Lebensm.-Wiss. u.-Technol.

, 32, 290}298 (1999)

Apple Juice Clari"cation Using Micro"ltration and

Ultra"ltration Polymeric Membranes
B. Girard* and L. R. Fukumoto

Agriculture and Agri-Food Canada, Paci"c Agri-Food Research Centre, 4200 Highway 97S,
Summerland, B.C., V0H 1ZO (Canada)
(Received July 7, 1998; accepted April 12, 1999)

The yux behavior of polyethersulfone (0.2 lm), polyvinylidene -uoride (0.2 lm), cellulose (0.2 lm, 100 kDa, 30 kDa, 10 kDa), and
polysulfone (0.2 lm, 100 kDa, 30 kDa, 10 kDa) membranes was examined during dead-end ,ltration of apple juice. Membranes with
molecular weight cut-o+ of 30 and 100 kDa had superior -ux performance to 0.2 lm or 10 kDa membranes. A cross-ow system
equipped with various tubular polymeric membranes was also used to clarify apple juice at a temperature of 503C, a crossyow velocity of
3.3 m/s and a transmembrane pressure of 414 kPa. Steady state yuxes increased as the molecular weight cut-ow increased from 9 to
200 kDa. When challenged with P. diminuta, log reduction values between 6 and 7 were obtained with the cross-ow tubular polymeric
system. Membranes between 20 and 200 kDa produced juices with similar initial characteristics that underwent comparable changes
during storage at 4, 25 and 353C over 4}16 wk. The impact of xltration through the 9 kDa membrane was however noticeable on the
physico-chemical properties since the apple juice had a green tint, lower soluble solids, lower yavanol content, and experienced minimal
changes in browning and turbidity.

Keywords: micro"ltration; ultra"ltration; polymeric membrane; apple juice; physicochemical properties; microbial challenge;
sensory evaluation

Introduction membranes. The 0.1 km membrane permeate was darker

and had higher turbidity, total solids and soluble solids.
The use of polymeric membranes is widespread for the Rao et al. (4) found that apple juice permeate
clari"cation of apple juice by ultra"ltration. Padilla and from a 30 kDa polyamide membrane contained more
McLellan (1) studied the "ltration of apple juice through volatiles than permeate from a 50 kDa polysulfone mem-
10, 50, 100 and 500 kDa polysulfone hollow "ber mem- brane.
branes. Initial #ux increased with molecular weight cut- Various membrane con"gurations of polymeric mem-
o! (MWCO) but the "ltered juice from all membranes branes have been reported for ultra"ltration of apple
had similar acidity, soluble solids and sensory quality. juice including hollow "ber (1, 3, 5), spiral-wound (3), and
Although phenolic content was lower with the 500 kDa, plate and frame (2). Although these con"gurations have
it generally increased with MWCO as did total solids. better packing densities, they usually have narrow chan-
The juice from the 100 and 500 kDa membranes had nels which make these designs more susceptible to mem-
higher turbidity and brown color than the other mem- brane fouling than tubular con"gurations. Porter (6) lists
branes. All juices became turbid and browned during various other advantages and disadvantages of the vari-
storage at 43 3C over 6 months but no signi"cant changes ous con"gurations. Since freshly pressed apple juice
were observed at 18 3C. Sheu et al. (2) used 20 and 50 kDa contains a considerable amount of particulate matter,
polysulfone membranes in a plate and frame con"gura- pre"ltration is often necessary when using designs with
tion to "lter apple juice. The average #ux was 20 to 40 narrow channels. RoK sch (7) found apple juice could be
L/m h higher for the 50 than the 20 kDa membrane. Wu concentrated approximately 70 times using a 25 mm
et al. (3) compared 0.1 km ceramic, 50 kDa poly- diameter tubular system.
sulfone hollow "ber and 5 kDa polysulfone spiral-wound Few studies have examined the e!ect of membrane ma-
terial on ultra"ltration of apple juice. This paper focused
on the application of polymeric membranes for apple
PARC Contribution No. 1051 juice "ltration and paralleled additional work on ceramic
* To whom correspondence should be addressed. membrane published elsewhere (8). Initially, a stirred cell
0023-6438/99/050290#09 $30.00 Article No. fstl.1999.0554
All articles available online at http://www.idealibrary.com on

