Annals of Botany 82 : 727–733, 1998 Article No. bo980732
In situ Mechanical Testing of Fully Hydrated Carrots (Daucus carota) in the Environmental SEM B. L. T H I EL* and A. M. D O N A L D Polymers and Colloids Group, Caendish Laboratory, Department of Physics, Madingley Road, Uniersity of Cambridge, Cambridge CB3 0HE, UK Received : 11 March 1998
Accepted : 16 July 1998
We demonstrate the use of the Environmental Scanning Electron Microscope (ESEM) for in situ observation of mechanical tests on carrot (Daucus carota) parenchymal tissue. The ESEM tolerates several Torr of water vapour in the specimen chamber, thus allowing fully hydrated specimens to be examined at high resolution, but without preparation. Three tests were performed, involving slicing, tension and compression. The manner in which stress was distributed in the tissue, crack propagation and cell wall distortion were all observed in real time. # 1998 Annals of Botany Company Key words : Environmental scanning electron microscopy, carrot, Daucus carota, mechanical testing.
INTRODUCTION The relationship between the mechanical properties and cellular structure of food plants is of great economic interest. Producers would like to reduce losses due to cracking or breakage and food product manufacturers seek to minimize the amount of effort required in processing. Often the mechanical requirements for each are in opposition. Ultimately, the products, both raw and processed, must have textures that are pleasing to the consumer. Consequently, the role of cellular structure in determining mechanical properties of plant tissues has generated much research interest for several years (Van-Buren, 1979 ; Gibson and Ashby, 1988 ; Vincent, 1990 a, b). Progress has been limited by the ability to image structural details at high resolution with minimal specimen preparation artifacts. McGarry (1995) performed a series of elegant studies relating the tensile properties, water content and cell structure in carrots. Structure was imaged by scanning electron microscopy, for which carrots were prepared by freezing in liquid nitrogen and coated with gold. While freezing and coating specimens is the accepted conventional means of preparation, it suffers from two limitations. First, one can never be entirely confident that preparation artefacts have not been introduced. Second, it is impossible to perform in situ dynamic experiments on fresh tissues. The development of controlled atmosphere scanning electron microscopes (SEMs) surmounts these problems. In particular, the Environmental Scanning Electron Microscope (ESEM ; FEI-Philips-Electroscan) combines a differential pumping system with a gaseous secondary electron detector to produce high resolution images of fully hydrated materials. A schematic of the instrument is given in Fig. 1, * For correspondence. Fax 44 (0)1223 337000, e-mail : bt202! cam.ac.uk
0305-7364\98\120727j07 $30.00\0
in which it can be seen that a series of pressure limiting apertures effectively isolates portions of the column, each of which can be pumped individually. Most of the beam path, and in particular the electron gun, can therefore be held at vacuum levels between 10−& and 10−( Torr (comparable to a conventional SEM), while the specimen chamber can sustain pressure levels of up to 50 Torr. If water vapour is held in the chamber, the saturated vapour pressure curve (Fig. 2) indicates that 100 % relative humidity can be achieved with moderate pressures and temperatures. For example, liquid water can be stabilized, and specimens can be examined without fear of dehydration, at 3 mC and 5 Torr. Scattering of the probe electrons in the comparatively high pressure of the specimen chamber is limited by using very small working distances, typically 7–10 mm. At a water vapour pressure of 5 Torr, the mean-free-path for elastic (i.e. high angle) scattering events is approx. 15 mm for 20 keV electrons. Accordingly, a high percentage of the electrons in the probe reach the specimen without being scattered. Those that are scattered are distributed over a broad angular range. The resulting distribution of electrons at the sample surface is that of a sharp (unbroadened) probe surrounded by a very broad and flat skirt. The probe is sufficiently sharp, and the background sufficiently flat, to produce an acceptable signal-to-background ratio, resulting in high resolution images (Danilatos, 1988). The final component necessary in this configuration is a detector capable of collecting secondary electron signals in the gaseous environment. Unlike the primary electrons, the low energy secondary electrons have a mean-free-path of the order of microns in the gas. Thus, conventional Everhart– Thornley detectors employing a Faraday cage are ineffective. The ESEM uses a Gaseous Secondary Electron Detector (GSED) which places a positively charged annular electrode directly below the pole piece of the objective lens (i.e. directly above the specimen). An electric field of the order of # 1998 Annals of Botany Company
728
Thiel and Donald—Mechanical Testing of Hydrated Carrots in the ESEM Gun
10–6 torr Rotary pump Rotary pump
Rotary pump
Diffusion pump Diffusion pump
pla 2
10–5 torr
pla 1
10–2 torr
Gas inlet Chamber
10 torr
Vent Increasing pressure
F. 1. Differential pumping system. The environmental scanning electron microscope (ESEM) uses a differentially pumped electron column. The gun and upper column are held at 10−' Torr by diffusion pumps. Pressure limiting apertures isolate subsequent portions of the column that are pumped by a combination of diffusion and mechanical pumps.
