The Impact of Agglomeration and Storage on Flavor and Flavor Stability of Whey Protein Concentrate 80% and Whey Protein Isolate

Authors

  • B.J. Wright,

    1. Authors are with Dept. of Food, Bioprocessing, and Nutritional Sciences, Southeast Dairy Foods Research Center, North Carolina State Univ., Box 7624, Raleigh, NC 27695, U.S.A. Direct inquiries to author Drake (E-mail: mdrake@unity.ncsu.edu).
    Search for more papers by this author
  • S.E. Zevchak,

    1. Authors are with Dept. of Food, Bioprocessing, and Nutritional Sciences, Southeast Dairy Foods Research Center, North Carolina State Univ., Box 7624, Raleigh, NC 27695, U.S.A. Direct inquiries to author Drake (E-mail: mdrake@unity.ncsu.edu).
    Search for more papers by this author
  • J.M. Wright,

    1. Authors are with Dept. of Food, Bioprocessing, and Nutritional Sciences, Southeast Dairy Foods Research Center, North Carolina State Univ., Box 7624, Raleigh, NC 27695, U.S.A. Direct inquiries to author Drake (E-mail: mdrake@unity.ncsu.edu).
    Search for more papers by this author
  • M.A. Drake

    1. Authors are with Dept. of Food, Bioprocessing, and Nutritional Sciences, Southeast Dairy Foods Research Center, North Carolina State Univ., Box 7624, Raleigh, NC 27695, U.S.A. Direct inquiries to author Drake (E-mail: mdrake@unity.ncsu.edu).
    Search for more papers by this author

  • The use of trade names does not imply endorsement or lack of endorsement of those not mentioned.

Abstract

ABSTRACT:  The impact of agglomeration on flavor and flavor stability of whey protein concentrates 80% (WPC80) and whey protein isolates (WPI) has not been widely addressed. This study examined the impact of agglomeration on the flavor and flavor stability of commercial WPC80 and WPI across 18 mo of storage. Duplicate agglomerated and nonagglomerated WPC80 and WPI were collected from 4 facilities and stored at 21 °C, 50% relative humidity. Volatile analysis using solid phase microextraction (SPME) with gas chromatography-mass spectrometry (GC-MS) and descriptive sensory analysis were conducted every 2 mo. Solubility index, bulk volume, dispersibility, moisture, and color (L, a, b) were tested every 3 or 6 mo. Consumer acceptance testing with protein beverages was conducted with fresh and stored whey proteins. Higher intensities and more rapid development of lipid oxidation flavors (cardboard, raisin/brothy, cucumber, and fatty) were noted in agglomerated powders compared to nonagglomerated powders (P < 0.05). Volatile analysis results confirmed sensory results, which indicated increased formation of aldehydes and ketones in agglomerated products compared to nonagglomerated powders (P < 0.05). Consumer acceptance scores for protein beverages were lower for beverages made with agglomerated WPC80 stored for 12 mo and agglomerated or nonagglomerated WPI stored for 18 mo compared to fresh products while trained panelists detected differences among beverages and rehydrated proteins earlier. Agglomeration with or without lecithin decreased the storage stability of whey proteins. These results indicate that the optimum shelf life at 21 °C for nonagglomerated whey proteins is 12 to 15 mo and 8 to 12 mo for agglomerated whey proteins.

Introduction

There are several value-added products that can be obtained through the processing of liquid whey including concentrated whey, whey powder, lactose, and whey protein concentrates and isolates (USDEC 2004; Nnanna and Wu 2007). Whey protein concentrates and isolates are often used as value-added functional and highly nutritious ingredients in high protein nutritional products (USDEC 2004). Whey protein concentrates (WPC) result from removing nonprotein components to achieve a protein content of 34% to 89%, whereas whey protein isolate (WPI) results from removing nonprotein components to achieve at least a 90% protein content (USDEC 2004).

Whey proteins may be agglomerated after processing to enhance functional characteristics. Agglomeration is the process of producing agglomerates, small clumps of several particles. Agglomeration increases dispersability and decreases dispersion time because of the increase in the particle size and porosity of the powder (Pietsch 2005; Turchiuli and others 2005). Industrially, the most common way to produce agglomerated powders is by spray drying followed by the rewetting and agglomeration of powders in fluidized beds (Pietsch 2005). Whey protein powder is fluidized on a bed by an upward hot air flow. A solvent, called a binder, is sprayed onto the powder from above or inside of the bed (Turchiuli and others 2005). The binder can be water/steam, whey solution, or lecithin solution. The binder partially dissolves the outer coating of the dried particle, making the particle sticky, allowing for binding with other particles (Pietsch 2005). Lecithin is sometimes used in this process to further increase dispersion properties (wettability) of the powder. Instantized whey protein powders generally refer to products agglomerated with lecithin. More recently, agglomeration has been conducted as the whey protein is spray-dried, eliminating the need for the re-wet agglomeration process. This newer process is called single pass agglomeration (Henning and others 2006).

Flavor analysis of whey proteins has been limited. Descriptive sensory analysis of freshly produced WPC80 and WPI has documented a variety of flavors at low intensities including sweet aromatic, cardboard/wet paper, pasta water, brothy, animal/wet dog, cucumber, and soapy flavors, astringent mouthfeel, and bitter taste (Carunchia-Whetstine and others 2005; Russell and others 2006). Concurrent volatile analysis has revealed a variety of heat-induced and lipid and protein oxidation compounds (Carunchia-Whetstine and others 2005; Wright and others 2006). Mahajan and others (2004) and Sithole and others (2006) documented flavor chemistry and stability of noninstantized whey powders through 9 mo of storage. They reported variability in flavor profile among different suppliers similar to previous studies with WPC80 and WPI, but no change in sensory profiles through 9 mo of storage. The purported shelf life of WPC80 and WPI varies from 12 to 18 mo depending on the supplier. Sound science-based information on shelf life of these ingredients is important for widespread application and usage. However, to our knowledge, few studies have addressed stability of WPC80 or WPI or the impact of agglomeration on the flavor and flavor stability of WPC80 and WPI. Javidipour and Qian (2008) evaluated instantized and noninstantized WPC80 from a single manufacturer under accelerated storage conditions and documented increases in relative abundance of selected volatile compounds including a range of aldehydes, suggestive of lipid oxidation. They observed that instantized product developed lipid oxidation products faster than noninstantized product. However, products from multiple manufacturers were not evaluated nor was WPI, and sensory properties were not evaluated. In this study, agglomerated (re-wet and single pass) and nonagglomerated samples of WPC80 and WPI from different facilities were analyzed for flavor properties (sensory and instrumental) and selected physical properties over the course of 18 mo.

Materials and Methods

Samples

Duplicate sets of commercially packaged samples (approximately 23 kg each) from 2 different production dates were received from 4 facilities (F1-F4) (Table 1). At each facility, agglomerated products were manufactured from the same lot of nonagglomerated samples, meaning that variability within each facility for comparison of agglomerated and nonagglomerated products was minimized. Proteins were received in commercial 2-ply Kraft paper bags with liners. Upon receipt and initial sampling, products were transferred to 5 gallon high density polyethylene (HDPE) lidded bins (M & M Industries, Chattanooga, Tenn., U.S.A.) with 2.5 mL modified polyfluorotetraethylene (PTFE) lidded bins liners (Welch Fluorocarbon, Dover, N.H.) to facilitate sampling across storage time without the effect of an opened package. Bins were stored at 21 °C, 50% relative humidity (typical storage conditions).

Table 1—.  Whey protein sources.
Sample codeFacilityTreatment abbreviationLocationTreatmentWhey source
  1. C = control nonagglomerated; L = lecithin agglomerated; S = steam agglomerated.

 14CMidwestWPC80 nonagglomeratedCheddar
 24LMidwestWPC80 lecithin agglomeratedCheddar
 34CMidwestWPI nonagglomeratedCheddar
 44LMidwestWPI lecithin agglomeratedCheddar
 53CNorthwestWPI nonagglomeratedCheddar
 63SNorthwestWPI steam agglomeratedCheddar
 72CMidwestWPC80 nonagglomeratedMozzarella
 82SMidwestWPC80 steam agglomeratedMozzarella
 91CWest coastWPC80 nonagglomeratedCheddar
101SWest coastWPC80 steam agglomeratedCheddar
111LWest coastWPC80 lecithin agglomeratedCheddar

Time zero was considered the time of product receipt, which was within 1 mo of processing. Solubility, bulk density, color, protein, moisture, ash, minerals, sensory analysis, and volatile analysis were conducted at time zero. Samples were evaluated for sensory and volatile compound profiles every 2 to 18 mo of storage. Solubility index, dispersability, bulk volume, and color measurements were taken after 0, 6, 9, 12, 15, and 18 mo of storage. Periodically, freshly manufactured samples (within 2 wk manufacture) were received as reference products for consumer sensory tests. These products were received and stored as previously described and used within 3 wk of receipt.

