the NPAM backside. The alumina barrier layer that blocks the pores bottom was removed by wet chemical etching in H3PO4 5% aqueous solution at room temperature during 100 minutes, thus resulting in a two-side opened porous structure for the NPAMs.

2.2. Surface functionalization by Atomic Layer Deposition

6

Atomic Layer Deposition (ALD) coating of the Ox NPAMs samples was performed in a Savannah 100 thermal ALD reactor from Cambridge Nanotech (USA). Appropriated precursors for the conformal layer deposition of different functional oxides were selected as indicated in Table 1, employing in all cases high purity Ar as the carrier gas. The NPAMs were exposed to the different precursors in a sequential ordering, using long exposure times in the range of 45 – 60 s, to allow gaseous precursors to diffuse through the high aspect ratio pores of the Ox NPAM sample. Between two subsequent precursor pulses, an extended purge (90 s) with Ar flow of 50 sccm was performed in order to evacuate from the ALD reaction chamber the excess of unreacted gaseous precursor, as well as the reaction by-products. The number of ALD cycles was adjusted according to the different growth rates of each metal oxide as is indicated in Table 1, in order to adjust the thickness of the deposited layer to around 4 nm. In the case of the double-layered ALD coatings, the samples were firstly covered with a 4 nm thick SiO2 layer as previously described. Subsequently, ALD precursors and deposition conditions were conveniently modified, in order to perform the deposition of a second layer (either Al2O3 or AZO) with a thickness of 1-2 nm.

Table 1: Precursors and deposition conditions employed for the ALD deposition coating of the Ox NPAMs pores with different functional oxides. Oxide coating Al2O3

ALD precursors

Precursor Substrate Growth rate temperature (ºC) temperature (ºC) (nm/cycle)

Ref.

H2O

60

200

0.106

27

180

0.06

11, 28

230

0.022

29

200

0.146

30, 31

Trimethylaluminum 20

SiO2

H2O

60

O3

20

(3-Aminopropyl)

100

Triethoxysilane O3

20

Ferrocene

100

H2O

60

Diethylzinc

20

Fe2O3

ZnO

7

H2O TiO2

Titanium

60

200

0.05

32

200

0.144

33

(IV) 75

tetraisopropoxide H2O

60

Diethylzinc

20

AZO (ZnO:Al)

Trimethylaluminum 20

2.3 Morphological characterization and chemical analysis The morphology of the Ox-based NPAMs samples was studied by SEM analysis performed in a JEOL 6610LV microscope, equipped with an Oxford INCA EDS microanalysis system. The SEM is fitted with a tungsten filament electron gun operating at an acceleration voltage of 20 kV. Top and bottom surface views and cross section images obtained after mechanical fracturing of the NPAMs were obtained for all samples, which were coated with a thin Au layer by sputtering deposition to ensure their electrical conductivity. The images were further analysed using ImageJ software [34], for determining the main basic lattice parameters of the membranes (pores size and length, together with the inter-pores distance). The chemical characterization of both the ALD covered and uncovered alumina membranes was performed by local EDS analysis and composition profile line-scans, which allows us for determining the elemental distribution of the ALD deposited layers along the pores of the Ox-based NPAMs. 2.4. Membrane potential measurements

Membrane potential, or the equilibrium electrical potential difference between two electrolyte solutions of different concentration (Cf and Cv) separated by a membrane, was measured in a dead-end test cell consisting of two glass half-cells, with a magnetic stirrer in the bottom of each half-cell to minimize the concentration-polarization at the membrane surfaces (stirring rate of 540 rpm). A Ag/AgCl reversible electrode was placed in each halfcell and connected to a digital voltmeter (Yokohama 7552, 1G input resistance) for the cell potential (E) measurements [35]. These measurements were carried out with different NaCl solutions (at 25 ± 2 ºC, pH = 5.9 ± 0.2) by keeping constant the concentration of the solution at one side of the membrane (Cf = 0.01 M) and gradually changing the concentration of the solution at the other side (0.002 M ≤ Cv ≤ 0.1 M). Membrane potential (mbr) values were 8

obtained for each pair of Cv/Cf concentrations by subtracting the electrode potential to cell potential values, that is, mbr = E - elect. After membrane potential measurements the samples were checked again by SEM imaging and EDS analysis, in order to confirm that no surface modification occurs due to the slightly acidic pH of the test solutions.