Vapour pressure (torr)
18 16 14
Condensing
12 100% relative humidity
10 8 6
Dehydrating
4 2 0
5
10 Temperature (°C)
15
20
F. 2. Saturated vapour pressure curve for water. Specimen temperature and water vapour pressure in the chamber can be set to control relative humidity (RH) at the sample surface. RH values less than 100 % result in increasingly severe dehydrating conditions, while temperature and pressure settings above the 100 % RH line cause condensation of water onto the sample.
100 V mm−" is created between the sample and the electrode. Secondary electrons leaving the specimen are accelerated by the field until their kinetic energy exceeds the ionization threshold of the gas (12n6 eV in the case of water vapour). Once this occurs, ionizing collisions can liberate more electrons, effectively amplifying the signal. This cascade current is collected by the electrode and can be used to form an image. Amplifying characteristics of this gaseous detection system have been discussed elsewhere (Fletcher, Thiel and Donald, 1997 ; Thiel et al., 1997). An additional benefit of the cascade amplification is that positive ions are also created by the ionization events. These are directed towards the sample by the electric field, where they serve to neutralize accumulated negative charge. Thus the need for a conductive coating on the specimen is obviated. The ESEM enables appropriate in situ mechanical tests to be performed on fully hydrated specimens. Materials exhibit three types of mechanical response to external stresses : elastic (recoverable) deformation ; visco-elastic (time-de-
pendent, partially-recoverable deformation ; and plastic (non-recoverable) deformation. In a cellular material, these are manifest by cell-wall bending (Gibson and Ashby, 1988 ; Vincent, 1990 b), flow of fluid between cells (Gibson and Ashby, 1988 ; Krokida, Maroulis and Marinos Kouris, 1998) and crack propagation (Gibson and Ashby, 1988 ; Vincent, 1990 b) respectively. It should be noted, however, that the cell wall material itself can exhibit each of these mechanical responses, under the right circumstances. As with all scanning electron microscopes, the ESEM offers potentially nanometre scale resolution and a very large depth of field. It is therefore possible to assess visually the manner in which stresses are accommodated and cracks propagate through the material. The ability to extract information is limited by the image quality, which, in turn, is governed by signal intensity. Although organic materials are sensitive to high intensity electron beams, conditions can be found that minimize radiation damage (Kitching and Donald, 1998). Our aim was to demonstrate the utility of this technique by describing the methodology for three types of mechanical measurements on carrot parenchymal tissue. Mechanical tests were performed with the aid of a custom-built stage for the ESEM, as described later. The effects of ageing are most readily observable in a slicing test wherein a scalpel blade is driven through the material within the field of view. This method is akin to the wedge tests described by Vincent (1991), wherein a wedge with a known angle is driven through tissue. The force required to move the wedge through the material is recorded with a transducer. In our experiment, however, distortion of the tissue and rupture of the cells at the leading edge of the blade can be observed directly in real time. Although our results are purely qualitative, they do provide a great deal of insight into how stresses are distributed in the material, and how this distribution changes as a function of ageing and water content. To obtain more quantitative information, one of the most commonly used mechanical measurements is the tensile, or load s. extension test. Placing the material in tension initially causes elastic stretching, and ultimately leads to failure by the opening of cracks. These cracks nucleate at structural flaws in the material at the mesoscopic (i.e. cellular) scale. Accordingly, several useful parameters can be obtained, including elastic modulus, yield stress, ultimate tensile strength, extensibility, fracture strength and fracture toughness. Tensile tests on carrots have been published by Verlinden et al. (1996) who