Sensory analysis

Ten percent solutions (w/v) of whey protein were prepared with deionized, deodorized water under subdued lighting. Approximately 30 mL of rehydrated product were dispensed into 58 mL plastic cups with lids (Sweetheart Cup Co., Owings Mills, Md., U.S.A.) labeled with random 3-digit codes. Samples were stored at 5 °C for 24 h and removed from refrigeration 2 h before evaluation.

Sensory flavor properties were evaluated by 10 panelists (8 female, 2 male; age range 25 to 46 y), with extensive training (>150 h each) in the sensory analysis of dairy products, using the Spectrum method™ and Spectrum™ 15-point intensity scale (Sensory Spectrum, New Providence, N.J., U.S.A.). Sensory properties were evaluated for each sample in duplicate using a lexicon adapted from Russell and others (2006)(Table 2).

Table 2—.  Sensory language for whey proteins (WPC80 and WPI).
TermDefinitionReference and preparation
  1. Adapted from Russell and others (2006).

Sweet aromaticSweet aromatic associated with grains, such as oatmealQuaker oatmeal (50 g soaked in 500 mL water), Vanilla cake mix
CardboardAroma associated with wet cardboard and brown paper2 × 2 cm piece of brown paper bag boiled in water for 30 min
Potato brothyAromatics associated with boiled potatoesDrained broth from canned with potatoes
Cabbage brothyAromatics associated with boiled cabbageCabbage leaf boiled in 500 mL water for 5 min
CucumberAromatic associated with freshly cut cucumbersOne small cucumber diced and placed in a covered bowl
FattyAromatic associated with old fryer oil(E,E)-2,4-decadienal (2 ppb in skim milk)
Raisin/SpicyAromatic associated with stewed raisinsBoil 50 g dark raisins in 500 mL water
AstringencyChemical feeling factor characterized by a drying or puckering of the oral tissuesSoak 6 black tea bags in 500 mL water for 10 min

Volatile analysis

Volatile compounds were extracted and characterized by headspace solid phase microextraction (SPME) with gas chromatography mass spectrometry (GC-MS) (Quach and others 1999). Samples were prepared under subdued lighting and tested in triplicate. Ten percent whey protein solutions (w/v) were prepared using deodorized, deionized water and heated in a heating block at 40 °C for 30 min similar to the procedure described by Quach and others (1999). A 3-phase fiber (DVB/Carboxen/PDMS Stable Flex)(Supelco, Bellfonte, Pa., U.S.A.) was exposed at a depth of 2 cm into the headspace for 30 min followed by a 5 min fiber injection at 3 cm depth into an injection port fitted with a SPME liner at 250 °C of a Varian Gas Chromatograph CP-3380 Saturn 2000 ion trap MS (Varian, Walnut Creek, Calif., U.S.A.). A fiber blank was run once per day of testing, and a water blank was run each time the water supply was changed. Separations were performed on a fused silica capillary column (Rtx-5 30 m length × 0.25 mm i.d. × 0.25 um df; Restek, Bellefonte, Pa., U.S.A.). Helium gas was used as a carrier at a constant flow of 1 mL/min. The oven temperature was programmed from 40 to 250 °C at a rate of 8 °C/min with initial and final hold times of 5 min. Compounds were detected with MSD with the transfer line maintained at 120 °C, manifold at 80 °C; ion trap at 150 °C. Ionization was carried out via automatic electronic ionization detecting a mass range of 35 to 350 m/z using EM voltage of 2135V and 1 s per scan.

For positive identifications, retention indices (RI), and mass spectra of unknowns were compared with those of authentic standard compounds analyzed under identical conditions. Tentative identifications were based on comparing mass spectra of unknown compounds with those in the Natl. Inst. of Standards and Technology (NIST MS Search 2.0, NIST/EPA/NIH Mass Spectral Library) mass spectral database or on matching the RI values of unknowns against those of authentic standards. For the calculation of retention indices, an alkane series (C8-C20 Fluka Chemie Sigma Aldrich, Switzerland) was used (Van den Dool and Kratz 1963).

Proximate analysis

Proximate analysis (fat, moisture, protein, and ash) was conducted by a commercial facility, in duplicate on each of the whey proteins using standard methods. Fat content was determined by the Mojonnier method (Mojonnier Bros. Co., Chicago, Ill., U.S.A.) (Hooi and others 2004). Moisture was determined by vacuum oven, ash by muffle furnace, and protein by the Kjeldahl method (Hooi and others 2004). A conversion factor of 6.38 was used to convert total nitrogen to protein for whey proteins (Hooi and others 2004).

Bulk density

The bulk density method was adapted from the Niro method Nr A2b (Anonymous 2006). Ten grams of product were placed into a 50 mL graduated cylinder and tapped lightly 3 times onto a countertop. Measurements were taken for the final volume. Bulk density was calculated by dividing weight by final volume. All samples were tested in triplicate.

Color

A 60 × 15 mm round polystyrene petri dish (Becton Dickinson Labware, Franklin Lakes, N.J., U.S.A.) was filled with powder. Four measurements of L*, a*, and b* were taken at random places on each petri dish with a handheld Minolta Chromameter CR-300 (Minolta Camera Co., Japan). Additionally, color measurements were taken at each timepoint on each of the samples rehydrated at 10% (w/v) in deionized water. All samples were tested in quadruplicate.

Dispersibility

The dispersibilty method was adapted from the Niro method Nr A5a (Anonymous 2005). Ten grams of powder were added to 100 mL deionized water in a 500 mL glass jar and shaken vertically by hand beginning with the hand parallel to the floor and raised 90 degrees until the hand was perpendicular to the floor. The number of shakes to fully disperse the protein into solution was determined in triplicate on each sample.

Solubility index

The solubility index was measured according to the method outlined by the ADPI (2002). Twenty grams of powder were added to 200 mL deionized water and 3 drops of Antifoam B emulsion A5757 (Sigma Chemical Co., St. Louis, Mo., U.S.A.) in a Waring 1-speed commercial blender 700 (Waring Products Div., New Hartford, Conn., U.S.A.) and mixed for 90 s. The sample was then allowed to sit for 5 min before mixing with a spoon for 5 s. Two 50 mL conical centrifuge tubes were immediately filled to 50 mL. The tubes were centrifuged for 5 min in a 40.6-cm dia swinging bucket centrifuge (Damon/IEC Div. Intl.) at 164 ×g. The supernatant was siphoned off and discarded. Twenty five milliliters of deionized water were added to each tube and the sediment was redispersed with a wire. The tubes were filled to 50 mL with deionized water and then centrifuged again. The milliliters of sediment remaining in the tube after centrifugation were reported as the solubility index. The process was conducted in duplicate on each sample.

Descriptive analysis and consumer acceptance of ingredient applications

Flavor carry-through of the WPC80 and WPI were examined using 2 beverage applications. For the WPC80, a peach flavored protein beverage was used. For the WPI, a fruit flavored clear acidified protein beverage was used. WPC80 from facilities 1 and 2 (Cheddar and Mozzarella WPC80, respectively) and WPI from facility 3 were selected for ingredient applications testing across storage time based on initial (time 0, 2, and 4 mo) trained panel sensory profiles of rehydrated proteins. Formulations were obtained from commercial sources (Table 3 and 4). For each panel, approximately 30 mL of product were dispensed into 58 mL plastic cups with lids (Sweetheart Cup Co., Owings Mills, Md., U.S.A.) labeled with random 3-digit codes. Samples were prepared the day prior to testing, stored at 5 °C overnight and served cold (5 °C).

Table 3—.  Peach flavored whey protein beverage formulation used for WPC80 consumer testing.
Ingredient% w/w
  1. aTate and Lyle, Decatur, Ill., U.S.A.

  2. bCasco Inc., Cardinal, Ontario, Canada.

  3. cMcCormick & Co. Inc., Hunt Valley, Md., U.S.A.

  4. dPeach #15197 (David Michael & Co., Philadelphia, Pa., U.S.A.).

Deionized water79.88
Fructosea 9.94
WPC 80 6.81
Corn syrup solidsb 2.23
Citric acida 0.77
Red #40c 0.01
Yellow food coloringc 0.04
Peach flavoringd 0.32
Table 4—.  Clear acidified protein drink formulation used for WPI consumer testing.
Ingredient% w/w
  1. aTate and Lyle, Decatur, Ill., U.S.A.

  2. bSour apple #105621, Mother Murphy's, Greensboro, N.C., U.S.A.

  3. cMcCormick & Co. Inc., Hunt Valley, Md., U.S.A.

