1: %\documentclass[12pt,preprint]{aastex}
2: %\documentclass[preprint2]{aastex}
3: \documentclass{aastex}
4: %%\usepackage[]{natbib}
5:
6: \begin{document}
7:
8: \title{The Physics of Protoplanetesimal Dust Agglomerates \\
9: II. Low Velocity Collision Properties}
10: \author{Doreen Langkowski}
11: \affil{Institut f\"ur Geophysik und extraterrestrische Physik,
12: Technische Universit\"at zu Braunschweig, Mendelssohnstr. 3,
13: D-38106 Braunschweig, Germany}
14:
15: \author{Jens Teiser}
16: \affil{Institut f\"ur Planetologie, Westf\"alische
17: Wilhelmsuniversit\"at M\"unster, Wilhelm-Klemm-Str. 10, D-48149
18: M\"unster, Germany} \email{j.teiser@uni-muenster.de}
19:
20: \author{J\"urgen Blum}
21: \affil{Institut f\"ur Geophysik und extraterrestrische Physik,
22: Technische Universit\"at zu Braunschweig, Mendelssohnstr. 3,
23: D-38106 Braunschweig, Germany} \email{j.blum@tu-bs.de}
24:
25: \slugcomment{{\bf Accepted by the Astrophysical Journal}}
26:
27: \begin{abstract}
28: For the investigation of collisions among protoplanetesimal dust
29: aggregates, we performed microgravity experiments in which the
30: impacts of high-porosity mm-sized dust aggregates into 2.5
31: cm-sized high-porosity dust aggregates can be studied. The dust
32: aggregates consisted either of monodisperse spherical, of
33: quasi-monodisperse irregular or of polydisperse irregular
34: micrometer-sized dust grains and were produced by random ballistic
35: deposition with porosities between 85\% and 93\%. Impact
36: velocities ranged from $\sim 0.1 ~\rm m~s^{-1}$ to $\sim 3 ~\rm
37: m~s^{-1}$ and impact angles were almost randomly distributed. In
38: addition to the smooth surfaces of the target aggregates formed in
39: our experiments, we ``molded'' target aggregates such that the
40: radii of the local surface curvatures corresponded to the
41: projectile radii, decreasing the targets porosities to $80-85$\%.
42: The experiments showed that impacts into the highest-porosity
43: targets almost always led to sticking, whereas for the less porous
44: dust aggregates, consisting of monodisperse spherical dust grains,
45: the collisions with intermediate velocities and high impact angles
46: resulted in the bouncing of the projectile with a mass transfer
47: from the target to the projectile aggregate. Sticking
48: probabilities for the impacts into the ``molded`` target
49: aggregates were considerably decreased. For the impacts into
50: smooth targets, we measured the depth of intrusion and the crater
51: volume and could derive some interesting dynamical properties
52: which can help to derive a collision model for protoplanetesimal
53: dust aggregates. Future models of the aggregate growth in
54: protoplanetary disks should take into account non-central impacts,
55: impact compression, the influence of the local radius of curvature
56: on the collisional outcome and the possible mass transfer between
57: target and projectile agglomerates in non-sticking collisions.
58: \end{abstract}
59:
60: \keywords{Solar Nebula, Planetesimals, Experimental Techniques,
61: Collision Physics, Solar System Origin}
62:
63: \section{Introduction}
64: Young stars are surrounded by accretion disks. It is now widely
65: accepted that the formation of planetesimals, the km-sized
66: precursors of the terrestrial planets and of the cores of the gas
67: planets, is initiated by the process of agglomeration of
68: protoplanetary dust in these disks. For the young Solar System,
69: the circumstellar disk is also known by the name of ''solar
70: nebula``. Due to the decreasing rate of accretion, the gas disk
71: cools down and fine dust particles condense. Typical initial grain
72: sizes, as evidenced by astronomical observations of circumstellar
73: accretion disks \citep{kesetal06,przetal03}, primitive meteorites
74: from our own Solar System \citep{ker93}, and recent measurements
75: from comets \citep{haretal05,keletal05,hoeetal06} are in the
76: (sub-)micrometer size range.
77:
78: The velocities of these dust grains relative to the gas disk are
79: caused by Brownian motion, gravity-induced drift motion and gas
80: turbulence and result in mutual collisions among the grains
81: \citep{weicuz93} which can lead to the growth of dust
82: agglomerates. When the collision velocities are sufficiently low,
83: the dust particles show a hit-and-stick behavior
84: \citep{popetal00,bluwur00,domtie97} and the dust agglomerates are
85: bound by weak van der Waals forces \citep{heietal99}. As a
86: consequence, fractal dust agglomerates are formed whose fractal
87: dimensions $D_{\rm f}$, defined by the relation between
88: agglomerate mass $m$ and size $s$, $m \propto s^{D_{\rm f}}$,
89: range from $D_{\rm f} \approx 1.4 \ldots 1.5$
90: \citep{bluetal00,krablu04,pasdom06} for Brownian motion-dominated
91: growth to $D_{\rm f} \approx 1.8 \ldots 1.9$ for growth caused by
92: drift motions \citep{bluetal98} or gas turbulence
93: \citep{wurblu98}. Due to the increasing agglomerate mass and the
94: consequentially increasing collision velocities for protoplanetary
95: dust \citep{weicuz93}, collisions will eventually lead to the
96: compaction of the agglomerates \citep{domtie97,bluwur00} so that
97: the fractal dimensions will increase to $D_{\rm f} = 3$
98: \citep{weicuz93,blu04}. Such agglomerates are, however, not
99: compact but can be quite porous. Simulation experiments suggest
100: that the volume filling factors, given by
101: \begin{equation}\label{vff}
102: \phi = 1 - p =\frac{\rho}{\rho_0}~,
103: \end{equation}
104: with $p$, $\rho$ and $\rho_0$ being the porosity, the mass density
105: of the aggregate and of the solid grain material, respectively,
106: should be as low as $\phi \le 0.20 \ldots 0.33$ even if the
107: collision velocities are as high as 50 $\rm m~s^{-1}$
108: (\citet{blusch04}; \citet{bluetal06}, hereafter referred to as
109: paper I). As long as the collision velocities are $\lesssim 1 \rm
110: m~s^{-1}$, the maximum volume filling factor for loose particle
111: ensembles is $\phi = 0.07 \ldots 0.15$ (\citet{blusch04}; paper
112: I). Very low collision velocities could lead to even lower volume
113: filling factors of $\phi \approx 0.01$ \citep{ormetal07}.
114:
115: For fractal dust agglomerates, concurring theoretical
116: \citep{domtie97} and experimental \citep{bluwur00} results show
117: that above a threshold velocity of $\sim 1 \rm \, m~s^{-1}$, dust
118: agglomerates do no longer stick together but bounce off and
119: fragment upon collision. For non-fractal dust aggregates, impact
120: experiments by \citet{wuretal05a} and \citet{wuretal05b}, who used
121: cm-sized porous projectiles and decimeter-sized porous targets,
122: showed that for high-porosity targets ($\phi = 0.12 \ldots 0.26$)
123: no sticking occurs in the velocity range $16.5 \ldots 37.5 \, \rm
124: m~s^{-1}$ and a crater is formed on the target which leads to a
125: mass loss from the target. For compacted targets ($\phi = 0.34$),
126: a mass gain of the target of $\sim 50$ percent of the projectile
127: mass could be found for impact velocities $\ge 13 \, \rm
128: m~s^{-1}$.
129:
130: Not much is known about the collision behavior of high-porosity
131: dust agglomerate in the velocity regime around $1 \, \rm
132: m~s^{-1}$. Earlier experiments by \citet{blumue93}, who used
133: mm-sized dust agglomerates with $\phi = 0.26$ and collision
134: velocities in the range $0.15 \ldots 3.8 \, \rm m~s^{-1}$, showed
135: that none of the collisions between (almost) equal-sized
136: agglomerates led to sticking, while at the highest collision
137: velocities fragmentation dominated. With our new technology for
138: the formation of high-porosity macroscopic dust agglomerates
139: (\citet{blusch04}; paper I), realistic macroscopic dust
140: agglomerates with volume filling factors in the range $\phi = 0.07
141: \ldots 0.15$ are now available with which the collision behavior
142: of protoplanetary dust agglomerates can be investigated.