studied these parameters in detail, but did not correlate them with microstructural observations. Finally, the converse experiment can also be performed wherein the specimen is placed under compression. The information obtained from these measurements will be slightly different from the tensile tests, as material failure here is not governed by flaws at the mesoscopic scale, but rather the bursting strength of cell walls. It should be noted that our intent here was not to obtain highly accurate, quantitative data on mechanical properties ; our conditions for testing were not sufficiently well controlled to allow such measurements. The load-extension curves obtained were semi-quantitative at best, and most
Thiel and Donald—Mechanical Testing of Hydrated Carrots in the ESEM
MATERIALS AND METHODS Experiments were performed in an ElectroScan Model 2010 Environmental SEM equipped with a tungsten filament. Water vapour at 5 Torr was used for the imaging gas. An accelerating voltage of 15 kV was used to provide the best compromise between image quality and radiation damage. Geometric constraints of the stage limited the working distance to no less than 8n5 mm. Our mechanical stage was custom-built by Oxford Instruments. It was equipped with a 25 kg-f load cell and interchangeable specimen grips. The specimen rested on a water-backed, Peltier-cooled, chilled block that held the specimen to a temperature of 3 mC, allowing it to remain hydrated. We did not attempt statistically significant systematic tests, but rather explored a range of experimental conditions. Results from ‘ typical ’ runs are given to illustrate our points, with the exact experimental parameters noted. Where two tests are compared directly, identical experimental conditions were used. A strain rate of 10 µm s−" was used for the tensile tests, while compression experiments were performed at 5 µm s−". For the slicing experiments, the blade was manually driven, so a deformation rate was not applicable. There is some degree of experimental error in all of the calculated stress and strain values. Practical considerations for supporting, gripping and cooling the specimen led to some uncertainty in determining the gauge volume in which the distortion is assumed to take place. ‘ Fresh ’ carrots (Daucus carota) were obtained from J. Sainsbury’s plc and either used immediately or stored in a plastic bag in a refrigerator. Longitudinal strips of phloem parenchyma were made using a kitchen vegetable slicer (VIC No. 2 ", Victory Engineering, St. Louis, MO, USA). For # tensile tests, a stainless steel die was used to cut ‘ dog bone ’ shapes from the strips as illustrated in Fig. 3. Width and thickness of the gauge section were 4 mm while the gauge length was 10 mm. All tensile specimens were notched to a depth of 1 mm in order to localize the point of failure. Specimens for other tests were 4 mm cubes cut from the strips using a scalpel. Test pieces were temporarily placed in a small dish of distilled water containing chopped carrot shavings to avoid dehydration prior to examination.
B
10 mm
A
25 mm
useful for obtaining relative values and observing trends. Rather, our aim was to gain insight into the manner in which the material responds to external stresses, observe failure mechanisms, and identify critical features within the microstructure that are most likely to be points where failure is initiated or propagated. It is a basic tenet of materials science that many of the properties (including mechanical behaviours such as fracture strength), of a material are limited by its flaws, and not by the ideal, bulk, homogenous properties (Davidge, 1979). We must point out, though, that we were measuring the properties of specific tissues. Information on these is critical for those attempting to model the mechanical responses of such systems. Plant organs taken as a whole will exhibit much more complex composite mechanical responses, which depend not only on the properties of the individual tissues, but on their distribution and connectivity as well.
729
4 mm
12 mm
F. 3. A, Tensile specimen geometry. A stainless steel die is used to cut tensile specimens from slices of carrot. Tapered ends of specimens wedge into grips (B), during extension testing.