Deionized water81.9
Sucrose11  
WPI6 
Citric acida 0.9
Flavorb  0.18
Red #40c   0.013
Yellow #5c   0.007

For each consumer test, WPC80 or WPI (fresh and stored) from different facilities were tested on different days. The goal was to compare stored and fresh products from the same facility rather than products from different facilities. Previous studies have already established that products from different facilities and ingredient applications made from products from different facilities have different flavors, flavor carry-through, and consumer preferences (Carunchia-Whetstine and others 2005; Drake 2006; Drake and others 2009). WPC80 beverages were evaluated after 6 (descriptive analysis), 9 (descriptive analysis), 12 (descriptive analysis and consumer testing), 15 (descriptive analysis), and 18 (descriptive analysis and consumer testing) mo storage. WPI beverages were evaluated after 6 (descriptive analysis), 9 (descriptive analysis), 12 (descriptive analysis and consumer testing), 15 (descriptive analysis), and 18 (descriptive analysis and consumer testing) mo storage. For each product set from each facility at each time point, beverages were also made from fresh products (<1 mo old) as a point of reference.

For the preparation of the peach flavored protein beverage, all dry ingredients were blended first. The dry ingredients were slowly added into the deionized water with agitation using an Oster 2614 hand blender (Sunbeam Products, Inc., Boca Raton, Fla., U.S.A.). The ingredients were agitated until blended evenly. Color and flavor were then added to the beaker and blended using an Oster 2612 hand blender. For the preparation of the clear, acidified protein beverage, all dry ingredients were blended first. The dry ingredients were slowly added into a 2000 mL glass beaker containing the deionized water with agitation from an Arrow 1750 electric stirrer with a setting of 3 on the speed dial until blended evenly. The product was then heated in a water bath to 80 °C for 1 min with constant agitation from an Arrow 1750 electric stirrer, and subsequently cooled in an ice bath to 5 °C. Color was then added to the beaker and blended by hand until evenly dispersed.

Descriptive analysis was conducted on each beverage using the trained panel and methods described previously. Attributes evaluated included fruit aroma, whey protein aroma, fruit flavor, whey protein flavor, sweet and sour tastes, and astringency. For consumer testing, self-reported beverage consumers (n= 75) were recruited from the university population through email listservs and fliers. Panelists signed informed consent forms that provided a listing of all possible ingredients in the beverages tested. Prior to tasting, participants were asked a series of questions to gather demographic data including gender, age, shopping habits, and meal replacement product consumption and purchasing habits. The order of presentation was randomized and balanced among the consumers each day. Consumers scored all products for overall acceptability on a 9-point hedonic scale where 9 = like extremely and 1 = dislike extremely. Consumers were provided with ambient temperature deionized water for palate cleansing between samples. Compusense®five, release 4.6 (Compusense, Inc., Guelph, ON, Canada) was used for data collection.

Statistical analysis

Previous studies have established distinctive flavor and volatile compound differences between products from different facilities (Carunchia-Whetstine and others 2005; Russell and others 2006; Drake and others 2008). Since our goal was to evaluate the impact of agglomeration and storage, analysis of variance (ANOVA) with means separation (Fisher's least significant difference) was performed on collected data to determine differences between treatments and timepoints within each facility (SAS, v 9.1, Cary, N.C., U.S.A.). Principle component analysis (PCA) was performed on descriptive sensory and volatile compound data to visualize changes with time across treatments within each facility (XLStat version 2006.5, Addinsoft USA, New York, N.Y., U.S.A.).

Results and Discussion

Sensory analysis

Flavor terms documented in whey proteins included sweet aromatic, cardboard, potato brothy, cabbage brothy, cucumber, raisin, and fatty, along with the feeling factor astringency (Table 5 and 6). The attribute sweet aromatic was not detected in WPI, but was documented in many of the WPC80. Previous research has also reported these flavors and flavor differences between WPC80 and WPI (Carunchia-Whetstine and others 2005; Russell and others 2006). Different flavor profiles were documented among the whey proteins from the different facilities, consistent with previous research (Carunchia-Whetstine and others 2005; Drake and others 2009). These differences have previously been purported to be due to different fluid whey sources as well as differences in specific processing procedures.

Table 5—.  Flavor profiles of agglomerated and nonagglomerated WPC80 across storage time.
Sample facility treatmentaTimeAroma intensitySweet aromaticCardboardPotato/brothyCucumberFattyRaisinAstringent
  1. aSample number refers to treatment and facility. C = control nonagglomerated; S = agglomerated with steam; L = agglomerated with lecithin. See Table 1. Means within a column within a facility that are different (P < 0.05) are followed by different letters. ND = not detected. Attribute intensities were scored using a 0 to 15-point universal Spectrum™ intensity scale where 0 = absence of the attribute and 15 = extremely high intensity of attribute. Most whey protein flavors fall between 0 and 4 (Russell and others 2006).

1 02.0a1.3a1.6cNDNDNDND2.1a
F4 22.0a1.3a2.0b1.0aNDNDND2.0a
C 42.1a1.2a2.1b0.9aNDNDND2.0a
 62.1a1a   2.1b0.9aNDNDND2.0a
 82.0a1a   2.0b1.0aNDNDND2.2a
102.0a0.5b2.2b1.1aND0.6b0.8a2.3a
122.1a0.6b2.1b0.9a1.0a1.0a0.9a2.2a
152.2a0.5b 2.5ab1.0a1.1a1.0a1.1a2.3a
182.2aND2.8a1.0a1.0a1.2a0.9a2.1a
2 02.1a1.3a2.0b1.2aNDNDND1.9a
F4 22.0a1.1a2.1b1.0aNDNDND2.0a
L 42.2a1.1a1.9b1.1aNDNDND2.0a
 62.1a1.0a2.0b1.0aND0.6b0.6b1.9a
 82.0aND2.0b1.0a0.6b0.5b0.7b2.0a
102.1aND2.1b1.1a1.0a0.8b0.7b2.1a
122.0aND2.2b1.0a1.1a1.0a0.7b2.0a
152.1aND2.6a1.0a1.0a1.0a0.7b2.1a
182.0aND2.7a1.0a1.0a1.0a1.0a2.0a
7 01.4a2.1a0.6cNDNDNDND1.9a
F2 21.4a2.0a0.7cNDNDNDND1.9a
C 41.3a2.0a1.1bNDNDNDND2.1a
 61.3a2.0a1.1bNDNDNDND2.1a
 81.2a1.9a1.2bNDNDNDND2.0a
101.4a1.5b1.2bNDNDNDND2.2a
121.3a1.0c1.3bNDND0.6cND2.0a
151.4a0.9c1.3bNDND0.5cND2.0a
181.5aND1.5b0.5b2.3a1.0b1.3a2.1a
8 01.3a 1.3bc0.6cNDNDNDND2.1a
F2 21.3a1.1c1.3bNDNDNDND2.0a
S 41.3a1.1c1.3bNDND 0.8bcND2.1a
 61.4a1.0c 1.5abNDND1.1bND2.0a
 81.3a1.0c1.7aNDND1.3aND2.2a
101.4a0.6d1.8aNDND1.4aND2.1a
121.5a0.5d1.7aNDND1.6aND2.0a
151.2aND1.7aND0.6c1.5aND2.2a
181.5aND2.0a1.3a1.1b1.5a1.6a2.2a
9 02.0a1.3a2.1d0.8bNDNDND2.2a
F1 22.2a1.1a2.0d1.0bNDNDND2.0a
C 42.3a0.9a2.0d1.0bNDNDND2.1a
 62.0aND2.0d 1.2abNDNDND2.1a
 82.1aND2.1d1.5aND0.5bND2.2a
102.3aND2.0d1.4aND0.9a0.5c2.2a
122.2aND2.3c1.5a0.6b1.1a0.6c2.0a
152.1aND 2.8bc1.5a0.5b1.0a0.5c2.0a
182.3aND 2.7bc1.5a0.6b1.1a0.6c1.9a
10 02.1aND2.4cNDNDNDND2.3a
F1 22.0aND2.5cNDNDNDND2.0a
S 42.1aND2.5cNDNDNDND2.0a
 62.2aND 2.7bcNDNDNDND2.2a
 82.0aND 2.7bcNDNDNDND2.2a
102.1aND2.8bNDNDNDND2.3a
122.2aND3.0bND0.7b1.1aND2.0a
152.0aND3.0bND1.0a1.0a1.0b2.0a
182.0aND3.5aND1.0a1.0a1.0b2.1a
11 01.6bND1.6eNDNDNDND2.0a
F1 22.0aND2.0dNDND1.0aND2.2a
L 42.0aND2.0dNDND1.0aND2.1a
 62.0aND2.4cND0.6b1.2a1.0b2.3a
 82.0aND2.3c1.3a1.1a1.2a1.4a1.9a
102.1aND2.3c1.1b1.3a1.3a1.4a2.2a
122.1aND2.3c1.1b1.3a1.3a1.3a2.0a
152.0aND2.5c1.1bND1.2a1.5a2.0a
182.1aND3.0b1.3aND1.3a1.5a2.3a
Table 6—.  Flavor profiles of agglomerated and nonagglomerated WPI across storage time.
Sample facility treatmentaTimeAroma intensityCardboardPotato/brothyCabbage/brothyCucumberFattyRaisinAstringent
  1. aSample number refers to treatment and facility. C = control nonagglomerated; S = agglomerated with steam; L = agglomerated with lecithin. See Table 1. Means within a column within a facility that are different (P < 0.05) are followed by different letters. ND = not detected. Attribute intensities were scored using a 0 to 15-point universal Spectrum™ intensity scale where 0 = absence of the attribute and 15 = extremely high intensity of attribute. Most whey protein flavors fall between 0 and 4 (Russell and others 2006).