143:
144: Sect. \ref{sectobj} gives an overview of the scientific objectives
145: of this work, Sect. \ref{sectsetup} presents the experimental
146: technology and the properties of the used dust agglomerates, Sect.
147: \ref{sectexpres} gives the experimental results on sticking
148: efficiencies, mass transfer, coefficients of restitution, energy
149: loss, tensile strength, crater formation and impact fragmentation.
150: Sect. \ref{sectdisc} discusses the low-velocity impact dynamics of
151: high-porosity dust agglomerates, Sect. \ref{sectsummary} gives a
152: summary of our experimental findings, and in Sect.
153: \ref{sectconclusion}, we draw conclusions from our work for the
154: formation of larger objects in the protoplanetary nebula.
155:
156: \section{\label{sectobj}Objectives}
157: This paper describes novel low-velocity impact experiments between
158: high-porosity dusty projectiles of diameters between $\sim 0.2$ mm
159: and $\sim 3$ mm and high-porosity dusty targets of 2.5 cm
160: diameter. From previous modelling and laboratory work it became
161: clear that above a certain agglomerate size the hitherto present
162: growth of fractal agglomerates is no longer feasible due to the
163: increasing collision energy with increasing agglomerate size
164: \citep{domtie97,bluwur00}. Thus, the agglomerates above a certain
165: size limit should be non-fractal but highly porous in structure
166: \citep{blu04,ormetal07}. With a new experimental method, we are
167: able to manufacture macroscopic, high-porosity dust agglomerates
168: (see Sect. \ref{sectsetup}) with which realistic collision
169: experiments for macroscopic protoplanetesimals are possible. The
170: experiments described below have the following main scientific
171: objectives:
172: \begin{enumerate}
173: \item Under what physical conditions do high-porosity dust
174: agglomerates stick?
175: \item What are the possible outcomes of collisions between
176: high-porosity dust agglomerates?
177: \item What is the influence of projectile mass, velocity,
178: impact angle, local radius of curvature, and
179: porosity on the outcome of a collision?
180: \item What are the sticking efficiencies in mutual collisions
181: between protoplanetesimal, non-fractal dust agglomerates?
182: \item What are the coefficients of restitution in non-sticking
183: collisions?
184: \end{enumerate}
185:
186: \section{\label{sectsetup}Experimental Setup, Samples and Procedures}
187: For the experimental simulations of dust-dust collisions between
188: mm-sized and cm-sized dust agglomerates, we used monolithic
189: dust-agglomerate samples of 2.5 cm diameter and $\sim 1$ cm height
190: as targets and smaller fragments of such samples as projectiles.
191: All samples were formed by random ballistic deposition (RBD; see
192: paper I and \citet{blusch04} for details of the RBD process).
193: Table \ref{table1} summarizes the physical properties of the
194: particles and the resulting RBD agglomerates. Three monomer
195: particle types were used: (1) monodisperse $\rm SiO_2$ spheres
196: with 1.5 $\rm \mu m$ diameter, (2) quasi-monodisperse, irregular
197: diamond particles with $\sim 1.5$ $\rm \mu m$ diameter, and (3)
198: polydisperse, irregular $\rm SiO_2$ grains with diameters in the
199: range 0.1 - 10 $\rm \mu m$. The motivation for the use of these
200: three samples was less in their cosmochemical relevance for
201: protoplanetary dust but lay in their span of morphologies and size
202: distributions and the question how these morphologies influence
203: the outcome of the collisions. As was shown by \citet{popetal00},
204: the impact of the material on the sticking behavior is smaller
205: than the influence of the individual particle morphology. Our
206: experiments span a wide range in morphological parameters so that
207: we consider them relevant for protoplanetary collision processes.
208: \placetable{table1}
209:
210: It is interesting to note that the volume filling factor of the
211: agglomerates (see Table \ref{table1}), which describes the
212: fraction of volume filled with particles, is dependent on the
213: particle morphology and the width of the particle size
214: distribution (paper I), and varies by a factor of two between
215: monodisperse, spherical monomers ($\phi = 0.15$) and polydisperse,
216: irregular grains ($\phi = 0.07$). The broader the size
217: distribution, the fluffier the agglomerates are. For RBD
218: agglomerates consisting of spherical monodisperse particles, one
219: expects a volume filling factor of $\phi = 0.15$
220: \citep{vol59,watetal97} which is exactly found in our samples (see
221: paper I and \citet{blusch04}).
222:
223: To simulate the collisional history of protoplanetary dust
224: aggregates and to test the influence of the local target surface
225: curvature on the outcome of a collision, we also performed a
226: series of impact experiments into ``molded'' targets. These
227: targets, consisting of $1.5~\rm \mu m$ diameter $\rm SiO_2$
228: spheres with an initial volume filling factor of $\phi = 0.15$,
229: were slightly locally compacted by a half-spherical mold with 1 mm
230: radius so that they ultimately consisted of asperities with local
231: radii of curvature of 1 mm (representing, e.g., sticking
232: projectiles from previous impacts). X-ray tomography of the
233: ``molded'' targets showed that the ``hills'' consisted of
234: uncompressed aggregated dust ($\phi = 0.15-0.17$), while the
235: ``valleys'' were slightly compressed to a volume filling factor of
236: $\phi = 0.17-0.20$. Examples of an unprocessed and a processed
237: target and several projectile agglomerates are shown in Fig.
238: \ref{targetprojectiles}.
239:
240:
241:
242: For the realization of collisions between dusty targets and dusty
243: projectiles we developed a setup whose functionality is depicted
244: in Fig. \ref{setup}.
245:
246:
247: The experiments are performed under microgravity conditions within
248: a pressurized capsule which is, prior to each impact experiment,
249: held at the top of the microgravity drop tower in Bremen. Inside
250: the experimental setup, a set of five typically mm-sized
251: projectile agglomerates is held in a device which is able to
252: simultaneously release the projectiles when an electrical current
253: is applied to two solenoid magnets. The projectiles' release is
254: done while the experiment capsule is still held at ambient
255: gravitational acceleration. Due to the gravitational acceleration,
256: the released projectiles gain vertical velocities proportional to
257: their time of flight. After some preselected travel distance (or
258: time of flight), the experiment capsule is released to free fall.
259: Thus, the relative velocity between the projectiles and the target
260: is frozen and the impacts happen at a residual acceleration level
261: of less than $10^{-5} \rm ~m~s^{-2}$, mimicking the conditions in
262: protoplanetary disks. The free-fall height of the projectiles can
263: be adjusted from $\sim 5$ cm to $\sim 45$ cm. Due to a limited
264: acceleration length of the projectiles of $\lesssim 45$ cm, the
265: {\it maximum} impact velocity is $\sim 3 ~\rm m~s^{-1}$, while the
266: {\it actual} impact velocity is only determined by the time lapse
267: between projectile release and start of the free-fall phase of the
268: whole experiment. For the simulation of random impacts, the target
269: can be tilted by 30 and 60 degrees relative to the velocity vector
270: of the projectiles. For the exclusion of aerodynamic effects
271: during the collisions, the experiment chamber is evacuated to less
272: than 20 Pa pressure.