RESULTS AND DISCUSSION Slicing Although slicing was the most qualitative test, it also provided the most dramatic visual results, and helped to convey a more tangible sense of the material response. The blade acted as a point source for a compression field radiating outward from the crack tip. Figure 4 shows a blade being driven through a ‘ fresh ’ carrot, and carrots that have been stored for 1 or 3 weeks. The most readily apparent change is in the size of the stress distribution field. In the ‘ fresh ’ tissue, turgor pressure is high, so cells cannot readily distort to accommodate stresses. This distortion is a conversion of strain energy into mechanical work. Without that avenue for dissipation, the strain energy remains concentrated at the crack tip resulting in brittle, catastrophic failure of the cell wall. A drop in turgor pressure caused by water loss is one of the first manifestations of ageing, mimicked by refrigerator storage in this study (Alberts et al., 1989). The added flexibility of the cells allows the strain energy to be distributed over a larger volume. In many materials, the stress field induced by compression propagates by buckling, with the wavelength of distortion being a function of the elastic moduli. In Fig. 4 B the wavelength is of the order of the cell size, indicating that stresses are still being accommodated at that structural level. During the final stages of ageing, enzymes begin to break down the supramolecular structure of the cell wall material itself (Alberts et al., 1989). The cell walls have degraded to the point where there is negligible resistance to the flow of fluid, and the open cellular structure collapses readily upon the application of force. The load is then borne entirely by the compressed cell wall material. This is evidenced by the much higher buckling frequency on the scale of the cell wall thickness, as seen in Fig. 4 C. By this stage, the deformation zone encompasses a large number of cells. The degradation of the cell wall
730
Thiel and Donald—Mechanical Testing of Hydrated Carrots in the ESEM A
B
C
F. 4. A scalpel blade is driven through carrots that were ‘ fresh ’ (A) or stored for 1 (B) or 3 (C) weeks. The blade is moving top to bottom. The liquid from the ruptured cells is forming a droplet at the blade in C.
600 Fresh
500 Stress (kPa)
Aged
400 300 200 100 0
0
0.1
0.2
0.3 Strain
0.4
0.5
0.6
F. 5. Experimental stress-strain plots for carrot stored for 1 week or ‘ fresh ’ carrot strained to failure. The elastic modulus for stored carrot is 1n03 MPa, and the ultimate tensile strength is 507 kPa. The ‘ fresh ’ carrot exhibits a higher modulus of 1n36 MPa, but retains roughly the same tensile strength of 504 kPa.
material is further illustrated by the ease with which it is shredded by the scalpel tip. Tensile testing A typical plot of the stress-strain behaviour for a carrot stored for 1 week is shown in Fig. 5. The three dips in the linear-elastic portion of the curve are likely to be artefacts due to the specimen slipping in the grips, rather than local failure events. In this example, the elastic modulus is 1n03 MPa and the ultimate tensile strength (peak load sustained by the specimen) is 507 kPa, both of which are consistent with values earlier reported (Verlinden et al., 1996). Figure 6 shows the carrot at various stages during the collection of stress-strain data. A lengthy and essentially linear elastic regime appears to be typical for these materials, and is accompanied by simple stretching of the cell walls. Figure 6 B shows the distortion of cells at the crack tip early in the elastic regime, while Fig. 6 C was taken just prior to catastrophic failure, near the ultimate tensile strength. Note that as the material at the head of the crack tip becomes distended, the crack tip is effectively blunted. Once the ultimate tensile strength is surpassed (i.e. at the peak of the
stress-strain curve) however, irreversible damage begins to take place. Cells delaminate and tissues tear, allowing the crack to propagate through the material. The applied force required to continue driving the crack forward decreases rapidly as the stress is concentrated in the remaining material. Even during failure (Fig. 6 D and E) the propagating crack is temporarily bridged by more ductile material, demonstrating significant local plasticity of the cell wall material. For comparison, the stress-strain curve for a ‘ fresh ’ carrot is also given in Fig. 5. The ultimate tensile strength of this specimen was 504 kPa, nearly identical to that of the carrot stored for a week. This indicates that the rupture strength of the cells has remained constant and that minimal degradation of the cell wall material and inter-cellular lamella has taken place. However, there are three significant differences between the two cases : (1) extensibility is notably increased, from approx. 45 % in ‘ fresh ’ carrot to 57 % for carrot stored for 1 week ; (2) the elastic modulus decreased from 1n36 MPa in ‘ fresh ’ carrot to 1n03 MPa in carrot stored for 1 week ; (3) ‘ fresh ’ carrot shows much less plastic deformation, as indicated by the sharpness of the failure regime immediately before and after the peak load. Systematic tests to determine the accuracy of these values were not performed ; however, all three observations are consistent with a loss of turgor pressure in the cells as the carrot ages during storage. Future experiments will be aimed at performing a more quantitative analysis of these effects, and observing the plastic deformation of the microstructure near the peak load. It is often instructive to consider mathematical models for material behaviour, and to try and identify connections between microstructural features and terms in the equations. This process can be an aid in interpreting experimental results, as well as designing future experiments. As a starting point for analysis, we used the Grifith–Orowan–Inglis treatment for crack propagation leading to failure in brittle and partially ductile engineering materials (see, for example, Davidge, 1979). The criterion for crack advance is taken to be when the elastic strain energy released per unit length equals the energy required to form two new surfaces, each of unit length. For an elliptical crack of length 2c in a brittle,