3 03.1a2.3bND1.3aNDNDND2.2a
F4 22.4b2.3bND1.3aNDNDND2.1a
C 42.4b2.2b1.1b1.4aNDND0.5b2.0a
 62.3b2.3b1.0b1.2aNDND0.6b2.1a
 82.4b2.3b1.0b1.3aNDND0.6b2.2a
102.2b2.4b1.1b1.5aNDND0.5b2.0a
122.5b2.4b1.0b1.3aNDND0.5b2.0a
152.3b2.7a1.5a1.2aNDND0.6b2.0a
182.3b2.7a1.7a1.4aNDND1.0a2.1a
4 02.8a2.4bNDNDNDNDND2.0a
F4 22.2b2.4b1.1bNDNDND0.6b2.0a
L 42.2b2.3b1.0b0.5b0.5bND1.0a2.2a
 62.3b2.3b1.0b0.6b0.5bND1.1a2.0a
 82.3b2.4b1.0b0.5b1.1aND1.0a2.2a
102.4b2.5b1.1b0.7b1.3a0.5b1.2a2.1a
122.2b2.3b1.1b0.6b1.1a0.6b1.0a2.0a
152.4b2.4b1.0b0.6b1.2a0.6b 0.9ab2.0a
182.3b2.5b1.0b1.0b1.1a1.1a1.1a2.0a
5 02.1a1.5c1.2bNDNDNDND1.8a
F3 22.0a1.5c1.2bNDNDNDND1.8a
C 42.0a1.8c1.1bNDNDND0.5b1.8a
 62.0a 2.0bc1.7a1.2bNDND 0.7ab1.9a
 82.0a1.9c1.6a1.0bND1.0a1.1a2.0a
102.0a2.1b1.6a1.1bND1.0a1.1a2.0a
122.0a2.3b1.5a1.0b0.5c1.1a1.0a2.0a
152.0a2.3b1.5a1.0b0.5c1.1a1.0a1.9a
182.0a2.3b1.4a2.0a1.5a1.0a1.0a2.0a
6 02.1a2.3bND0.8bNDND0.6b1.9a
F3 22.1a2.3bND1.0bNDND 0.8ab1.9a
S 42.2a2.2b0.5c1.1bNDND1.0a2.0a
 62.3a2.4b0.5c0.9b0.6cND1.1a1.9a
 82.2a2.3b0.5c1.1b1.1bND1.0a1.9a
102.1a 2.5ab 0.8bc0.9b1.0b0.5b1.0a2.0a
122.2a 2.5ab0.5c1.1b1.1b0.5b1.1a2.0a
152.3a2.7a1.0b1.0b1.0b0.6b1.0a1.8a
182.3a2.7a1.0b2.0a1.5a0.5b1.0a2.0a

Within each facility, there were differences in the initial flavor profiles of agglomerated and nonagglomerated products although these differences were not consistent between facilities (Table 5 and 6). Similar trends were observed for WPC80 and WPI. The process of agglomeration with or without addition of lecithin does require an additional process step with added potential heat to some or all of the product (added steam to all products in the case of re-wet process or recirculated fines in the case of single pass process). Small changes in initial flavor profile of agglomerated and nonagglomerated products from the same lots would thus be expected. Lecithin may also be a source of flavor. The initial distinct flavor profile of lecithin-agglomerated WPC80 from facility 1 (Table 5, Figure 1) suggests that the addition of lecithin may alter initial flavor profile of whey protein.

Figure 1—.

PCA biplot of sensory changes in nonagglomerated and agglomerated WPC80 from Facility 1 over time. Numbers indicate treatment type and storage time. The letter “C” represents the nonagglomerated control. “S” represents the steam-agglomerated samples. “L” represents the lecithin-agglomerated samples. The number next to each letter represents the storage in months at the time of analysis.

The flavor profiles of whey proteins from each facility changed with storage time (P < 0.05; Table 5 and 6). As products aged, cardboard flavors increased in intensity and sweet aromatic flavors, if present initially, decreased in intensity. Cucumber, fatty, and raisin flavors were not detected in any of the fresh samples but were detected in many samples with storage time, suggesting that these flavors were related to changes that occurred during product storage. Potato/brothy flavor was detected in some samples initially and in others following storage time. Cabbage/brothy flavor, which was detected only in WPI, followed a similar pattern and was not necessarily specific to stored products. Previous research has documented the source of this flavor and that it is due to breakdown of sulfur-containing amino acids (Wright and others 2006). As such, steps during whey protein processing as well as Maillard reactions during product storage may account for the presence of this flavor as well as potato/brothy flavor (which is caused by the sulfur-containing compound methional) initially and/or during product storage. In samples 7 and 8 (Table 5, Figure 2), cucumber and potato/brothy flavors were not detected through 15 mo of storage. These whey proteins (nonagglomerated and steam agglomerated WPC80) were from a Mozzarella source and displayed a distinct initial flavor profile from the Cheddar whey proteins which continued throughout storage (Table 5, Figure 1 and 2).

Figure 2—.

PCA biplot of sensory changes in nonagglomerated and agglomerated WPC80 from Facility 2 over time. Numbers indicate treatment type and storage time. The letter “C” represents the nonagglomerated control. “S” represents the steam-agglomerated samples. The number next to each letter represents the storage in months at the time of analysis.

Changes in flavor with storage time were also different for the agglomerated whey proteins compared to nonagglomerated whey proteins within a facility (P < 0.05, Table 5 and 6). Principal component biplots of sensory profile changes with storage time were also examined to determine trends. Biplots from 2 different facilities are shown as representative changes with storage time (Figure 1 and 2). In products from facility 1, which manufactured nonagglomerated, agglomerated, and lecithin-agglomerated WPC80 from Cheddar whey, cardboard flavor increases with storage time were more dramatic in agglomerated and lecithin-agglomerated samples than in the nonagglomerated samples (Table 5, Figure 1). Cucumber and fatty flavors increased over time, and these flavors generally developed more quickly in the lecithin agglomerated samples. Fatty flavor was detected earliest in agglomerated samples 2, 4, and 11, which were sourced from different facilities and were WPC80 and WPI, but were all agglomerated with lecithin (Table 5 and 6). Suriyaphan and others (2001a, 2001b) documented fatty and grassy flavors and lipid oxidation products contributed by lecithin in fermented milks. In contrast, potato/brothy flavor increased over storage time in agglomerated and nonagglomerated products (Table 5 and 6). Some of the samples, particularly the agglomerated and instantized (lecithin-agglomerated) whey proteins, also developed cucumber flavors (Table 5 and 6). Many of the flavors that developed with storage (cardboard, cucumber, fatty) suggested the presence of lipid oxidation. The agglomerated and lecithinated samples generally developed these flavors faster than the unagglomerated samples. Storage-related flavors were not detected sooner in WPC80 compared to WPI.

Volatile analysis

Volatiles identified and quantified in the WPC80 and WPI were diacetyl, dimethyl disulfide, dimethyl trisulfide, toluene, trans-2-hexenal, hexanal, heptanal, octanal, nonanal, 2-heptanone, benzaldehyde, 1-octen-3-ol, 2-pentyl furan, 2-ethyl-1-hexanol, 2-nonanone, and 2-undecanone (Table 7 and 8). These compounds were selected because they represented a range of Maillard, lipid oxidation, or fermentation compounds and they were consistently detected within quantification limits from 3 or more whey proteins. The volatiles 2-heptanone and 2-nonanone are products of free fatty acids and have been associated with a stale, cardboard flavor, while 1-octen-3-ol is a product of lipid oxidation and has a mushroom aroma (Quach and others 1999). Dimethyl disulfide and dimethyl trisulfide are characterized by onion or cabbage aromas (Mahajan and others 2004), and dimethyl trisulfide has been implicated as the source of cabbage off-flavor in WPI (Wright and others 2006). Dimethyl disulfide and dimethyl trisulfide are degradation products of sulfur-containing amino acids (Yvon and Rijnen 2001; Wright and others 2006). Benzaldehyde is thought to be produced from the reduction of benzoic acid, a compound that is naturally occurring in milk (Quach and others 1999) or by amino acid degradation (Yvon and Rijnen 2001; Gomez-Ruiz and others 2002). Diacetyl, known for its buttery aroma, is derived from citrate fermentation by lactic acid bacteria (Avila and others 2005).