273:
274: During the free-fall time of 4.74 seconds, the impacts are
275: observed by a set of different cameras and illumination schemes
276: (see Fig. \ref{setup}): (1) A high-speed (462 frames per second
277: [fps]), high-resolution (1k $\times$ 1k pixels) camera with a
278: field of view (FOV) of $30 \times 30 ~ \rm mm^2$ is arranged such
279: that it observes the impacts tangential to the target surface; the
280: illumination is provided by a synchronized Xe flash lamp with
281: $\sim 1 ~\rm \mu s$ flash duration. Due to the limited resolution,
282: this camera can only detect particles $> 30 \rm \mu m$. (2)
283: Another high-speed (220 fps), low-resolution (256 $\times$ 256
284: pixels) camera with a FOV of $56 \times 56 ~ \rm mm^2$ can observe
285: the impacts perpendicular (in the case of non-normal impacts) or
286: almost perpendicular (in the case of normal impacts) to the target
287: surface; additional illumination is provided by a laser curtain
288: with 30 mm width and 1 mm thickness parallel and close to the
289: surface of the target; this camera-illumination combination was
290: specifically installed for detecting small fragments that are
291: otherwise invisible. (3) An additional low-speed (25 fps),
292: low-resolution (720 $\times$ 576 pixels) video camera with a FOV
293: of $52 \times 41 ~ \rm mm^2$, located 30 degrees from the target
294: normal, observes the target surface prior and after the impacts; a
295: point-source illumination for the determination of, e.g., crater
296: depths is provided by a halogen lamp located 32 degrees above the
297: target's ``horizon''.
298:
299: Data analysis consists of a thorough image analysis including the
300: determination of projectile sizes, impact velocities and impact
301: angles with respect to the local target normal, and the
302: determination of the outcomes of the collisions (sticking, rebound
303: or fragmentation) as well as parameters determining the effects
304: during the collisions. These encompass, among others, projectile
305: and fragment sizes, depths of intrusion, crater depths, and
306: fragment velocities.
307:
308: The variable parameters in the experiment are the projectile
309: velocity, the impact angle, the physical agglomerate properties
310: (see Table \ref{targetprojectiles}) and the projectile mass. Fig.
311: \ref{experimentalparametersa} gives an overview of the projectile
312: masses, the impact velocities and the impact angles for the three
313: agglomerate compositions described in Table
314: \ref{targetprojectiles} and impacts into unprocessed targets. For
315: a better statistical representation, the impact angle $\theta$
316: (relative to the target normal) is replaced by the squared sine of
317: the above-defined angle, $\sin^2(\theta)$, so that each
318: $\sin^2(\theta)$-interval has the same statistical probability for
319: random collisions. The original projectiles had sizes of typically
320: 1 mm and, thus, masses of $m_{\rm proj} \sim 10^{-6}$ kg. During
321: the release of these projectiles, a small fraction of their mass
322: fragmented off and in most cases also hit the target. This means
323: that we could observe, in addition to the original projectiles,
324: several collisions between projectiles of sub-mm sizes and the
325: target agglomerates. The lower limit of agglomerate masses of
326: $\sim 10^{-9}$ kg is due to the finite resolution of the cameras
327: (see above). The target masses are in the range $m_{\rm targ}
328: \approx 1.0 \ldots 1.5$ g so that $m_{\rm proj} \ll m_{\rm targ}$
329: for all impacts. Thus, our impact experiments are valid for all
330: target masses $m_{\rm targ} \gg m_{\rm proj}$ and are not
331: restricted to target sizes of centimeters. Impact velocities
332: ranged from $\sim 0.1~\rm m~s^{-1}$ to $\sim 3~\rm m~s^{-1}$, with
333: a slight systematic increase in velocity with increasing
334: projectile mass. This effect -- with a typical difference in
335: impact velocity between the largest and the smallest projectiles
336: of $\sim 0.5~\rm m~s^{-1}$ -- is due to some residual friction of
337: the travelling projectiles with the rarefied-gas atmosphere of
338: $\sim 20$ Pa pressure. The total number of microgravity
339: experiments performed for these experiments is 45. The experiments
340: were carried out in three series between November 2003 and October
341: 2004.
342:
343: Fig. \ref{experimentalparametersc} shows the parameter space of
344: the impact experiments of $\rm SiO_2$ dust aggregates with $\phi =
345: 0.15$ into ``molded'' targets with $\phi = 0.15-0.20$. These data
346: were collected in a drop-tower campaign comprising 9 flights in
347: April 2006. Due to the morphology of the target surface (see Fig.
348: \ref{targetprojectiles}), the variation of the impact angle is
349: irrelevant so that all impacts were carried out normal to the
350: target surface.
351:
352:
353: \section{\label{sectexpres}Experimental Results}
354:
355: \subsection{\label{stpr}Sticking Properties}
356: For dust aggregates consisting of irregular $\rm SiO_2$ and
357: diamond particles the sticking probability in the mass -- velocity
358: -- impact angle range is very close to unity (see Fig.
359: \ref{experimentalparametersa}), with only a few projectile
360: aggregates bouncing from the target. For dust aggregates
361: consisting of spherical, monodisperse $\rm SiO_2$ grains, i.e. for
362: the densest projectiles and targets, the distribution of the
363: sticking (open circles in Fig. \ref{experimentalparametersa}) and
364: non-sticking collision events (full circles in Fig.
365: \ref{experimentalparametersa}) in the parameter space is not
366: random. It is evident that the non-sticking collisions into the
367: flat targets occur preferentially for intermediate velocities of
368: $\sim 1-2~\rm m~s^{-1}$, for larger impact angles, and for more
369: massive projectiles. A determination of the sticking probability
370: as a function of projectile mass and the components of the impact
371: velocity normal and tangential to the target surface is shown in
372: Fig. \ref{stickingprobability}. Sticking probabilities were
373: derived by sliding averaging over 13 data points sorted in
374: projectile mass and normal/tangential impact velocity. From Figs.
375: \ref{experimentalparametersa} and \ref{stickingprobability} it is
376: clearly visible that (1) the sticking probability is $\beta = 1$
377: for aggregates with $m \lesssim 10^{-7}$ kg and falls steadily to
378: values $\beta \approx 0.5$ for the highest aggregate masses of $m
379: \approx 5\cdot10^{-6}$ kg, (2) for both, very low ($v \lesssim 0.5
380: \rm ~ m~s^{-1}$) and very high ($v \gtrsim 2 \rm ~ m~s^{-1}$)
381: normal impact velocities, the sticking probability is $\beta
382: \approx 1$ (however, mind that projectile velocity and mass are
383: not independent parameters and that the slowest impact velocities
384: stem from the smallest projectiles), whereas the sticking
385: probability drops to $\beta \approx 0.5$ for normal impact
386: velocities in the range $0.5 {\rm ~ m~s^{-1}} \lesssim v \lesssim
387: 1.5 {\rm ~ m~s^{-1}}$, (3) the sticking probability decreases
388: steadily for increasing tangential impact velocity and reaches
389: values as low as $\beta \approx 0.4$ for the highest tangential
390: velocities of $\sim 1.7~\rm m~s^{-1}$.
391:
392:
393:
394: For the ``molded'' and slightly compressed targets, sticking is
395: even the exception (see Fig. \ref{experimentalparametersc}), while
396: for the flat and fluffy targets most collisions result in mass
397: gain of the target. We find a sticking probability of $\beta
398: \approx 0.2$ when the target is slightly compacted and has local
399: radii of curvature comparable to the projectile radii. It was
400: observed that the projectiles stuck to the target only when they
401: accidentally hit a ``valley''.
402:
403: Fig. \ref{movie_examples} shows example movies into unprocessed
404: and ``molded'' target aggregates (online version only).
405:
406:
407:
408: \subsection{\label{sectmasstrans}Mass transfer in collisions}
409: Due to the low number of non-sticking collisions for the
410: agglomerates consisting of irregular monomers, we could not
411: perform a statistical analysis of the properties of the bouncing
412: agglomerates for non-spherical monomers. Thus, this and the
413: following subsections will mainly deal with the analysis of the
414: non-sticking collisions of agglomerates consisting of spherical,
415: monodisperse $\rm SiO_2$ particles into flat soft targets of the
416: same material.
417:
418: Fig. \ref{masstransferimage} displays an example of a collision in
419: which the impinging agglomerate bounced off after the collision.
