Rheological study of different mashed potato preparations using large amplitude oscillatory shear and confocal microscopy

Journal of Food Engineering 169 (2016) 326e337

Contents lists available at ScienceDirect

Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng

Rheological study of different mashed potato preparations using large amplitude oscillatory shear and confocal microscopy Helen S. Joyner (Melito)*, Alexander Meldrum University of Idaho, 875 Perimeter Dr., MS 2312, Moscow, ID 83843, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 24 March 2015 Received in revised form 26 August 2015 Accepted 28 August 2015 Available online 1 September 2015

Texture is a major driver for mashed potato products. Measuring food texture via descriptive sensory analysis can be expensive and timeconsuming, so rheometry is often used as a method of estimating food texture. The objective of this study was to determine small- and largestrain rheological behavior of different mashed potato formulations. Rheological data, including large amplitude oscillatory shear data, and confocal microscopy images were collected. Comparison of rheological data and confocal images showed relationships between sample rheological behavior and sample structural damage were found. With the exception of freeze-dried samples, samples exhibiting greater starch damage exhibited more linear, fluid-like rheological behavior. Freeze-dried samples showed microfractures on the surface of the starch granules. It was hypothesized that these microfractures allowed amylose strands to leach from the granules without destroying structural integrity. Based on these results, mashed potato rheological behavior, and therefore texture, can be controlled by controlling extent of starch damage. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Large amplitude oscillatory shear Potatoes Rheology Structure

1. Introduction Texture plays an important role in consumer acceptance of mashed potatoes and other potato products (Faulks, 2014). In cooked potatoes, most consumers prefer a soft, mealy potato compared to a non-mealy, sticky or waxy potato (van Marle et al., 1997). A high solids content can result in increased product firmness due to the increase in structural stability. Free amylose can form gels, also resulting in increased product firmness. Amylose can be released from the starch granule if the granule's surface has been damaged (Linehan and Hughes, 1969). The current gold standard for evaluating textural attributes is descriptive sensory analysis, which uses a trained panel to evaluate food products. However, developing and maintaining a trained panel can be time-consuming and expensive. An alternative way to obtain information about food texture is to relate texture to structure through mechanical behavior. Mechanical properties of food can be acquired through rheological techniques. Traditional rheology focuses on small-strain mechanical properties. Small-strain tests are useful for describing initial responses to stress and strain in a sample. However, small-strain tests are

* Corresponding author. E-mail address: [email protected] (H.S. Joyner). http://dx.doi.org/10.1016/j.jfoodeng.2015.08.032 0260-8774/© 2015 Elsevier Ltd. All rights reserved.

limited to the linear viscoelastic region, or the region in which the sample displays ideal behavior. Strains placed on foods during processing or chewing are generally much larger. It has been found that large-strain mechanical behavior relates more closely to oral processing and sensory behavior than small strain behavior (Çakir et al., 2012). Unfortunately, it can be difficult to measure fundamental mechanical behavior of foods under large strains due to violations of the assumptions of linearity made in the calculation of fundamental parameters or difficulties in preparing or maintaining samples in the required shape for testing. Over the last several decades, there has been increased effort to develop novel techniques to measure nonlinear viscoelastic behavior of foods. One such technique is large amplitude oscillatory shear (LAOS). Similar to traditional small amplitude oscillatory testing, samples are placed under an oscillating torque or deformation. In small-strain testing, the strain placed on the samples usually does not exceed 1%. In LAOS, the strain amplitude is much larger, ranging from 25% to >100%. These large strains can result in significant nonlinear behaviors, which can be viewed in a plot of stress versus strain, or a Lissajous plot, as deviations from an elliptical shape. Ewoldt et al. (2008) developed a technique to quantify the extent and type of nonlinear behavior shown in these Lissajous plot. Briefly, a mathematical technique is used to deconvolve the output stress waveform into various harmonics. Further mathematical manipulation allows the third-order harmonic data

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

to be used to determine both extent and type of nonlinear behavior. The extent of nonlinear behavior can be measured using the third 0 00 harmonic viscoelastic moduli, G3 and G3 . The type of nonlinear behavior can be identified by using two new viscoelastic moduli and two new instantaneous viscosities defined by Ewoldt et al. (2008). These new parameters can be seen in the Lissajous plots 0 in Fig. 1. The ratio between the large-strain elastic modulus (GL ) and 0 the minimum-strain elastic modulus ðGM ) can be compared to find 0 0 the extent of strain-softening (GL =GM < 1) or strain-hardening 0 0 (GL =GM > 1) behaviors. The ratio of the instantaneous viscosities 0 0 at maximum shear rate (hL ) and at minimum shear rate ( hM ) can be used in a similar manner to determine the extent of shear-thinning 0 0 0 0 (hL =hM < 1) and shear thickening behaviors (hL =hM > 1) (Melito et al., 2012b). Various LAOS techniques have been used to characterize nonlinear viscoelastic behavior of polymers for several decades. These techniques can also be used to measure nonlinear viscoelastic behavior of food products, such as whey protein gels (Melito et al., 2012a), gluten gels (Ng et al., 2011), hydrocolloid solutions (Melito et al., 2013a), and cheese (Melito et al., 2013a). Understanding how and to what extent food structures deform under strain can yield valuable insight on how food structure and structural changes impact food texture. Therefore, the objective of this

327

study was to evaluate small- and large-strain rheological behaviors of different mashed potato preparations, and determine the relationships between starch structure and rheological behavior by comparing rheological data with confocal microscopy. 2. Material and methods 2.1. Materials Burbank potatoes were purchased from a local retailer and kept in a controlled atmosphere for up to 48 h before use. Instant potato flakes (Idahoan, Lewisville, Idaho, USA) were also purchased from a local retailer and kept in a desiccator to prevent moisture uptake. Potato starch was donated by AVEBE (Food Grade Potato Starch, AVEBE; Veendam, Netherlands). Samples were stained for confocal imaging with EMD Millipore HARLECO® Safranin Solution for Gram Staining (65092B-95, Merck KGaA, Darmstadt, Germany). Potato samples evaluated were termed “whole potato”, “freezedried potato”, “instant potato” and “potato starch”. Whole potato refers to mashed potatoes made with a raw potato; freeze-dried potato refers to mashed potatoes prepared from whole potato samples that were freeze-dried and rehydrated; instant potato refers to mashed potatoes prepared from commercial instant potato

Fig. 1. Lissajous plots showing a. linear, elastic behavior; b. nonlinear, elastic behavior; c. linear viscous behavior; d. nonlinear viscous behavior. The red dashed line indecates Chebyshev polynomials of the first kind, a secondary tool to evaluate nonlinear behavior and the small dotted lines are the slope. In a. and b., G’L ¼ slopes of the solid lines and G’M ¼ slopes of the red dotted lines. In c. and d., h’L ¼ slopes of the solid lines and h’M ¼ slopes of the dotted lines (Melito et al., 2012b). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

