Cunyu Li,Yun Ma,Hongyang Li,Guoping Peng,2,*
1Pharmacy College,Nanjing University of Chinese Medicine,Nanjing 210023,China
2Jiangsu Collaborative Innovation Center of Chinese Medicinal Resources Industrialization,Nanjing 210023,China
1.Introduction
Ultra filtration(UF)has been applied in various types of industries such as sewage treatment,seawater desalinization,biology and medicine.For example,Harmen J.Zwijnenberg used a UF membrane to produce a higher quality protein from potato fruit juice[1].Akon Higuchi investigated the use of immobilized DNA membranes for chiral separation of phenylalanine[2].Wanget al.[3]developed a membrane-based cost-effective process for the separation and purification of superoxide dismutase(SOD)from garlic.The membrane molecular weight cut-off(MWCO)and its distribution are very important parameters for membrane quality and membrane transport mechanisms.The pore size distribution dominates the separation characteristics of asymmetric membranes,which can be used to predict the MWCO of porous membranes and the rejection for different solute molecules or particles.
Different experimental methods can be employed to evaluate membrane pore size,such as microscopic,bubbling test,liquid displacement and thermoporometry[4–6],these methods mentioned above all have certain advantages.Furthermore,existing detection techniques for the MWCO of UF membranes calls for the use of standard solutes with known molecular weight such as dextran,protein or polyethylene glycol(PEG)[7,8].These methods mentioned above have their specific characteristics for various membranes with different pore sizes,and also exhibits some limitations in membrane type and pore size.One of the obvious defects was membrane fouling,for example,dextran fouling was visualized using atomic force microscopy(AFM)and quantified by ATR-IR spectroscopy and by the mass balance in simultaneous diffusion–adsorption measurements(SDAM)[9].The fouling mechanisms of PEG involved in the UF were investigated by Velaet al.[10],and the membrane pollution characters of BSA were detected by dye tests[11],electron spin resonance(ESR)spectroscopy[12],etc.
The MWCO of UF membranes is defined as the molecular weight of a solute that has a rejection value of 0.9,which might be used as selection parameters in mixture separation[13].In the progress of membrane MWCO testing,the membrane retention rate can be critically affected by the standard solutes molecular weight distribution.But the molecular weight distribution of standard solutes varies with different factories,in other words,the same molecular weight of dextran,BSA and PEG had different molecular weight distribution,respectively.And eventually the quality of standard solutes had an impact on the accuracy of membrane MWCO,and the above method was complicated in operation,detection and analysis.Therefore,a convenient and accurate method was urgently needed for the determination of UF membranes MWCO.
After abundant experiments based on UF separation of compounds,Panax notoginseng saponins,a low-costand accessible water extract was fromPanax notoginsengor Americanginseng,was selected as standard substance,which had four representative elements:Notoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1andGinsenoside Rd.The four solutes had similar molecular weight 932–1108 with different surface activity and existential state in aqueous solution and showed different retention rate with series of MWCO UF membranes.In this paper,the membrane operating condition and concentration polarization were discussed,and the relationship between solutes retention rate and membrane MWCO was analyzed by the “three-step”method.The method was developed to predict the MWCO of different manufacturer UF membrane.In addition,the measuring accuracy and convenience of this method were also measured and compared.
2.Materials and Methods
2.1.Materials
In order to determine the MWCO of UF membranes,series of regenerated cellulose membranes were selected as standard MWCO membranes in this paper,which manufactured by Millipore Co.in USA,and these membranes had thesame properties,such as morphology,except pore sizes.The detail membrane properties are listed in Table 1.The UF apparatus is shown in Fig.1.
Master flex®L/S™peristaltic pump(pumphead:easy-load®Model 7017-21,Millipore Corporation,USA)was used to circulate the feed solution,which can provide the constant flux at different feed pressure.Notoginsenoside R1(C47H80O18,CAS:#80418-24-2),Ginsenoside Rg1(C42H72O14,CAS:#22,427–39-0),Ginsenoside Rb1(C54H92O23,CAS:#41,753–43-9)andGinsenoside Rd.(C48H82O18,CAS:#52,705–93-8)were purchased from National Institute for the Control of Pharmaceutical and Biological Products in China,Panax notoginseng saponins(Cat.No.:090701,from Yunnan Yuxi Wanfang natural medicines Co.,Ltd.in China)was used to characterize the rejection coefficient of the hollow fiber membranes in salt-free Milli-Q water.Six untested commercial UF membranes and series of molecular weight dextrans were selected to verify the feasibility and superiority of this method and the detail membrane properties are listed in Tables 2 and 3.