420: From the comparison between the first and the last image of the
421: sequence (see inset) it is evident that the size of the projectile
422: agglomerate changed. We determined the mass ratio of the
423: projectile immediately after and before the impact by
424: \begin{equation}\label{masstransfer}
425: \mu = \frac{m^{\prime}}{m} \approx
426: \left(\frac{s^{\prime}_{\rm max} \cdot s^{\prime}_{\rm min}}
427: {s_{\rm max} \cdot s_{\rm min}}\right)^{3/2} \, ,
428: \end{equation}
429: in which $s^{\prime}_{\rm max}$, $s^{\prime}_{\rm min}$, $s_{\rm
430: max}$, and $s_{\rm min}$ denote the maximum and minimum linear
431: extension of the agglomerate after and before the impact,
432: respectively. Fig. \ref{masstransferhistogram} shows the
433: distribution of the derived mass ratios. Although the statistics
434: is still somewhat poor, it can be seen that the mass transfer
435: between target and projectile agglomerate can be considerable and
436: obtains mostly values in the range $\mu = 1 \ldots 4$. The mean
437: mass ratio is $\bar{\mu}=2.1$. This means that on average a
438: non-sticking impact leads to a considerable mass loss of the (more
439: massive) target agglomerate, a process which has not been
440: considered before.
441:
442:
443: \subsection{\label{cor}Coefficient of restitution and energy loss}
444: In the case of non-sticking the kinetic energy is not fully
445: absorbed within the projectile and target agglomerates. A usual
446: method to describe the amount of plasticity in a collision is by
447: using the coefficient of restitution, defined by
448: \begin{equation}\label{coefficientofrestitution}
449: \epsilon = \frac{v^{\prime}}{v} \, .
450: \end{equation}
451: Here $v^{\prime}$ and $v$ denote the relative velocity between the
452: projectile and the target after and before the collision. Fig.
453: \ref{coefficientofrestitutionfig} shows the coefficients of
454: restitution for all 18 non-sticking impacts of agglomerates
455: consisting of $\rm SiO_2$ spheres as a function of impact velocity
456: and projectile mass. The data points are scattered between
457: relative low $\epsilon < 0.1$ and rather high values $\epsilon >
458: 0.4$ with no apparent dependence of the coefficient of restitution
459: on the impact velocity and on the projectile mass. The mean value
460: of the coefficient of restitution is $\bar{\epsilon} = 0.20$ and
461: the root mean square value is $\sqrt{\bar{\epsilon^2}} = 0.16$.
462:
463:
464:
465:
466: The coefficient of restitution plays an important role in the
467: dense dust-dominated subdisk or inside condensations caused by the
468: streaming instability, in which mutual collisions among the dust
469: aggregates can act as ``cooling'' \citep{johetal07} . For the
470: physical processes inside colliding dust aggregates, it is,
471: however, more interesting to consider the ratio of total kinetic
472: energy after and before the collision (taking into account the
473: mass transfer from target to projectile) as a function of the
474: squared impact parameter. Fig. \ref{energyloss} shows this data.
475: Also plotted in Fig. \ref{energyloss} is the linear relation
476: \begin{equation}
477: \label{transrot}
478: \frac{E^{\prime}_{\rm kin}}{E_{\rm kin}} = \epsilon^2(0) +
479: \epsilon^2(1) \cdot \sin^2{\theta}
480: \end{equation}
481: which is for $\epsilon^2(0) = 0$ (perfectly inelastic central
482: collisions) and $\epsilon^2(1) = (5/7)^2 = 0.51$ (pure frictional
483: transition from translational to rotational motion and no
484: plasticity for glancing collisions) an upper limit for
485: dust-aggregate collisions \citep{blumue93}. It is evident that our
486: data fall much below the upper limit given by Eq. \ref{transrot}.
487: This means that plasticity does not only play a role for the
488: normal but also for the tangential component of the collision,
489: i.e. the shear strength of the aggregate material is overcome
490: during the impacts. However, the data in Fig. \ref{energyloss}
491: shows an increasing trend of $\frac{E^{\prime}_{\rm kin}}{E_{\rm
492: kin}}$ with increasing $\sin^2{\theta}$, which was also observed
493: by \citet{blumue93}, i.e. the total amount of plasticity decreases
494: with increasing impact parameter.
495:
496:
497: \subsection{Semi-elastic rebound and tensile strength}
498: Even for the cases in which the projectile agglomerates stuck to
499: the target after the collision, the total kinetic energy was
500: initially not fully dissipated into plastic deformation of the
501: aggregates. An example is given in Fig. \ref{semielasticrebound}.
502: After impacting the target, the projectile (consisting of
503: irregular $\rm SiO_2$ particles) is rebounding but is not able to
504: escape from the target due to a too strong inter-particle
505: attraction. Fig. \ref{reboundingparticle} shows the temporal
506: dependence of the displacement of the projectile from its deepest
507: penetration during the rebounding phase of its trajectory.
508:
509:
510:
511:
512: The example in Fig. \ref{semielasticrebound} also demonstrates how
513: the mass transfer described in Sect. \ref{sectmasstrans} works: on
514: the way out of the target, the projectile gets decelerated (see
515: Fig. \ref{reboundingparticle}) by the adhesion forces between the
516: projectile and the target. As long as the adhesion is larger than
517: the tensile strength (see below), the projectile is attached to
518: the target and drags target material along. In the cases of
519: non-sticking, the tensile strength of the macroscopic dust
520: aggregate (see Table \ref{table1}) is overcome with some of the
521: target material sticking to the projectile agglomerate. It is
522: clear that the tensile strength is almost reached by the
523: rebounding particle in Fig. \ref{semielasticrebound}. Thus, we can
524: use the data in Fig. \ref{reboundingparticle} to estimate the
525: dynamic tensile strength of the high-porosity agglomerate and
526: compare these value to the static measurements of the same
527: material. We can fit a parabolic function
528: \begin{equation}\label{parabola}
529: l(t) = \frac{1}{2} ~ a_0 ~ t^2 + v_0 ~t \, ,
530: \end{equation}
531: to the first three data points in Fig. \ref{reboundingparticle}
532: and get $a_0 = -185 \, \rm m~s^{-2}$ and $v_0 = 0.76 \, \rm
533: m~s^{-1}$ for the maximum initial acceleration and velocity of the
534: rebounding projectile (solid curve in Fig.
535: \ref{reboundingparticle}). In addition to that, we fit a decaying
536: exponential function of the form
537: \begin{equation}\label{exponential}
538: l(t) = l_0 ~ \cdot ~ \left(1-\exp\left[-\frac{t}{\tau}\right]\right) \, ,
539: \end{equation}
540: with $l_0 = 2.2 \cdot 10^{-3}$ m and $\tau = 3.1 \cdot 10^{-3}$ s,
541: to all data points shown in Fig. \ref{reboundingparticle} (dashed
542: curve). Both functions are not motivated by physical
543: considerations but only help to estimate the initial acceleration.
544: Differentiations of the function in Eq. \ref{exponential} yield
545: the initial velocity and acceleration of $v_0 = 0.71 \, \rm
546: m~s^{-1}$ and $a_0 = -227 \, \rm m~s^{-2}$. With an estimated
547: cross section of the particle of $A = 1.5 \cdot 10^{-6} \, \rm
548: m^2$ and an estimated mass of the rebounding agglomerate of $m =
549: 1.7 \cdot 10^{-6}$ kg (including the mass transfer from the target
550: to the impinging projectile), we get for the lower limit of the
551: tensile strength, $T \stackrel{>}{\sim} m a_0 / A$, values of
552: \begin{equation}
553: \label{tensile1}
554: T \stackrel{>}{\sim} 210 \, \rm N~m^{-2}
555: \end{equation}
556: for the parabolic function and of
557: \begin{equation}
558: \label{tensile2}
559: T \stackrel{>}{\sim} 260 \, \rm N~m^{-2}
560: \end{equation}
561: for the decaying exponential function, respectively. These values
562: are in good agreement with the static values $T = 300 \, \rm
563: N~m^{-2}$ (see paper I and Table \ref{table1}).