328

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

flakes; and potato starch refers to mashed potatoes prepared from commercial potato starch that was rehydrated to give the starch similar moisture content to mashed potatoes made from whole potatoes. It should be noted that although these samples are differentiated by preparation method, two of the four samples (instant potatoes and potato starch) were commercial products and were used as-is. Because this study focuses on the differences among the samples due to starch damage rather than the effects of other process-related differences or different potato cultivars, the decision was made to use commercial instant potatoes and potato starch rather than prepare these samples from whole potatoes in the lab. 2.2. Sample preparation Whole potatoes were peeled and sliced into cubes by hand with 13 ± 2 mm sides. The potatoes were boiled at 97.1  C in water for 15 min. The potatoes were drained, placed in a blender (Waring Commercial; Torrington, Connecticut, USA) and blended at 18,000 RPM for 10 s. To prepare the freeze-dried potato samples, 100 g of the blended potatoes were placed in a 80  C freezer for 24 h. The frozen potatoes were then freeze-dried in a Labconco 4.5 L freeze-dryer (Kansas City, Missouri, USA) for 48 h. Instant potatoes and freeze-dried potatoes used for rheological testing were rehydrated to 82% moisture content (wet basis) using boiling water at 97.1  C. The rehydrated samples were blended for 10 s at 18,000 RPM. Potato starch was rehydrated in a centrifuge tube to 82% moisture content (wet basis) using boiling water. The centrifuge tube containing the rehydrated potato starch was capped and placed in a beaker of boing water for 5 min to allow the starch to gelatinize. All rehydrated samples and the whole potato sample were allowed to cool at room temperature to 25  C after preparation. Samples were tested immediately after cooling. Preliminary differential scanning calorimetry testing showed that all samples were completely gelatinized after preparation (data not shown). Since rheological properties can vary between samples due to compositional and structural differences. Samples were prepared with similar levels of moisture content (Table 1) to allow proper comparison of structural aspects.

Ash contents were determined in triplicate by a standard method (Wehr and Frank, 2004). Protein content was considered negligible. Carbohydrate content was calculated by subtracting percents moisture and ash from 100 percent.

2.4. Rheological testing All rheological data was collected on a DHR-3 rheometer (TA Instruments; New Castle, Delaware, USA) using a parallel plate system (25 mm diameter) with a gap height of 1 mm. A thin layer of Vaseline was applied to outside edge of the samples to reduce moisture loss in the sample. Tests were conducted at 25  C. All samples were tested in triplicate. Strain sweeps was conducted from 0.01 to 100% strain. The end of the linear viscoelastic region (LVR) for each sample was determined as the strain at which the complex moduli deviated more than 2% from the region of constant moduli. Critical stress and strain were taken as the stress and strain at the end of the LVR. Frequency sweeps and creep were performed at 75% of the smallest critical strain and stress determined, respectively, to ensure that testing was conducted within the LVR. Frequency sweeps were conducted at 0.03% strain (75% of the critical strain of the weakest sample) from 0.1 to 100 rad/s, with 5 points per decade of frequency collected. Creep tests were conducted at a stress of 7 Pa with a duration of 180 s. Stress relaxation tests were conducted at 0.03% strain with a duration of 180 s. The Deborah number was calculated by dividing the relaxation time by the process time.

NDe ¼

lrel tprocess

The relaxation time was calculated from the stress relaxation tests. Relaxation time was equal to the time when the stress fell to 36.8% of the initial stress. Process time was equal to the time required to complete one oscillation (inverse of the frequency). Large amplitude oscillatory testing was conducted by collecting raw waveform data at 1%, 10%, 25%, 50%, and 100% strain at 3 different frequencies (0.2, 2, and 20 rad/s). All testing was performed using strain-control mode. Waveform data were evaluated for slip during testing (e.g. appearance of truncated peaks), and data showing slip were discarded.

2.3. Moisture content The moisture content of each sample was determined by placing a pre-weighted sample in a forced draft oven (Yamato DKN400; Yamato, Japan) at 135  C for 4 h. The sample was weighted again after the 4-h incubation time. Moisture content (MC) was calculated by:

  ðweight wet sample  weight dry sampleÞ MC ¼ 100% weight wet sample

2.5. Confocal scanning laser microscopy All samples were stained with safranin at a concentration of 0.375% in water overnight, then rinsed in water before viewing. Zseries were collected on an Olympus Fluoview 1000 laser scanning confocal microscope (Olympus Corporation; Center Valley, PA, USA) using 10, 20 and 60 objectives and a 551 nm laser.

(Eq. 1) 2.6. Data analysis

Table 1 Sample composition (wet basis)a. Sample

Water (%)

Carbohydrate (%)

Ash (%)

Starch Instant Freeze-dried Whole potato

83.0 83.1 82.8 82.5

16.0 16.8 16.0 16.4

0.00 0.10 1.12 1.13

a

Proteins and fat contents were considered to be negligible.

LAOS data were analyzed using the MITlaos program (MITlaos beta) created by Ewoldt et al. (free software download can be requested at [email protected]) (2007) designed for MATlab (MathWorks Natick, Massachusetts, USA). Moisture contents and selected rheological data were analyzed using ANOVA followed by Tukey's test using SAS 9.3 (SAS Institute, Cary, NC, USA). Confocal images were quantitatively analyzed for granule area, width, length, circularity, and roughness using NIS Elements Confocal software (Nikon Instruments Inc., Melville, NY, USA).

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

3. Results and discussion 3.1. Microscopy images Confocal microscopy images showed differences in the structural integrity of the starch granules among the samples (Fig. 2). Whole potato and freeze-dried samples showed the least amount of structural damage; the granules contained no large holes and had shapes typical of potato starch. However, the freeze-dried samples appeared to absorb more dye than the whole potato samples (Fig. 2b, circled area 1), and freeze-dried sample granules appeared more textured than the whole potato sample granules. This phenomenon may be attributed to tiny fractures on the outside of the starch granule, likely caused by the freeze-drying process. Unlike the whole and freeze-dried potato samples, both the instant and starch samples showed visible structural damage. The most notable damage to the instant sample granules was the deep fissures across the center of many of the granules, indicating extensive granule damage (Fig. 2c, circled area 2). Starch sample granules had the greatest amount of granule damage, with many broken pieces and few intact granules. The extent of granule damage can alter the structure of starch solutions, and those structural changes can have significant impact on rheological behavior. As the degree of visible starch damage increased the average size and size variability of the granules decreased (Table 2). The decreased granule size likely reflected damage due to processing, as

329

whole potato samples had the largest granule size while the potato starch, which underwent significant processing, had the smallest granule size. Surface damage was analyzed by comparing sample roughness. All samples had similar average roughness, although the standard deviations varied. The whole potato sample exhibited the least variability in roughness, likely due to limited surface damage to the granules. On the other hand, the starch, instant, and freezedried samples exhibited larger variation in roughness. These variations in roughness were an indication of damage to the surface structures. The circularities of the samples were similar, with the exception of the potato starch sample, which had a higher circularity and lower standard deviation than the other samples. Size, shape, and surface roughness can all affect rheological properties, as granules differing in these parameters will move past each other in different ways, particularly under large forces or deformations. Thus, shape diversity of the starch granules in the different samples can yield unique rheological behaviors and it may be possible to control rheological behaviors by producing certain microstructures.

