Photoplastic effect on ice - APS Link Manager

2 downloads 0 Views 223KB Size Report
May 23, 1994 - (1) The Peierls model (for basal plane glide in ice). This model has found wide application for semiconduc- tors [7]. Strong evidence in favor of ...
VOLUME 72, NUMBER 21

PH YSICAL REVI EW LETTERS

23 MAY 1994

Photoplastic El'ect on Ice Niyaz N. Khusnatdinov

and Victor

F. Petrenko

Thayer School of Engineering, Dartmouth College, HanoverN, ew Hampshire (Received 7 February 1994)

03755

This paper examines the effect of ultraviolet illumination on the plastic deformation of ice single crystals. The study of ice plasticity is essential for both fundamental and applied reasons. For instance, plastic deformation of ice is a major factor in the How of the giant ice caps in Antarctica and Greenland and of mountain glaciers. Snow densification, sintering, and ice friction are also phenomena in which plastic deformation has a role. Since ice has a very unusual atomic structure, in which oxygen atoms are ordered but protons are disordered, the motion of dislocations through ice cannot occur in the same way as in ordinary crystals. This is the intriguing subject of the research here. PACS numbers:

61.72. Hh, 82. 50.Fv

lce plasticity has been studied intensively during the last few decades [1]. Both macroscopic deformation and the mobilities of various individual dislocations have been investigated in detail. Nevertheless, so far there is no satisfactory physical model capable of explaining such important experimental results as the strong anisotropy of the ice plasticity [2], the softening of ice by doping with small concentrations of acids [3-5], strong evidence for a Peierls barrier to the motion of dislocations in the basal (0001) plane, and the absence of such a barrier to the motion of dislocations in nonbasal planes [6]. Four models have been suggested to describe ice plasticity and the motion of individual dislocations in ice: (1) The Peierls model (for basal plane glide in ice). This model has found wide application for semiconductors [7]. Strong evidence in favor of the Peierls model is the fact that dislocations quickly moved into a hexagonal formation after deformation was started and then glided as straight segments [6]. Yet the experimental data show dislocations are able to move more easily than expected on the basis of this model [6]. This model also does not take into account proton disorder and therefore cannot explain the effect of dopants on the plasticity. (2) Glen's model [8]. Glen first supposed that proton disorder can be the obstacle for dislocation glide in ice. If so, then the dislocation velocity id should be reciprocal to the reorientation time of a water molecule in the ice lattice, i.e. , the Debye relaxation time, rD [9-11]. Since a-o where ~ is the ice's high-frequency conductivity [12, 13], the dislocations mobility should be proportional to a concentration of the majority charge carriers. Glen's model explains qualitatively the effects of doping with HF and HCl on plasticity, but gives a value lower than the experimental one for vd. (3) Perez et al. invented a model based on a noncrystalline dislocation core [14], which uses several adjustable parameters to obtain the correct velocity of dislocations. Nevertheless it contradicts the motion of dislocations as straight segments, which indicates a crystalline character of the core. (4) Weertman proposed another mechanism [15] by

which the disorder of the protons in ice will inhibit the motion of dislocations: anelastic loss, which arises from reorientation of molecules in the stress field of the dislocation as it moves. Any such effect will constitute a barrier in addition to those already considered, and seems

likely to be comparatively small. None of these models can explain all experimental data, which are in part contradictory. 1ndeed, it is well known that doping of ice with hydrofluoric acid HF and hydrochloric acid HC1 leads to softening of ice [4,5], while doping with ammonia NH3 leads to its hardening

[4]. These changes in plasticity are in relation to the changes of Debye relaxation time according to Glen's model. Investigation of the dislocation velocity in HFdoped ice [16] showed that the influence of doping is much less than it could be according to the theory. Recent observations of HCl-doped ice [17] showed that doping had no significant efl'ect on vd, while rid changed by more than a factor of 10. The softening of ice could result from the difference between initial densities of dislocations. For instance, it is known that doped crystals are much less perfect [18]. Petrenko and Schulson [19] found an alternative way of changing the concentrations of charge carriers in ice without changes in the dislocations' density or introduction of foreign atoms: By applying electric fields to thin specimens undergoing deformation in shear, they have shown that a reduction in the leads to a corresponding conductivity high-frequency reduction in the creep rate. Their result supports the idea that plastic deformation is related to the proton disorder and the rate of reorientation of hydrogen bonds. There is still room for uncertainty because of the very small thickness of their samples (50 pm) and the potential influence of boundary layers on plasticity in such a deformation geometry. The photoconductivity (PC) of ice recently found both in pure [20] and doped [21] single crystals brings new hope to attempts to verify a correlation between ice plasticity and conductivity. The PC of ice is excited under light with wavelengths shorter than 190 nm. A modified "autoionization" reaction was suggested to explain the

