Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

rstb.royalsocietypublishing.org

Research Cite this article: Filippov AE, Popov VL, Gorb SN. 2015 The functional significance of density and distribution of outgrowths on co-opted contact pairs in biological arresting systems. Phil. Trans. R. Soc. B 370: 20140032. http://dx.doi.org/10.1098/rstb.2014.0032 One contribution of 19 to a discussion meeting issue ‘Cell adhesion century: culture breakthrough’. Subject Areas: biomechanics, theoretical biology, computational biology Keywords: attachment, adhesion, arresting system, structure, biomechanics, numerical simulation Author for correspondence: Stanislav N. Gorb e-mail: [email protected]

Electronic supplementary material is available at http://dx.doi.org/10.1098/rstb.2014.0032 or via http://rstb.royalsocietypublishing.org.

The functional significance of density and distribution of outgrowths on co-opted contact pairs in biological arresting systems Alexander E. Filippov1,2, Valentin L. Popov3 and Stanislav N. Gorb1 1 Department of Functional Morphology and Biomechanics, Zoological Institute of the University of Kiel, Am Botanischen Garten 1–9, 24098 Kiel, Germany 2 Donetsk Institute for Physics and Engineering, National Academy of Sciences, 83114 Donetsk, Ukraine 3 FG Systemdynamik und Reibungsphysik, Technische Universita¨t Berlin, Institut fu¨r Mechanik, Sekr. C8-4, Raum M 122, Straße des 17. Juni 135, 10623 Berlin, Germany

Microstructures responsible for temporary arresting of contacting surfaces are widely distributed on surfaces in different organisms. Recent morphological studies show that these structures have different density of outgrowths and not ideal distribution pattern on both complementary parts of the contact. One can suggest that this difference is optimized by natural selection to get stronger mechanical arrest within the system. In this paper, we simulate such a system numerically, both in the frames of continuous contact and discrete dynamical models to prove this hypothesis and elucidate other aspects of optimization of such mechanical adhesive systems.

1. Introduction Specialized pairs of microstructure-covered fields of arthropod cuticle often function for mechanical adhesion (attachment) between different body parts within an organism [1–5]. These systems are often called probabilistic fasteners due to the fact that they are composed of cuticular micro-outgrowths of different origin [6] and because the size, shape and density of the outgrowths on both surfaces do not correspond exactly to each other. Thus, mechanical interactions between the micro-outgrowths of both surfaces take place without precise positioning of both surfaces [1]. Probabilistic fasteners prevent sliding of surfaces, when in contact. Mechanical adhesion (attachment), in this case, must depend on the size, shape, distribution density and the material properties of outgrowths (called elements in this paper), and is fast, precise and reversible. Individual elements are not necessarily hook-shaped (figure 1). The mechanism of attachment in such systems is definitely different from the typical hook-and-loop Velcro principle, but has been only marginally studied for the case of probabilistic fasteners with parabolic contact elements [5]. Because this kind of attachment structure occurs most widely among insects: in the unguitractor plate [8], intersegmental fixators of leg joints [7], synchronizing mechanisms of contralateral legs in cicada [9], dragonfly head-arresting system [10], and especially in wing-attachment devices [2,3,11,12], it is of the importance for the functioning of numerous biomechanical systems of insects. The major similarity of these devices is that co-opted fields of cuticular outgrowths are present on two separate parts of the body. Sometimes, as in the case of the forewing fixators of beetles, more than just one pair (up to eight pairs) of co-opted structures are involved in the proper functioning of the mechanical system. In such cases, each pair favours movement of the closed wings in one preferred direction. The head arrester of adult dragonflies immobilizes the head in some critical moments when it may experience strong mechanical disturbances [10]. Different shapes of outgrowths in functionally corresponding fields have been previously described [3]: (i) hook-like or cone-shaped or parabolic elements (outgrowths) and mushroom-like elements, (ii) cone-shaped or parabolic elements on both surfaces,

& 2014 The Author(s) Published by the Royal Society. All rights reserved.

Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

(a)

2

(c)

SPC

rstb.royalsocietypublishing.org

d a

TS

CS

15 µm

250 µm (d)

(b) d m MT TS

120 µm

20 µm

Figure 1. An example of the head-arresting system in the dragonfly Aeshna mixta (Anisoptera, Aeshnidae) [7]. Post-cervical sclerite of the neck (a,c) and the microtrichia field of the rear surface of the head (b,d) show both corresponding fields of cuticle outgrowths involved in the functioning of the arrester system. a, anterior direction; d, dorsal direction; m, medial direction; CS, campaniform sensilla; MT, microtrichia; SEC, eucervical sclerite; SPC, post-cervical sclerite; TS, trichoid sensilla.

(iii) clavate elements on both surfaces and (iv) plate-like elements on both surfaces [5].

2. Some structural properties observed in biological systems As mentioned in §1, surface outgrowths of probabilistic fasteners vary in different biological systems and even between corresponding fields within the same system. However, in the pair of corresponding surfaces, one could expect some constraints in size and distribution of co-opted areas of outgrowths. In the dragonfly head arrester, the size of the outgrowths is usually larger in large species, and their density is lower, but the density of outgrowths on co-opted fields of one species is surprisingly never the same (figure 1). Similar observation has been made for the co-opted microtrichia fields of the forewing fixators of the beetles [2]. In the example of the dragonfly arrester, the density is higher on the neck sclerite than on the rear surface of the head [7]. The difference in the microtrichia density on the corresponding structures is more distinct in the larger species [7]. An increase in size of single microtrichia in larger species may be explained by possible correlation with an increasing cell size in

larger species. It is more difficult to explain the difference in the outgrowth density on the corresponding structures. In addition, the distribution of the outgrowths shows a roughly hexagonal pattern, which is generally accepted to represent the highest package density of structures, but this pattern is far from an ideal one. It remains unknown whether these features have any functional significance on the function of co-opted fields. Understanding the role of dimensions and distribution of outgrowths of co-opted fields is crucial for understanding the mechanical behaviour of such systems in contact. To the best of our knowledge, there is only the model of frictional behaviour of two single parabolic pins in sliding contact [5], but there are no models predicting the behaviour of arrays of outgrowths in such systems. This study was undertaken to establish simple models that realistically describe behaviour of co-opted fields of probabilistic fasteners in contact. Especially, we were interested in understanding the role of distribution of outgrowths on corresponding fields. To the best of our knowledge, the main question raised in this paper has not previously been discussed in the literature: how does the difference in densities and in the type of distribution of single elements influence the contact forces generated by the system of co-opted fields?

Phil. Trans. R. Soc. B 370: 20140032

SEC

Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

3

rstb.royalsocietypublishing.org

3 z

3

(b)

(a)

2

2 z 1

1

0 0

0 0

y

5 2

6

8

y

5

x

10 0

2

4

6 x

8

Figure 2. Two typical examples (a,b) of ordered and partially randomized contacting surfaces with different periods of asperities. Dark colour of the greyscale map corresponds to maxima and white colour to minima of the down surface. Reverse greyscale map is used for the upper surface.

3. Continuous contact model First, we analysed contact properties of a continuous numerical model. The conceptual structure of the model is shown in figure 2. We constructed numerically two flexible contacting surfaces with regular, or partially randomized, arrays of the outgrowths having different lengths, periods of distribution, symmetry and widths of the asperities on both upper and lower surfaces. For a definiteness and numerical simplicity, we simulate the outgrowths by the arrays of Gaussians ( ) [(x  xk )2 þ (y  yj )2 ] kj G (x, y; {xk , yj }) ¼ Gkj exp  , (3:1) D2kj k ¼ 1, 2, . . . , Nx; j ¼ 1, 2, . . . , Ny with numerically regulated height Gk,j, width D2kj and individual positions of their centres fxk, yjg. By manipulating these parameters and by varying distances between outgrowths of upper and lower surfaces, we were able to simulate different combinations of forms of outgrowths on both contacting surfaces. It should be noted that the majority of the structures observed in biological systems possess more or less randomized hexagonal ordering [5,10,12]. Such a type of structure is well expected intuitively, as a self-organized densely packed array of relatively rigid outgrowths repelling one another by their cores. However, in many specific cases in real biological systems, these structures have a lower degree of symmetry and rhombic or striped quasi-periodic ordering rather than the ideal hexagonal one. In any case, one can generally suggest that the real ordering as a result of natural selection is some sort of compromise between self-organized dense packing of outgrowths and the specific functional requirements of the particular system; for example, its arresting efficiency, generated friction forces in contact and probably other functional properties, which can demand a symmetry specific for the required function. This hypothesis has not been proven rigorously in an experiment before, in spite of some experimental studies showing that this type of structure is generally very promising for the purposes of biomimetics [5,13–16]. Unfortunately, it is rather difficult to approach this problem experimentally because of the difficulty of designing structures with a low grade of symmetry for comparison with structures with a high grade of the symmetry. This is why we decided to build and analyse a numerical model, which allows simulation of how particular ordering of the outgrowths satisfies different prescribed functional demands.