3. Results and discussion

3.1. Microstructure and morphological parameters of alumina membranes

The average values of pore radius, rp, and inter-pore distance, Dint, were obtained from SEM and further image analysis. Fig. 1 shows SEM top (a and c), and bottom (b and d) views of the Ox NPAM samples together with the Ox+ZnO ones. SEM images evidence the hexagonal, highly ordered nanoporous structure characteristic of NPAMs. The pore radii dispersion histograms, shown as insets in Fig. 1 (a-d), evidence the narrow pore size distribution, as well as the pore radii modification induced through the ALD surface functionalization. From a comparison between images corresponding to the as-anodized and uncoated (Fig 1 a and b) respect to the ALD coated NPAM samples (Fig 1 c and d), it becomes evident the reduction of about 8 nm in the pore diameter, associated with the conformal ALD deposited ZnO layer. Furthermore, top and bottom SEM views display a slightly different pore size and porosities for both, uncoated and ALD coated samples, which might be related to the differential exposure to the gaseous precursors of the bottom and top NPAM surfaces during ALD processes. Thus, the mean pore size is obtained by averaging the values determined from both sides of each NPAM sample. The morphological parameters obtained for each Ox NPAM sample are displayed in Table 2. The as-prepared NPAMs have an average pore radii of 16 ± 2 nm, which is reduced down to about 12-13 nm after surface modification by ALD deposition of a single layer of the different oxides studied in this work. Additional ALD deposition, as in the case of the double layered systems, causes a decrease of the average pore radii to around 11 nm. Despite the variation of pore diameter, the inter-pore distance keeps constant at around 105 nm, and therefore the porosity, , decreases from ~ 8 % for the un-coated NPAM down to around ~ 4 % for the ALD modified NPAMs ( = (2/√3)(rp/Dint)2, [15]). The thickness of the different

9

NPAMs takes approximately constant values of around 60 m (average thickness, <xm> = 60 ± 4 m). Table 2: Morphological parameters characteristic of the studied Ox NPAMs: pore radii (rp), inter-pore distance (Dint), and estimated average porosity (), calculated according to ref. [15]. Dint (nm)

%)

Ox

16 ± 2

105 ± 5

8

Ox+Al2O3

12 ± 2

105 ± 5

5

Ox+SiO2

12 ± 2

105 ± 5

5

Ox+Fe2O3

12 ± 1

105 ± 5

4

Ox+ZnO

12 ± 1

105 ± 5

5

Ox+TiO2

13 ± 2

105 ± 5

6

Ox+SiO2+Al2O3

11 ± 1

105 ± 5

4

Ox+SiO2+AZO

11 ± 1

105 ± 5

4

b)

60 40 20 0

12.5 15.0 17.5 20.0 22.5 25.0

Relative frecuency (%)

rp (nm)

Relative frecuency (%)

a)

Membrane

60 40 20 0

Pore radii (nm)

Pore radii (nm)

500 nm

d)

40

20

0

10.0

12.5

15.0

17.5

20.0

Relative frecuency (%)

Relative frecuency (%)

500 nm

c)

12.5 15.0 17.5 20.0 22.5 25.0

Pore radii (nm)

60

40

20

0

7.5

10.0

12.5

15.0

17.5

Pore radii (nm)

500 nm

500 nm

Fig. 1: SEM top (a and c) and bottom (b and d) surface views corresponding to the asanodized Ox NPAM sample (a and b) and the Ox+ZnO sample after being covered with an oxide layer by ALD (c and d). The respective insets show the pore radii distribution obtained from each SEM micrograph. 10