3.2. Strain sweep results Results from the strain sweeps are shown in Table 3. Whole potato samples showed the greatest critical stress, while the starch sample had the lowest critical stress. Critical stresses were related to the amount of starch granule damage. As the amount of granule damage increased, the critical stress decreased. Critical strain, on the other hand, was less affected by starch granule damage. Instant,

Fig. 2. Safranin stained confocal images at 10 magnification; Images are 1500um  1500 um; a. whole potato, b. freeze-dried potatoes, c. instant potatoes, and d. starch. Circle 1 gives an example of a starch granule that has absorbed a higher level of safranin stain. Circle 2 gives an example of a damaged starch granule.

330

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

Table 2 Quantitative confocal analysis results. Sample

Area (mm2)

Starch Instant Freeze-dried Whole potato

1.74 19.8 23.4 35.4

   

103 103 103 103

(1.62 (12.7 (18.1 (19.8

   

103) 103) 103) 103)

Length (mm)

Width (mm)

Circularity

Roughnessa

56.7 183 200 257

26.7 96.7 103 127

0.858 0.809 0.810 0.803

0.956 0.968 0.963 0.952

(28.8) (68.2) (90.3) (62.6)

(10.8) (40.2) (44.1) (46.1)

(0.005) (0.133) (0.122) (0.063)

(0.046) (0.028) (0.044) (0.005)

Data are presented as mean (standard deviation). a Roughness is a measurement of surface irregularities and is measured using the change in reflection intensity across a surface, where 1.0 indicates a surface with no change in intensity across a plane (smooth surface).

Table 3 Critical values of the linear viscoelastic region for all samplesa. Sample

Critical stress (Pa)

Critical strain (%)

G* (kPa)

Phase angle (degrees)

Relaxation time (s)

Starch Instant Freeze-dried Whole potato

8.6 16.3 23.7 51.1

0.036 0.077 0.056 0.077

24.3 21.3 42.3 66.8

33.6 29.8 19.3 22.8

0.106 0.139 0.328 0.412

(0.77) (2.35) (3.11) (7.72)

C BC B A

(0.005) (0.000) (0.007) (0.000)

C A B A

(1.5) C (3.1) C (3.0) B (10.4) A

(2.44) (0.17) (1.11) (1.87)

A A B B

(0.050) (0.073) (0.009) (0.159)

A A B B

Data are presented as mean (standard deviation). a Letters in each column that are different indicate significant differences.

freeze-dried and whole potato samples did not have significantly different critical strains. The starch sample showed significantly lower critical strain compared to the other samples, possibly due to the total collapse of the granule and structure mainly being provided by free amylose strands (Perera et al., 1997). The trends observed in the critical stress values were reflected in the complex modulus (G*) values (Table 3). G* values increased with increasing critical stress within the significance between samples. These results indicated that granule structure impacted moduli values, with a decreased G* value, or less rigid structure, corresponding to increased granule damage. All samples showed solidlike viscoelastic behavior, with phase angles below 45 . Freezedried samples had the lowest phase angle, while starch samples had the highest. In general, strain sweep results showed that samples with a greater extent of granule damage were less rigid and exhibited more fluid-like behavior. 3.3. Frequency sweep results Frequency sweep results are shown in Fig. 3. All samples showed greater values for the elastic modulus (G0 ) as compared to the loss modulus (G00 ), indicating elastic-dominant behavior. These results were expected based on the strain sweep results: all samples showed elastic-dominated behavior in the LVR. Elastic moduli values increased with decreased starch granule damage, which is in agreement with previous rheological results. Viscous moduli values were fairly similar among samples. These

Fig. 3. Frequency sweep results.

results indicate that the elastic behavior was more affected by starch granule damage than viscous behavior.

3.4. Creep results Creep results are shown in Fig. 4. Freeze-dried samples had the smallest compliance values, while starch samples had the largest compliance values. As with the strain sweep results, these results correspond to the amount of observed granule damage, with an increase in granule damage corresponding to increased compliance values. These results are also in agreement with the phase angle values: samples with higher phase angles showed higher compliance values. These results were expected, since larger compliance values indicate a greater extent of long-term flow, which is indicative of increased viscous behavior. However, it should be noted that materials can display viscous-type behaviors (e.g. flow, permanent deformation) while still exhibiting elastic-dominated behavior. Although all samples showed elastic-dominated behavior in the frequency sweep, the fact that their phase angles were greater than zero indicated that they exhibited some degree of viscous behavior, even if that behavior is not dominant over the elastic behavior. Interestingly, the freeze-dried samples showed lower compliance values than the whole potato samples, although they showed greater starch granule damage. It is possible that the

Fig. 4. Creep results.

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

higher strains and frequencies, indicating more elastic-dominant behavior. On the other hand, the shapes of the Lissajous plots for instant, freeze-dried, and whole potatoes at strains of 10% indicated viscous-type behavior, which was in agreement with the phase angle data (Table 4). At first, these results appear to contradict the frequency sweep data, which show elastic-dominated behavior over the entire frequency range covered in LAOS testing. However, these frequency sweeps were performed at 0.03% strain, which is significantly lower than the lowest strain used for LAOS testing (1%). The shift from elastic-dominated to viscous-dominated behavior during LAOS testing may be explained by looking at the phase angles for the samples at different strains and frequencies (Table 4). As the strain placed on the samples increased, the phase angles increased, shifting the dominant behavior in most samples from elastic to viscous. This shift is likely a result of damage to the sample microstructure such as network disruption, leading to viscous-type behaviors such as permanent deformation and flow. It should be noted that the phase angles reported in Table 4 were calculated using the MITlaos program, which accounts for nonlinearity in the output stress wave when calculating phase angle. Therefore, the phase angles reported should be an accurate representation of the ratio of elastic to viscous behavior, regardless of strain and frequency. The relative amounts of nonlinear behavior can be estimated from the full Lissajous curves by the amount of distortion from an elliptical shape. In general, the amount of distortion from an elliptical shape increased with increased strain and frequency for all samples. Starch samples showed the least distortion from an elliptical shape compared to the other samples. Among the other samples, no one sample showed (qualitatively) more nonlinear

microfractures on the surface of the freeze-dried granules may have allowed amylose strands to leach from the granule and form a gel network upon cooling. This network would yield a more solidlike structure, and a corresponding decreased in compliance values. Although the magnitudes of the compliance values differed among samples, all samples displayed similar compliance curve shapes (Fig. 4). After a sharp initial increase, compliances leveled off to an equilibrium value. These results indicated elastic-dominant behavior, which was in agreement with the strain and frequency sweep results. 3.5. LAOS results