We started with analysis of the contact properties of two surfaces with an ideal hexagonal distribution of their outgrowths: Zu,d (x, y) ¼

Nx,Ny X

Gu,d,kj (x, y; {xu,d k , yu,d j , Gu,d kj , Du,d kj }),

(3:2)

k¼1,j¼1

where arrays of the coordinates fxk, yjg belong to the regular hexagonal lattice fx0k, y0jg with all constant Gaussian parameters for each of the two (upper and lower) surfaces: Gu ¼ G 0u ¼ const., Gd ¼ G 0d, Du ¼ D0u, Dd ¼ D0d. Despite the perfect symmetry of each surface, we allow different periods of outgrowths on upper and lower surfaces l 0u, l 0d. Other fixed parameters of the surfaces can also be different: l 0u = l 0d, G 0u = G 0d, D0u = D0d. This corresponds to our biological observations [2,7]. Below, after the study of the perfect symmetry case, we shall modify the surfaces by randomizing some of the mentioned parameters, or in the general case, all of them ( xu,d k ¼ xu,d 0k (1 þ zu,d x,k ) (3:3) yu,d j ¼ yu,d 0j (1 þ zu,d y,j ): Here, zx,k and zy,j are d-correlated random numbers, which are independent for every surface, position and coordinate fx,yg: kzu x,k l ¼ kzd x,k l ¼ kzu y,k l ¼ kzd y,k l ¼ 0 kzu x,k zd x,k l ¼ 0, kzu x,k zu y,k l ¼ 0 kzu x,k zu x,k0 l ¼ Du x dkk0 , {x ¼ y; u ¼ d}:

(3:4)

The intensity of the randomization can easily be regulated by choosing amplitudes Du,d x,y of the random noise z. It is obvious that in the limit Du,d x,y ! 0 corresponding surfaces degenerate to the perfectly ordered ones, as well as in the opposite limit Du,d x,y  1 the system becomes absolutely disordered. This freedom in choice of the constants Du,d x,y allowed us to model different combinations of weakly and strongly disordered surfaces as well as to study various contacts between completely ordered and differently randomized surfaces. Typical realizations of ordered and intermediately randomized surfaces are shown in figure 2 ((a, less randomized) and (b, more randomized), respectively). When the surfaces contact one another, they deform proportionally to their elasticity ku,d. If both are made of the same material, their elasticities coincide ( ku ¼ kd ; k) and they deform equally. In a numerical simulation, it means that, when converging surfaces Zu(x, y) and Zd(x, y) formally intersect

Phil. Trans. R. Soc. B 370: 20140032

10 0

4

Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

(a)

(b)

/max()

/max()