290
 lwt/vol. 32 (1999) No. 5

system was used for an examination of the e!ect of disc membranes (47 mm diam.) of 0.2 km pore size were
membrane material on apple juice "ltration. Apple juice obtained from Gelman Sciences Inc. (Montreal, PQ,
was also "ltered through a pilot plant tubular system Canada). Mixed cellulose esters (CE, MF type, GSWP-
using ultra"ltration polymeric membranes. The e!ects of 04700) and hydrophilic polyvinylidene #uoride (PVDF,
cross#ow velocity and transmembrane pressure on #ux Durapore, GVWP-04700) disc membranes (47 mm
were determined on the tubular system to optimize these diam.) of 0.2 km pore size were supplied by Millipore
operating parameters. A microbial challenge was carried Corp. (Mississauga, ON, Canada). PS type membrane
out to establish the e!ectiveness of selected tubular discs (43 mm diam.) of 10 kDa (PM10-13122) and
polymeric ultra"ltration membranes in removing micro- 30 kDa (PM30-13222), and cellulosic type membrane
organisms. Batches of juice were also "ltered through discs (43 mm diam.) of 10 kDa (YM10, 13622), 30 kDa
di!erent membranes to determine the e!ect of MWCO (YM30, 13722) and 100 kDa (YM100, 14422) were pro-
on #ux behavior and on juice properties. Filtered juices vided by Amicon Canada Ltd. (Oakville, ON, Canada).
were membrane sterilized, aseptically packaged and then PS membrane discs of 100 kDa (GR40PP) were cut from
stored at 4, 25 and 35 3C for up to 16 wk to examine their DDS (De Danske Sukkerfabrikker) Lab 20 model mem-
storage stability. branes (Dow Danmark A/S, Separation Systems,
Nakskov, Denmark). Membrane discs that were not de-
signed for the Amicon system were carefully cut and
Materials and Methods "tted for use.
Juice (50 mL) was "ltered through the stirred cell at
Apple juice room temperature with 276 kPa N pressure and a stirrer
A blend of McIntosh, Spartan and Red Delicious apples 
rotation of 750 rpm. The headspace in the stirred cell
(Malus domestica Borkh.) was obtained from the Sum- was #ushed with N before "ltration. The weight
merland Research Centre orchards and from other or- 
of permeate obtained from the cell was monitored over
chards in the British Columbia Okanagan Valley. The time to calculate #ux. The membrane and fouling layer
apples were stored at 0 to 2 3C for several weeks until resistances were determined using the following equa-
processed. Apples tested negative for starch using the tion:
method of Lau (9).
Apples were brought to room temperature before pro- *P
J" Eqn [1]
cessing. They were manually washed and then hammer- k(R #R )
milled through a 1.3 cm aperture screen. The mash was K D
held for 60 min at room temperature to promote oxida- where J is the average #ux (kg/ms) observed be-
tion prior to juice extraction with a screw press (Vetter tween concentration factors of 7.5 and 11, *P is the
Model BA6006 Type 1/2, Postfach, Germany). Pectinex transmembrane pressure (Pa), k is the permeate viscosity
Ultra SP (0.012 mL/100 mL) and Pectinex 100L (Pa ) s), R is the membrane resistance (m/kg), and R is
K D
(0.006 mL/100 mL) enzymes (Novo Nordisk Biochem the fouling layer resistance (m/kg). Water and juice
North America Inc., Franklinton, NC, U.S.A.) was added #ux for the CE and PS micro"ltration membranes
and the juice was kept overnight at 4 3C in a stainless (0.2 km) were repeated six times for evaluating coe$-
steel tank after #ushing the headspace with N . Before cients of variation. All other measurements were done in

"ltration, complete depectinization was ensured by in- duplicates.
creasing the temperature of the juice to 50 3C for 2 h
using a tube-in-shell heat exchanger.
The depectinization procedure was monitored using Tubular polymeric membrane xltration system
an alcohol test and a centrifugation test. The alcohol A B1 twin-entry tubular membrane module (PCI Mem-
test consisted of mixing equal volumes of juice with brane Systems Ltd., Eden Prairie, MN, U.S.A.) consisting
ethanol (95 g/100 mL) in a test tube. The presence of of two parallel sets of nine tubes in series was connected
a precipitate indicated that pectin was still present. The to a membrane "ltration unit (APV Membrane Systems,
centrifugation test was modi"ed from the method of Ishii Tonawanda, NY, U.S.A.). The tubes had a diameter of
and Yokotsuka (10) and consisted of centrifuging 40 mL 12.5 mm and the total surface area for all 18 tubes was
of juice at 2500;g for 5 min at 10 3C. The supernatant 0.9 m. Several PCI membranes of di!erent MWCOs
was "ltered through Whatman No. 41 "lter paper and and composition were tested including 200 kDa PVDF
the turbidity measured using a Hach Ratio/XR tur- (FP200), 100 kDa PVDF (FP100), 25 kDa PES (ES625),
bidimeter (Model 43900, Hach Co., Loveland, CO, 20 kDa PS (PU120) and 9 kDa PES (ES209). Due to the
U.S.A.). All batches of juices were depectinized to the limited commercial availability, membranes made with
same extent. di!erent material but the same MWCO could not be
compared. The "ltration unit was setup to run in a batch
mode for a feed and bleed circuit. The feed temperature
Stirred cell system was regulated by a heat exchanger connected to the feed
Depectinized apple juice was "ltered through a Model 52 tank. Both feed and recirculation pumps had variable
stirred cell of 43 mm diameter (Amicon Canada Ltd., drives.
Oakville, ON, Canada). Polyethersulfone (PES, Supor For #ux studies, depectinized juice (50 L) was recir-
200-60301) and polysulfone (PS, HT Tu!ryn 200-66199) culated through either 25 kDa PES or 200 kDa PVDF