Table 7—.  Concentrations of selected compounds (ppb) in agglomerated and nonagglomerated WPC80 during storage.
Sample facility treatmentaTimediacetyldimethyl disulfidetoluenetrans-2-hexenalhexanal2-heptanoneheptanalbenzaldehydedimethyl trisulfide1-octen-3-ol2-pentyl furanoctanal2-ethyl-1-hexanol2-nonanonenonanal2-undecanone
  1. aSample number refers to treatment and facility. See Table 1. C = control nonagglomerated; S = agglomerated with steam; L = agglomerated with lecithin. Means within a column within a facility that are different (P < 0.05) are followed by different letters. ND = not detected.

1 00.65 c0.76 cND6.62 d6.56 d1.05 d0.65 c0.10 d0.24 b1.69 c2.76 d0.42 d0.12 b0.78 c0.67 b0.20 b
F4 21.07 c0.84 cND3.84 d3.86 d0.76 d0.77 c0.01 d0.31 b4.31 c0.81 d0.41 dND0.66 c1.26 b0.21 b
C 40.54 c0.76 cND3.39 d3.42 d0.39 d0.33 c0.01 d0.13 b1.73 c0.16 d0.16 d0.03 b0.22 c0.38 b0.08 b
 616.40 b37.57 bND52.71 c52.71 c14.93 c9.32 c1.26 d3.44 ab30.77 b2.43 d4.1 c1.61 b6.93 b11.76 a1.79 a
 867.06 a159.45 aND135.84 b137.02 b35.34 b18.21 ab2.72 d8.63 a73.40 a16.44 c6.27 c2.71 b9.25 b13.73 a0.20 b
1021.71 b78.76 b5.88 a28.64 c0.22 e41.51 ab17.51 ab27.07 b10.70 a44.42 ab3.94 d0.79 d1.28 b15.65 a10.89 a2.58 a
1217.42 b76.74 b6.32 a42.03 c0.25 e37.77 ab17.89 ab11.71 cd6.84 ab42.00 ab4.37 d1.00 d0.37 b15.39 a11.06 a2.36 a
1519.19 b144.73 a2.76 a66.76 cND21.99 b10.82 b10.18 cd5.16 b9.02 c1.78 d0.39 dND8.38 b7.74 ab1.42 a
1811.85 b58.12 b3.67 aNDND23.31 b10.5 b16.05 dc5.44 b25.74 ab3.74 d1.11 d6.13 a8.86 b6.00 ab1.36 a
2 00.66 c0.65 cND13.48 d13.48 d1.08 c1.03 c0.91 d0.29 b2.52 c2.74 d0.69 d0.05 b0.92 c1.16 b0.25 b
F4 20.31 c1.19 cND7.98 d8.04 d0.94 c0.98 c0.61 d0.40 b2.32 c1.59 d0.76 dND0.80 c1.36 b0.23 b
L 40.38 c2.06 cND12.36 d12.47 d0.92 c0.77 c0.60 d0.27 b1.69 c0.86 d0.40 d0.06 b0.44 c0.65 b0.12 b
 627.53 b72.33 b1.28 b522.48 a524.43 a43.86 ab20.72 ab14.66 c9.64 a20.44 b46.66 b8.55 b1.63 b11.11 ab10.56 a1.96 a
 820.73 b73.82 b4.99 a563.43 a565.19 a57.21 a28.12 a13.78 c11.94 a42.86 ab65.08 a13.53 a1.96 b18.42 a17.84 a3.44 a
1212.32 b99.66 b5.20 a52.28 c0.34 e38.42 b21.92 ab40.77 a11.29 a35.88 ab6.55 d1.17 d0.31 b15.72 a10.91 a2.29 a
159.16 b57.54 b2.89 aND c0.04 e25.05 b13.52 b19.62 bc8.82 a23.00 b4.14 d1.43 d4.71 a8.83 b6.49 ab1.27 a
1811.35 b96.29 b3.01 aND c0.07 e28.50 b15.42 b30.94 ab10.80 a22.02 b3.83 d1.54 cd5.90 a10.51 ab7.78 c1.66 a
7 00.13 b0.01 bND1.92 e1.93 c0.13 d0.19 cNDND0.89 c0.06 c0.08 cND0.17 d0.20 c0.10 b
F2 20.26 b0.03 bND1.01 e1.01 c0.12 d0.12 cNDND4.33 c0.03 c0.03 cND0.10 d0.14 c0.03 b
C 434.10 b4.97 bND78.07 b78.07 b10.03 bc11.18 cNDND61.44 b5.55 bc3.15 bND6.26 d6.53 b4.75 b
 615.46 b4.97 bND75.88 b75.88 b10.57 bc8.05 c0.07 b0.27 a13.74 c5.68 bc2.57 bND5.74 d4.71 b2.38 b
10111.41 a96.01 a48.84 a1017.87 aND179.96 a177.9 a20.2 a5.31 a116.26 a30.23 a5.20 a4.99 a143.55 a94.19 a44.45 a
129.09 b3.53 b1.35 bNDND6.84 bc6.13 c0.54 b0.09 a4.25 c0.92 c0.75 bc3.42 a3.45 d3.93 bc0.95 b
154.70 b3.35 b1.83 bNDND7.10 bc5.95 c1.03 b0.17 a5.15 c1.22 c0.04 c4.92 a4.16 d3.05 bc1.19 b
185.10 b3.79 b2.25 bNDND8.29 bc6.41 c0.78 bND6.52 c1.44 c0.12 c4.95 a5.24 d3.86 bc1.64 b
8 00.39 b0.03 bND3.77 e3.77 c0.32 c0.37 cNDND2.54 c0.17 c0.09 cND0.38 d0.44 c0.22 b
F2 20.30 b0.05 bND4.33 e4.34 c0.21 c0.28 cNDND7.74 cND0.08 cND0.12 d0.16 c0.03 b
S 419.96 b5.23 bND332.06 b331.9 a12.99 bc16.58 c0.05 b0.03 a17.22 c11.96 b3.39 bND5.86 d6.75 b2.68 b
 69.14 b3.84 b0.54 b354.82 b355.03 a14.24 bc14.26 cNDND3.82 c7.36 bc4.86 aND5.77 d6.71 b1.38 b
 85.84 b11.12 b8.31 b44.95 dND37.4 b28.56 c1.54 b0.44 a35.75 bc4.55 bc4.25 ab6.03 a19.61 cd9.79 b3.95 b
1011.38 b14.64 b9.01 b99.23 d0.12 c49.61 b46.4 b2.64 b0.47 a43.10 bc7.07 bc1.4 bc5.36 a28.20 c16.03 b6.42 b
122.23 b3.77 b1.19 bNDND12.85 bc9.84 c0.29 b0.07 a10.19 c1.44 c0.22 c6.37 a5.92 cd4.10 b1.23 b
152.84 b4.66 b2.33 bND0.01 c15.38 bc11.35 c0.69 b0.15 a13.73 b2.14 bc0.34 c9.47 a8.56 cd4.22 c1.69 de
185.93 b4.97 b2.13 bNDND17.53 cd11.52 c0.35 b0.09 a16.79 b2.21 gND9.61 bc9.22 cd4.70 c2.05 de
9 00.72 e0.22 dND20.60 g20.70 f0.78 d1.31 c0.01 c0.08 c3.92 c0.49 b0.74 bND0.85 d1.44 c0.27 b
F1 21.41 e0.49 dND34.67 g34.87 f0.87 d1.19 c0.02 c0.11 c2.70 c0.46 b0.52 bND0.55 d1.14 c0.19 b
C 631.43 d32.93 cND511.44 b514.34 c19.05 c15.26 c0.06 c2.25 c30.69 c2.65 b4.92 aND6.15 cd7.45 c0.39 b
 830.24 d47.63 cND508.13 b509.99 c22.29 c15.6 cND4.64 c22.21 c10.17 b5.85 a0.20 b7.80 cd9.70 c1.38 b
10239.57 b539.31 a36.97 a242.18 e2.41 f250.55 a181.86 ab15.89 a57.45 a171.7 a37.64 a3.11 a7.99 b129.81 a94.11 ab19.61 a
1211.18 de45.30 c3.46 b34.56 g0.15 f25.39 c18.02 c1.38 bc4.27 c18.79 c5.36 b2.43 a0.51 b12.93 cd9.61 c1.88 b
1513.63 de39.85 c1.71 bNDND16.09 c10.49 c1.33 bc3.53 c9.70 c2.45 b0.50 b7.13 b7.72 cd5.79 c1.13 b
188.29 e38.57 c2.12 bNDND16.14 c10.29 c2.55 bc3.76 c11.17 c2.27 b0.44 b7.44 b7.80 cd5.05 c1.18 b
10 01.65 e0.76 dND12.15 g12.17 f1.32 d1.38 c0.01 c0.50 c7.02 c1.22 b0.82 bND1.47 d2.01 c0.38 b
F1 23.90 e1.62 dND109.46 f109.98 e3.04 d4.75 c0.01 c0.73 c8.20 c2.35 b2.17 aND2.03 d4.55 c0.80 b
S 629.96 d46.83 cND390.62 cd392.78 d18.84 c15.48 c0.53 c3.82 c35.91 c6.91 b5.01 aND6.4 cd7.16 c0.30 b
 830.18 d49.24 cND357.64 d346.43 d19.74 c15.41 cND h6.01 c21.93 c10.75 b5.97 a0.09 b6.96 cd9.97 c0.97 b
10418.05 a499.76 a39.04 a405.93 c2.74 f251.39 a209.14 a13.04 a66.31 a183.83 a39.85 a4.28 a38.02 a127.6 a108.12 a19.38a
129.55 e42.03 c2.91 b34.34 g0.27 f18.70 c14.21 c2.40 bc5.23 c12.69 cf3.08 b0.21 b1.22 b9.57 cd8.02 c1.39 b
1513.02 de35.47 c1.45 bND0.06 f12.71 c9.17 c1.59 bc3.84 c7.63 c2.04 b0.58 b5.97 b5.14 d4.86 c0.75 b
1810.87 de45.90 c2.25 bND0.08 f16.63 c10.78 d5.57 b5.70 c10.99 c2.31 b0.49 b8.38 b6.33 cd5.43 c0.89 b
11 00.95 e0.29 dND8.03 g8.08 f0.87 d0.72 c0.02 c0.19 c3.45 c0.60 b0.38 bND0.71 d0.76 c0.19 b
F1 20.60 e0.32 dND17.58 g17.71 f0.64 d0.58 cND0.13 c2.39 c0.38 b0.28 bND0.39 d0.65 c0.11 b
L 660.70 d60.30 cND642.26 a646.56 a27.83 c22.87 c0.02 c3.91 c86.77 bc6.86 b7.27 aND9.24 cd12.48 c0.33 b
 829.68 d46.29 c0.42 b570.17 b572.27 b22.24 c17.43 c0.05 c4.52 c42.26 c6.60 b6.78 aND7.30 cd10.95 c1.46 b
10131.50 c361.98 b23.97 ab243.57 e3.36 f156.69 b140.82 b5.05 b40.77 b103.98 b30.27 a5.09 a2.30 b82.91 b77.09 b12.23 ab
1210.82 de31.54 c2.64 b30.69 g1.4 f19.19 c19.00 c1.52 bc4.31 c14.67 c4.98 b0.55 b0.75 b11.67 cd11.21 c1.3 b
1513.82 de37.29 c1.62 bND0.22 f17.48 c13.03 c1.04 bc4.36 c10.81 c6.69 b1.18 ab8.23 b9.16 cd8.09 c1.14 b
189.92 e36.83 c1.98 bND0.33 f16.93 c12.90 c3.12 bc4.98 c11.65 c4.13 b0.66 b9.81 b8.54 cd7.00 c1.11 b
Table 8—.  Concentrations of selected compounds (ppb) in agglomerated and nonagglomerated WPI during storage.
Sample facility treatmentaTimediacetyldimethyl disulfidetoluenetrans-2- hexenalhexanal2-heptanoneheptanalbenzaldehydedimethyl trisulfide1-octen-3-ol2-pentyl furanoctanal2-ethyl-1- hexanol2-nonanonenonanal2-undecanone
  1. aSample number refers to treatment and facility. See Table 1. C = control nonagglomerated; S = agglomerated with steam; L = agglomerated with lecithin. Means within a column within a facility that are different (P < 0.05) are followed by different letters. ND = not detected.