0031-9007/94/72(21)/3363 (4) $06.00 1994 The American Physical Society

3363

PHYSICAL REVIEW LETTERS

VOLUME 72, NUMBER 21

photoconductivity

of ice [22]:

2H20+

H20*+ H20

hv

H30++OH + (e + V+0), V is a vacancy and D is a D defect. According to reaction (I), the illumination adds such mobile charge carriers as ions H30+ and D defects which in turn change the Debye relaxation time rD. Since this change is not accompanied by doping with foreign atoms or changing the dislocation density, the PC could be used as a convenient tool in distinguishing among different models of dislocation motion. It was also important for the motivation of this study that in nature the plastic deformation of ice occurs under intensive solar illumination. We have used pure, filtered de-ionized water from a Milli-Q Water System, model ZD40 115 95 of Millipore Corp. The water was degassed before the ice growing The specific resistance of water was p procedure. =1.83&10 Dcm at room temperature. For doping of water we have used a pure grade of 2N HCI solution and 45% potassium hydroxide solution KOH from EM Science, a Division of EM industries, Inc. To grow single crystals of pure and doped ice we have used the express method [23] in which single crystals of ice grow during intensive cooling of the water s surface by the evaporation of water in a vacuum. The procedure usually takes 10-15 min to obtain a crystal of 10 cm in diameter with a thickness of 0.5 cm. One advantage of this method is the ability to grow single crystals with a We have not controlled high concentration of impurities. the initial density of dislocations in the crystals grown by this method. Together with pure single crystals we have of grown doped single crystals with a concentration 10' cm n of HCI and KOH. The amounts of incorporated impurities were checked by measuring electric conductivity. The quality of single crystals and the direction of the c axis were controlled in crossed polarizers. The crystals were first cut by a band saw and then thinned with a microtome to obtain 1 x 7 x 20 mm samples to the larger face as with the c axis directed at shown in Fig. 1. Such an orientation produces conditions favorable for dislocation slip in the basal (0001) plane. Dislocation bands parallel to the basal plane were visible All ice samples were dein all our deformed crystals. formed in the creep regime under constant stress using a four-point bend deformation geometry. The experimental setup is shown in Fig. 1. Displacements of the upper (8) quartz plane' were measured with a strain gauge with a sensitivity of k =0.25 pm/mV. The 8 quartz plane was fixed; a =15 mm and b=9 mm. A set of ~eights from 10 to 500 g was used to produce constant stress. All the results described below were obtained in a steady state of creep, which usually corresponded to 1% & e& 5%. For calculation of strain e for a particular geometry of deformation we used the following approximate expression:

23 MAY 1994

A

Quartz

B

Quartz

where



-45'

3364

/ / / / / / / / / /

FIG. I. Experimental setup for four-point a =9 mm; b =15 mm.

bending

of ice

samples:

4h b2

2

where g is the displacement of an A quartz plane and h is I mm). the thickness of a crystal To measure the electric conductivity of ice, we used two stainless steel electrodes, a guarding ring and a lockin amplifier (EG&G Princeton Applied Research, model 5209), which has a built-in ac generator. Two sources of light were used for excitation of photoplastic eITect (PPE): a deuterium lamp, L2196 from Hamamatsu Corp. , with a window that is transparent to UV light with a wavelength k & 160 nm, and a 75 %' xenon lamp, model 6251, from Oriel Corp. To measure spectral dependence of PPE, we used a monochromator, model 82-405, of Fisher Scientific Corp. We used two hxed 1 mm slits. Reciprocal dispersion and eAective aperture were 5 nm/mm and f/6. 0 In addition to th.e monochromator, optical filters were used. Figure 2 sho~s the plastic deformation of a pure single crystal under constant load, P =50 g and surface stress, o =1.0 MPa at T = —) 5'C. The sample purity is characterized by its low-frequency conductivity, +0=i.4 x 10 0 ' cm ', and high-frequency conductivity,

(-

tT

=2X10

0 'cm at T= —15 C.