0.8 0.6 0.4 0.2 0.2 0.3 angle, f/p

0.4

0.8 0.6

m

1.0

0.4 0.2

0

0.1

0.2 0.3 angle, f/p

0.4

0.5 0

0.1

0.2 0.3 angle, f/p

0.4

Figure 3. Angle-dependent mean values of gradientkjrZjl, deviationkjdZjl and friction coefficient m ¼ kjrZjjdZjl=kjdZjl of regular hexagonal lattices shown for commensurate lu ¼ ld and incommensurate lu/ld ¼ t cases by thin and bold lines in the subplots (a), (b) and (c), respectively. one another, in the regions of intersection they form some equilibrium contact surface, which coincides with the mean value: Zcontact (x, y) ¼ (Zu (x, y) þ Zd (x, y))=2. The deformations of both surfaces cause local elastic forces, which in this case are equal to f elastic (x, y) ¼ 2k(Zu (x, y)  Zd (x, y))=2 ; kdZ(x, y). These forces correspond to the load acting on both contacting surfaces. Even at an unknown coefficient k, the load can be characterized by the deviation of the surfaces dZ(x,y) from their equilibrium positions. Nevertheless, from the observations of biological structures, one can estimate typical deformations dZ(x, y) for dragonfly species depicted in figure 1 in the interval 10–20 mm and corresponding elastic forces f elastic(x, y) ¼ kdZ(x, y) of about 10–20 mN at the typical spring constants of insect micro-outgrowths of 1 N m21 [17]. This estimation will be useful to validate the proposed model in further experimental studies. If both surfaces represent ideal hexagonal lattices, one can expect dZ(x, y) to depend on a relation between their periods lu,d and mutual orientation between the surfaces (angle w between their axes of symmetry). It is clear that the minima of dZ(x,y) should appear at some special orientations w, when asperities of upper and lower surfaces fall between one another. It is especially predictable when the periods lu,d are commensurate or simply equal (lu ¼ ld). The opposite situation is expected for the strongest incommensurate relation lu/ld ¼ t, where pffiffiffi t ¼ (1 þ 5)=2 ¼ 1:618::: is the so-called golden ratio. For example, in the dragonfly head arrester, this relationship ranges from 1.5 to 4.0. Thus, it is interesting to note that this relation lu/ld ¼ t between the periods of outgrowths on both surfaces is quite close to some previously found relations in biological systems [7]. It means that the hypothesis that some incommensurateness between the periods lu,d is biologically important seems to be very plausible. To prove it, we have performed calculation of angle-depending dZ(x, y; w) configurations for different lu,d. As expected, at some angles w, regular patterns of dZ(x, y; w) were observed for the commensurate cases (especially for lu ¼ ld). However, at the majority of arbitrary angles w, projections of both lattices do not coincide, and rather complex contact patterns appear. From a biological point of view, such an integral characteristic as the mean absolute value of dZ(x, y; w) averaged over spatial coordinates kjdZ(x, y; w)jljx,y ;kjdZjl should be important. We calculated it for different values of lu,d. The typical angle dependence of kjdZjl normalized to the maximum values for the two most important and representative cases lu ¼ ld and lu/ld ¼ t

are shown in figure 3a. It is seen directly from figure 3 that despite naive expectations, the curves in both commensurate and incommensurate cases have their own fine structure of minima corresponding to some special orientations w. According to the general definition [18], to estimate the local impact on friction between two rough surfaces, one also needs an absolute value of the inclination (or mathematical gradient) of the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi contact surface jrZ(x, y)j ¼ (@ Z(x, y)=@ x)2 þ (@ Z(x, y)=@ y)2 . Being integrated over the whole system at different mutual orientations w and normalized per unit area, this gives the angledependent mean gradient kjrZ(x, y; w)jljx,y ; kjrZjl. The results of the integration for two representative cases lu ¼ ld and lu/ld ¼ t are shown in figure 3b. Local impact to friction is proportional to the product of both values: local gradient jrZ(x, y)j and deviation jdZ(x, y)j determining local load, which is proportional to the elastic force f elastic(x, y) ¼ kdZ(x, y). The mean friction coefficient m is given by a sum of these products jrZ(x, y)jjdZ(x, y)j normalized to the total load



kkjrZ(x, y)jjdZ(x, y)jl , kjdZ(x, y)j

(3:5)

and at constant k, it does not depend on the elasticity: m ¼ kjrZjjdZjl=kjdZjl. As we found, for all relations lu/ld, this practically important combination is much less dependent on the orientation w than on solitary values of both kjrZjl and kjdZjl, and what is even more important, there is a big difference between commensurate and incommensurate cases. This is seen directly from figure 3c, where the thin line, representing the case lu ¼ ld, has a well pronounced deep minimum, but the bold curve corresponding to lu/ld ¼ t is practically flat (or has small maxima at some angles). As mentioned in §2, the distribution of outgrowths on biological surfaces is not perfectly hexagonal, but partially randomized. The same calculations as for the cases of regular distributions were repeated for the randomized hexagonal lattices xu,d k ¼ xu,d 0k (1 þ zu,d x,k ), yu,d j ¼ yu,d 0j (1 þ zu,d y,j ) with conserved equal amplitudes Du x,y ¼ Dd x,y of the random noises kzu x,k zu x,k0 l ¼ Du x dkk0 , ... fx ! y; u ! dg. It was found that even for quite intensive randomization Du,d x,y ¼ 0:15, the curves for jrZ(x, y)j and jdZ(x, y)j still have well pronounced minima. However, physically meaningful friction coefficients for both commensurate lu ¼ ld and incommensurate lu/ld ¼ t cases demonstrate relatively small