564:
565: \subsection{Intrusion and crater depths}
566: For those impact events in which the projectiles stuck to the
567: target agglomerate we could determine the depth of intrusion
568: $d_{\rm i}$ perpendicular to the surface. It turned out that the
569: depth of intrusion was generally larger for higher impact
570: energies. Fig. \ref{doienergy} shows the data for all projectile
571: and target materials as a function of the kinetic impact energy of
572: the projectiles.
573:
574: In order to penetrate into the target agglomerate, the projectiles
575: need to overcome the compressive strength of the target. The data
576: in Fig. \ref{doienergy} suggest that a threshold energy $E_{\rm
577: min}$ is required to yield a finite penetration depth. For the
578: determination of this value, we assume a linear dependence between
579: the depth of intrusion and the logarithm of the impact energy, as
580: suggested by the data in Fig. \ref{doienergy}. A least squares fit
581: of
582: \begin{equation}\label{intrusion2}
583: d_{\rm i} = x \log\left(\frac{E_{\rm kin}}{E_{\rm
584: min}}\right)\,
585: \end{equation}
586: (for $E_{\rm kin} \ge E_{\rm min}$) results in $x = 0.54~\rm mm$
587: and $E_{\rm min} = 3.1 \cdot 10^{-9} ~ \rm J$ for spherical $\rm
588: SiO_2$, $x = 0.47~\rm mm$ and $E_{\rm min} = 2.8 \cdot 10^{-10} ~
589: \rm J$ for irregular diamond, and $x = 0.61~\rm mm$ and $E_{\rm
590: min} = 8.0 \cdot 10^{-10} ~ \rm J$ for irregular $\rm SiO_2$,
591: respectively (solid lines in Fig. \ref{doienergy}).
592:
593:
594:
595:
596: If we only consider the projectiles whose energies are close to
597: the threshold energy $E_{\rm min}$, we find that their masses are
598: $m \approx 10^{-9}$ kg. With aggregate densities of 300 $\rm
599: kg~m^{-3}$, 390 $\rm kg~m^{-3}$, and 182 $\rm kg~m^{-3}$ for
600: spherical $\rm SiO_2$, diamond, and irregular $\rm SiO_2$, we get
601: typical projectile volumes at the onset of intrusion of $V = 3.3
602: \times 10^{-12}~\rm m^3$, $V = 2.6 \times 10^{-12}~\rm m^3$, and
603: $V = 5.5 \times 10^{-12}~\rm m^3$, respectively. Using the
604: above-derived minimum impact energies for intrusion, $E_{\rm
605: min}$, we can derive the critical impact pressure $p_{\rm min}
606: =E_{\rm min} / V = 940$ Pa, $p_{\rm min}=100$ Pa, and $p_{\rm
607: min}=150$ Pa, respectively, which is remarkably close to the
608: static compressive strengths of the respective target aggregates
609: (see Table \ref{table1}). Thus, for the onset of penetration, a
610: minimum impact pressure equivalent to the compressive strength is
611: required.
612:
613: In addition to the depth of intrusion, we also derived for the
614: non-sticking events the crater depths. This was either done by a
615: determination of the shadow lengths inside the craters for oblique
616: illumination or by a measurement of the length of the escaping
617: projectile agglomerate. The crater depth is then approximated by
618: the difference of the length of the escaping projectile (which is,
619: due to the mass transfer between target and projectile, larger
620: than before the impact; see Sect. \ref{sectmasstrans}) and the
621: length of the part of the projectile protruding out of the target
622: at closest approach. A comparison of the crater depth and the
623: depth of intrusion for the case of spherical $\rm SiO_2$ monomers
624: (only in that case we have sufficiently many rebounding
625: projectiles) shows that the crater depths are a factor $\sim 3$
626: larger than the depths of intrusion. However, the mass transfer is
627: less than a factor of $\sim 3$ due to only a partial penetration
628: of the projectiles.
629:
630: \subsection{Impact fragmentation and the
631: Influence of Local Surface Curvature and Density Enhancement} In a
632: few cases, the impinging projectile agglomerate fragmented upon
633: impact. Part of the residual fragments left the target
634: agglomerate, the other fragments stuck to the target. Most
635: fragmentation events occurred when the projectile either hit a
636: previously captured projectile or another surface irregularity
637: with low radius of curvature. Thus, we systematically investigated
638: the influence of the local radius of curvature of the target
639: aggregate on the outcome of a collision. For this, we used
640: previously ``molded'' target aggregates (see Sect. \ref{sectsetup}
641: and Fig. \ref{targetprojectiles}c) and made a series of impacts of
642: mm-sized, high-porosity projectile aggregates.
643:
644: As already mentioned in Sect. \ref{stpr} and visible in Fig.
645: \ref{experimentalparametersc}, the sticking probability is
646: considerably reduced by the surface sculpting. Only 5 out of 25
647: projectiles stuck to the target, and sticking was apparently
648: restricted to those cases in which the projectile hit a ``valley''
649: on the target. In contrast to that, the fragmentation efficiency,
650: while negligible for impacts into smooth flat targets, increases
651: to 24\% (6 projectiles). However, most of the projectiles (14)
652: were semi-elastically rebounding after the impact, which makes the
653: rebound probability with 56\% the most likely single collisional
654: outcome for the impacts into ``molded'' targets.
655:
656: It is clear from these experiments that, besides impact velocity,
657: impact angle, aggregate mass and aggregate packing density, the
658: local radius of curvature and, thus, the collisional history of
659: protoplanetary dust aggregates plays an important role for the
660: outcome of mutual collisions.
661:
662: \section{\label{sectdisc}Discussion: Impact Dynamics of
663: High-Porosity Dust Agglomerates} From the results presented in the
664: previous section, we can derive some fundamental (although
665: approximate and preliminary) dynamical properties of high-porosity
666: dust aggregates. If we assume for simplicity that the aggregates
667: are fully elastic during collisions, we can use the well-known
668: Hertzian equation of motion \citep{hertz1882} for the collision
669: between a spherical particle of radius $s$ and mass $m$ and an
670: infinitely large sphere
671: \begin{equation}\label{hertz1}
672: m \frac{{\rm d}^2 \delta}{{\rm d} t^2} + \frac{4}{3} s^{1/2}
673: E^* \delta^{3/2},
674: \end{equation}
675: with $E^*$ being the elasticity parameter, defined by
676: \begin{equation}\label{hertz2}
677: E^* = \frac{1}{2} \frac{E}{1 - \nu^2}
678: \end{equation}
679: for like materials. Here, $E$ and $\nu$ are the modulus of
680: elasticity and Poisson number of the materials, respectively. As
681: our aggregates very likely have Poisson numbers $\nu \approx 0$
682: (i.e. they are highly compressive), Eq. \ref{hertz2} reduces to
683: \begin{equation}\label{hertz3}
684: E^* = \frac{1}{2} E .
685: \end{equation}
686:
687: With Eq. \ref{hertz3}, Eq. \ref{hertz1} can be solved for the
688: maximum penetration
689: \begin{equation}\label{hertz4}
690: \delta_{\rm max} = \left( \frac{15 m v^2}{8 s^{1/2} E} \right)^{2/5} ,
691: \end{equation}
692: when $v$ is the initial collision velocity. If we identify
693: $\delta_{\rm max}$ with the measured intrusion of the projectiles,
694: $d_{\rm i}$, and plot these double-logarithmically as a function
695: of the parameter $\frac{m v_{\rm n}^2}{s^{1/2}}$, motivated by the
696: functionality in Eq. \ref{hertz4}, we see that we can indeed find
697: a slope close to 2/5 as predicted by Eq. \ref{hertz4} (Fig.
698: \ref{hertzplot}).