3 00.59 c2.62 dND0.79 a0.79 a0.06 c0.06 a0.68 d0.81 d0.61 d0.09 d0.07 b0.10 a0.09 c0.20 f0.09 c
F4 22.67 c12.37 dND0.27 a0.27 a0.56 c0.28 a2.72 d4.39 cd6.63 bcd0.36 d0.41 b0.42 a0.53 c1.73 ef0.48 bc
C 40.32 c4.14 dND0.02 a0.02 a0.03 c0.02 a0.42 d0.78 d0.81 cd0.04 d0.06 b0.02 a0.10 c0.34 f0.13 c
 68.39 bc59.93 cND0.32 a0.27 a0.96 c0.40 a1.92 d6.95 bcd3.60 bcd0.65 cd0.36 b0.20 a0.62 c1.84 ef1.05 abc
 814.45 b92.25 cd0.65 a0.73 a0.73 a2.09 bc1.14 a2.45 d13.00 bc12.22 ab2.08 bcd1.20 ab0.63 a1.38 bc5.40 a1.40 abc
1024.62 a109.32 c2.41 a6.46 aND2.34 bc1.11 a19.52 ab15.21 bc2.32 cd0.73 cd0.27 bND1.92 cdef4.26 bc1.46 abc
1211.63 b127.32 c2.86 a8.07 aND2.54 bc1.23 a24.37 a14.26 bc2.50 bcd0.91 bcd0.40 bND2.02 abc4.23 bc1.45 abc
1510.94 b90.29 cd1.05 aNDND1.38 bc0.76 a10.34 cd10.87 bc1.31 cd0.67 cd2.07 aND1.15 bc3.38 cd0.75 bc
186.71 c91.44 cd1.70 aNDND1.42 bc0.71 a15.59 cb11.18 bc1.51 cd0.69 cd1.30 abND1.29 bc2.80 cdef0.90 abc
4 00.53 c1.83 dND2.88 a2.87 a0.14 c0.17 a0.80 d0.52 d0.96 d0.56 cd0.19 b0.09 a0.12 c0.47 f0.10 c
F4 22.78 c8.26 dND4.88 a4.92 a1.03 c0.48 a2.21 d2.05 cd3.79 bcd0.75 cd0.54 b0.25 a0.56 bc1.90 ef0.45 bc
L 40.33 c3.19 dND4.00 a4.03 a0.19 c0.10 a0.33 d0.80 d0.40 d0.20 d0.11 b0.03 a0.10 c0.27 f0.11 c
 612.91 b122.16 cND2.33 a2.33 a4.1 abc1.04 a2.61 d9.74 c17.32 a1.50 bcd0.73 b0.89 a1.11 bc2.48 def0.89 abc
 87.56 c75.75 efND16.84 a16.84 a3.81 bc1.39 a2.41 d6.53 c7.71 bc0.79 cd1.12 ab0.46 a1.08 bc3.31 cd1.03 abc
1212.89 b338.23 a2.62 a32.80 aND7.98 a2.37 a24.32 a31.89 a5.10 bcd4.72 ab0.37 b0.46 a3.38 a4.36 bc2.09 a
1511.22 b244.29 b1.02 a10.74 aND4.86 abc2.11 a13.32 bc15.52 bc3.20 bcd6.44 a1.43 ab0.11 a1.70 bc3.25 cde0.90 abc
188.23 bc103.85 c1.03 aNDND3.81 bc1.07 a13.17 bcd8.95 c2.26 cd1.20 bcd0.85 b0.40 a1.37 bc2.6 def0.70 bc
5 01.84 c12.69 c0.33 c4.04 b4.07 c0.24 e0.25 b0.44 c3.74 gh2.40 cd0.37 d0.27 e2.61 c0.39 fg0.88 e0.16 cd
F3 48.23 b154.35 b0.04 c2.07 b2.06 c1.96 de1.09 b1.96 c29.81 cd12.08 ab3.20 b1.25 de4.26 b1.78 cdef4.99 abc2.27 b
C 610.54 b174.75 b0.57 c7.63 b7.63 c2.77 de1.71 b2.31 c38.39 cb15.62 a3.97 abc1.73 de9.12 a2.61 cd5.67 abc4.83 a
 86.99 bc218.80 b7.22 ab19.87 bND3.30 de2.28 b14.67 ab52.61 ab5.37 bcd1.62 bcd0.61 de0.02 d3.54 abc5.77 abc1.50 bc
109.04 b386.17 a9.76 a27.57 bND6.41 cd3.43 b17.20 a63.83 a8.37 abc1.66 bcd0.44 e0.17 d4.90 a5.38 abc1.56 bc
122.45 c47.27 c0.95 cNDND0.53 e0.28 b1.99 e7.55 fgh0.75 cd0.38 d2.71 d0.03 d0.51 fg1.28 de0.18 cd
187.61 b163.07 b4.51 bcNDND2.24 de1.24 b10.34 b32.51 bcd3.67 bcd3.10 bcd12.31 aND2.24 cde3.96 cde0.77 c
6 00.55 c2.53 c0.01 c1.02 b1.02 c0.06 e0.07 b0.69 c0.82 h0.66 d0.09 d0.08 e0.09 d0.10 g0.22 e0.09 d
F3 40.52 c2.64 cND0.03 b0.03 c0.09 e0.06 b0.67 c1.35 gh1.22 cd0.16 d0.10 e0.03 d0.19 fg0.38 e0.27 cd
S 62.72 c84.37 c2.92 c31.68 b31.68 c2.07 de0.99 b1.58 c16.82 defg7.05 bcd1.03 cd0.82 de3.46 bc1.10 efg2.62 de1.80 bc
 89.41 b195.04 b5.79 b19.62 b0.18 c4.01 cde2.13 b9.02 b33.73 bcd4.63 bcd1.18 cd0.87 de0.43 d2.51 cde4.79 bcd1.20 cde
1010.06 b284.22 ab8.58 a25.83 b0.31 c5.75 cd2.60 b12.22 ab43.95 b7.20 bcd1.76 cd1.20 de0.29 d3.35 bc5.38 abc1.38 cde
1218.19 a144.39 b2.34 c201.70 a202.42 a12.96 a8.75 a3.26 c17.65 def11.99 ab5.98 a6.79 c0.24 d4.62 a6.66 a0.70 c
1517.57 a169.10 b4.36 cde139.30 a127.41 b9.06 b7.82 a9.03 b25.42 de16.53 a6.15 a8.98 b0.50 d4.16 ab6.27 ab0.85 c
185.23 c190.06 b5.6 bNDND4.21 cde1.68 b8.46 b32.31 bcd4.97 bcd1.00 cd7.90 bc0.60 d1.97 cdef3.13 cd0.69 c