Thiel and Donald—Mechanical Testing of Hydrated Carrots in the ESEM
731
B
A
C
D
E
F. 6. Opening of a crack tip during tensile loading. A, Notched region before loading ; B, elastic deformation of cells at approx. 30 % strain ; C, crack tip immediately prior to failure ; D, ductile bridging of crack during propagation ; E, immediately after failure.
thin plate of modulus E, it can be shown that the stress required for the crack tip to advance is approximately σf l
9 : Eγ c
" #
(1)
where γ is the surface energy. In the present case of a cellular vegetable, γ represents either the breaking of cell walls, or intercellular delamination. The dominant process
will be the one with the lower value for γ. As illustrated in Fig. 6, performing the tests in situ allows the crack path to be followed, and shows the relevant failure mechanism. This concept can be extended to include partially ductile materials if an additional term is introduced for the plastic or viscous work done, γp. Note that γp is not a true surface energy that comes from breaking bonds as in γ, but rather represents the additional irreversible work that accompanies extension of the crack. Because it is still an energy required
732
Thiel and Donald—Mechanical Testing of Hydrated Carrots in the ESEM Compressive stress (kPa)
0 –100 –200 –300 –400 –500 –600
0
200
400
600
800 1000 1200 1400 1600 1800 2000
Step number F. 8. Cyclic compression data. Negative slopes indicate the specimen is being compressed, while positive slopes indicate the grips are being opened. The step size along the ordinate corresponds to 4 µm.
F. 7. A crack that has clearly propagated through the walls of several adjacent cells. No evidence of cellular delamination can be seen.
per unit crack length, γp is dimensionally identical to γ. The new form of eqn (1) becomes σf l
) 9E(γjγ : c
"
p #
(2)
Distortion of the cell shape, and diffusion of liquid cell contents is reflected in γp, and will therefore change with ageing and turgor pressure. This is also the point at which strain rate becomes relevant, through the ability of the material to redistribute stresses by the flow of liquid. While eqn (2) gives an indication of the failure criterion for stress at the crack tip, it is also necessary to consider how the macroscopic stress, σ, is concentrated at the crack tip. The contribution of Inglis to the model was to show that the maximum stress at the crack tip σm, is inversely related to the crack tip radius, ρ, by σm l 2σ
0ρc1
" #
(3)
Crack propagation occurs when σm l σf. This implies that when a material can respond to an externally applied stress by plastically deforming near the crack tip, the local stress will be less concentrated. In this case, the material will be less likely to fail by brittle crack propagation. In Fig. 6 A–C, the cells at the crack tip are able to distort, effectively increasing the crack tip radius. Conversely, in Fig. 4, the ESEM provides visual confirmation that in the stored carrots the strain field at the scalpel tip is much larger, which actually serves to reduce the stress concentration at the crack tip. Compression Figure 7 shows a cell wall of a ‘ fresh ’ carrot in the process of bursting under a compressive load. The crack has
initiated within the wall, presumably at a weak point in the cellulose network, and propagates rapidly. Equations (1)–(3) still govern this crack propagation, but now the material constants are those for just the cell wall material, rather than for the entire assembly. Turgor pressure affects the distribution of stresses in the cell, making the cell behave in an analogous fashion to a pressure vessel. If the specimen is not stressed to failure, but is compressively cycled, then time dependent visco-elastic responses can be studied. Recoverable cell deformation and elastic buckling modes of cell walls can be observed. Figure 8 shows the material response to two compression– relaxation cycles, followed by compression to failure. The ordinate has been left as ‘ Step ’ number, where each step corresponds to a displacement of p4 µm (2n5 steps sec−"). The sign of the displacement is given by the slope of the curve ; positive for compression, negative for relaxation. Images shown in Fig. 9 were collected at the start of the experiment, at each of the load maxima, and at failure. Some damage is present at the second maximum. The similarity of the first two cycles is consistent with the deformation being entirely elastic, and therefore recoverable. The vertical gaps occurring between the cycles represent relaxation that took place during the 60 sec necessary to acquire a high resolution image on the microscope. This relaxation of the material is evidence of a time-dependent, or visco-elastic response. In future experiments this effect could be exploited to investigate viscoelastic properties and possibly relate them to the diffusivity of fluid through cell walls.