2000

2000

1000

1000

1000

0

-1000

0

-1000

-2000

-50

0

50

100

-50

-2000

-100

-50

0

50

100

-100

1000

-1000

-50

0

50

100

0

-1000

-2000

-100

Stress (Pa)

2000

1000

Stress (Pa)

2000

1000

0

-50

Strain (%)

0

50

100

-50

0

50

100

-100

1000

500

500

500

500

-500

-1000

-1500 -50

0

Strain (%)

Starch

50

100

0 -500

-1000

-1500 -100

Stress (Pa)

1500

1000

Stress (Pa)

1500

1000

Stress (Pa)

1500

1000

-1000

-50

0

50

100

100

-50

0

50

Strain (%)

Instant

0 -500

-1500 -100

Strain (%)

0

-1000

-1500 -100

-50

Strain (%)

1500

-500

50

0

Strain (%)

0

100

-2000

-100

Strain (%)

0

50

-1000

-2000

-100

0

Strain (%)

2000

0

-50

Strain (%)

1000

Stress (Pa)

Stress (Pa)

100

2000

-2000

Stress (Pa)

Frequency (rad/s)

50

Strain (%)

-1000

0.2

0

0

-1000

-2000

-100

Strain (%)

2

0

-1000

-2000

-100

Stress (Pa)

2000

1000

Stress (Pa)

2000

Stress (Pa)

Stress (Pa)

3.5.1. General Lissajous comparison As previously discussed, nonlinear behavior can be viewed in Lissajous plots, or plots of stress versus strain. While decomposition and analysis of the stress wave can yield quantitative information of large-strain viscoelastic behavior (see the next section), it is also possible to analyze the full stress wave for nonlinear behavior. Fig. 5 shows Lissajous figures for all potato samples at different frequencies. Each sample displayed a distinct Lissajous pattern, particularly at higher frequencies. The differences in the Lissajous plots highlight the differences in viscoelastic behavior among the potato samples. Starch samples had higher peak stress values than the other potato samples at all frequencies. The shapes of the starch sample Lissajous plots were also more indicative of linear behavior than the other samples, as the minor axes of the ellipses were shorter in length. These results were in agreement with phase angle data for each strain-frequency combination (Table 4). Phase angles for the starch samples were lower than the other samples, particularly at

20

331

Freeze-dried

100

-100

-50

0

50

100

Strain (%)

Whole

Sample Fig. 5. Lissajous plots for potato samples at different frequencies and strains. Strains shown are 1%, 10%, 25%, 50% and 100%, with the area under the curve increasing with increased strain.

332

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

Table 4 Phase angles for samples under LAOSa. Frequency (rad/s)

Strain (%)

Phase angles (degrees) Starch

0.2

2

20

1 10 25 50 100 1 10 25 50 100 1 10 25 50 100

21.0 10.0 10.0 20.0 37.0 22.7 15.9 14.6 16.1 21.3 28.7 17.5 16.5 15.8 15.8

(4.94) (4.30) (1.61) (9.60) (7.85) (6.12) (0.35) (2.02) (2.25) (2.58) (5.56) (2.50) (2.79) (2.10) (0.67)

Instant BC C C BC A BC BC C BC BC AB BC BC BC BC

24.1 62.4 69.9 75.0 77.7 35.3 50.9 52.0 56.3 59.0 38.2 44.5 47.5 50.7 53.4

(1.90) (4.11) (3.06) (2.01) (1.83) (10.8) (2.37) (4.26) (3.66) (2.79) (1.44) (1.52) (2.96) (3.83) (1.42)

H BC AB A A GH CDE CDE CDE BCD FG EFG DEF CDE CDE

Freeze-dried

Whole

27.3 (13.16) C 75.4 (3.31) BA 77.6 (3.58) BA 80.4 (2.51) BA 81.9 (1.29) A 21.3 (2.7) C 73.6 (0.86) BA 77 (0.89) BA 77.5 (4.42) BA 82.2 (1.13) A 25.0 (10.8) C 59.7 (14.0) BA 59.0 (10.7) B 73.5 (10.9) BA 77.8 (8.44) BA

22.6 75.5 78.1 80.7 81.8 30.2 63.8 68.6 70.9 73.1 20.7 50.6 54.2 61.5 66.9

(4.68) (2.03) (1.64) (2.30) (1.35) (2.24) (0.43) (1.32) (0.74) (0.53) (1.63) (2.67) (4.34) (2.23) (1.75)

I ABC AB A A H EF CDE CD BCD I G G F DEF

Data are presented as mean (standard deviation). a Letters in each column that are different indicate significant differences.

behavior at frequencies of 2 and 20 rad/s, but the freeze-dried and whole samples at 0.2 rad/s showed more distortion from an elliptical shape. 3.5.2. Elastic and viscous Lissajous comparison The LAOS analysis developed by Ewoldt et al. (2008) allows quantitative comparison of the extent and type of nonlinear 0 0 behavior. The ratio of G3 to G1 gives a measure of the extent of nonlinear behavior for samples where elastic behavior is dominant, with increased values corresponding to increased nonlinear 0 0 behavior. A value of <0.01 G3 =G1 has been found to indicate linear 0 0 viscoelastic behavior, while a value of >G3 =G1 0.01 indicates nonlinear viscoelastic behavior (Melito et al., 2012b). For samples 00 00 with viscous-dominant behavior, the ratio of G3 to G1 is used to determine the extent of nonlinear behavior, with the same cutoff value for linear versus nonlinear viscoelastic behavior. 0 0 The ratio of GM to GL is a measure of elastic-related nonlinear 0 0 behavior. A value of GL =GM less than 0.9 indicates strain softening, 0 0 while a value of GL =GM 1.1 indicates strain hardening (Ewoldt et al., 0 0 2008). Similarly, the ratio of hM to hL is a measure of viscous-related 0 0 nonlinear behavior. A value of hM =hL <0.9 indicates shear thinning, 0 0 while a value of hM =hL 1.1 indicates shear thickening (Ewoldt et al., 2007, 2008). LAOS parameters for all samples are shown in Tables 5 and 6. 0 0 00 00 There appears to be a general trend of increased G3 =G1 and G3 =G1 with increased strain for all samples. Interestingly, the trend of 0 0 increased G3 =G1 with increased strain is clearer at lower strains and 00 00 frequencies, while the trend of increased G3 =G1 with increased strain is clearer at higher strains and frequencies. Further examination of the LAOS data revealed a shift in several samples from significant elastic nonlinear behavior to significant viscous nonlinear behavior as strain increased. As previously discussed, these results appear to contradict the frequency sweep data; however, the shift from elastic-dominated to viscous-dominated behavior during LAOS testing may be explained using phase angle data (Table 4). Again, the probable cause of this elastic-to-viscousdominated behavior is permanent deformation and flow due to microstructural damage (network disruption) caused by increased strain. All samples but the starch sample showed a change from elastic-dominated to viscous-dominated behavior above certain strains. The change in behavior was unexpected based on the other rheological data and the common conception that mashed potatoes are more solid than fluid. The materials used in most published LAOS studies that incorporate the analytical technique used in this