699:
700: The two straight lines in Fig. \ref{hertzplot} formally give
701: $E=430$ Pa and $E=17,800$ Pa. Taking the geometric mean of these
702: two values, the agglomerates penetrate as deep as if they had a
703: modulus of elasticity of $\sim 2,800$ Pa. As the collisions are
704: clearly dominated by plasticity, we expect the compressive
705: strength to be of the same order as the modulus of elasticity,
706: i.e. $C \approx 2,800$ Pa. A comparison with Fig. 4 from paper I
707: shows that such compressions should lead to a compaction of $\phi
708: \approx 0.2$. Mind, however, that $E$ in Eq. \ref{hertz4} needs
709: not to be constant. As was shown by \citet{blusch04}, the
710: compressive stress $p_{\rm c}$ and the volume filling factor
711: $\phi$ are related through $p_{\rm c} \propto (\phi -
712: \phi_0)^{\beta}$ for $\phi_0 \le \phi \lesssim 0.22$, with $\phi_0
713: = 0.15$ and $\beta = 0.8$. Thus, we expect a similar relation
714: between $E$ and $\phi$ to be existent.
715:
716: If we define the crater volume by
717: \begin{equation}
718: V = p ~ d_{\rm i}^2 ~ (s - \frac{d_{\rm i}}{3})
719: \label{cratervolume1}
720: \end{equation}
721: for $d_{\rm i}<s$ and
722: \begin{equation}
723: V = p ~ s^2 ~ d_{\rm i} ~ - \frac{\pi}{3} ~ s^3
724: \label{cratervolume}
725: \end{equation}
726: for $d_{\rm i}\geq s$, with $s$ being the radius of the
727: projectile, we can approximately derive the dynamic impact
728: pressure $p_{\rm dyn}$ for fully plastic collisions
729: \citep{johnson1985}, defined by
730: \begin{equation}
731: \label{impactpressure}
732: p_{\rm dyn} = \frac{E_{\rm kin}}{V} .
733: \end{equation}
734: In Fig. \ref{cvenergy}, we plotted the crater volume as a function
735: of the normal component of the impact energy, $E_{\rm kin,n}$. It
736: is evident that there is a strong correlation between these two
737: quantities. We fit a power-law $V \propto E_{\rm kin,n}^{\alpha}$
738: to the data in Fig. \ref{cvenergy}, with $\alpha = 0.75 \pm 0.03$.
739: Thus, the dynamic impact pressure slightly increases with impact
740: energy from $\sim 200$ Pa to $\sim 2,000$ Pa for the impact energy
741: range between $\sim 3 \times 10^{-9}$ J and $\sim 2 \times
742: 10^{-5}$ J (see Fig. \ref{pressureenergy}). The slightly
743: increasing values of $p_{\rm dyn}$ towards larger impact energies
744: is probably caused by the compaction of the aggregate volume
745: during impact. For the minimum impact energy for which intrusion
746: was found (see above), $E_{\rm min} = 3.1 \cdot 10^{-9} ~ \rm J$,
747: a minimum dynamic pressure of $\sim 300$ Pa is required to cause
748: the formation of a crater and, thus, compaction of the agglomerate
749: material in the contact zone between projectile and target
750: agglomerate. This pressure is very close to the minimum stress
751: under which the material yields (see Fig. 4 in paper I).
752:
753:
754:
755:
756: With this data, we can also explain the mass transfer observed in
757: many non-sticking collisions (see Sect. \ref{sectmasstrans} and
758: Figs. \ref{masstransferimage} and \ref{masstransferhistogram}).
759: During the impact, the projectile and part of the target volume
760: are slightly compacted. As was shown in paper I, aggregate
761: compaction increases the tensile strength of the material. Thus,
762: the compacted parts in the collisional volume have higher inner
763: cohesion so that they are more likely to keep sticking together
764: after the impact.
765:
766: \section{\label{sectsummary}Summary}
767: We performed oblique impact experiments of typically 0.2-3 mm
768: diameter projectile agglomerates into 2.5 cm diameter target
769: agglomerates of identical composition. We used three different
770: particle types (monodisperse spherical $\rm SiO_2$ with 1.5 $\rm
771: \mu m$ diameter, irregular diamond with 1-2 $\rm \mu m$ diameter,
772: irregular $\rm SiO_2$ with 0.1-10 $\rm \mu m$ diameter) and
773: produced agglomerates by the random ballistic deposition process
774: \citep{blusch04}. Depending on the particle type, the agglomerates
775: had volume filling factors of $\phi = 0.15$ (for monodisperse $\rm
776: SiO_2$), $\phi = 0.11$ (for diamond), and $\phi = 0.07$ (for
777: irregular $\rm SiO_2$), respectively. Impact velocities ranged
778: between 0.1 and 3 m/s.
779:
780: From the results of our experimental investigation presented in
781: the previous sections, we can draw the following conclusions:
782: \begin{enumerate}
783: \item For a similar distribution of projectile masses, impact
784: velocities and impact angles, the volume filling factor (porosity)
785: of projectile and target has a considerable influence on the
786: collisional outcome. Whereas very porous dust aggregates
787: ($\phi \lesssim 0.10$) almost always stick, the sticking
788: probability decreases with increasing filling factor for $\phi
789: \gtrsim 0.15$ (Figs. \ref{experimentalparametersa} -
790: \ref{stickingprobability}).
791:
792: \item For dust aggregates with volume filling factors
793: $\phi \gtrsim 0.15$, projectiles with diameters $\gtrsim 1$ mm do only
794: stick in near-central collisions, while near-grazing impacts
795: lead to a rebound of the projectile (see Fig.
796: \ref{experimentalparametersa}). The reason for the mass
797: dependence of the sticking behavior
798: is the interplay between the energy of the bouncing
799: aggregates (which is, due to an almost mass-independent
800: coefficient of restitution (see Fig.
801: \ref{coefficientofrestitutionfig}), proportional to the aggregate mass)
802: and the contact energy (which is roughly proportional to the
803: aggregates' cross section). For small aggregates, the inertial
804: energy of the rebounding aggregates is not capable of breaking
805: the contacts, whereas large aggregates easily bounce off. For
806: large aggregates, central collisions lead to a deeper
807: penetration of the projectile aggregate into the target and
808: thus to a larger contact area with stronger binding forces.
809:
810: \item Collisions which do not lead to sticking between projectile
811: and target agglomerate result, on average, in a mass transfer
812: from the larger to the smaller collision partner, i.e. to a
813: mass loss of the target agglomerate (see
814: Figs. \ref{masstransferimage} and \ref{masstransferhistogram}). This can be explained by an
815: impact compaction of material comprising the projectile and
816: part of the target and a resulting higher cohesion (tensile strength) of the
817: compacted material.
818:
819: \item In the case of non-sticking, most of the kinetic energy
820: of the projectile is dissipated in the collision. The residual
821: energies of the bouncing projectiles (including the
822: mass-transfer effect) are very small for near-central collisions
823: and increase with increasing impact
824: angle up to a few ten percent of the pre-collision energy
825: (see Fig. \ref{energyloss}).
826:
827: \item The tensile strength of the agglomerates made of
828: irregular $\rm SiO_2$ particles, as estimated in one
829: experiment in which the projectile almost rebound after a
830: collision (see Fig. \ref{semielasticrebound}), $T \gtrsim 260$ Pa,
831: is in good agreement with the static measurements
832: (see Table \ref{table1}). The tensile strength is responsible
833: for aggregate cohesion and sticking upon impact.
834:
835: \item If we identify the compressive strength of the agglomerates
836: made of spherical $\rm SiO_2$ particles with the estimated
837: dynamic impact pressure (see Eq. \ref{impactpressure}
838: and Fig. \ref{pressureenergy}), we see that the compressive
839: strength increases with impact energy, as was already
840: indicated by the static measurements in paper I. For impact
841: velocities $\lesssim 3\, \rm m s^{-1}$ and projectile sizes
842: $\sim 1$ mm, we get a maximum compressive strength of $C
843: \approx 2,000$ Pa.