As it is shown in Table 2, the ALD modification of the Ox NPAM with an unique metal oxide causes a reduction in the pore size of the as-obtained anodic alumina membrane around 25 % and 35 % in porosity, while the pore size reduction in the Ox NPAM samples subjected to a double process of the ALD functionalization is around 30 % and 50 %, respectively. As it can be appreciated, there is a good agreement between measured pore size and expected nominal film thickness of the deposited metal oxide layer indicated in the experimental section, taking into account the error margins arising from the limited resolution of SEM images. 3.3. Chemical analysis along the nanopores of the studied membranes

In order to determine the distribution of the ALD coating materials into the NPAMs, EDS microanalysis was performed in the cross sections of the ALD metal oxide functionalized NPAMs. Fig. 2 shows cross section images of samples Ox+ZnO (a) and Ox+TiO2 (c), together with the respective EDS analyses performed along the lines indicated in the SEM figures. For both ZnO (Fig. 2 b) and TiO2 (Fig. 2 c) ALD functionalized samples, the presence of the oxide coating layer in the internal open surfaces of the NPAMs (pore walls) through the whole membrane thickness was confirmed. In addition, a higher signal from the coating materials (i.e., Ti and Zn) was detected near the top and bottom surfaces of the ALD coated Ox NPAMs, indicating a preferential deposition on these surfaces in comparison with the internal pore ones, which can be due to the long diffusion path for the gaseous precursor molecules needed to reach the internal pore surfaces.

11

b)

Aluminum Ka1 Oxygen Ka1

Zinc Ka1 120

4 90 60

2

30 0

0 10

20 30 40 50 Membrane thickness (m)

d)2

Aluminum Ka1 Oxygen Ka1

60

Titanium Ka1

90

60 1 30

Counts Ti

Counts Al, O /1000

c)

Counts Zn

Counts Al, O /1000

a)

0

0 10

20 30 40 Membrane thickness (m)

50

Fig. 2: SEM cross section views (a, c) and the respective EDS elemental profiles (b, d) obtained from line-scans along the lines indicated in yellow colour, for samples Ox+ZnO (a, b) and Ox +TiO2 (c, d).

In order to discard any chemical dissolution effect on the metal oxide ALD covered membranes during membrane potential measurements, the SEM/EDS analysis was repeated after these electrochemical measurements, evidencing no relevant modification of the nanostructure and chemical composition of the samples.

3.3. Study of the diffusive transport across the nanoporous alumina membranes

Once the morphological (pore size/porosity) and chemical changes in the nanopores of the Ox-based NPAMs due to metal oxides coverage was confirmed by SEM and EDS elemental analysis, its effect on the diffusive transport through the ALD functionalized nanoporous alumina membranes was studied. As it is well known, the pore size is a key parameter in the transport across porous membranes, but the presence of charges on the membrane surfaces (both external planar surface and internal surface or pore-walls) is also a 12

factor of great importance in the transport of electrolyte solutions and charged species [3638]. The effective fixed charge (Xf) may act favoring the transport of counter-ions (ions of opposite sign as the membrane charge) and reducing the transport of co-ions [9, 11]; consequently, it should affect the transport number of the ions in the membrane, ti, which represents the fraction of the total current crossing the membrane transported for one ion (ti = Ii/IT) [39]. Furthermore, since ion transport numbers are related with other diffusive parameters such as the ionic mobility (ui) and diffusion coefficient (Di) by electrochemical relationships [39], their determination provides the basic electrochemical information on the membrane behavior. The effective fixed charge and ion transport numbers can be determined from membrane potential values (∆mbr) by using the Teorell-Meyer-Sievers (TMS) model [40, 41]. According to this model, the membrane potential can be considered as the sum of: a) two Donnan potentials (one at each membrane-solution interface), which are associated to co-ions exclusion; b) a diffusion potential in the membrane caused by the different mobility of the ions inside the membrane pores; that is: mbr = ∆øDon(I) + ∆ødif + ∆øDon(II). Then, the membrane potentials can be expressed as [37]:

 mbr

 4 y v2  1  wU RT  cf  U ln  ln 2 wz F  cv 4 y f  1  wU 

4 y v2  1  w   2 4 y f  1  w 

(1)