study displayed either primarily elastic or primarily viscous behavior (Ewoldt et al., 2007). In this study, however, a shift in the dominance of elastic to viscous behavior was observed. This shift is likely due to the onset of flow of the starch granules past each other and network breakdown at larger strains. Because of this shift in viscoelastic behavior, both the elastic and viscous Lissajous curves and viscoelastic parameters were used to analyze the nonlinear behavior of the samples. Elastic behavior is dominant for all samples at 1% strain. At 1% strain, samples showed little to no nonlinear elastic behavior at all frequencies (Table 5). These results may also be seen in the elastic Lissajous plots (Figs. 6e9). No distortion from an elliptical shape is visible at 1% strain for any of the samples, indicating ideal visco0 0 elastic behavior. GL =GM values at 1% strain and all frequencies for all samples were all between 0.9 and 1.1, supporting the observation of ideal viscoelastic behavior. Above 1% strain, samples differ in their extent and type of nonlinear behavior (Table 5, note that higher standard deviations are not unexpected when evaluating samples under larger strains and frequencies). Freeze-dried potato samples generally displayed a greater extent of nonlinear elastic behavior at all strains and 0 0 0 0 0 0 frequencies, based on G3 =G1 and GL =GM values (Table 5). GL =GM values for freeze-dried samples indicated significant strainhardening behavior at strains above 1% and all frequencies. Additionally, the freeze-dried samples exhibited a greater extent of nonlinear viscous behavior than the other samples at frequencies of 00 00 0 0 0 0 0.2 and 2 rad/s, based on G3 =G1 and hL =hM values (Table 6). GL =GM 0 0 and hL =hM values indicated strain-hardening and shear-thinning 00 00 behavior, respectively. Interestingly, although the G3 =G1 values 0 0 indicated significant nonlinear behavior, hL =hM values did not indicate significant shear rate-dependent behavior. These results were confirmed in the elastic and viscous Lissajous plots, respectively (Fig. 6). It was unexpected that the freeze-dried samples would display more nonlinear behavior than the other samples, since the microstructures of these samples were not as damaged as the instant and starch potato samples. It is possible that the surface damage to the freeze-dried starch granules, likely caused by the freeze-drying process, was the cause of the increase in nonlinear behavior, especially since the freeze-dried granules were otherwise similar to the whole potato granules (Martens and Thybo, 2000). Unlike the other samples, the freeze-dried potato samples showed secondary loops in their viscous Lissajous plots at strains of 50% and 100% at frequencies of 2 and 20 rad/s. These loops were indicative of viscoelastic overshoot, which is analogous to the stress

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

333

Table 5 LAOS elastic parameters for all samplesa,b. Frequency (rad/s)

Strain (%)

Starch 0

0.2

2

20

Frequency (rad/s)

0

G3 =G1

1 10 25 50 100 1 10 25 50 100 1 10 25 50 100

1.06 1.04 1.04 1.03 0.99 1.07 1.04 1.04 1.06 1.08 1.04 1.03 1.02 1.06 1.12

(0.036) AB (0.02) AB (0.02) AB (0.06) AB (0.08) B (0.05) AB (0.01) AB (0.01) AB (0.02) AB (0.03) AB (0.01) AB (0.01) AB (0.01) AB (0.01) AB (0.01) A

0.017 0.011 0.010 0.013 0.016 0.017 0.010 0.010 0.013 0.019 0.012 0.009 0.008 0.013 0.026

Strain (%)

Freeze-dried 0

0.2

2

20

Instant

GL =GM

1 10 25 50 100 1 10 25 50 100 1 10 25 50 100

0

0

0

0

0

GL =GM (0.009) (0.005) (0.004) (0.004) (0.006) (0.013) (0.004) (0.003) (0.002) (0.004) (0.005) (0.002) (0.001) (0.001) (0.001)

AB AB AB AB AB AB AB AB AB AB AB AB B AB A

1.04 1.95 2.64 10.4 23.3 1.10 1.20 1.20 1.25 1.31 1.04 1.16 1.11 1.10 1.08

0

G3 =G1

(0.01) (0.52) (0.74) (7.57) (16.5) (0.10) (0.07) (0.04) (0.06) (0.03) (0.01) (0.02) (0.01) (0.03) (0.01)

B B B AB A B B B B B B B B B B

0.011 0.148 0.187 0.312 0.392 0.026 0.050 0.049 0.064 0.078 0.011 0.042 0.034 0.034 0.031

(0.003) (0.047) (0.039) (0.066) (0.096) (0.025) (0.002) (0.010) (0.013) (0.008) (0.003) (0.003) (0.002) (0.003) (0.001)

D BC B A A D CD CD CD CD D CD D D D

Whole

GL =GM

0

G3 =G1

0

0

GL =GM

0

0

G3 =G1

0

0

1.19 25.4 26.3 15.0 5.09 1.06 8.53 46.8 10.6 5.64 1.04 1.67 1.39 4.30 5.65

(0.23) A (34.1) A (31.6) A (13.1) A (2.95) A (0.02) A (2.19) A (42.9) A (7.94) A (2.82) A (0.02) A (0.59) A (0.474) A (2.69) A (4.66) A

0.035 (0.038) D 0.6 (0.182) AB 0.555 (0.219) AB 0.719 (0.245) A 0.766 (0.171) A 0.015 (0.003) D 0.441 (0.007) ABC 0.399 (0.016) ABCD 0.526 (0.086) AB 0.697 (0.089) A 0.01 (0.003) D 0.114 (0.083) CD 0.076 (0.074) CD 0.235 (0.157) BCD 0.264 (0.162) BCD

1.07 11.6 57.4 11.5 7.59 1.05 1.56 1.59 1.79 1.91 1.01 1.16 1.15 1.14 1.15

(0.015) B (4.43) AB (64.5) A (9.89) AB (4.63) AB (0.01) B (0.05) B (0.08) B (0.11) B (0.08) B (0.003) B (0.04) B (0.04) B (0.05) B (0.04) B

0.02 0.51 0.46 0.61 0.65 0.01 0.12 0.11 0.14 0.16 0.00 0.04 0.04 0.04 0.05

(0.004) C (0.10) AB (0.10) B (0.14) AB (0.11) A (0.002) C (0.008) C (0.014) C (0.014) C (0.011) C (0.001) C (0.008) C (0.008) C (0.013) C (0.013) C

Data are presented as mean (standard deviation). 0 0 0 0 a Absolute values of G3 =G1 and GL =GM values were used. b Letters in each column that are different indicate significant differences.