Phil. Trans. R. Soc. B 370: 20140032

0.1

4

1.5

rstb.royalsocietypublishing.org

1.0

0

(c) 1.0

Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

(a)

(b)

/max()

/max()

0.8 0.6 0.4 0.2 0.2 0.3 angle, f/p

0.4

0.8 0.6

m 1.0

0.4 0.2

0

0.1

0.5 0

0.2 0.3 0.4 angle, f/p

0.1

0.2 0.3 angle, f/p

0.4

Figure 4. Angle-dependent mean values of gradientkjrZjl, deviationkjdZjl and friction coefficient m of randomized hexagonal lattices shown for commensurate lu ¼ ld and incommensurate lu/ld ¼ t cases by thin and bold lines in the subplots (a), (b) and (c), respectively. fluctuations. These fluctuations are related only to the limited size of the numerically simulated systems, so each individual calculated line varies from one realization to another. All the results obtained for the values kjrZjl, kjdZjl, and friction coefficients m of randomized system, are summarized in figure 4a–c. It is seen directly from figure 4c that the friction coefficient in the incommensurate case lu/ld ¼ t is definitely higher than in the commensurate case lu ¼ ld.

4. Discrete model and dynamic simulations Unfortunately, our continuous model may only partially reflect the real situation. Besides, in many cases, it also becomes time consuming to calculate dependencies of the system properties on the varied parameters, especially in the case of the dynamics simulation. It would be useful to create an alternative model which will allow the same system to be studied by numerical simulation of Brownian dynamics of simple mutually interacting subsystems. Below, we have developed and studied such a discrete model. The conceptual structure of the discrete model is presented in figure 5. We simulated two contacting surfaces of probabilistic fasteners by interacting arrays of outgrowths, which are attached by one of their ends to the upper or to the lower flat surfaces. The outgrowths are allowed to rotate around the attachment roots with some flexibility and interact one with another according to simple rules, defined by corresponding interaction potentials. These properties of the model are qualitatively represented in figure 5, where two arrays of the outgrowths are depicted by two different kinds of the lines with attachment roots shown by the black and open circles, respectively. Owing to the interactions one with another and their flexibility, the outgrowths appear instantly bent as is shown in figure 5. As in the above continuous model, we can distribute the roots of the outgrowths differently. In particular, they can be ordered into a perfect or partially randomized hexagonal ‘dense packed’ lattice, or be placed in some other order. By default, the outgrowths are vertically oriented at equilibrium a0 ¼ p/2. The simplest way to simulate the deviations of the outgrowths from perfectly vertical positions is to use an effective elastic interaction fw ¼ B( p/2 2 a), where a is the instant inclination angle. When the deviations are relatively small, the difference between a and p/2 is proportional

f

b

Fexternal

6 5

1.0 z 0.5 0 0

4 3 2

y

2 x

4

1

6 8 0

Figure 5. Conceptual view of the discrete model. Fixed bases of asperities of the upper and lower surfaces are marked by the dark and open circles, respectively. Movable fibres are shown by the black lines. Angle b is calculated starting from one of the coordinates y coinciding with any of the symmetry axes. to the distance between the projections of instant positions of the outgrowth ends frjg and their root coordinates {r 0j }. Here, the discrete index j ¼ 1, 2, . . . N runs over all N outgrowths and {r 0j } is the two-dimensional vector {r 0j } ¼ {x0j , y0j }. Thus, in the following, we simulated elasticity of the outgrowths by the simple elastic relation fjelastic ¼ keff (r 0j  r j ),