2.2.Membrane characterization
The size of the pores as well as pore size distribution is an important parameter deciding the separation performance[14].We obtained volumetric permeation flux and solute retention rate during solute separation experiments with filtration membranes.Both the flux and retention rate are strongly dependent on the structure of the membrane.Therefore,if the relationship between the flux and retention rate and the membrane structure is known,we can characterize the membrane structure,such as pore size,pore size distribution,pore density,and so on.The relationship is founded on the molecular transport through the membrane.We can also interpret diffusion and sieving measurements in terms of the pore size distribution.
2.2.1.The influence of operating pressure
All UF experiments were performed in a thin channel module described in Fig.1 with an effective membrane area of 0.5 m2,adjustable feed flow,and pressure.Prior to a UF experiment,deionized water was circulated in the test loop until steady state.
Pressure influence measurements were carried out withPanax notoginseng saponins.The test solutions were prepared by dissolving reweighed amounts ofPanax notoginseng saponinsin deionized water at a concentration of 10 g·L-1.Test conditions were pressures of 0.01,0.05,0.1,0.15,0.2 and 0.25 MPa,circulation velocity was regulated byMillipore peristaltic drive pump.When operating pressures steady state was reached in UF,the concentrations of the bulk and permeate were analyzed with high performance liquid chromatography,Agilent 1100.And the retention rate was calculated according to Eq.(1),whereCfandCpare the solute concentrations in feed and permeate solution.Each measurement was performed in triplicate.
Table 1The properties of the standard membranes
Fig.1.UF process diagram.1.Stock solution tank;2.inlet pipe;3.peristaltic pump;4.pressure gage;5.UF membrane;6.pressure gage;7.rejected solution;8.speed regulator valve;9.ultra filtrate;10.ultra filtrate tank;11.HPLC;12.chromatogram;13.membrane pores.
Table 2Parameters of being-measured UF membranes
Table 3Dextran standards used in the feed solution
2.2.2.MWCO determination experiments
Before starting the MWCO measurement,each membrane module was stored in 25 mmol·L-1sodium hydroxide aqueous solutions to protect the membranes from the contaminant of bacteria,etc.and rinsed by deionized water until the pH value was neutral before using.
The same composition and concentration ofPanax notoginseng saponinwas used for tests.In order to ensure that the separation performance of the membranes was not changed during filtration experiments,feed solution, filtrate and retention rate solution were repeated and compared to the initial results.During the experiment,the pipeline of feed solution, filtrate,and rejected solution were placed in the same tube.Before sampling analysis,membrane module was pressurized at the test pressure for minimum of2h to reach the steady state conditions.The retention rate was calculated using Eq.(1)with series of standard membranes in Table 1.Subsequently,several diagrams were designed using the retention rates as the ordinate and MWCOs similar to the abscissa to find the preliminary relationship.
2.2.3.Standard curvefitting
The concentration polarization at the membrane surface and the interference in pore transport betweenPanax notoginseng saponinare not considered.That is to say,this computing method can only be applied to analysis the pore size distribution of the UF membranes without concentration polarization.The regression curve of solutes retention rate and membrane MWCO was analyzed using chart tools of Excel 2003,the algorithm was “exponential or logarithmic trend line”.
2.2.4.Method verification
2.2.4.1.Feasibility verification.The six commercial UF membranes were installed in apparatus as described in Fig.1 to verify the feasibility of the above method.The feed solution and filtrate sample were collected and measured using a HPLC method,and the retention rates were calculated by Eq.(1),each membrane was performed in triplicate and average values are given.Afterwards,the MWCO of different membranes were calculated by the standard regression curve with the corresponding retention rate.
2.2.4.2.Dextran rejection verification.The feed solution consisting of 1 g·L-1dextrans with various molecular weight was prepared according to the concentrations given in Table 3.The dextran standards were dissolved in 0.05 mol·L-1reagent grade KH2PO4.The pH of the buffer was adjusted to 7.0±0.1 by adding HPLC grade NaOH.The pH was measured using a Sartorius Model PB-10 pH meter.The same buffer was used as the mobile phase for the gel permeation chromatography(GPC)analysis described below.