overshoot observed during initiation of viscous flow (Ewoldt and McKinley, 2010). The loops were due to strong nonlinear elastic 0 0 behavior (Bird et al., 1977), which was confirmed by the G3 =G1 and 0 0 GL =GM values (Table 5). Specifically, the instantaneous elastic stress response for elastic behavior was greater than the rate of deformation in the sample during the period of strain application in which the loop was observed, resulting in a net reduction in the stress response at larger deformations. Ewoldt et al. (2007) provides a full explanation of this phenomenon. This strong elastic response is in agreement with other rheological results, which indicated that the freeze-dried potato samples exhibited increased elastic behavior compared to the other samples. Instant potato samples displayed both significant nonlinear elastic behavior and significant nonlinear viscous behavior, particularly at frequencies of 2 and 20 rad/s (Fig. 7). Phase angles for the instant potato samples indicated viscous-dominated behavior for all strains at a frequency of 0.2 rad/s, 10% and higher strain at a frequency of 2 rad/s, and 25% and higher strain at a frequency of 20 rad/s. The dominance of elastic behavior at lower strains and higher frequencies may be explained by the Deborah number, which is a ratio of the relaxation time of the sample and the time required to perform a complete oscillation. A Deborah number greater than 1 indicates elastic-dominant behavior; a Deborah number less than 1 indicates viscous-dominant behavior. The Deborah number for the whole potato, instant, freeze-dried, and starch samples tested at 1 rad/s were 7.16, 9.39, 22.2, and 27.8, respectively. Thus, the samples at this (and higher) frequency exhibited elastic-dominant behavior, according to the Deborah

number. However, at higher strains, permanent deformation (flow) can result in viscous-dominated behavior, since the structure of the sample is disrupted. Since the Deborah number calculation is dependent on the assumption of ideal viscoelastic behavior, it is not an accurate indicator of viscous versus elastic behavior at high strains. In addition, the phase angles for instant potato samples at the frequencies were closer to 45 (the crossover point for elasticdominated to viscous-dominated behavior) than the phase angles for the other samples (Table 4). This relative balance between elastic- and viscous-dominated behavior is a likely explanation of why the instant potato samples exhibited both elastic and viscous nonlinear behavior. The extent of nonlinear viscoelastic behavior for the instant potato samples increased with both increased strain and frequency. At strain-frequency combinations where instant potato samples 0 0 0 0 showed elastic-dominant behavior, G3 =G1 and GL =GM values indicated fairly linear viscoelastic behavior at 1% strain and strainhardening behavior at 10% strain (Table 5). The elastic Lissajous plots at these strain-frequency combinations supported these results (Fig. 7a). At strain-frequency combinations where instant 00 00 potato samples showed viscous-dominant behavior, G3 =G1 and 0 0 hL =hM values indicated significant nonlinear viscous (shear-thinning) behavior only at frequencies at 20 rad/s (Table 6). Again, these results were confirmed by the shapes of the viscous Lissajous plots (Fig. 7b). Plots at 20 rad/s showed clear distortions from an elliptical shape compared to plots at lower frequencies. Viscous Lissajous plots at lower frequencies displayed a lesser extent of nonlinear behavior, although the distortion of the elliptical shape increased

334

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

Table 6 LAOS viscous parameters for all samplesa,b. Frequency (rad/s)

Strain (%)

Starch 0

0.2

2

20

Frequency (rad/s)

2

20

Instant 00

00

G3 =G1

1 10 25 50 100 1 10 25 50 100 1 10 25 50 100

0.98 (0.02) AB 0.84 (0.04) BCDE 0.94 (0.06) ABC 0.99 (0.05) AB 1.01 (0.03) A 0.95 (0.03) ABC 0.9 (0.03) ABCD 0.87 (0.05) ABCD 0.87 (0.05) ABCD 0.82 (0.06) CDE 0.91 (0.13) ABCD 0.77 (0.02) DE 0.78 (0.03) DE 0.76 (0.04) DE 0.68 (0.02) E

0.008 0.037 0.015 0.006 0.009 0.014 0.025 0.031 0.032 0.054 0.021 0.042 0.037 0.045 0.098

Strain (%)

Freeze-dried 0

0.2

0

hL =hM

1 10 25 50 100 1 10 25 50 100 1 10 25 50 100

0

(0.004) (0.007) (0.014) (0.003) (0.007) (0.007) (0.009) (0.010) (0.013) (0.024) (0.028) (0.004) (0.008) (0.005) (0.003)

CD BCD CD D CD CD BCD BCD BCD B BCD BCD BCD BC A

00

00

hL =hM

0

G3 =G1

1.03 1.12 1.09 1.03 0.99 1.01 0.97 0.95 0.93 0.91 1.01 0.88 0.82 0.77 0.74

(0.01) B (0.01) A (0.01) A (0.02) B (0.01) BCD (0.04) BC (0.01) CDE (0.02) DEF (0.01) EF (0.01) FG (0.003) BC (0.01) G (0.02) H (0.02) I (0.01) I

0.004 0.013 0.014 0.006 0.019 0.008 0.012 0.015 0.024 0.031 0.001 0.028 0.039 0.057 0.070

(0.003) (0.005) (0.003) (0.006) (0.005) (0.005) (0.002) (0.004) (0.006) (0.000) (0.000) (0.002) (0.008) (0.008) (0.003)

G EFG DEFG FG CDEF FG EFG DEFG CDE BC G BCD B A A

(0.016) (0.027) (0.008) (0.007) (0.003) (0.001) (0.000) (0.003) (0.002) (0.003) (0.001) (0.004) (0.005) (0.000) (0.002)

ABDC A DEC ABC AB BCDE E E DE BCDE BCDE E E BCDE AB

Whole 00

00

hL =hM

0

G3 =G1

1.22 0.95 1.01 0.95 0.93 1.13 1.05 1.03 0.98 0.90 1.06 1.02 0.97 0.83 0.69

(0.17) A (0.02) BCD (0.02) BCD (0.004) BCD (0.01) CD (0.01) BA (0.02) BAC (0.01) BC (0.05) BCD (0.002) CD (0.01) BAC (0.03) BC (0.03) BCD (0.10) DE (0.12) E

0.044 0.087 0.021 0.047 0.050 0.027 0.031 0.008 0.038 0.050 0.013 0.007 0.011 0.043 0.074

0

(0.032) BCD (0.013) A (0.007) CD (0.006) BCD (0.007) ABC (0.004) CD (0.009) CD (0.003) D (0.002) BCD (0.000) ABC (0.001) CD (0.003) D (0.004) CD (0.02) BCD (0.028) BA

00

00

hL =hM

0

G3 =G1

1.15 1.02 1.03 0.96 0.94 1.08 1.05 1.02 1.00 0.97 1.07 1.03 1.00 0.93 0.84

(0.08) A (0.07) BCD (0.02) BC (0.02) CDE (0.01) DE (0.01) AB (0.01) BC (0.02) BCD (0.01) BCDE (0.01) CDE (0.002) AB (0.02) BCD (0.03) BCDE (0.00) EF (0.01) F

0.033 0.054 0.012 0.037 0.041 0.018 0.001 0.004 0.007 0.016 0.016 0.003 0.006 0.022 0.038

Data are presented as mean (standard deviation). 00 00 0 0 a Absolute values of G3 =G1 and hL =hM values were used. b Letters in each column that are different indicate significant differences.