Curves 1 and 2 represent both measured time diagrams of the displacement of the upper quartz plane 8 and the sample's plastic creep rate i, respectively. The "up" and "down" arrows indicate the times at which the light was switched on and off. In this particuof lar test a monochromatic light with a wavelength 1=260 nm and an intensity of 0.2 pW/cm2 was used. It follows that the illumination of ice leads to its hardening, i.e. , a photoplastic effect. One can observe that PPE has reversible and i rreversible corn ponen ts. The ratio of these components varies from sample to sample. In most of our

PH YSICAL REVIEW LETTERS

VOLUME 72, NUMBER 21

20-

0.3

-10

C

23 MAY 1994

0.2~

5

10-

a ~

4)

0. 1—

CO

74

4~

~

~

0

2

~

t

4

t

~

I

6

~

0

~

t

10

S

N

2

Time, 10 s FIG. 2. Photoplastic effect in pure ice for monochromatic light X=260 nm at T= —l5'C. Curve 1 represents the creep (displacement of A plane) of a single crystal under constant load P =50 g, which corresponds to the stress at the samples' surface +=1.0 MPa; curve 2 is a strain rate i.

tests the reversible part was very small. We examined the action of the UV illumination on the rate of plastic strain, ~, of fifteen samples prepared from eight pure ice varied from single crystals. The observed change in 19% to 0% when ice was illuminated with light at A. =260 nm. In four samples, the PPE was altogether absent for unknown reasons. The irreversibility of PPE leads to loss of the ice sample's sensitivity to illumination after some dose has been accumulated by the ice. Figure 3 shows the decrease (in percent) of slope changes on the creep curve due to subsequent irradiation pulses of 100 s duration. We found that preliminary deformation of a crystal does But prolonged not change its sensitivity to illumination. irradiation prior to deformation extinguishes the PPE. J/cm at A, =260 nm Light exposure of about 8x lO makes a sample "blind. Because of the very low intensity of the light used, we have never observed any heating effects such as appearance of internal melting (Tyndall figures [12]) in ice samples. The spectral dependence of PPE on pure ice is shown in Fig. 4, curve 1. This spectrum was normalized to the spectral dependence of the light source. As a reference point we chose k =300 nm. To eliminate the influence of the dose accumulation, we averaged the data for four wavelength dependencies taken from different samples which were measured starting at different wavelengths. Several tests were performed on doped single crystals. Two crystals doped by KOH (with the molecular concentration n 10' 0 ' cm ', a ao =1.7 x 10 ' =7 x l 0 0 cm ') were not sensitive to illumination. But three out of four crystals doped by HCl (n

i

"



—10' cm,

0

'cm

')

FIG. 3. Dose dependence of PPE as a function of 100 s irradiation cycles with a monochromatic light intensity 0. 2 pW/cm at X=260 nm.

dence of PPE on HC1-doped ice. This spectrum was also subjected to the normalization procedure described above for samples of pure ice. While these findings are not yet suScient to define a we are physical mechanism for this new phenomenon, able to draw several useful conclusions. First of all, the observed changes in ice's plastic strain rate did not correlate with the corresponding changes in the rate of water molecules' reorientation in the ice bulk rD ' as was supposed in Glen's model [8]. Indeed, while in our never exceeded experiments hs/s reaches 44%, Ae /rr 0. 1%, even when A. (190 nm. Moreover, they have the opposite sign. The wavelength dependencies of the PC and the PPE are also different; see Fig. 5. It is worthwhile to mention at this point that due to the strong interaction between inhomogeneous elastic strains near dislocations and the protonic charge carriers, OH and H30+ ions, D and L defects, the defects' concentrations in ice bulk can differ from those in the dislocations' vicinity. A similar problem was resolved for cracks in ice [24] and should be analyzed for dislocations in the future. Without such an analysis we cannot say that Glen's mod-

i

~~,

I

I

I

I

I

I

I

I

s

I

I

I

~

T= -15 C C

4

CO

cm,

ao=3 7x10

0

'cm

'

I

a~=8x10

showed the photoplastic effect even more than the crystals from pure ice at A, =230-270 nm. Curve 2 in Fig. 4 represents the wavelength depen-

I

180

~

I

~

I

~

I

~

I

220 260 nm Wavelength,

T

300

FIG. 4. Normalized pure (curve

I) and

wavelength dependence of the PPE in HCl-doped (curve 2) ice at T= —15 C.

3365

PH YSICAL REV I EW LETTERS

VOLUME 72, NUMBER 21

10

a

I

I

1

s

I

s

I

s

I

a

Grant

No. DMR-

Petrenko and R. W. Whitworth, CRREL Special Report {to be published).

U. S. Army

National Science Foundation 91 22 l 922.

I

23 MAv 1994

[I] V. F.

under

[2] J. Muguruma,

0

~

I

180

I

~

1

220

/

I

I

~

300

260

Wavelength,

nm

FIG. 5. Wavelength dependencies of PPE (curve conductivity [20] (curve 2), and excitation spectrum luminescence at X=420 nm [25] (curve 3).