The six commercial UF membranes were equilibrated for 1 h under permeate recycle using the UF apparatus at a flow rate of 0.1 ml·min-1,as shown in Fig.1.These conditions have been shown to minimize concentration polarization and gel layer formation[15].All experiments were conducted in triplicate and average values are given.After equilibration,multiple permeate and feed samples were collected and analyzed using an Agilent Technologies Model 1050 HPLC system using a Shimadzu Technologies RID-6A refractive index detector.The gel permeation column used was a TSKGEL G3000SWxL.Column temperature was maintained at(25±1)°C.The HPLC was calibrated using solutions of the narrow molecular weight dextran standards listed in Table 2.Dextran retention rates were obtained from the GPC data by Eq.(1).
3.Results and Discussion
3.1.Membrane operating pressure
Clarifying the operating pressure effect of different MWCO membranes is very important for obtaining reliable retention rate results.As the pressure was more than 0.15 MPa the solutes retention rate had been flattened,the solutes retention rate at a given pressure decreased with increasing MWCO of the membrane.Hwang and Sz[16]studied the relationship between trans membrane pressures and resistance of dextran,and found that all filtration resistances increased linearly with increasing trans membrane pressure and the pore size reduction is caused mainly by dextran adsorption.Meanwhile,the conclusion was not consistent with results as shown in Fig.2,mainly because of the fouling features ofNotoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1andGinsenoside Rdwith regenerated cellulose membrane.This suggests that concentration polarization can be neglected as the operating pressure reached 0.15 MPa,so in this experiment,the operating pressure was controlled at this design pressure.
Fig.2 shows the retention rate dynamic changes ofNotoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1andGinsenoside Rdwith seven standard membranes.Moreover,as Fig.2(a)and(c)indicates,although the retention rate of the solutes increased with the operating pressure, the retention rate curves difference betweenNotoginsenoside R1andGinsenoside Rb1was significant.This can be explained by the difference in the interfacial or surfactant effect ofNotoginsenoside R1andGinsenoside Rb1with UF membranes,as seen in Fig.2(b)and(d),the retention rate difference was disappearing as the membrane MWCO increasing,which indicates all retention rate data applicable to molecular sieving theory[17].
Fig.2.Effects of operating pressure on solutes retention rate for seven standard membranes.(a)Notoginsenoside R1,(b)Ginsenoside Rg1,(c)Ginsenoside Rb1and(d)Ginsenoside Rd.
3.2.Standard curve calculation
3.2.1.Cumulative retention rate for MWCO preliminary judgments
The HPLC chromatographic peak area change ofNotoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1andGinsenoside Rdwith parts MWCO UF membrane were shown in Fig.3.It can be seen that when the membrane MWCO was 100000,the retention rate did not change with the four chemical compositions,indicating that test range of MWCO did not exceed 100000.
Fig.3.High performance liquid chromatogram of Notoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1and Ginsenoside Rd with 10000 filtrate,30000 filtrate,50000 filtrate,100000 filtrate and original solution.
Fig.4.Panax notoginseng saponin cumulative retention rate of series membranes.
Cumulative retention rate was the sum of retention rate,which can be more intuitive appreciation for the separation differences among UF membranes.Meanwhile,the cumulative retention rate provided a simple way to predict the MWCO range of tested UF membranes.Fig.4 showsPanax notoginseng saponincumulative retention rate of series membranes. The separation behaviors of the four solutes have a significant difference. However, the four main components inPanax notoginseng saponin, Notoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1andGinsenoside Rd,which had similar structure and MW.This special separation phenomenon provided a new way to calculate MWCO of UF membrane,and find the relationship between existence states of the components and molecular cutting separation process.Panax notoginseng saponinwas chosen as standard substance,but the surface activity was different,for instance,theNotoginsenoside R1andGinsenoside Rg1were hemolytic,asGinsenoside Rb1andGinsenoside Rdwere non-hemolytic[18].Therefore,the critical micelle concentration(CMC)ofGinsenoside Rb1andGinsenoside Rdwere lower thanNotoginsenoside R1andGinsenoside Rg1,so it's easier for micelles to be born.In membrane separation progress,the actual rejected solutes may be the micelles with different MW.Another way to understand Fig.4,compared withNotoginsenoside R1andGinsenoside Rg1,Ginsenoside Rb1andGinsenoside Rdhad greater molecular weight and harder permeation through the porous membranes.And this retention rate difference ultimately results in a useful method to preliminary judgments of untested membrane pore sizes range.