with increasing strain. Interestingly, the decrease in nonlinear viscous behavior at lower frequencies for the instant potato samples corresponded to an increase in nonlinear elastic behavior. Lissajous plots for instant starch samples at 0.2 and 2 rad/s displayed significant strain0 0 hardening behavior (Fig. 7), which was confirmed by GL =GM 0 0 values above 1.1 (Table 5). G3 =G1 values increased with increased strain, as expected, but they also increased with decreased frequency. Thus, instant potato samples displayed a greater extent of nonlinear behavior at lower frequencies and higher strains. As previously stated, increased strain can cause increased sample disruption, and a corresponding increase in nonlinear behavior. The increased nonlinear behavior at lower frequencies may be explained by considering the relationship of frequency to viscoelastic behavior. As previously stated, increased frequency results in an increase in the solid-like behavior of a material because the time to complete an oscillation is less than the relaxation time of the material (Winter, 1997). Provided that structural damage is minimal, i.e. there is little permanent deformation and flow in the sample, the material should act increasingly like an ideal solid at a given strain as the frequency increases. The potato starch sample showed elastic-dominant behavior for all strains and frequencies tested, according to phase angle data 0 0 (Table 4). Starch samples displayed lower G3 =G1 values compared with the other samples (Table 5), indicating a lower degree of nonlinear viscoelastic behavior. This result was also observed in the Lissajous plots (Figs. 5 and 8). While there is a small amount of distortion in the elliptical shape at large strains and frequencies, the

starch sample Lissajous plots remained closer to an ideal elliptical 0 0 shape than the plots for the other samples. GL =GM values for the starch samples also indicated fairly linear viscoelastic behavior; the 0 0 only GL =GM value indicating nonlinear behavior for starch samples was the value at 100% and 20 rad/s. It was unsurprising that the samples exhibited nonlinear viscoelastic behavior at this strainfrequency combination, as it was a large deformation at high 0 0 speed. However, the GL =GM value was only slightly above the cutoff for strain-hardening behavior. Overall, the starch samples showed little nonlinear elastic behavior at all strains and frequencies tested. On the other hand, starch samples showed significant nonlinear 00 00 0 0 viscous behavior, based on their G3 =G1 and hL =hM values (Table 6). The extent of nonlinear behavior increased with increased strain 0 0 and frequency; hL =hM values indicated shear-thinning behavior. These results were unexpected, as the starch samples displayed the lowest phase angles of all samples (Table 4). However, it is possible that the dominant elastic behavior lessened the effect of the nonlinear viscous behavior. Examining the shape of the starch Lissajous in Fig. 5, the distortions in the elliptical shape may have been due to the viscous behavioral component. Since both the distortions and extent of nonlinear behavior increased with increased strains and frequencies, this is a plausible explanation for the presence of nonlinear viscous behavior. Whole potato samples shifted from elastic-dominated to viscous-dominated behavior between 1% and 10% strain based on phase angle data (Table 4). They displayed nonlinear elastic viscous behavior at most strain-frequency combinations, based on the 0 0 00 00 G3 =G1 and G3 =G1 values, respectively (Tables 5 and 6). These results

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

Fig. 6. Elastic (a) and viscous (b) Lissajous plots for freeze-dried potato samples.

were observed in the distortion of the elastic and viscous Lissajous plots (Fig. 9). Whole potato samples displayed strain-hardening behavior at nearly all strain-frequency combinations, particularly at lower frequencies and higher strains (Table 5). Although they displayed significant nonlinear viscous behavior, whole potato samples generally did not display significant shear-thinning or shear-thickening behavior (Table 6). These results may be due to the structural features of the whole potato samples. As previously stated, the whole potato samples had the least amount of structural damage. It is possible that the undamaged granules were able to slide past each other at higher strains, resulting in viscous flow. The lack of damage to the granules would enable them to move smoothly past each other without entangling or flocculating regardless of shear rate; hence, no shear-dependent behavior would be present. Nonlinear behavior would therefore manifest as strain hardening (elastic type) behavior, as granules likely jammed against each other at high shear rates and were unable to flow (Kaur et al., 2002). These results show that the full Lissajous from a large-amplitude oscillatory test can be used as a fingerprint for differentiating different potato samples based on rheological behavior. Decoupling the Lissajous into elastic and viscous components can provide quantitative analysis of large-strain viscoelastic behavior. To properly understand viscoelastic behavior under large strain, both the elastic and viscous behavior should be analyzed together, as samples with elastic-dominated behavior may have a significant viscous component at different strains and frequencies, or vice versa. There has been significant effort to correlate rheological behaviors to structural features, textural attributes, and/or oral

335

Fig. 7. Elastic (a) and viscous (b) Lissajous plots for instant potato samples.

processing behaviors, (e.g. Çakır et al., 2012b, Çakır et al., 2012a; Melito et al., 2013b; Kokini, 1987; Richardson et al., 1989; Janhoj et al., 2009; Sonne et al., 2014.). However, the relationships between these features are complex; often, there are many interrelationships between texture and mechanical behaviors. Furthermore, a high degree of correlation between two or more features does not necessarily mean that one feature is the cause of the other. While there are certainly relationships between structural features, textural attributes, and oral processing behaviors, correlational analysis is not necessarily the best method to determine these relationships and caution should be used when determining how one feature impacts or relates to another. Nevertheless, although rheological behaviors may not directly relate to or cause certain textural attributes, particularly in samples with complex structures, rheology (in particular nonlinear rheology) may be used as a screening tool to reduce the number of samples requiring analysis by sensory panel. Rheology may also be used to understand how different structures deform and break down; this deformation and breakdown behavior often given an indication of expected textural attributes.

4. Conclusions Each potato sample showed different LAOS behavior, and the differences in those behaviors was found to be related to structural differences. In general, increased starch granule damage resulted in an increase in viscous-type rheological behavior, as well as a reduction in nonlinear behavior under large strains. It was hypothesized that damage to the surface structure resulted in

336

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

Fig. 9. Elastic (a) and viscous (b) Lissajous plots for whole potato samples. Fig. 8. Elastic (a) and viscous (b) Lissajous plots for potato starch samples.

differences in granule circularity and size, allowing for granules to more easily flow past each other and yielding more viscous-like behavior. Starch granules with less structural damage had greater increases in phase angle as strain increased. With the exception of the potato starch sample, all samples showed shifts from elastic-to viscous-dominant behavior at large strains. This behavioral change may cause changes in perceived texture. Therefore, when analyzing LAOS results using this elastic and viscous behavior decomposition technique, both elastic and viscous nonlinear behavior should be considered. Phase angles should be used to assist in data interpretation, as samples may shift from elastic-to viscous-dominated behavior or vice versa at certain strains and frequencies. These results also highlight the importance of starch granule damage on food rheological behavior. It should be noted that the presence of structural damage does not necessarily indicate undesirable rheological or sensory attributes. Rather, controlling the amount of starch granule damage during the production of starchbased foods can result in products with a variety of textures. Quantification of starch damage may not give sufficient information to understand finished product texture; coupling quantification of starch damage with LAOS analysis yields a better understanding of how structural features impact mechanical behaviors. This coupled evaluation has potential applications for the industry. For example, it may be used as a screening procedure to reduce the number of samples requiring sensory evaluation or to compare the effects of two different processes on finished product attributes. It may also be used to provide a better match of product

mechanical behavior by adjusting the process or formulation of a new product to match the Lissajous plot shape of an existing product. Two products with similar mechanical properties, particularly in the nonlinear region, are likely to have similar sensory attributes. Overall, evaluation of both starch damage and nonlinear mechanical behavior is a useful, cost-effective tool for product development. Nomenclature 0