844:
845: \item Above a threshold value for the impact energy, $E_{\rm min}
846: \approx 3\cdot 10^{-10} \ldots 3\cdot 10^{-9}$ J,
847: a crater is formed. The crater volume scales with the normal
848: component of the impact
849: energy as $V \propto E_{\rm kin,n}^{0.75}$.
850: The corresponding
851: dynamical pressure, $p_{\rm dyn} = E_{\rm kin}/V$ is
852: only weakly dependent on impact energy, $p_{\rm dyn} \propto
853: E_{\rm kin}^{0.25}$ and ranges from $\sim 200$ Pa to
854: $\sim 2,000$ Pa in our
855: experiments. The threshold value for crater formation of
856: $p_{\rm dyn} \approx 200$ Pa corresponds well with the onset of
857: compaction measured by \citet{blusch04} and paper I.
858: The largest
859: crater volumes were found for the highest impact energies
860: close to the transition between sticking and non-sticking.
861: These crater volumes (see Fig. \ref{cvenergy}) are typically
862: $V = 10^{-8} ~ \rm m^3$, Comparison with the masses of the
863: largest (and, thus, most energetic) projectile agglomerates
864: and the projectile
865: densities (see data in Table \ref{table1}) show that the
866: corresponding projectile volumes are also $V_{\rm proj}
867: \approx 10^{-8} ~ \rm m^3$. Thus, the amount of compaction
868: of the target and projectile agglomerates in these cases is
869: not negligible. From \citet{blusch04} we can see that at
870: a compression of $2,000$ Pa, the volume filling
871: factor increases from $\phi = 0.15$ to $\phi = 0.20$.
872: If these static compaction
873: measurements are transferrable to the dynamic problem,
874: we conclude that a volume
875: at least as large as twice the projectile's
876: is compressed to $0.20/0.15 = 1.33$
877: of its previous volume filling factor.
878:
879: \item Impact fragmentation was only rarely observed for impacts
880: into flat and high-porosity targets. When the target surface
881: was artificially roughened with a local radius of curvature
882: of 1 mm (which resulted in a slight compaction of the target
883: agglomerate to $\phi \approx 0.2$), the sticking
884: probability was considerably reduced,
885: and bouncing and fragmentation were the dominating processes.
886: Earlier experiments by \citet{blumue93} also found impact
887: fragmentation when two similar-sized dust aggregates collide at
888: velocities of a few m/s. Thus, the local radius of
889: curvature of two colliding dust aggregates plays a dominating
890: role in the outcome of the collision.
891: \end{enumerate}
892:
893: \section{\label{sectconclusion}Conclusions}
894: In paper I, we have shown that macroscopic protoplanetary dust
895: aggregates are expected to be very porous. Depending on the
896: collisional history and the size of the aggregates, we expect the
897: dusty objects to have volume filling factors between $\phi \approx
898: 0.1$ (for all objects whose collision velocities never exceeded
899: $\sim 1~\rm m s^{-1}$, i.e. for sizes $\lesssim$ cm) and $\phi
900: \approx 0.3$ (for all objects with sizes $\gtrsim 1$ m). In this
901: paper, we confirmed that collisions among fluffy protoplanetary
902: dust aggregates with velocities $\sim 1~\rm m s^{-1}$ lead to
903: impact pressures of $\lesssim 2,000$ Pa and, thus, to a moderate
904: increase in volume filling factor. Collisions between
905: (sub-)mm-sized dust aggregates and cm- to dm-sized fluffy objects
906: are very abundant in the solar nebula and the impact velocities
907: range around $1~\rm m s^{-1}$ \citep{weicuz93}. Thus, our
908: microgravity experiments match the solar-nebula conditions very
909: closely and are directly applicable to growth models.
910:
911: We find that the sub-mm sized dust aggregates always stick to much
912: larger target aggregates, independent of impact velocity and
913: impact angle. The larger, mm-sized aggregates, however, behave
914: differently. Sticking is restricted to the higher velocities and
915: to the smaller impact angles. This trend was found for dust
916: aggregates consisting of monodisperse spherical particles, for
917: quasi-monodisperse irregular particles as well as for irregular
918: monomer particles with a wide size distribution, although
919: quantitative differences in the sticking probabilities exist.
920: Thus, for random impacts in the solar nebula, the sticking
921: probability in collisions between mm-sized dust aggregates and
922: cm-dm sized dusty bodies is below unity. Moreover, non-sticking
923: (and even fragmentation) is favored when the local radius of
924: curvature at the point of impact of the larger body is similar to
925: the projectile radius. This can be the case for collisions between
926: similar-size dust aggregates or between projectiles and targets
927: with irregular, non-flat surface textures. Another interesting
928: feature we found in our experiments is that non-sticking impacts
929: (i.e. those with higher impact angles) lead to a mass transfer
930: from the larger to the smaller body.
931:
932: As a consequence, future growth models for protoplanetary dust
933: should take into account that the outcome of a single collision
934: between dust aggregate A and dust aggregate B is not only
935: dependent on the size of A and B and the mutual collision
936: velocity, but is also influenced by the collisional history (e.g.
937: the distribution of local radii of curvature on the surfaces of A
938: and B; compaction) and the (random) impact angle of the particular
939: collision. In addition to that, due to the possible occurrence of
940: mass transfer from a larger body A to a smaller body B (in
941: bouncing collisions) or fragmentation of the smaller object B
942: (e.g. when it hits a surface part of A with a small local radius
943: of curvature), the numerical description of the collisional
944: outcome (and its use in e.g. Smoluchowski's growth equation)
945: becomes rather complex.
946:
947: From our experimental findings, we consider it rather unlikely
948: that protoplanetary bodies can grow beyond dm-sizes in a direct
949: and simple hit-and-stick manner. Although our experiments suggest
950: that small projectiles stick at higher velocities than large
951: projectiles, this trend does not imply that km-sized bodies can
952: form by the accumulation of very small dust aggregates or single
953: particles. Recent experiments by Schr\"apler \& Blum (unpublished)
954: show that at impact velocities $\gtrsim 15~\rm m~s^{-1}$ dust
955: aggregates cannot grow by the accumulation of single grains (and,
956: therefore, also not by the accumulation of small dust aggregates).
957: However, m-sized bodies in protoplanetary disks possess relative
958: velocities with respect to small grains of the order of $50~\rm
959: m~s^{-1}$ \citep{weicuz93}. Moreover, experiments by
960: \citet{wuretal05a} suggest that impacts between dusty projectiles
961: into fluffy dusty targets around $\sim 10~\rm m s^{-1}$ never lead
962: to an accumulation of mass on the larger body but lead to strong
963: fragmentation of the projectile and cratering (i.e. mass loss) of
964: the target. A possible way out of this dilemma could be the
965: indirect effect of projectile or fragment capturing by aerodynamic
966: \citep{wuretal01a,wuretal01b} or electrostatic \citep{blu04}
967: forces or by gravitational collapse in locally overdense regions
968: in the midplane of protoplanetary disks \citep{johetal07}.
969:
970: \acknowledgments
971: We are indebted to the German Space Agency DLR
972: for supporting this work (grant no. 50 WM 0336) and providing us
973: with the drop tower flights. We thank the staff at the ZARM drop
974: tower facility for their help and hospitality during our
975: campaigns.
976:
977: \begin{thebibliography}{99}\label{sec:TeXbooks}
978:
979: \bibitem[Blum \& M\"unch (1993)]{blumue93} Blum, J., \&
980: M\"unch, M.\ 1993, \icarus, 106, 151.
981:
982: \bibitem[Blum et al.(1998)]{bluetal98} Blum, J., Wurm, G.,
983: Poppe, T., \& Heim, L.-O. 1998, Earth, Moon and Planets, 80, 285
984:
985: \bibitem[Blum \& Wurm (2000)]{bluwur00} Blum, J., \& Wurm, G.\
986: 2000, \icarus, 143, 138
987:
988: \bibitem[Blum et al.(2000)]{bluetal00} Blum, J., et al.