291
lwt/vol. 32 (1999) No. 5

membranes. Only two tubes in parallel were used corres- Microbial challenge test
ponding to a membrane area of 0.1 m. Both the per- The microbial retention of the 9 kDa and 25 kDa PES
meate and retentate (bleed) were returned to the feed membranes was tested with Pseudomonas diminuta
tank which was continuously #ushed with N . The feed (Brevundimonas diminuta) ATCC 19146 (American

pump was set to deliver 6 L/min and the recirculation Type Culture Collection, Rockville, MD, U.S.A.). The
pump was increased over 15 min to the desired cross#ow culture was propagated in nutrient broth at 30 3C for
velocity (CFV). Cross#ow velocities of 3.3 and 6.7 m/s 24 h. The cells were separated by centrifugation at
corresponded to recirculation rates of 49 and 98 L/min 5 000;g for 15 min at 4 3C and were resuspended in 50 L
and pressure di!erentials of 83 and 276 kPa, respectively. of 0.1 g/100 mL peptone to challenge a membrane with
The recycle #ux was continuously monitored using 1;10 cfu/cm. The bacterial test solution was circulated
a graduated cylinder and stopwatch. The system was through the membrane at 25 3C, 2 m/s CFV and 207 kPa
allowed to equilibrate for 30 to 50 min before increasing TMP with the feed pump delivering 6 L/min. The solu-
the transmembrane pressure (TMP) by adjusting the tion was recycled for 60 min before withdrawing per-
retentate outlet valve. meate and retentate samples. CFV and TMP were then
Prior to batch concentration runs and microbial increased to 3.3 m/s and 414 kPa, respectively, and sam-
challenge tests, the membranes were sanitized by circula- ples were taken again after 60 min.
ting a 50 mg/L chlorine solution at 50 3C for 30 min Aliquots of retentate were serially diluted with
then rinsing with distilled water. The permeate tank 0.1 g/100mL peptone and spread plated in duplicate on
and lines underwent a sterilization treatment at 121 3C nutrient agar. Plates were incubated at 25 3C for 72 h.
and 104 kPa for 15 min. Sterile N was used to Two 1000 mL samples of permeate were membrane "l-

maintain a positive pressure in the system after tered through sterile 0.45 km membranes. The mem-
cooling. branes were incubated on absorbent pads soaked with
Batches of apple juice (200 to 240 L) were "ltered through nutrient broth at 25 3C for 72 h.
the above mentioned membranes. The operating condi-
tions for these batch concentration runs were 3.3 m/s
CFV, 414 kPa TMP, and 50 3C. The feed pump provided Analytical measurements
a #owrate of 6 L/min and the feed tank was continuously Samples were analysed for color, turbidity, viscosity,
#ushed with N . The #ux was monitored using an elec- soluble solids, titratable acidity, #avanols and protein.