I),

photo-

of photo-

to PPE in ice, since one should compare a possible correlation between dislocation velocity and the concentration of defects near the dislocation core, but not the concentration of defects in the ice bulk. Let us now compare the excitation spectra of the photoplastic effect (see Fig. 5), the PC [20] and the photoluminescence (PLM) at X=420 nm [25], which are also shown in Fig. 5. The PPE and PLM spectra are similar for X& 200 nm, but PC is absent in this wavelength range. In electronic semiconductors with dislocations, a similar difference in spectra among the PPE, PC, and PLM is well known and is explained by a redshift of optical transitions in the vicinity of the dislocation core and by transitions between electron energy bands and dislocation levels [26]. The same might be true in such a protonic semiconductor as ice. An interesting point is why the photoplastic eAect exists in pure ice and is even more pronounced in HCldoped ice, while it is absent in KOH-doped samples. It is well known that HCl doping increases concentrations of H30+ ions and L defects and decreases the number of ions and D defects, while KOH doping decreases OH the number of H30+ ions and D defects and increases ions and L defects [12, 13]. [f these the number of OH of defects produce the changes in the concentration diA'erence in the observed reaction of pure and doped ice, then it should be H30+ ions which are necessary for the photoplastic eAect in ice. Under the UV illumination, H30+ ions somehow multiply, impeding the dislocation motion. But this is a subject for future research. These materials are based upon work supported by the

el is not applicable

3366

S. Mae, and A. Higashi, Philos. Mag. 13, 625 (1966). [3] S. 3. Jones, Phys. Lett. 25A, 366 (1967). [4l S. 3. 3ones and J. W. Glen, Philos. Mag. 19, 13 (1969). [5] T. Nakamura and S. 3. 3ones, Scr. Met. 4, 123 (1970). [6] C. Shearwood and R. W. Whitworth, Philos. Mag. A 64„ 289 (1991). [7] J. P. Hirth and J. Lothe, Theory of Dislocations (Wiley, New York, 1982), 2nd ed. [8] J. W. Glen, Phys. Condens. Mater. 7, 43 (1968). [9] R. W. Whitworth, J. G. Paren, and J. W. Glen, Philos. Mag. 33, 409 (1976). [10] H. J. Frost, D. J. Goodman. and M. F. Ashby, Philos. Mag. 33, 951 (1976). [11] R. W. Whitworth, Philos. Mag. A 41, 521 (1980). [12] P. V. Hobbs, lce Physics (Clarendon, Oxford, 1974). [13] V. F. Petrenko, U. S. Army CRREL Special Report 1993-20, 1993. [14] J. Perez, C. Ma'i, J. Tatibouet, and R. Vassoille, 3. Glaciol. 25, 133 (1980). in Ice and Snow; Properties. Processes and Applications, edited by W. D. Kingery (MIT Press. Cambridge, MA, 1963), pp. 28-33. [16] C. Mal, J. Perez, J. Tatibouet, and R. Vassoille, 3. Phys. (Paris) 39, 307 (1978). [17] C. Shearwood and R. W. Whitworth, Philos. Mag. A 65, 85 (1992). [18] M. Oguro, in Lattice Defects in Ice Crysrais, edited by A. Higashi (Hokkaido Univ. Press, Sapporo, 1988), p. 27. [19] V. F. Petrenko and E. M. Schulson, Philos. Mag. A 67,

[15] 3. Weertman,

173 (1993). [20] N. N. Khusnatdinov, V. F. Petrenko, and A. N. Turanov, Phys. Status Solidi (a) 118, 401 (1990). [21] N. N. Khusnatdinov and V. F. Petrenko, in Physics and Chemistry of Iece, edited by N. Maeno and T. Hondoh (Hokkaido Univ. Press, Sapporo, 1992), p. 170. [22] V. F. Petrenko and N. N. Khusnatdinov, 3. Chem. Phys. (to be published). [23] N. N. Khusnatdinov and V. F. Petrenko, in Physics and Chemistry of Ice (Ref. [21]), pp. 395-8. [24] V. F. Petrenko, Philos. Mag. B 67, 301 (1993). [25] T. I. Quickenden, R. A. J. Litjens, C. G. Freeman, and S. M. Trotman, Chem. Phys. Lett. 114, 164 (1985). [26] Yu. A. Ossipyan, V. F. Petrenko, A. V. Zaretskii, and R. W. Whitworth, Adv. Phys. 35, 115 {1986).