PSI - Issue 12

A. Papangelo et al. / Procedia Structural Integrity 12 (2018) 265–273 A. Papangelo / Structural Integrity Procedia 00 (2018) 000–000

268

4

(where notice that we have to assume a PO >> b to be consistent with the thin layer assumption), and hence substituting into (5)

E ∗ LR 1 / 2 (2 b ) 1 / 4

w E ∗

3 / 4

8 3

(7)

P PO = −

whereas the average stress in the contact at pull-o ff is

√ 2

b

1 / 2

√ 2

E ∗ w

P PO A PO

2 3

2 3

K Ic √ b

(8)

= −

= −

σ PO =

where K Ic is toughness of the contact. Hence, notice that the JKR solution simply gives the Gri ffi th condition imposed by a Stress Intensity Factor which scales only with the size the layer b and not any other length scale (like the radius of the punch). The interesting result is that as b → 0 the limit of the force also goes to ∞ . Eq. (8) can be written in dimensionless form as

√ 2

E ∗ σ th l 1 / 2 a

2 3

σ PO σ th

(9)

= −

where l a = w / E ∗ b is a dimensionless adhesion parameter. Figure 2 shows how increasing l a (for a given set of material constants this implies a reduction in the layer thickness b ) the average pull-o ff stress is increased. Since σ PO will be bounded by theoretical strength, the situation is analogous to the well known case of a fibrillar structure in contact with a rigid halfspace, like that discussed for Gecko and many insects who have adopted nanoscale fibrillar structures on their feet as adhesion devices (Gao & Yao, 2004). In our case, to have a design insensitive to small variations in the tip shape, we would simply need to go down in the scale of the layer thickness. In fact, imposing σ PO /σ th = − 1, we obtain a critical value for l a , namely l a , cr , above which the theoretical strength of the material is reached, which also defines, for fixed material properties, the order of magnitude of the ”critical” thickness of the layer below which we expect theoretical strength

9 8

σ th E ∗

E ∗ w σ 2 th

2

8 9

l a , cr =

→ b cr =

(10)

Taking w = 10 mJ / m 2 , σ th = 20 MPa and E ∗ = 1 GPa, like done in (Gao & Yao, 2004), we estimate l a , cr = 4 . 5 × 10 − 3 10 9 10 − 2 ( 20 × 10 6 ) 2 = 22 nm, which is of the same lengthscale of the estimate (of a di ff erent geometry) of 64 nm robust design diameter of the fiber of the fibrillar structure. Hence, with this size of layer of nanoscopic scale, we would be able to devise a quite strong attachment for any indenter. In the halfplane limit case, from Barquins (1988), Chaudhury et al. (1996) we have for the cylinder or b cr = 8 9