The presence of hexanal, heptanal, octanal, nonanal dimethyl disulfide, and dimethyl trisulfide in whey or whey protein powders has been previously reported (Quach and others 1999; Mahajan and others 2004; Carunchia-Whetstine and others 2005; Wright and others 2006). Additionally, Quach and others (1999) reported the presence of 1-octen-3-ol, 2-pentyl furan, benzaldehyde, 2-heptanone, 2-nonanone, and 2-undecanone in fresh WPC80, and Mahajan and others (2004) identified diacetyl in fresh sweet whey powder. To our knowledge toluene, trans-2-hexenal, and 2-ethyl-1-hexanol have not been reported in whey or whey protein powders. However, these compounds have been reported in other dairy products. Trans-2-hexenal, and toluene have been detected in butter, cultured dairy spreads, and cheese (Lund and Holmer 2001; Carunchia-Whetstine and others 2003; Povolo and Contarini 2003; Lozano and others 2007), and toluene and 2-ethyl-1-hexanol have been reported in several kinds of cheeses (Gomez-Ruiz and others 2002; Carunchia-Whetstine and others 2003; Avila and others 2005; Coda and others 2006; Povolo and others 2007).

On average, hexanal was the most abundant compound quantified in the fresh WPC80 followed by trans-2-hexenal (Table 7). Hexanal is an oxidation product from n-6 fatty acids while trans-2-hexenal is an oxidation product from n-3 fatty acids (Lund and Holmer 2001). Similar to our results, Quach and others (1999) demonstrated that hexanal was the most abundant aldehyde in the headspace of fresh WPC80. The compound 1-octen-3-ol was also abundant in the fresh WPC80. In general, the Mozzarella whey proteins (samples 7 and 8) had lower concentrations of all compounds over storage time than the Cheddar whey proteins. One exception was 1-octen-3-ol, which tended to be present in higher concentrations in the Mozzarella whey proteins. On average, the fresh WPI were most abundant in dimethyl disulfide and dimethyl trisulfide, followed by hexanal and trans-2-hexenal (Table 8). WPI generally had lower concentrations of the selected volatiles compared to WPC80, consistent with previous studies (Carunchia-Whetstine and others 2005).

Volatile compound results, similar to sensory results, indicated differences between samples from different facilities (Table 7 and 8). For example, the Mozzarella whey proteins (samples 7 and 8 from facility 2) were lower in concentration of most of the identified volatiles than all of the other samples, except 1-octen-3-ol. These results are consistent with their distinct sensory profiles. Mozzarella fluid whey is different in flavor profile from fluid Cheddar whey (Gallardo Escamilla and others 2005) and WPC80 from these sources would similarly be expected to be distinct. The WPI did not exhibit initial overt volatile compound differences although there were differences in specific compound concentrations between products from different facilities (Table 8). Previous research has demonstrated that WPC80 and WPI from different facilities had different sensory and volatile compound profiles (Carunchia-Whetstine and others 2005; Drake and others 2009). Sensory and volatile compound results are consistent with these previous studies.

Volatile compound changes occurred with storage time, similar to documented sensory results. Many of these compounds, particularly the aldehydes and ketones, confirmed the evolution of lipid oxidation during storage, which was also suggested by sensory results and previous studies (Carunchia-Whetstine and others 2005). Whitfield (1992) similarly hypothesized that lipid oxidation as well as the Maillard reaction initiated flavor development in whey products. WPC80 and WPI undergo heating during spray drying during which residual lactose can react with the protein to form Maillard browning or caramelization reaction products (Carunchia-Whetstine and others 2005). In addition, there is evidence that degradation of amino acids, including aromatic amino acids, branched-chain amino acids, and methionine, is a major process in the formation of aroma compounds in dairy products (Yvon and Rijnen 2001).

As the samples reached 12 to 18 mo of storage, further volatile compound changes were observed. Decreases in relative abundance of many of the selected compounds (diacetyl, trans-2-hexenal, hexanal, 2-heptanone, 2-nonanone) were observed (Table 7 and 8; Figure 3 and 4). Many lipid oxidation compounds are highly reactive and may be breaking down or reacting with other compounds. These results suggest that degradation has occurred in whey proteins, regardless of agglomeration treatment by this time. Javidipour and Qian (2008) found similar results during an accelerated storage study. Their research also showed that hexanal and heptanal concentrations initially increased in WPC80 followed by a subsequent decrease during storage.

Figure 3—.

PCA biplot of selected volatile compounds in nonagglomerated and agglomerated WPC80 from Facility 1 over time using solid phase microextraction. Numbers indicate treatment type and storage time. The letter “C” represents the nonagglomerated control. “S” represents the steam-agglomerated samples. “L” represents the lecithin-agglomerated samples. The number next to each letter represents the storage in months at the time of analysis.

Figure 4—.

PCA biplot of selected volatile compounds in nonagglomerated and agglomerated WPC80 from Facility 2 over time using solid phase microextraction. Numbers indicate treatment type and storage time. The letter “C” represents the nonagglomerated control. “S” represents the steam-agglomerated samples. The number next to each letter represents the storage in months at the time of analysis.

Proximate analysis

All WPC80 and WPI met the minimum required protein content; 80% dry weight for WPC80, and 90% dry weight for WPI. The WPC80 ranged from 4.29% to 6.83% fat while the WPI contained 0.11% to 0.55% fat. There were not consistent differences in fat contents among agglomerated compared with nonagglomerated products within the same protein content (P > 0.05) nor were differences in fat content within WPC80 or WPI correlated to selected lipid oxidation flavors or volatiles. Sensory and volatile component differences between WPC80 and WPI have been previously linked to differences in composition such as the differences in fat content documented in this study (Carunchia-Whetstine and others 2005). Moisture ranged from 3.8% to 5.1% for WPC80 and 3.1% to 4.4% for WPI. Ash similarly varied among samples and differences were not specific to product type (agglomerated compared with nonagglomerated or WPC80 compared with WPI), but all values were within previously published ranges (USDEC 2004; Carunchia-Whetstine and others 2005; Russell and others 2006). Moisture did not change significantly with storage time (P > 0.05).

Bulk density

Bulk density ranged from 0.255 to 0.460 g/mL for WPC80 and 0.245 to 0.279 g/mL for WPI. The bulk density of agglomerated products was lower than their unagglomerated counterparts (P < 0.05) (results not shown) although these differences were much larger for WPC80 compared to WPI. Agglomeration increases the volume of powders because it increases the particle size of the powder, and these large particles cannot fit as compactly in a given amount of space as smaller unagglomerated particles. Thus, the weight of an agglomerated powder that fits into a given volume is less than a nonagglomerated powder, decreasing the bulk density. Bulk density results for each sample did not change over storage time (P > 0.05). The bulk density of bagged agglomerated whey powder can increase if too much weight is stacked on top of the bags. However, for this project the powders were stored in plastic bins and thus were not subjected to any excess weight to compress the powder. A change in bulk density over time was not otherwise expected.

Color

L*, a*, and b* values of the whey proteins and reconstituted whey proteins were distinct for products from different facilities (P < 0.05) with less apparent and inconsistent differences due to agglomeration (results not shown). L* represents the lightness value (black [0] to white [100]), and a*, b* represent chromaticity coordinates (b* blue [−] and yellow [+], a* green [−] to red [+]). Most samples changed over time in L* and b* values (P < 0.05), although not with any apparent pattern. In the reconstituted samples, the a* value was more sensitive to change than the L* and b* values. Similarly, Sithole and others (2005), reported that Hunter L* and a* values appeared to be most sensitive to change during storage of sweet whey powder. They hypothesized that the color change was due to Maillard reactions occurring during storage. However, WPC80 and WPI should be less sensitive to change than unconcentrated whey powder because they contain less lactose, and color changes observed in this study were smaller than those reported by Sithole and others (2005).