G1 00

G1 0

G3 00 G3 0 GL 0

GM 0

hL

0

hM

fundamental (first harmonic) elastic modulus, normally reported by rheometer software (Pa) fundamental (first harmonic) viscous modulus, normally reported by rheometer software (Pa) third harmonic elastic moduli (Pa) third harmonic viscous moduli (Pa) large-strain elastic modulus, determined by the slope of the line from the origin to the maximum strain in a Lissajous plot of stress versus strain (Pa) minimum-strain elastic modulus, determined by the slope of the tangent line on a Lissajous at the origin of stress versus strain (Pa) instantaneous viscosity at maximum shear rate, determined by the slope of the line from the origin to the maximum strain in a Lissajous plot of stress versus strain rate (Pa s) instantaneous viscosity at minimum shear rate, determined by the slope of the tangent line on a Lissajous at the origin of stress versus strain rate (Pa s)

H.S. Joyner, A. Meldrum / Journal of Food Engineering 169 (2016) 326e337

References Bird, R.B., Armstrong, R.C., Hassager, O., Curtiss, C.F., 1977. Dynamics of Polymeric Liquids. Wiley, New York. Çakir, E., Koc, H., Vinyard, C.J., Essick, G., Daubert, C.R., Drake, M., Foegeding, E.A., 2012. Evaluation of texture changes due to compositional differences using oral processing. J. Texture Stud. 43, 257e267. Çakır, E., Daubert, C.R., Drake, M.A., Vinyard, C.J., Essick, G., Foegeding, E.A., 2012a. The effect of microstructure on the sensory perception and textural characteristics of whey protein/k-carrageenan mixed gels. Food Hydrocoll. 26, 33e43. Çakır, E., Vinyard, C.J., Essick, G., Daubert, C.R., Drake, M., Foegeding, E.A., 2012b. Interrelations among physical characteristics, sensory perception and oral processing of protein-based soft-solid structures. Food Hydrocoll. 29, 234e245. Ewoldt, R.H., McKinley, G.H., 2010. On secondary loops in LAOS via self-intersection of LissajouseBowditch curves. Rheol. Acta 49, 213e219. Ewoldt, R. H., Winter, P. and McKinley, G. H. MITlaos Version 2.1 Beta for MATLAB. MATLAB-based data analysis software for characterizing nonlinear viscoelastic responses to oscillatory shear strain. (self-published, Cambridge., MA). Ewoldt, R.H., Clasen, C., Hosoi, A., McKinley, G.H., 2007. Rheological fingerprinting of gastropod pedal mucus and synthetic complex fluids for biomimicking adhesive locomotion. Soft Matter 3, 634e643. Ewoldt, R.H., Hosoi, A., McKinley, G.H., 2008. New measures for characterizing nonlinear viscoelasticity in large amplitude oscillatory shear. J. Rheology (1978ePresent) 52, 1427e1458. Faulks, R.M., Agricultural Research Council, F. R. I., Colney Lane, Norwich, Griffiths, N. M. and Agricultural Research Council, F. R. I., Colney Lane, Norwich, 2014. Influence of variety, site and storage on physical, sensory and compositional aspects of mashed potato. J. Sci. Food Agric. 34, 979e986. Janhoj, T., Frosta, M.B., Prinzb, J., Ipsena, R., 2009. Sensory and instrumental characterization of low-fat and non-fat cream cheese. Int. J. Food Prop. 12, 211e227. Kaur, L., Singh, N., Sodhi, N.S., Gujral, H.S., 2002. Some properties of potatoes and their starches I. Cooking, textural and rheological properties of potatoes. Food Chem. 79, 177e181. Kokini, J.L., 1987. The physical basis of liquid food texture and texture-taste interactions. J. Food Eng. 6, 51e81.

337

Linehan, D., Hughes, J., 1969. Texture of cooked potato I.dIntroduction. J. Sci. Food Agric. 20, 110e112. Martens, H.J., Thybo, A.K., 2000. An integrated microstructural, sensory and instrumental approach to describe potato texture. LWT-Food Sci. Technol. 33, 471e482. Melito, H.S., Daubert, C.R., Foegeding, E.A., 2012a. Creep and large amplitude oscillatory shear behavior of whey protein isolate/.-carrageenan gels. Appl. Rheol. 22, 63691. Melito, H.S., Daubert, C.R., Foegeding, E.A., 2012b. Validation of a large amplitude oscillatory shear protocol. J. Food Eng. 113, 124e135. Melito, H., Daubert, C., Foegeding, E., 2013a. Relationships between nonlinear viscoelastic behavior and rheological, sensory and oral processing behavior of commercial cheese. J. Texture Stud. 44, 253e288. Melito, H.S., Daubert, C.R., Foegeding, E.A., 2013b. Relationships between nonlinear viscoelastic behavior and rheological, sensory and oral processing behavior of commercial cheese. J. Texture Stud. 44, 253e288. Ng, T.S., McKinley, G.H., Ewoldt, R.H., 2011. Large amplitude oscillatory shear flow of gluten dough: a model power-law gel. J. Rheol. (1978ePresent) 55, 627e654. Perera, C., Hoover, R., Martin, A., 1997. The effect of hydroxypropylation on the structure and physicochemical properties of native, defatted and heat-moisture treated potato starches. Food Res. Int. 30, 235e247. Richardson, R.K., Morris, E.R., Ross-Murphy, S.B., Taylor, L.J., Dea, I.C.M., 1989. Characterization of the perceived texture of thickened systems by dynamic viscosity measurements. Food Hydrocoll. 3, 175e191. Sonne, A., Busch-Stockfisch, M., Weiss, J., Hinrichs, J., 2014. Improved mapping of inmouth creaminess of semi-solid dairy products by combining rheology, particle size, and tribology data. Food Sci. Technol. 59, 342e347. van Marle, J.T., Stolle-Smits, T., Donkers, J., van Dijk, C., Voragen, A.G., Recourt, K., 1997. Chemical and microscopic characterization of potato (Solanum tuberosum L.) cell walls during cooking. J. Agric. Food Chem. 45, 50e58. Wehr, H., Frank, J., 2004. Standard Methods for the Examination of Dairy Products. American Public Health Association, Washington, DC. Winter, H.H., 1997. Analysis of dynamic mechanical data: inversion into a relaxation time spectrum and consistency check. J. Newt. Fluid Mech. 68, 225e239.