tromagnetic #owmeter (Model M053724010R100A, ABB Brown color was measured at an absorbance of 420 nm
Kent-Taylor, Rochester, NY, U.S.A.) as the permeate was using a Beckman DU 640 spectrophotometer (Beckman
separately collected in a SS tank. The juice was processed Instruments, Inc., Fullerton, CA, U.S.A.). Color was also
to a concentration of three times for the 9 kDa mem- assessed spectrophotometrically with a CIE L*, a* and
brane and 10 times for the other membranes. Prior to b* Color Determination and Matching program Version
juice "ltration, the water #ux was measured under the 1.0 (Beckman Instruments, Inc.). Values were calculated
same conditions. The membrane and fouling layer resist- for standard illuminant C and the CIE 1964 Supple-
ances were calculated using Eqn [1]. The "ltration unit mentary Standard Observer. Turbidity was determined
was cleaned-in-place using a solution of 0.1 to 0.15 with a Hach Ratio/XR turbidimeter (Model 43900, Hach
g/100 mL Ultrasil 10 (Klenzade, Mississauga, ON, Co., Loveland, CO, U.S.A.). A Brook"eld LVDV-II#
Canada) with 100 to 200 mg/L free chlorine. Following viscometer with a UL adapter (Brook"eld Engineering
a rinse with water, the cleaning cycle was repeated a sec- Laboratories, Inc., Stoughton, MA, U.S.A.) was used to
ond time. measure viscosity at a shear rate of 73 s\. Soluble solids
were assessed with a Reichert Abbe Mark II digital
refractometer (AO Scienti"c Instruments, Bu!alo, NY,
U.S.A.). Juice samples (10 g) were titrated with 0.1
Bottling and storage N NaOH to an endpoint of pH 8.1 using a Metrohm 686
Immediately before bottling, the clari"ed juice was Titroprocessor and 665 Dosimat (Metrohm Ltd. Switzer-
"ltered through a sterile 0.45 km Supor DCF capsule land) and titratable acidity was expressed as g malic acid
(Gelman Sciences Inc., Montreal, PQ, Canada). The per 100 mL juice. Flavanol content was determined by
capsule had a 0.8 km pre"lter and a 0.45 km hydrophillic the acidi"ed vanillin method of Broadhurst and Jones
PS "lter. This "ltration step ensured that microbial (11) using (-)-epicatechin (Sigma Chemical Co., St. Louis,
growth would not occur during storage. The juice was MO, U.S.A.) as standard. Initial protein content was
bottled into 250 mL glass jars with twist cap lids in measured using the Kjeldahl method. Juice samples (5 g)
a laminar #owhood. The jars and lids were steam steriliz- were dried at 70 3C in a vacuum oven (104 kPa) prior to
ed prior to use. The headspace of the jars was #ushed digestion. Nitrogen was measured with a Technicon
with N "ltered through a 0.2 km PTFE membrane Hi- AutoAnalyzer II (Technicon Industrial Systems,

Flo Sol-Vent capule (Gelman Sciences). Juices were Tarrytown, NY, U.S.A.) using a colorimetric assay based
stored at 35, 25 and 4 3C in the dark. At 35 3C, samples on the reaction of ammonia, sodium salicylate, sodium
were removed after 1, 2, 3 and 4 wk storage. At 25 and nitroprusside and sodium hypochlorite in a bu!ered al-
4 3C, samples were removed after 4, 8, 12 and 16 wk kaline medium which produced an ammonia-salicylate
storage. Initial and stored samples were frozen at complex with an absorbance maximum at 660 nm. A fac-
!30 3C until analysed. tor of 6.25 was used to convert nitrogen concentration to

292
 lwt/vol. 32 (1999) No. 5

protein concentration. For all analyses, measurements

were made on two replicate samples.

Sensory Analysis
Di!erences between juices "ltered through the various
membranes were compared using triangle tests. Eighteen
judges were selected from sta! at the Paci"c Agri-Food
Research Centre. Juices were stored at !30 3C after
initial processing. Prior to sensory evaluation, the juices
were thawed at 4 3C. Juice samples (30 mL) were pre-
sented to each judge in covered black wine glasses with
random three-digit codes. The juices were allowed to
equilibrate at room temperature for 30 min before pre-
sentation. Judges were asked to smell and then taste a set
of three glasses in the order given before selecting the
odd sample. The triangle tests were completely random-
ized for the odd sample and the order of presentation.
The level of signi"cance was determined according to
Larmond (12).