where R and F are the gas and Faraday constants, and T is the temperature of the system; U is a parameter related to the ions transport numbers (for 1:1 electrolytes, U = ((D+ - D-)/(D+ + D-) = (t+ - t-) = (2t+ - 1)), yi = KsjCi/│Xf│, being Ksj the partition coefficient or membrane/solution concentration ratio [42]. It should be also pointed out that other factors (dielectric, steric…) described in the literature [43-46] for very narrow nanopores are not considered in this work. Fig. 3 shows the membrane potential as a function of the concentration ratio for the studied samples. For comparison, membrane potentials for an ideal anion-exchanger membrane (t- = 1 and t+ = 0, solid line), which means total exclusion of co-ions (cations in this case), and the diffusion potentials for the NaCl solution (dashed line, calculated by solution ion transport number [47]), are also represented in Fig. 3. As it can be observed, differences in mbr values depending on both the membrane pore size and the material on the membrane surface were obtained. From a qualitative point of view, lower absolute values were obtained when a SiO2, TiO2 or Fe2O3 layer is covering the nanopores of the Ox NPAM membrane (for both Cv < Cf and Cv > Cf), but more negative values were obtained in the case 13

of Al2O3 or ZnO layer (mainly for Cv > Cf). Since all these nanoporous membranes have practically the same pore size and porosity, these results are a clear indication of the influence of the surface material on diffusive ion transport. In the case of Ox+ SiO2+ Al2O3 and Ox+SiO2+AZO membranes, which exhibit the higher differences with the original Ox NPAM membrane, both effects should exist. ideal anion-exchanger ideal anion-membrane

20 20

40

(b) (a)

solution solutiondiffusion diffusionpotential potential

0 0

mbr (mV)

(mV)  (mV) mbr mbr

40 40

-20 -20

20

(b)

solution diffusion potential

0

-20

-40 -40 -60

ideal anion-exchanger

-40

C = 0.01 0.01 NaCl M NaCl Cff =

-2

-1

00

11

22

Cf = 0.01 M NaCl

-2

ln(C /Cf)/C ) ln v(C v f

-1

0

1

2

ln (Cv/Cf)

Fig. 3: Variation of membrane potential with the NaCl solutions concentration ratio for membranes: (a) Ox (□), Ox+Al2O3 (∆), Ox+SiO2 (●), Ox+Fe2O3 (◄), Ox+TiO2 (►), Ox+ZnO (x), Ox+SiO2+Al2O3 (♦) and Ox+SiO2+AZO (*); (b) Ox (□), Ox+Al2O3 (∆), Ox+SiO2+Al2O3 (◊) and Sf (■). The effect of pore size on membrane potentials (without influence of surface material) is presented in Fig. 3.b, where the results obtained for the original Ox NPAM sample, and after surface coverage by Al2O3 and SiO2+Al2O3 layers are shown. Moreover, for comparison reasons, membrane potentials for an alumina membrane fabricated also by the two-step anodization method but using sulfuric acid (Sf NPAM) instead of oxalic acid, and consequently presenting lower geometrical parameters than the Ox NPAM sample, are also shown in Fig. 3.b (for Sf NPAM sample: rp = 11 ± 2 nm,  = 10 % and inter-pore distance Dint = 65 ± 2 nm [11]). According to these results, the reduction in pore size seems to amplify the membrane-ions electrical interactions, increasing the exclusion of the cation (co-ion). Estimation of Xf and t+ values for each alumina membrane was performed by means of Eq. (1), fitting the curves shown in Fig. 3 according to the procedure explained elsewhere [11, 35], and using the pore radius determined from SEM micrographs as a fitting parameter. The 14

obtained values are indicated in Table 3. Fig. 4 shows a comparison of experimental values obtained for the Ox+ZnO membrane (points) and calculated ones using the values for Xf and t+ indicated in Table 3, where the contributions of both Donnan and diffusion potentials are separately indicated. According to the results shown in Fig. 4, membrane potentials at low concentration ratio are basically due to the Donnan potential, but its contribution is reduced at high concentration ratio when diffusion potential presents a predominant contribution. As it can be observed, a very good agreement between experimental and calculated values for the Ox+ZnO membrane, as well as for the other analyzed samples, was obtained, being the differences between experimental and calculated values lower than 7.5 %.