3.2.2.Correlation analyses between retention rate and MWCO
The rejected behavior ofNotoginsenoside R1with series MWCO was similar withGinsenoside Rg1.The retention rates were decreased with the membrane MWCO increasing.The growth rate of rejection with MWCO increasing was slowed as MWCO＞10000.Meanwhile,Ginsenoside Rb1andGinsenoside Rdwere found to have different rejection characteristics fromNotoginsenoside R1andGinsenoside Rg1as the four solutes had similar structures and molecular weights,the growth rate of rejection was obviously increased as MWCO＞10000.To improve the calculation accuracy of membranes MWCO,the range of 1000–100000 membranes and retention rates were selected to fit standard curves.The standard curves of the four solutes were calculated by membrane MWCOs asX-axis and changes of retention rate during the experimentsasY-axis.Theregressionequations areshowninTable4,whereRwas solute retention rate,andMwas membrane MWCO.
3.3.Method validation
The results of cumulative retention rate with the untested commercial UF membranes are shown in Fig.5.As can be seen,the MWCO ranges were 10000(membrane I),5000–10000(membrane II),30000(membrane III),50000(membrane IV),50000(membrane V)and 50000(membrane VI)by compared toPanax notoginseng saponincumulative retention rate in Fig.4.
Fig.5.Panax notoginseng saponin cumulative retention rate of being-measured UF membranes.
Theexperimental MWCO of UF membranes wascalculated by linear equation in Table 4 and the results were listed in Table 5.Also in this case,the experimental MWCO was in consistent with the labeled MWCO of membrane I,it can be seen that the hydrophilic material of polyethersulfone and modified cellulose has similar trans-membrane behavior.Membrane II has a lager determined MWCO,this can be explained by the membrane damage of long usage period,as reported in Table 2.The phenomenon of MWCO variation caused by membrane fouling,material loss and so on is very common in normal operations.However,if the leakage and breakage points are not found in time,it will result in great deviation.It can be seen that both membrane III and IV generally have broader range than the labeled MWCO,due to membrane preparation.The results of membrane IV,V and VI with same labeled MWCO show that the membrane pore size ranges were different,due to standard substance or membrane preparation of the three manufacturers.
Fig. 6 show that the retention rate curves of the six being-measured UF membranes. In plotting the retention rate curves, the labeled MWCO 50000 of Vwas very similar to VI, the results was verified in Table 5. The membrane MWCO was a very important parameter for membrane transport mechanisms, so the rejection characteristics can be selected as index to compare the membrane difference. The retention rate 90% as index,we can found that, the MWCO of membrane IV was the largest, VI, V, III, I and the II gradually become smaller. The similar results were shown in Table 5,so we ensure the experimental MWCO byPanax notoginseng saponinis close to the measured modal with series of dextran.
As a standard material for membrane MWCO analysis,dextran has wide applications in membrane industry.However,a limitation of the dextran method,as acknowledged by the researchers,is the difficulty involved in obtaining experimental results.The MWCO of UF membrane was evaluated and obtained in terms of a diagram with retention rate and series of dextrans.In other words,three or more molecular weights of dextran were needed to test one UF membrane with retention ratecurves.Compared with the dextran,Panax notoginseng saponinhas more advantages which can help researchers improve testing cost and period.In addition,UF membrane MWCO of 1000 to 100000 can just be detected byPanax notoginseng saponin.But some limitations were found in the studies,such as the detecting range was con fined to 1000–100000,the structure stability ofPanax notoginseng saponinwas hydrolyzed to aglycone and sugars in the condition of strong acid and strong alkaline and so on.Therefore,this method does not completely meet the MWCO detected requirement of UF membrane of 1000–1000000,and the suitable solution environment was as well as the guarantee for the accurate results.
Table 4The correlation between MWCO and retention rate
Table 5Fitting MWCO of being-measured UF membranes
Fig.6.Retention rate curves of being-measured UF membranes with dextran.
4.Conclusions
We used retention rate as a parameter by a simple natural plant ingredient to calculate membrane MWCO and its range.In membrane separation industry,new membrane material is the prevailing trend[19],meanwhile,a convenient and feasible method was supplied for comparing membrane MWCO difference and predicting MWCO range,and some conclusions can be made as follows:
1.Three steps for prediction the MWCO and MWCO range of UF membranes.Firstly,standardized operation;secondly,the MWCO preliminary judgment by cumulative retention rate;thirdly,the specific MWCO and MWCO range were calculated by exponential and logarithmic equation.