Results and Discussion

Stirred cell
Micro"ltration (MF) membranes of 0.2 km pore size had
similar steady state #ux for apple juice "ltered through
the stirred cell. Based on water and juice #ux from cellu-
lose (25267$2648 L/mh and 67.0$6.6 L/mh, respec-
tively) and polysulfone (2588$243 L/mh and 64.3$
7.2 L/mh, respectively) MF membranes, coe$cients of
variation varied between 9.4 and 11.1%. Delineation Fig. 1 Relative #ux of depectinized apple juice in a stirred cell
between membranes were more apparent based on their (TMP, 276 kPa; stirrer rotation, 750 rpm; temperature, 233C)
relative #ux (permeate #ux/pure water #ux) (Fig. 1A). through micro"ltration (A) membranes (}䉬}) 0.2 km cellulose;
PVDF and PS comparatively gave higher values than (}䉱}) 0.2 km polyvinylidene #uoride; (*;*) 0.2 km polysulfone;
(}䊏}) 0.2 km polyethersulfone; and ultra"ltration, (B) mem-
PES. These di!erences were consistent with the results of branes (}䉬}) 100 kDa cellulose; (}䉱}) 30 kDa cellulose; (*;*)
Riedl et al. (13) and have been associated with membrane 10 kDa cellulose; (}䊏}) 100 kDa polysulfone; (}䊉}) 30 kDa
surface morphology rather than membrane surface hy- polysulfone; (}"}) 10 kDa polysulfone.
drophobicity. Smooth membranes (e.g. PS and nylon)
produced a dense surface layer whereas this same layer
on rougher membranes (PES and PVDF) was more Flux studies with tubular membranes
open (13). Typical behavior for cross#ow UF membranes is
As with MF membranes, the initial relative #ux of generally noted by an initial #ux increase followed by
100 kDa membranes were high and the rates of per- a plateau as TMP is ramped from low to high values. The
meate production decreased rapidly (Fig. 1B). Ultra"ltra- pressure independent region of #ux is due to concentra-
tion (UF) membranes of 30 kDa had initial #ux declines tion polarization and the formation of a gel layer (14).
intermediate to 100 and 10 kDa. Relative #ux of all UF Using the cross#ow tubular membrane system, the high-
membranes were also higher than that of MF membranes er CFV (6.7 m/s) led to a large increase in recycle #ux
by at least "ve to ten fold (Fig. 1A and B). Once with the 200 kDa PVDF membrane, but this #ux
the membrane resistances were taken into account, less improvement was not observed with the 25 kDa PES
fouling was experienced by UF membranes of 30 and membrane (results not shown). In cross#ow membrane
100 kDa as evidenced by lower fouling layer resistances "ltration processes, CFV is known to improve #ux due to
(Table 1). These results indicated that an optimum higher shear rates promoting the removal of components
apple juice "ltration #ux exists between 0.2 km deposited on the membrane surface (14). At higher
and 10 kDa membranes. Flux was expected to be lower MWCO (200 kDa), increasing CFV up to 414 kPa im-
for the 10 kDa due to high membrane resistance. Apple proved #ux because the fouling layer was a limiting
juice particulates of less than 0.2 km likely caused factor. At lower MWCO (25 kDa), the pore size rather
pore plugging and fouling in the MF membrane lead- than the fouling layer was more restrictive and increased
ing to increased #ux reduction compared to UF mem- turbulence at the membrane surface was therefore not as
branes. e!ective.

293
lwt/vol. 32 (1999) No. 5

Table 1 Membrane resistance (R ), fouling layer resistance (R ) and total resistance (R ) for apple juice "ltration
K D 2
with a stirred cell system.

Membrane Resistance?@

R R R
K D 2
(;10 m/kg) (;10 m/kg) (;10 m/kg)

0.2 km cellulose (Millipore) 3.9 1054.7 1058.6

0.2 km polyethersulfone (Gelman) 17.5 1233.7 1251.2
0.2 km polysulfone (Gelman) 38.4 1064.1 1102.4
0.2 km hydrophilic polyvinylidene #uoride (Millipore) 38.8 1088.6 1127.4
100 kDa polysulfone (DDS) 189.1 624.4 813.5
30 kDa polysulfone (Amicon) 143.5 760.9 904.4
10 kDa polysulfone (Amicon) 492.0 1050.4 1542.4
100 kDa cellulose (Amicon) 86.1 747.2 833.3
30 kDa cellulose (Amicon) 216.2 701.7 917.9
10 kDa cellulose Amicon) 1118.3 345.2 1463.5

? Transmembrane pressure, 276 kPa; stirrer speed, 750 rpm; temperature, 23 3C

@ Values for 0.2 km cellulose and polysulfone membranes are averages of six replicates while the remaining values are averages of
two replicate samples