(4:1)

with an effective elastic constant keff. Real biological outgrowths consist of sufficiently stiff polymers and in the nearest approximation can be modelled by rigid outgrowths, which can rotate around their root positions, but practically do not bend in their middle segments. Owing to the interaction one with another, they are repulsed by their hard cores and cannot penetrate one into another closer than to a characteristic distance corresponding to their widths. The widths of the outgrowths w can vary from one system to another, and can also be different for upper wu and lower wd surfaces. In real biological systems, these widths wu,d normally correlate with the density of outgrowths, or in other words with the periodicity lu,d of the structure

Phil. Trans. R. Soc. B 370: 20140032

0.1

5

1.5

rstb.royalsocietypublishing.org

1.0

0

(c) 1.0

Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

12

Fmax

8 6 4 2 1.2

2.0

1.6

2.4

lu/ld

Figure 6. Dependence of the friction force on the relationship between the periods lu/ld. Vertical dashed line marks golden ratio lu/ld ¼ t. Black and open symbols, as well as bold and thin fitting lines, correspond to the different elastic constants ka ¼ kb ¼ 10 and ka ¼ kb ¼ 1, respectively. Circles and diamonds correspond to regular and randomized Du,d x,y ¼ 0:15 lattices, respectively. General tendency of the friction force evolution with the randomization of the lattices is shown by the arrow. wu,d  lu,d . In the following, we estimated this choice of parameters in the interaction between the outgrowths. The simplest way to simulate numerically the repulsion between the outgrowths is to use Gaussian interaction potentials

Ukja,b

¼

U0a,b

8 " # 9 < r a,b  r a,b 2 = k exp  , : ; Sab

(4:2)

where k ¼ 1, 2, . . . , Nx, j ¼ 1, 2, . . . , Ny; U0a,b and S2ab ¼ (wa þ wb )=2 are intensities and decays of the interactions, and indices a, b can independently take the values a ¼ u, a ¼ d, b ¼ u and b ¼ d, corresponding to upper and lower surface lattices. To perform a numerical experiment, the centre of mass R of the upper lattice was pulled by an external elastic force F external ¼ K(Vt  R) with velocity V ¼ {V sin (b), V cos (b)}. If the system has hexagonal (or any other symmetry), the angle b is calculated starting from one of the coordinates y coinciding with any of the symmetry axes. This model has been used to perform a number of experimental runs at different pulling angles b, for different symmetries and mutual orientation w of upper and lower lattices, amplitudes of the randomization of the outgrowth distributions Du,d x,y , and their rotational stiffness ka,b. These simulations extend and generally confirm the results obtained in the study of our continuous model. In figure 6, we summarize the most important results for the representative general orientation w ¼ p/5 and pulling direction b ¼ 3p/10 obtained for two strongly different stiffness constants equal for both surfaces, ka ¼ kb ¼ 10 and ka ¼ kb ¼ 1. The simulations for all mentioned sets of the parameters were done at varied relations lu/ld between the periods of the perfect or randomized lattices in the interval 1 , lu/ld , 2.5. The numerical experiment was organized as follows. At different lu/ld, the external force F external pulls the upper plate from its static initial position, while after some transient process, it starts to move from stationary. We record a maximum friction force Fmax ¼ max (jF external j) which, according to the definition, determines the static friction threshold, and show its dependence on the relation lu/ld in figure 6.

5. Biological and biomimetic significance of obtained results A previous analytical model of a pair of sliding parabolic outgrowths in contact showed that the pull-off force which can be generated by probabilistic fasteners comprising individual outgrowths, is of the same order of magnitude as the loading force needed to bring such arrays of individual outgrowths into contact [5]. Furthermore, that analytical model is valid only for the small deformations at the initial phase of the contact formation between sliding outgrowths. A full analysis considering outgrowth deformation, even for a single pair of outgrowths, can only be made numerically. However, in reality, the forces will strongly depend on the collective behaviour of numerous outgrowths, and will therefore be strongly dependent on their spatial distribution on co-opted lattices. This hypothesis can only be tested numerically. It has been previously shown that in biological systems the density of outgrowths correlates to the length, width and especially the distance between them, so that the longest and widest outgrowths are usually sparsely distributed within the field [2]. However, there is only a weak correlation between length and width of outgrowths: longer outgrowths are not always wider. Distances between tips of longer and wider outgrowths are larger. The interplay of these parameters may result in different behaviour of contact element assemblages. This paper studied two numerical models predicting the behaviour of the probabilistic fasteners with different density