mbr (mV)

40

Ox+ZnO membrane

20 Donnan Contribution

0 Diffusion contribution

-20 -40 -2

-1

0

1

2

Ln(Cv/Cf) Fig. 4: Experimental values obtained for the Ox+ZnO membrane (*); calculated values (solid line) using the following values in Eq. (1): Xf = 1.7x10-2 M, t+ = 0.197 and t- = 0.803. Calculated values for Donnan potential: dashed line; calculated values for diffusion potential: dashed dotted line. On the other hand, taking into account the definition of ion transport number (ti = Ii/IT), the following relationship exists in the case of single salts: t+ + t- = 1, which allows us the determination of t- values. From these results, the anionic permselectivity of each membrane (P-) can also be determined by [39]: P- = (t- - t-o)/ t+o

(2)

where t-o and t+o represent the ion transport numbers in solution. PCl- values for the different membranes are also indicated in Table 3. As expected, the membranes with higher effective fixed charge show higher permselectivity, even for a similar pore size. 15

Table 3: Membrane effective fixed charge (Xf), cation transport number (t+), cationic permselectivity (P+), cationic (D+) and anionic (D-) diffusion coefficients. t+

PCl- (%)

D+ (m2/s)

0.228

40.8

3.5x10-10

1.23x10-9

1.3x10-2

0.173

55.0

3.0x10-10

1.25x10-9

Ox+SiO2

0.7x10-2

0.327

15.0

7.0x10-10

1.25x10-9

Ox+Fe2O3

0.8x10-2

0.306

20.5

5.0x10-10

1.25x10-9

Ox+ZnO

1.7x10-2

0.197

48.8

2.5x10-10

1.00x10-9

Ox+TiO2

0.9x10-2

0.276

28.3

4.8x10-10

1.25x10-9

Ox+SiO2+Al2O3

1.6x10-2

0.146

62.0

2.5x10-10

1.25x10-9

Ox+SiO2+AZO

2.8x10-2

0.088

77.0

1.1x10-10

1.30x10-9

Membrane

Xf (mol)

Ox

1.0x10-2

Ox+Al2O3

D- (m2/s)

Additionally, since the NaCl diffusion coefficient (Ds) in the original Ox NPAM sample was previously determined from salt diffusion experiments (Ds = 4.7x10-10 m2/s [48]), then taking into account the correlation between salt diffusion coefficient and ionic diffusion coefficients (Ds = 2(D+ - D-)/(D+ + D-) for 1:1 electrolytes), and between ion transport numbers ratio and ion diffusion coefficients ratio (t+/t- = D+/D-), the values for the anion and cation diffusion coefficients (D+ and D-) in the pores of the Ox NPAM were also estimated. Moreover, assuming the same order of magnitude for the ionic diffusion coefficients in the ALD modified alumina membranes, D+ and D- values through the pores of the different samples were also estimated, being the values obtained for these parameters also shown in Table 3. It should also be indicated that an equilibrium value around 2x10-10 m2/s for the Na+ diffusion coefficient in the pores of a NPAM, with practically a similar pore size than the Ox+Al2O3, was already determined from Fickian diffusion measurements using labelled radiotracer Na+ [49], and its similarity with the one obtained in this work can be considered as a prove of the reliability for the values shown in Table 3. A correlation of both, cation diffusion coefficient and salt diffusion coefficient in the membrane pores with the effective fixed charge concentration is shown in Fig. 5. The value of NaCl diffusion coefficient in solution (Dso) is also indicated in Fig. 5.b, which clearly shows the significant control of Xf on the diffusive transport across these ALD functionalized membranes.