Table 2 Challenge tests of tubular polyethersulfone tion of microorganisms in "nished product compared to
membranes with Pseudomonas diminuta levels that may be present in untreated juice. A tangential
#ow system using mineral membranes has been reported
Sampling time and conditions? Microbial counts@ for the production of cold sterile apple juice (15). The use
of a "nal dead-end "lter prior to bottling could however
Retentate Permeate ensure microbial stability for both polymeric and ce-
(cfu/L) (cfu/L) ramic membranes.
9 kDa membrane
Initial 14.0;10 *
After 1 h at 25 3C, 2 m/s Batch concentration with tubular membranes
CFV and 207 kPa TMP 13.3;10 1.2;10 A batch concentration study entails the monitoring of
After 1 h at 25 3C, 3.3 m/s #ux as the retentate returns to the feed tank and the
CFV and 414 kPa TMP 11.0;10 3.3;10
permeate is removed. The stabilized #ux increased with
25 kDa membrane membrane MWCO notwithstanding the di!erences in
Initial 12.1;10 *
After 1 h at 25 3C, 2 m/s
membrane material and manufacture (Fig. 2). The high-
CFV and 207 kPa TMP 11.6;10 3.5;10 est steady state #ux (236 L/mh) was obtained with the
After 1 h at 25 3C, 3.3 m/s 200 kDa PVDF membrane and the 9 kDa PES mem-
CFV and 414 kPa TMP 7.9;10 1.2;10 brane had the lowest #ux (14 L/mh). Relative #ux for the
UF membranes under cross#ow conditions were in the
? CFV, cross#ow velocity; TMP, transmembrane pressure
@ Results are the averages of two replicate samples
same range as those obtained with the stirred cell system.
Although PVDF membranes were of larger pore size
(100 kDa and 200 kDa), their #ux resistances due to the
Microbial challenge test fouling layer (R ) were smaller than the 20 kDa PS and
D
During the challenge tests of the 9 and 25 kDa PES 9 kDa PES (Table 3). Comparison of absolute #ux values
membranes, the concentrations of P. diminuta in the with other studies is often di$cult because of di!erences
retentate decreased slightly with time as cells adhered to in operating conditions, depectinization treatment and
the membrane and piping surfaces (Table 2). The low membrane type. The trends were consistent with that of
number of colony forming units (cfu) detected in the Padilla and McLellan (1), although #ux results
permeates indicated that the membranes "ltered out were lower for all membranes, except the 200 kDa mem-
most microorganisms. As P. diminuta is one of the brane.
smallest known bacteria (around 0.2 km), the 9 and Turbidity, viscosity, titratable acidity, and nitrogen (pro-
25 kDa membranes were e!ective in removing microor- tein) content were similar in all juices (Table 4). The
ganisms to log reduction values between 6 and 7. As the yellow/brown color of the juices decreased with mem-
CFV and TMP were increased, more microorganisms branes of smaller MWCOs. The visual appearance of the
passed through the membrane system. With a UF ce- juice "ltered through the 9 kDa membrane was distinctly
ramic membrane module, a log reduction value greater di!erent in that a slight green tint was evident. This
than 9 can be obtained (8). The tubular polymeric mem- membrane also retained more sugars and #avanols than
brane system used in this study can nevertheless meet the the other membranes as noted by reduced soluble solids
present FDA proposal of achieving a 100 000 fold reduc- and #avanol content. Along with the #avanols, the 9 kDa

294
 lwt/vol. 32 (1999) No. 5

Table 5 Triangle test comparisons of apple juice "ltered

through various ultra"ltration membranes

Comparison Correct Level of

responses? signi"cance

200 vs. 100 kDa 9 NS@

200 vs. 20 kDa 13 P(0.001
200 vs. 9 kDa 14 P(0.001
100 vs. 20 kDa 7 NS
100 vs. 9 kDa 14 P(0.001
20 vs. 9 kDa 15 P(0.001

? a panel of 18 judges was used

@ n.s."not signi"cant

membrane retained phenolic compounds responsible for

the decrease in yellow color (A , b*). The green tinge
 
possibly associated with the presence of chlorophyll
likely became apparent once the masking e!ect of the
yellow pigment was partially removed.
Triangle tests were carried out to compare #avor charac-
teristics of the juices "ltered through various membranes
Fig. 2 Permeate #ux and relative #ux of depectinized apple (Table 5). Although the juices from the 200 and 20 kDa
juice through ultra"ltration tubular membranes (TMP,
414 kPa; CFV, 3.3 m/s; temperature, 50 3C). (}䉬}) 200 kDa membranes were signi"cantly di!erent from each other,
polyvinylidene #uoride; (;) 100 kDa polyvinylidene #uoride; their respective di!erences with the 100 kDa membrane
(}䉱}) 20 kDa polysulfone; (}䊏}) 9 kDa polyethersulfone were too small to be detected. All other tested membrane

Table 3 Membrane resistance (R ), fouling layer resistance (R ) and total resistance (R ) for apple juice "ltration
K D 2
with a cross#ow tubular system

Membrane Resistance?@

R R R
K D 2
(;10 m/kg) (;10 m/kg) (;10 m/kg)

200 kDa polyvinylidene #uoride 128.5 322.6 451.1

100 kDa polyvinylidene #uoride 164.3 772.3 936.6
20 kDa polysulfone 442.3 1817.4 2259.7
9 kDa polyethersulfone 699.7 6670.4 7370.1

? Transmembrane pressure, 414 kPa; cross#ow velocity, 3.3 m/s; temperature, 50 3C

@ Results are averages of two replicate samples

Table 4 Properties of apple juice "ltered through di!erent tubular polymeric membranes

Property? Tubular polymeric membrane@

200 kDa PVDF 100 kDa PVDF 20 kDa PS 9 kDa PES

A 0.44 0.38 0.34 0.18

 
L* 95.9 96.2 96.6 99.0
a* !5.4 !4.9 !4.3 !4.8
b* 30.6 27.7 24.8 14.3
Turbidity (NTU) 0.12 0.09 0.16 0.15
Viscosity at 253C (cps) 1.38 1.37 1.38 1.36
Soluble solids (3Brix) 12.7 12.3 12.2 11.4
Titratable acidity 0.44 0.37 0.34 0.39
(g malic acid/100 mL juice)
Flavanol content (mg/L) 66 80 108 17
Protein content (mg/L) 29 29 23 34