Phil. Trans. R. Soc. B 370: 20140032

0

6

rstb.royalsocietypublishing.org

10

It is seen directly from figure 6 that for relatively stiff outgrowths the results for the frictional behaviour of the discrete dynamic model are in a very good agreement with those of the continuous model. The function Fmax(lu/ld) has its maximum around lu/ld  t. The vertical dashed line in the plot indicates the golden ratio lu/ld ¼ t. With an increased degree of randomization, the curve Fmax(lu/ld) generally shifts to higher values. In particular, the dependence Fmax(lu/ld), calculated at Du,d x,y ¼ 0:15, is shown in figure 6 (the curve is marked by the black diamonds). Good agreement of this behaviour and the above results of our continuous model probably arises from proper selection of the outgrowth rigidity ka,b ¼ 10, allowing the outgrowths to remain almost vertical in the course of motion. This should be expected, because the continuous model completely ignored horizontal motion of the outgrowths, which were represented by Gaussian maxima with fixed centres. It is interesting that in the system with softer outgrowths, friction behaviour is not much affected. For ka,b ¼ 1, both curves Fmax(lu/ld) obtained in cases of perfect and randomized distributions (figure 6) are slightly shifted to higher values of Fmax at relatively small lu/ld  1.5 values and to lower values of Fmax for larger lu/ld  2 values. Hence an optimal relationship between the periods of outgrowths of co-opted surfaces, corresponding to the maximum of Fmax(lu/ld), moves towards smaller values of lu/ld  1.45. Perhaps it happens because at extremely low stiffness of outgrowths it is easier for the fibres from one surface to adapt from the very beginning to the positions between the outgrowths of another surface. The same adaptation reduces a difference between the curves for perfect and randomized cases, because almost freely rotating ends of the outgrowths can leave their ideal lattice positions more easily.

Downloaded from http://rstb.royalsocietypublishing.org/ on March 29, 2015

Acknowledgements. We thank V. Kastner (Tu¨bingen, Germany) for linguistic corrections of the manuscript. Funding statement. This study was partly supported by the SPP 1420 priority programme of the German Science Foundation (DFG) ‘Biomimetic Materials Research: Functionality by Hierarchical Structuring of Materials’ ( project GO 995/9-2) to S.N.G.

References 1. 2.

3.

4. 5.

6.

7.

Nachtigall W. 1974 Biological mechanisms of attachment. Berlin, Germany: Springer. Gorb SN. 1998 Frictional surfaces of the elytra to body arresting mechanism in tenebrionid beetles (Coleoptera: Tenebrionidae): design of co-opted fields of microtrichia and cuticle ultrastructure. Int. J. Insect Morphol. Embryol. 27, 205–225. (doi:10.1016/S0020-7322(98) 00013-0) Gorb SN. 1998 Functional morphology of the head-arrester system in Odonata. Zoologica 148, 1– 132. Scherge M, Gorb SN. 2001 Biological micro- and nanotribology. Berlin, Germany: Springer. Gorb SN, Popov VL. 2002 Probabilistic fasteners with parabolic elements: biological system, artificial model and theoretical considerations. Phil. Trans. R. Soc. Lond. A 360, 211 –225. (doi:10.1098/ rsta.2001.0926) Richards AG, Richards PA. 1979 The cuticular protuberances of insects. Int. J. Insect Morphol. Embryol. 8, 143– 157. (doi:10.1016/00207322(79)90013-8) Gorb SN. 2001 Attachment devices of insect cuticle. Dordrecht, The Netherlands: Kluwer.

8.

9.

10.

11.

12.

13.