16

-9

DNa+ (m /s)

1.0x10

Ds (m /s)

(a)

-9

Ds

(b)

o

2

2

1,5x10

-9

1,0x10

-10

5.0x10

-10

5,0x10

(b) 0,0 0,00

0.0 0.01

0.02

Xf (M)

0.03

0,01

0,02

0,03

Xf (M)

Fig. 5: Na+ (a) and NaCl (b) diffusion coefficients in the nanopores of the different aluminabased membranes as a function of the effective fixed charge concentration on the membrane surface. Conclusions

Nanoporous alumina membranes with high aspect ratio of around 2000 were successfully morphological and chemically modified through the ALD technique, by covering them with different metal oxides inside the nanopores, according to SEM images and EDS analysis. The effect of both, pore size reduction (25-30 %) and chemical nature modification of membrane surface on the characteristic transport parameters of a typical model electrolyte (NaCl) was determined by analyzing membrane potentials. The obtained results show the electropositive character of all the studied alumina-based membranes. In addition, the significant effect of the membrane effective fixed charge on anion permselectivity and salt diffusion coefficients for membranes with pore size in the range of these nano-engineered samples have also been demonstrated. Consequently, the thin layer metal oxide coverage by ALD technique seems to be an easy and efficient way of membrane modification by both pore size reduction and selective control of diffusive transport. Furthermore, the study performed in double layered coated samples shows the great interest of sequential modification and its effect on the membrane performance. Acknowledgments

Financial support through Spanish MINECO (Research Projects CTQ2011-27770, MAT2013-48054-C2-2-R and MAT2016-76824-C3-3-R), together FICyT and Consejería de Economía y Empleo from Principality of Asturias under project Nº GIC-FC-15-GRUPIN1417

085, are gratefully acknowledged. The technical assistance provided by the ScientificTechnical Services (SCTs) of the University of Oviedo is also recognized. Dr. R. Zierold, at University of Hamburg (Germany) is acknowledged for his support on ALD deposition. References [1] J. Hohlbein, M. Steinhart, C. Schiene-Fisher, A. Benda, M. Hof, Ch. G. Hübner, Small 3 (2007) 380-385. [2] X. Jiang, N. Mishra, J.N. Turner, M.G. Spencer, Nonofluid. 5 (2008) 695-701. [3] R. Kennard, W.J. de Sisto, M.D. Mason, Appl. Phys. Lett., 97 (2010) 213701. [4] R. Karnik, R. Fan, M. Yue, D. Li, P. Yang, A. Majumdar, Nano Letters 5 (2005) 943-948. [5] Ch. Wang, J.J. Xu, H.Y. Chen, X.H. Xia, Sci. China Chem. 55 (2012) 453-468. [6] C. Boss, E. Meurville, J-M. Sallese, P. Ryser, J. Membr. Sci. 401/402 (2012) 217-221. [7] Q. Hu, Y.J. Choi, C.J. Kang, H.H. Lee, T.-S. Yoon, J. Ind. Eng. Chem. 24 (2015) 293– 296 [8] D. Hua, P.Pa. Li, Y. Wu, Y. Chen, M. Yang, J. Dang, Q. Xie, J. Liu, X.-Y. Sun, J. Ind. Eng. Chem. 19 (2013) 1395–1399. [9] E.A. Bluhm, N.C. Schroeder, E. Bauer, J.N. Fife, R.M. Chamberlin, K.D. Abney, J.S. Young, G.D. Jarvinen, Langmuir 16 (2000) 7056-7060. [10] S.J. Kim, Y-A Song, J. Han, Chem. Soc. Rev. 39 (2010) 912-922. [11] V. Romero, V. Vega, J. García, R. Zierold, K. Nielsch, V.M. Prida, B. Hernando, J. Benavente, ACS Appl. Mater. Interfaces 5 (2013) 3556-3564. [12] A.A. Belkova, A.I. Sergeeva, P.Y. Apel, M.K. Beklemishev, J. Membr. Sci. 330 (2009) 145-155. [13] J.A. Quinn, J.L. Anderson, W.S. Ho, W.J. Petzny, Biophys. J. 12 (1972) 990-1007. [14] I. Chlebny, B. Doudin, J.-Ph. Ansermet, Nanostruct. Mater. 6 (1993) 637-642. [15] H. Masuda, K. Fukuda, Science 268 (1995) 1466-1468. [16] R. Martín, C.V. Manzano, M. Martín-González, Mesopor. Mat. 151 (2012) 311-316. [17] V. Romero, V. Vega, J. García, V.M. Prida, B. Hernando, J. Benavente, J. Colloids Interface Sci. 376 (2012) 40-46. [18] M.H. Kim, A. Ayral Ch-B. Park, J-H. Choy, J-M. Oh, J. Nanosci. Nanotechnol. 11 (2011) 1656-1659. [19] A. Mehrparvar, A. Rahimpour, J. Ind. Eng. Chem. 28 (2015) 359-368. [20] E. Magnone, H.J. Lee, J.W. Che, J.H. Park, J. Ind. Eng. Chem. 42 (2016) 19-22.