? Results are the averages of two replicate samples

@ PVDF, polyvinylidene #uoride; PS, polysulfone; PES, polyethersulfone

295
lwt/vol. 32 (1999) No. 5

Fig. 3 E!ect of storage temperature and time on turbidity of apple juice "ltered through (a) 200 kDa PVDF, (b) 100 kDa PVDF,
(c) 20 kDa PS, and (d) 9 kDa PES membranes. (}䉬}) 35 3C; (}䊏}) 25 3C; (}䊉}) 4 3C

combinations were also found to be signi"cantly di!er- ing storage, changes in #avanols paralleled increases in
ent. Judges noticed the largest di!erences between juices turbidity and yellow/brown color. The 9 kDa membrane
"ltered through the 9 kDa membrane and the other initially retained more #avanols than the other mem-
membranes. They commented that the 9 kDa membrane branes and the juice consequently had the lowest changes
juice was thin (low body/mouthfeel/viscosity), had less in turbidity.
#avor and had an o!-#avor (medicinal/metallic). Besides The juices from all membranes had similar color cha-
retaining sugars and #avanols, the 9 kDa membrane may nges as measured using the absorbance at 420 nm
also have retained other #avor components. (Fig. 5). A pattern characterized by an initial decrease
of about 0.1 absorbance unit followed by an increase
of more than 0.1 unit was observed for most juices
Storage study stored at 25 and 35 3C. The rates at which the changes
The viscosity, soluble solids and titratable acidity of the occured were three times faster at 35 3C than at 25 3C.
juices did not change with either storage temperature or The oxidation and/or polymerization reactions involving
time. In addition, no microbial growth was observed the phenolic compounds may have created intermediate
during storage of the cold pasteurized juices. However, molecular species which absorbed less at 420 nm for
turbidity increased over time with higher storage temper- a period of time. The length of this period was dependent
atures and membranes of larger MWCOs (Fig. 3). Cha- on temperature. All juices stored at 4 3C consistently
nges were more pronounced at 35 3C as the rates of decreased in yellowness but did not increase again during
chemical reactions generally increase with temperature. the 15 wk period. The degradation products which pro-
The 200 kDa membrane had the most turbidity develop- gressively regained the partial loss of absorbance at
ment whereas the 9 kDa membrane had the least turbid- 420 nm were not generated at 4 3C probably due to
ity formation. Turbidity has previously been reported to a slower reaction rate.
increase with storage of juice (1).
Haze formation in UF clari"ed apple juice is a poten-
tial problem. Polymerization of phenolics and interac- Conclusions
tions with other components (e.g. proteins) can lead
to turbidity in fruit juice products (16, 17). Evidence Initial testing of di!erent membrane types during dead-
that phenolics could be involved in turbidity develop- end "ltration of apple juice indicated that PVDF and PS
ment is supported by the #avanol content changes ob- membranes gave higher relative #ux than PES and CE
served during storage (Fig. 4). Flavanol content in all membranes. In addition, 0.2 km or 10 kDa membranes
membrane "ltered juices decreased in a similar fashion had higher fouling layer resistances than 30 and 100 kDa
with storage temperature and time. On a comparative membranes. With tubular polymeric UF membranes, the
basis, #avanol content values among juices were an indi- #ux for depectinized apple juice improved as the
cator of the degree of #avanol polymerization (18). Dur- membrane MWCO increased from 9 to 200 kDa. Juices

296
 lwt/vol. 32 (1999) No. 5

Fig. 4 E!ect of storage temperature and time on #avanol content of apple juice "ltered through (a) 200 kDa PVDF, (b) 100 kDa
PVDF, (c) 20 kDa PS, and (d) 9 kDa PES membranes. (}䉬}) 35 3C; (}䊏}) 25 3C; (}䊉}) 4 3C

Fig. 5. E!ect of storage temperature and time on browning (A ) of apple juice "ltered through (a) 200 kDa PVDF, (b)
 
100 kDa PVDF, (c) 20 kDa PS, and (d) 9 kDa PES membranes. (}䉬}) 35 3C; (}䊏}) 25 3C; (}䊉}) 4 3C

"ltered through 200, 100 and 20 kDa membranes had change for all juices. Under these conditions, the cold
similar properties, but juice "ltered through a 9 kDa pasteurized juices were microbiologically stable. How-
membrane had lower soluble solids, #avanols and yel- ever, turbidity increased with storage especially for the
low/brown pigments. During storage at 4, 25 and 35 3C, higher MWCO membranes. A slight decrease in #avanol
the viscosity, soluble solids and titratable acidity did not content and a slight increase in browning were also