Gorb SN. 1996 Design of insect unguitractor apparatus. J. Morphol. 230, 219–230. (doi:10.1002/ (SICI)1097-4687(199611)230:2,219::AID-JMOR 8.3.0.CO;2-B) Gorb SN. 2004 The jumping mechanism of cicada Cercopis vulnerata (Auchenorrhyncha, Cercopidae): skeleton-muscle organisation, frictional surfaces, and inverse-kinematic model of leg movements. Arthr. Str. Dev. 33, 201–220. (doi:10.1016/j.asd. 2004.05.008) Gorb SN. 1999 Evolution of the dragonfly headarresting system. Proc. R. Soc. Lond. B 266, 525 –535. (doi:10.1098/rspb.1999.0668) Schrott A. 1986 Vergleichende Morphologie und Ultrastruktur des Cenchrus-Dornenfeldapparates bei Pflanzenwespen (Insecta: Hymenoptra, Symphyta). Ber. Nat. Med. Verein Innsbruck 73, 159–168. Gorb SN. 1999 Ultrastructure of the thoracic dorsomedial field (TDM) in the elytra-to-body arresting mechanism in tenebrionid beetles (Coleoptera: Tenebrionidae). J. Morphol. 240, 101 –113. (doi:10. 1002/(SICI)1097-4687(199905)240:2,101::AIDJMOR2.3.0.CO;2-7) Lilienthal K, Stubenrauch M, Fischer M, Schober A. 2010 Fused silica ‘glass grass’: fabrication and

14.

15.

16.

17.

18.

utilization. J. Micromech. Microeng. 20, 025017. (doi:10.1088/0960-1317/20/2/025017) Pang C, Kang D, Kim T, Suh K-Y. 2012 Analysis of preload-dependent reversible mechanical interlocking using beetle-inspired wing locking device. Langmuir 28, 2181–2186. (doi:10.1021/ la203853r) Pang C, Kim SM, Rahmawan Y, Suh K-Y. 2012 Beetle-inspired bi-directional, asymmetric interlocking using geometry-tunable nanohairs. ACS Appl. Mater. Interfaces 4, 4225–4230. (doi:10.1021/ am3009289) Pang C, Kwak MK, Lee C, Jeong HE, Bae W-G, Suh K-Y. 2012 Nano meets beetles from wing to tiptoe: versatile tools for smart and reversible adhesions. Nano Today 7, 496–513. (doi:10.1016/j.nantod. 2012.10.009) Niederegger N, Gorb S, Jiao Y. 2002 Contact behaviour of tenent setae in attachment pads of the blowfly Calliphoravicina (Diptera, Calliphoridae). J. Comp. Physiol. A 187, 961 –970. (doi:10.1007/ s00359-001-0265-7) Popov VL. 2010 Contact mechanics and friction. Physical principles and applications. Berlin, Germany: Springer.

7

Phil. Trans. R. Soc. B 370: 20140032

The Velcro principle was transferred to industrial applications from biological observations. Recent patent databases contain a large number of ideas dealing with applications of existing hook-and-loop fasteners. They are widely used in the textile industry, medicine, sports, transport, etc. However, the vast majority of these applications use the same types of hook-like tapes. Indeed, these systems generate very high attachment forces, if compared with the applied load. Their function is quite long-term attachment, whereas probabilistic systems, discussed in this paper, might be of importance for designing short-term fast attaching and detaching devices, as previously described for macroscopic cases [5] and for microscopic systems [13–16]. For the purpose of optimization of such technical devices, knowledge concerning the optimal relationship between distributions of single outgrowths on co-opted lattices in biological systems, and the numerical study presented here might be of great importance.

rstb.royalsocietypublishing.org

of single outgrowths and different regularity of their distributions in co-opted lattices. The main finding of our virtual experiments is that there is an optimal relationship of densities on co-opted lattices for generation of the strongest contact force. Another important result is that the optimal distribution of the outgrowths should be not absolutely perfect (grid-like), but slightly randomized. Both these features are observed in various biological probabilistic fastener systems capable of fixing body parts as a result of friction enhancement. Our finding that co-opted lattices within the same organism have different distribution of individual outgrowths was initially rather surprising for us [7], because the organism can definitely keep similar density of outgrowths on both surfaces by controlling the size of the cells responsible for development of such outgrowths. Furthermore, by keeping the same cell distribution on both surfaces, a highly ordered type of outgrowth distribution can be quite easily generated as, for example, in the case of the tarsal adhesive hairs in the fly. Numerical simulations, presented here, provide a strong argument for the functional significance of both certain difference in outgrowth density on co-opted lattices and certain imperfections in outgrowth distribution on each lattice in biological probabilistic fasteners.

The functional significance of density and distribution of outgrowths on co-opted contact pairs in biological arresting systems.

Microstructures responsible for temporary arresting of contacting surfaces are widely distributed on surfaces in different organisms. Recent morpholog...
1MB Sizes 2 Downloads 4 Views