CONCLUSIONS The results demonstrate the versatility of in situ mechanical testing in the ESEM. Having established the methodology, it is now appropriate to perform more extensive studies on the effects of storage. A more systematic study could attempt to manipulate the values of the material parameters used in eqns (1)–(3). Specifically, turgor pressure could be artificially altered to change the elastic modulus and the ability of the cells to deform. Cooking can also be used to
Thiel and Donald—Mechanical Testing of Hydrated Carrots in the ESEM A
B
C
D
733
F. 9. Micrographs showing compression cycling of carrot : before loading (A) ; at the two points of maximum loading indicated in Fig. 8 (B and C) ; after compressive failure (D).
weaken cell walls and cause deterioration of the intercellular pectic solids. Conversely, chelating agents could be used to increase stability of the pectic solids, all of which can affect the brittle and plastic surface energies. A C K N O W L E D G E M E N TS The authors thank the BBSRC for funding this work, and Drs Mike Gidley and Katherine Burrows of Unilever for valuable discussions. LITERATURE CITED Alberts B, Bray D, Lewis J, Raff M, Roberts K, Watson D. 1989. Molecular biology of the cell. New York : Garland. Danilatos GD. 1988. Foundations of environmental scanning electron microscopy. Adances in Electronics and Electron Physics 71 : 109–250. Davidge RW. 1979. Mechanical behaiour of ceramics. Cambridge : Cambridge University Press, Cambridge Solid State Science Series. Fletcher AL, Thiel BL, Donald AM. 1997. Amplification measurements of potential imaging gases in environmental SEM. Journal of Physics D : Applied Physics 30 : 2249–2257.
Gibson LJ, Ashby MF. 1988. Cellular solids : structure and properties. Oxford : Pergamon Press. Kitching S, Donald AM. 1998. Beam damage of polypropylene in the environmental scanning electron microscope. Journal of Microscopy 30 : 357–365. Krokida MK, Maroulis ZB, Marinos Kouris D. 1998. Viscoelastic behavior of dehydrated carrot and potato. Drying Technology 16 : 687–703. McGarry A. 1995. Cellular basis of tissue toughness in carrot (Daucus carota L.) storage roots. Annals of Botany 75 : 157–163. Thiel BL, Bache IC, Fletcher AL, Meredith P, Donald AM. 1997. An improved model for gas amplification in the environmental SEM. Journal of Microscopy 187 : 143–157. Van-Buren JP. 1979. The chemistry of texture in fruits and vegetables. Journal of Texture Studies 10 : 1–23. Verlinden BE, De Barsy T, De Baerdemaeker J, Deltour R. 1996. Modeling the mechanical and histological properties of carrot tissue during cooking in relation to texture and cell wall changes. Journal of Texture Studies 27 : 15–28. Vincent JFV. 1990 a. Fracture properties in plants. Adances in Botanical Research 17 : 235–287. Vincent JFV. 1990 b. Structural biomaterials. Princeton : Princeton University Press. Vincent JFV. 1991. The wedge fracture test : a new method for measurement of food texture. Journal of Texture Studies 22 : 45–57.