297
lwt/vol. 32 (1999) No. 5

observed over time. Storage at refrigerated temperature 6 PORTER, M. C. Ultra"ltration. In: Porter, M. C. (Ed.), Hand-
(4 3C) minimized these changes. Although the juice from book of Industrial Membrane ¹echnology. New Jersey:
Noyes Publications, Park Ridge, Ch. 3., p. 136}259 (1990)
the 9 kDa membrane was physicochemically more stable 7 ROG SCH, E. Experience with ultra"ltration in the 1984 sea-
during storage, #ux performance was comparatively low son. Confructa Studien, 29(1), 27}31 (1985)
and the sensory characteristics were a!ected due to in- 8 FUKUMOTO, L. R., DELAQUIS, P. AND GIRARD, B. Micro"l-
creased membrane retention of sugar, phenolic and other tration and ultra"ltration ceramic membranes for apple
#avor components. juice clari"cation. Journal of Food Science, 63, 845}850
(1998)
9 LAU, O. L. Harvest guides for B.C. apples. B.C. Orchardist,
7(7), 1A}20A (1985)
Acknowledgements 10 ISHII, S. AND YOKOTSUKA, T. Susceptibility of fruit juice to
enzymatic clari"cation by pectin lyase and its relation to
This study was done in collaboration with Sun-Rype pectin in fruit juice. Journal of Agricultural and Food Chem-
istry, 21(2), 269}272 (1973)
Products Limited of Kelowna, B.C. with the assistance of 11 BROADHURST, R. B. AND JONES, W. T. Analysis of conden-
the Industrial Research Assistance Program. The authors sed tannins using acidi"ed vanillin. Journal of the Science of
would like to thank Dr M. Cli!, Dr P. Delaquis and Food and Agriculture, 29, 788}794 (1978)
Mr T. Kopp for their assistance. 12 LARMOND, E. ¸aboratory Methods for Sensory Evaluation of
Written on behalf of the Department of Agriculture and Food. Ottawa: Publication 1637, Food Research Institute,
Agriculture Canada, Ontario p. 63 (1977)
Agri-Food, Government of Canada.  Minister of Pub- 13 RIEDL, K., GIRARD, B. AND LENCKI, R. W. In#uence of
lic Works and Government Services Canada 1999 membrane structure on fouling layer morphology during
apple juice clari"cation. Journal of Membrane Science, 139,
155}166 (1998)
References 14 CHERYAN, M. Modelling of ultra"ltration processes. In:
;ltra,ltration Handbook. Technomic Publishing Co., Inc.,
1 PADILLA, O. I. AND MCLELLAN, M. R. Molecular weight Lancaster, PA. Ch. 4, p. 73}125 (1986)
cut-o! of ultra"ltration membranes and the quality and 15 GROHMANN, M. AND FEUERPEIL, H. P. Cold sterile apple
stability of apple juice. Journal of Food Science, 54, juice-cold sterile "ltrates with tangential #ow "ltration.
1250}1254 (1989) Confructa Studien, 31, 171}174 (1987)
2 SHEU, M. J., WILEY, R. C. AND SCHLIMME, D. V. Solute and 16 NAGEL, C. W. AND SCHOBINGER, U. Investigation of the
enzyme recoveries in apple juice clari"cation using ultra"l- origin of turbidity in ultra"ltered apple and pear juice
tration. Journal of Food Science, 52, 732}736 (1987) concentrate. Confructa Studien, 29(1), 16}22 (1985)
3 WU, M. L., ZALL, R. R. AND TZENG, W. C. Micro"ltration 17 VAN BUREN, J. P. Causes and prevention of turbidity in
and ultra"ltration comparison for apple juice clari"cation. apple juice. In: Downing, D. L. (Ed.), Processed Apple Prod-
Journal of Food Science, 55, 1162}1163 (1990) ucts. New York: Van Nostrand Reinhold, Ch. 5, p. 97}120
4 RAO, M. A., ACREE, T. E., COOLEY, H. J. AND ENNIS, R. W. (1989)
Clari"cation of apple juice by hollow "ber ultra"ltration: 18 SINGLETON, V. L. Wine Phenols. In: Linskens, H. F. and
#uxes and retention of odor-active volatiles. Journal of Food Jackson, J. F. (Eds.), =ine Analysis. Berlin: Springer-Verlag,
Science, 52, 375}377 (1987) p. 198 (1988)
5 HEATHERBELL, D. A., SHORT, J. L. AND STRUBI, P. Apple
juice clari"cation by ultra"ltration. Confructa, 22, 157}169
(1977)

298