18

[21] F. Li, Y. Yang, Y. Fan, W. Xing, Y. Wang, J. Membr. Sci. 397/398 (2012) 17-23. [22] R.A. Mulvenna, J.L. Weidman, B. Jing, J.A. Pople, Y. Zhu, B.W. Boudouris, W.A. Phillips, J. Membr. Sci. 470 (2014) 246-256. [23] L.J. Small, D.R. Wheeler, E.D. Spoerke, Nanoscale 7 (2015) 16909-16920. [24] S. M. George, Chem. Rev., 110 (2010) 111–131. [25] A. Spende, N. Sobel, M. Lukas, R. Zierold, J.C. Riedl, L. Gura, I. Schubert, J.M. Montero Moreno, K. Nielsch, B. Stühn, Ch. Hess, Ch. Trautmann, M.E. Toimil-Molares, Nanotechnology 26 (2015) 335301. [26] V.M. Prida, K.R Pirota, D. Navas, A. Asenjo, M. Hernández-Vélez, M. Vázquez, J. Nanosci. Nanotechnol. 7 (2007) 272-285. [27] R. L. Puurunen, J. Appl. Phys. 97 (2005) 121301. [28] J. Bachmann, R. Zierold, Y. T. Chong, R. Hauert, C. Sturm, R. Schmidt-Grund, B. Rheinländer, M. Grundmann, U. Gösele, K. Nielsch, Angew. Chem. Int. Ed. 47 (2008) 6177 –6179. [29] J. Bachmann,J . Escrig, K. Pitzschel, J. M. Montero Moreno, J. Jing, D. Görlitz, D. Altbir, K. Nielsch, J. Appl. Phys. 105 (2009) 07B521. [30] E. Guziewicz, I. A. Kowalik, M. Godlewski, K. Kopalko, V. Osinniy, A. Wójcik, S. Yatsunenko, E. Łusakowska, W. Paszkowicz, M. Guziewicz, J. Appl. Phys. 103 (2008) 033515. [31] J.H. Heo, H. Ryu, W.-J. Lee, J. Ind. Eng. Chem. 19 (2013) 1638-1641. [32] X. Meng, M. Norouzi Banis, D.Geng, X. Li, Y.Zhang, R.Li,H. Abou-Rachid, X. Sun, Applied Surface Science 266 (2013) 132–140. [33] T. Tynell, M. Karppinen, Semicond. Sci. Technol. 29 (2014) 043001. [34] C. A. Schneider, W.S. Rasband, K.W. Eliceiri, Nature Methods 9 (2012) 671-675. [35] V. Romero, M .I. Vázquez, J. Benavente, J. Membr. Sci. 433 (2013) 152-159. [36] K. Sollner, J. Macromol. Sci-Chem A3 (1969) 1-86. [37] H. Strathmann, J. Membr. Sci. 9 (1981) 121-189. [38] Y. Kimura, H-L. Lim, T. Iijima, J. Membr. Sci. 18 (1984) 285-296. [39] N. Lakshminarayanaiah, Academic Press, New York, USA, (1969). [40] K.H. Meyer, J.F. Sievers, Helv. Chim. Acta 19 (1936) 649-664. [41] T. Teorell, Discuss. Faraday Soc. 21 (1956) 9-26. [42] M. Mulder, Kluwer Academic Publishers, Dordrecht, 1992. [43] A. Szymczyk, P. Fievet, J. Membr. Sci. 252 (2005) 77-88. [44] A. Yaroshchuk, O. Zhukova, M. Ulbricht, V. Ribitsch, Langmuir 21 (2005) 6872-6882.