2.The theoretical prediction of the MWCO and MWCO range of UF membranes by two steps is first developed and verified with hollow fiber,spiral and plain membranes of commercial brand.
3.The MWCO range of UF membranes can be divided into two zones mainly due to the retention rate difference amongNotoginsenoside R1,Ginsenoside Rg1,Ginsenoside Rb1andGinsenoside Rd.Zone I,1000–10000;Zone II,10000–100000.
[1]Y.Zhou,Z.L.Dai,D.Zhai,C.J.Gao,Surface modification of polypiperazine-amide membrane by self-assembled method for dye wastewater treatment,Chin.J.Chem.Eng.23(6)(2015)912–918.
[2]H.Q.Zhang,Z.X.Zhong,W.X.Li,W.H.Xing,W.Q.Jin,River water purification via a coagulation-porous ceramic membrane hybrid process,Chin.J.Chem.Eng.22(1)(2014)113–119.
[3]J.Wang,J.Liu,J.Lu,Z.Cui,Isolation and purification of superoxide dismutase from garlic using two-stage ultra filtration,J.Membr.Sci.352(2010)231–238.
[4]J.Ren,Z.Li,F.S.Wong,A new method for the prediction of pore size distribution and MWCO of ultra filtration membranes,J.Membr.Sci.279(2006)558–569.
[5]H.de Balmann,R.Nobrege,The deformation of dextran moleculares.Causes and consequences in ultra filtration,J.Membr.Sci.40(1989)311–327.
[6]G.Schock,A.Miquel,R.Birkenberger,Characteristization of ultra filtration membranes:cut-off determination by gel permeation chromatography,J.Membr.Sci.41(1989)55–67.
[7]S.R.Wickramasinghe,S.E.Bower,Z.Chen,A.Mukherjee,S.M.Husson,Relating the pore size distribution of ultra filtration membranes to dextran rejection,J.Membr.Sci.340(2009)1–8.
[8]R.Rohani,M.Hyland,D.Patterson,A re fined one- filtration method for aqueous based nano filtration and ultra filtration membrane molecular weight cut-off determination using polyethylene glycols,J.Membr.Sci.382(2011)278–290.
[9]H.Susanto,S.Franzka,M.Ulbricht,Dextran fouling of polyethersulfone ultra filtration membranes—causes,extent and consequences,J.Membr.Sci.296(2007)147–155.
[10]M.C.V.Vela,S.Á.Blanco,J.L.García,E.B.Rodríguez,Analysis of membrane pore blocking models applied to the ultra filtration of PEG,Sep.Purif.Technol.62(2008)489–498.
[11]S.Platt,M.Nyström,Amido black staining of ultra filtration membranes fouled with BSA,Desalination214(2007)177–192.
[12]K.C.Khulbe,T.Matsuura,S.Singh,G.Lamarche,S.H.Noh,Study on fouling of ultrafiltration membrane by electron spin resonance,J.Membr.Sci.167(2000)263–273.
[13]C.Zhao,X.Zhou,Y.Yue,Determination of pore size and pore size distribution on the surface of hollow fiber filtration membranes:A review of methods,Desalination129(2000)107–123.
[14]H.Kamusewitz,M.Schossig-Tiedemann,M.Keller,D.Paul,Characterization of polymeric membranes by means of scanning force microscopy(SFM)in comparison to resultsofscanningelectronmicroscopy(SEM),Surf.Sci.377-379(1997)1076–1081.
[15]A.L.Zydney,A.Xenopoulos,Improving dextran tests for ultra filtration membranes:effects of device format,J.Membr.Sci.291(2007)180–190.
[16]K.J.Hwang,P.Y.Sz,Effect of membrane pore size on the performance of cross- flow micro filtration of BSA/dextran mixtures,J.Membr.Sci.378(2011)272–279.
[17]J.K.Leypoldt,Determining pore size distributions of ultra filtration membranes by solute sieving-mathematical limitations,J.Membr.Sci.31(1987)289–305.
[18]H.X.Sun,Q.F.Zheng,Haemolytic activities and adjuvant effect ofGynostemma pentaphyllumsaponins on the immune responses to ovalbumin in mice,Phytother.Res.19(2005)895–900.
[19]Y.Q.Zhu,Z.Wang,C.X.Zhang,J.X.Wang,S.C.Wang,A novel membrane prepared from sodium alginate cross-linked with sodium tartrate for CO2capture,Chin.J.Chem.Eng.21(10)(2015)1098–1150.