Dispersibility

Dispersibility is defined as the breaking up of the powder mass into the primary particles in a liquid (Pietsch 2005). A powder high in dispersibility takes less time or force to disperse into solution than a powder with low dispersibility. In this study, dispersibility was inversely related to the shake number value reported. Thus, a powder with a low shake number value was highly dispersible. Nonagglomerated powders required more agitation to disperse than either the steam or lecithin agglomerated powders (>90 shakes for nonagglomerated WPC80 compared with 15 to 42 for agglomerated/instantized WPC80; >35 shakes for nonagglomerated WPI compared with 11 to 25 shakes for agglomerated/instantized WPI) (P < 0.05). There was not a difference between dispersability of steam-agglomerated powders compared to the lecithin-agglomerated samples and values did not change with storage time (results not shown) (P > 0.05). Gaiani and others (2007) found the wetting time for freshly agglomerated WPI to be much lower than nonagglomerated WPI. Additionally, Sithole and others (2006) found the wetting time for agglomerated sweet whey powder to be lower than nonagglomerated powder. Wetting time is inversely correlated to dispersability, as wetting time increases dispersablity goes down. The primary reason for agglomeration is to increase and ease dispersion of the powder into liquid. Sometimes, powders are agglomerated with lecithin, which is an emulsifier and wetting agent, and further increases dispersion. Thus, agglomerated powders should be higher in dispersibility (lower in wetting time) than nonagglomerated powders. Previous research indicated that the wettability of whey powders did not change significantly over 9 mo of storage (Sithole and others 2006), similar to results from this study.

Solubility index

Solubility index values ranged from 0.2 to 0.5 mL sediment (results not shown), did not change with storage time, and were not distinct between agglomerated and nonagglomerated samples (P > 0.05). A high solubility index indicates low solubility. Previous studies indicated that there was not a significant change in solubility index in skim milk powders over 29 y of storage (Lloyd and others 2004). Thus, the solubility index of similar dairy products, WPC80 and WPI, was not expected to change significantly over 18 mo of storage. However, Sithole and others (2005) reported a change in solubility index values of some sweet whey powders after 9 mo of storage, using a similar centrifugation method to that used in this study. Formation of these Maillard reaction products during storage can influence solubility and may account for the changes observed with sweet whey powder, which contains more lactose than WPC80 and WPI.

Ingredient applications

A 6% protein peach beverage was selected as the ingredient application evaluated since studies have indicated that this application is liked by consumers and is also sensitive to ingredient flavor carry-through (Drake and others 2009) while a clear acidic beverage was chosen for WPI since this is a common beverage application for WPI (Childs and others 2007; Drake and others 2009). WPC80 from facilities 1 and 2 (Cheddar and Mozzarella WPC80, respectively) and WPI from facility 3 were selected for ingredient applications testing across storage time. Trained panelists detected off-flavors in beverages made from stored agglomerated WPC80 sooner than from nonagglomerated WPC80 (P < 0.05) (results not shown). Cardboard and brothy flavors were detected in all beverages initially, consistent with previous studies (Childs and others 2007), but significant increases in these flavors were noted in the beverages made from agglomerated products (with or without lecithin) following 12 (facility 1) or 15 mo (facility 2) storage. In contrast, increases in these flavors were documented in beverages made from nonagglomerated samples following 15 (facility 3) or 18 (facility 1 and 2) mo storage. Similar to WPC80, trained panelists detected differences among WPI beverages made from fresh and stored WPI from the same facility with differences detected as early as 9 mo of storage (results not shown). Beverages made using the 9 and 12 mo stored WPI were significantly higher in cardboard flavor than beverages made using fresh WPI. Also, beverages using the stored agglomerated WPI (9 mo and longer) were significantly higher in cabbage and brothy flavors than beverages made with the stored nonagglomerated or fresh agglomerated WPI.

There were differences in consumer liking for beverages made from fresh products from different facilities (P < 0.05), consistent with previous studies (Drake and others 2009). For this reason, fresh and stored products from different facilities were tested on separate days. Consistent differences in consumer acceptability among beverages from fresh and stored agglomerated WPC80 were documented after 12 mo storage (Table 9). Beverages made using 18 mo WPI (agglomerated or nonagglomerated) scored lower than beverages made from fresh WPI (Table 10). To our knowledge, previous studies have not addressed the impact of whey protein storage on sensory quality in ingredient applications.

Table 9—.  Mean overall liking scores (n= 75 consumers) for 6% protein peach flavored beverages made from WPC80 stored for various times.
FacilitySampleOverall liking
  1. Liking was scored on a 9-point hedonic scale where 1 = dislike extremely and 9 = like extremely.

  2. Different letters within each facility indicate significant differences in liking (P < 0.05).

Facility 1Fresh nonagglomerated4.7a
Fresh steam agglomerated4.5a
Fresh lecithin-agglomerated4.4ab
Nonagglomerated 12 mo4.3ab
Nonagglomerated 18 mo4.1b
Steam agglomerated 12 mo4.2b
Steam agglomerated 18 mo3.8c
Lecithin agglomerated 12 mo3.9c
Lecithin agglomerated 18 mo3.6c
Facility 2Fresh nonagglomerated5.5a
Fresh steam agglomerated5.4a
Nonagglomerated 12 mo5.3a
Nonagglomerated 18 mo5.0ab
Steam agglomerated 12 mo4.8b
Steam agglomerated 18 mo4.6b
Table 10—.  Mean overall liking scores (n= 75 consumers) for 6% protein fruit punch flavored acidic beverages made from WPI stored for various times.
FacilitySampleOverall liking
  1. Liking was scored on a 9-point hedonic scale where 1 = dislike extremely and 9 = like extremely.

  2. Different letters indicate significant differences in liking (P < 0.05).

Facility 3Fresh nonagglomerated4.2a
Fresh steam agglomerated4.0a
Nonagglomerated 9 mo4.2a
Nonagglomerated 18 mo3.6b
Steam agglomerated 9 mo4.2a
Steam agglomerated 18 mo3.6b

Previous research has demonstrated that off-flavors in WPC80 and WPI can carry through into ingredient applications and negatively affect consumer acceptance, but that the potential for flavor carry-through is dependent on the nature of the off-flavor as well as the specific ingredient application (Drake 2006; Childs and others 2007; Drake and others 2009). Storage-related flavor changes were detected by trained panelists in directly rehydrated whey proteins sooner than in beverage applications. Flavor differences in dried dairy ingredients would be expected to be more apparent in directly rehydrated products compared to when incorporated into an ingredient application (Caudle and others 2005; Drake 2006). More differences were also found between beverages by trained panelists than by the consumers. This can easily be explained by the fact that a trained descriptive panel is trained to be much more sensitive to sensory differences than the average consumer (Meilgaard and others 1999; Drake 2007). It is possible that some of the flavor differences documented by trained panelists between the fresh and stored products were not discernable for consumers and/or they were readily masked by the beverage application. Childs and others (2007) also hypothesized that consumers had more difficulty distinguishing among products that were not liked (<5 on the 9-point hedonic scale) which might explain why differences were not noted in consumer acceptance for WPI beverages until after 18 mo of storage. The presence of storage-related off-flavors in ingredient applications is undesirable whether or not they are detectable by consumers. Most ingredient applications must have a commercial shelf life and maintain fresh flavor across this shelf life. The use of whey protein ingredients that already display off-flavors by descriptive analysis and volatile lipid oxidation products by instrumental analysis would certainly not be conducive to an application with an optimum shelf life, even if the application initially masked off-flavor carry-through.

Conclusions

Sensory and volatile analyses indicated that WPC80 and WPI changed in flavor over storage time, including an increase in lipid oxidation flavors and lipid oxidation volatiles. These changes were more rapid in the agglomerated and instantized products than in the nonagglomerated products. Off-flavors were detected by trained panelists in rehydrated steam and lecithin agglomerated whey proteins as early as 4 to 6 mo of storage and in nonagglomerated whey proteins within 6 to 10 mo storage. Flavor differences were documented by trained panelists between beverage applications made using fresh or stored WPC80 and WPI as early as 9 mo. There were significant differences in consumer acceptance between beverages containing fresh WPC80 and 12-mo-old agglomerated powders and between beverages made from fresh WPI and WPI stored for 18 mo. Since whey protein quality is often evaluated by sensory quality and ingredient applications are generally expected to have a commercial shelf life, these results suggest that the optimum shelf life for nonagglomerated WPC80 and WPI stored at 21C is 12 to 15 mo, and 8 to 12 mo for steam-agglomerated or lecithin-agglomerated WPC80 and WPI.

Acknowledgments

This study was funded in part by Dairy Management, Inc. Paper number FSR 08-24 of the journal series of the Dept. of Food, Bioprocessing, and Nutritional Sciences, North Carolina State Univ.

Ancillary