19

[45] A. Escoda, Y. Lanteri, P. Fievet, S. Déon, A. Szymczyk, Langmuir 26 (2010) 1462814635. [46] J. Benavente, V. Silva, P. Prádanos, L. Palacio, A. Hernando, G. Jonson, Langmuir 26 (2010) 11841-11849. [47] R.A. Robinson, R.H. Stokes, Butterworths, London (1959). [48] M.I. Vázquez, V.Romero, V. Vega, J. García, V.M. Prida, B. Hernando, J. Benavente, Nanomaterials 5 (2015) 2192-2202. [49] V. Romero, M.I. Vázquez, S. Cañete, V. Vega, J. García, V. M. Prida, B. Hernando, J Benavente, J. Phys. Chem. C 117 (2013) 25513-25518.

20

Table 1: Precursors and deposition conditions employed for the ALD deposition coating of the Ox NPAMs pores with different functional oxides. Oxide coating Al2O3

ALD precursors

Precursor Substrate Growth rate temperature (ºC) temperature (ºC) (nm/cycle)

Ref.

H2O

60

200

0.106

27

180

0.06

11, 28

230

0.022

29

200

0.146

30, 31

200

0.05

32

200

0.144

33

Trimethylaluminum 20

SiO2

H2O

60

O3

20

(3-Aminopropyl)

100

Triethoxysilane O3

20

Ferrocene

100

H2O

60

Diethylzinc

20

H2O

60

Fe2O3

ZnO

TiO2

Titanium

(IV) 75

tetraisopropoxide H2O

60

Diethylzinc

20

AZO (ZnO:Al)

Trimethylaluminum 20

21

Table 2: Morphological parameters characteristic of the studied Ox NPAMs: pore radii (rp), inter-pore distance (Dint), and estimated average porosity (), calculated according to ref. [15]. Membrane

rp (nm)

Dint (nm)

%)

Ox

16 ± 2

105 ± 5

8

Ox+Al2O3

12 ± 2

105 ± 5

5

Ox+SiO2

12 ± 2

105 ± 5

5

Ox+Fe2O3

12 ± 1

105 ± 5

4

Ox+ZnO

12 ± 1

105 ± 5

5

Ox+TiO2

13 ± 2

105 ± 5

6

Ox+SiO2+Al2O3

11 ± 1

105 ± 5

4

Ox+SiO2+AZO

11 ± 1

105 ± 5

4

22

Table 3: Membrane effective fixed charge (Xf), cation transport number (t+), cationic permselectivity (P+), cationic (D+) and anionic (D-) diffusion coefficients. t+

PCl- (%)

D+ (m2/s)

0.228

40.8

3.5x10-10

1.23x10-9

1.3x10-2

0.173

55.0

3.0x10-10

1.25x10-9

Ox+SiO2

0.7x10-2

0.327

15.0

7.0x10-10

1.25x10-9

Ox+Fe2O3

0.8x10-2

0.306

20.5

5.0x10-10

1.25x10-9

Ox+ZnO

1.7x10-2

0.197

48.8

2.5x10-10

1.00x10-9

Ox+TiO2

0.9x10-3

0.276

28.3

4.8x10-10

1.25x10-9

Ox+SiO2+Al2O3

1.6x10-2

0.146

62.0

2.5x10-10

1.25x10-9

Ox+SiO2+AZO

2.8x10-2

0.088

77.0

1.1x10-10

1.30x10-9

Membrane

Xf (mol)

Ox

1.0x10-2

Ox+Al2O3

23

D- (m2/s)