989: 2000, \prl, 85, 2426
990:
991: \bibitem[Blum(2004)]{blu04} Blum, J.\ 2004, in ASP Conf. Ser.
992: 309, Astrophysics of Dust, ed. A. N.Witt, G. C. Clayton, \& B. T.
993: Draine (San Francisco: ASP), 369
994:
995: \bibitem[Blum \& Schr\"apler(2004)]{blusch04} Blum, J., \&
996: Schr{\"a}pler, R.\ 2004, \prl, 93, 115503
997:
998: \bibitem[Blum et al. (2006)]{bluetal06} Blum, J., Schr{\"a}pler,
999: R., Davidsson, B.~J.~R., \& Trigo-Rodr{\'{\i}}guez, J.~M.\ 2006,
1000: \apj, 652, 1768 (paper I)
1001:
1002: \bibitem[Dominik \& Tielens(1997)]{domtie97} Dominik, C., \&
1003: Tielens, A.~G.~G.~M. 1997, \apj, 480, 647
1004:
1005: \bibitem[Harker et al.(2005)]{haretal05} Harker, D.~E., Woodward,
1006: C.~E., \& Wooden, D.~H.\ 2005, Science, 310, 278
1007:
1008: \bibitem[Heim et al.(1999)]{heietal99} Heim, L.-O., Blum, J.,
1009: Preuss, M., \& Butt, H.-J.\ 1999, \prl, 83, 3328
1010:
1011: \bibitem[Hertz(1882)]{hertz1882} Hertz, H.\ 1882, J. Reine Angew.
1012: Mathematik, 92, 156
1013:
1014: \bibitem[H\"orz et al.(2006)]{hoeetal06} H{\"o}rz, F., et al.\
1015: 2006, Science, 314, 1716
1016:
1017: \bibitem[Johansen et al.(2007)]{johetal07} Johansen, A., Oishi,
1018: J.~S., Mac Low, M.-M., Klahr, H., Henning, T., \& Youdin, A. 2007,
1019: Nature, 448, 1022
1020:
1021: \bibitem[Johnson et al.(1985)]{johnson1985} Johnson, K.~L.\ 1985,
1022: Contact Mechanics (Cambridge: Cambridge University Press)
1023:
1024: \bibitem[Keller et al.(2005)]{keletal05} Keller, H.~U., et al.\
1025: 2005, Science, 310, 281
1026:
1027: \bibitem[Kerridge (1993)]{ker93} Kerridge, J.~F.\ 1993,
1028: Icarus, 106, 135
1029:
1030: \bibitem[Kessler-Silacci et al. (2006)]{kesetal06} Kessler-Silacci, J.,
1031: et al.\ 2006, \apj, 639, 275
1032:
1033: \bibitem[Krause \& Blum(2004)]{krablu04} Krause, M., \& Blum, J.\
1034: 2004, \prl, 93, 021103
1035:
1036: \bibitem[Ormel et al. (2007)]{ormetal07} Ormel, C.~W., Spaans, M.,
1037: \& Tielens, A.~G.~G.~M.\ 2007, \aap, 461, 215
1038:
1039: \bibitem[Paszun \& Dominik(2006)]{pasdom06} Paszun, D., \&
1040: Dominik, C.\ 2006, Icarus, 182, 274
1041:
1042: \bibitem[Poppe et al.(2000)]{popetal00} Poppe, T., Blum, J.,
1043: \& Henning, T.\ 2000, \apj, 533, 454
1044:
1045: \bibitem[Poppe \& Schr\"apler(2005)]{popsch05} Poppe, T., \&
1046: Schr{\"a}pler, R.\ 2005, \aap, 438, 1
1047:
1048: \bibitem[Przygodda et al.(2003)]{przetal03} Przygodda, F., van
1049: Boekel, R., {\`A}brah{\`a}m, P., Melnikov, S.~Y., Waters,
1050: L.~B.~F.~M., \& Leinert, C.\ 2003, \aap, 412, L43
1051:
1052: \bibitem[Vold(1959)]{vol59} Vold, M.~J.\ 1959, J. Colloid Interface Sci., 14,
1053: 168
1054:
1055: \bibitem[Watson et al.(1997)]{watetal97} Watson, P.K., Mizes, H.,
1056: Castellanos, A., \& P\'erez, A. 1997, in Powders \& Grains 97, ed.
1057: R. Behringer \& J. T. Jenkins (Rotterdam: A. A. Balkema), 109
1058:
1059: \bibitem[Weidenschilling \& Cuzzi(1993)]{weicuz93} Weidenschilling,
1060: S.~J., \& Cuzzi, J.~N.\ 1993, in Protostars and Planets III, ed.
1061: E. H. Levy \& J. I. Lunine (Tucson: Univ. Arizona Press), 1031
1062:
1063: \bibitem[Wurm \& Blum(1998)]{wurblu98} Wurm, G., \& Blum, J.\
1064: 1998, \icarus, 132, 125
1065:
1066: \bibitem[Wurm et al. (2001a)]{wuretal01a} Wurm, G., Blum, J., \&
1067: Colwell, J.~E.\ 2001a, \icarus, 151, 318
1068:
1069: \bibitem[Wurm et al. (2001b)]{wuretal01b} Wurm, G., Blum, J., \&
1070: Colwell, J.~E.\ 2001b, \pre, 64, 046301
1071:
1072: \bibitem[Wurm et al.(2005a)]{wuretal05a} Wurm, G., Paraskov, G., \&
1073: Krauss, O.\ 2005a, \pre, 71, 021304
1074:
1075: \bibitem[Wurm et al.(2005b)]{wuretal05b} Wurm, G., Paraskov, G., \&
1076: Krauss, O.\ 2005b, Icarus, 178, 253
1077:
1078: \end{thebibliography}
1079:
1080: \setlength{\hoffset}{-7mm}
1081: %****** Table 1 ******
1082: \begin{deluxetable}{lcccl}
1083: \tabletypesize{\footnotesize}
1084: %\rotate
1085: \tablecaption{\label{table1}Physical parameters of the spherical
1086: $\rm SiO_2$ particles, the diamond grains, and the irregular $\rm
1087: SiO_2$ grains as well as of the resulting RBD agglomerates
1088: thereof. The reference numbers refer to [A] manufacturer
1089: information, micromod Partikeltechnologie GmbH, [B]
1090: \citet{blusch04}, [C] \citet{popsch05}, [D] \citet{heietal99}, [E]
1091: \citet{popetal00}, [F] \citet{bluwur00}, [G] manufacturer
1092: information, Saint-Gobain Diamantwerkzeuge GmbH \& Co. KG, [H]
1093: manufacturer information, Sigma-Aldrich Chemie GmbH, [I]
1094: \citet{bluetal06}.} \tablehead{\colhead{Physical property} &
1095: \colhead{Symbol} & \colhead{Value} & \colhead{Unit} &
1096: \colhead{Reference}} \startdata
1097: {\underline{Spherical $\rm SiO_2$ particles}}\\
1098: Material && $\rm SiO_2$, non-porous && [A]\\
1099: Morphology && spherical && [A]\\
1100: Molecular arrangement && amorphous & & [A]\\
1101: Density & $\rho_0$ & $2,000$ & $\rm kg~m^{-3}$ & [B]\\
1102: Radius & $s_0$ & $0.76 \pm 0.03$ & $\rm \mu m$ & [C]\\
1103: Mass & $m_0$ & $(3.7 \pm 0.4) \cdot 10^{-15}$ & kg &\\
1104: Surface molecules && Si-OH && [A]\\
1105: Surface energy && 0.014 & $\rm J~m^{-2}$ & [D]\\
1106: Adhesion force & $F_{\rm stick}$ & $(67 \pm 11) \times 10^{-9}$ & N & [D]\\
1107: Adhesion energy & $E_{\rm stick}$ & $(2.2 \pm 0.4) \times 10^{-15}$ & J & [E,F]\\
1108: Rolling-friction force & $F_{\rm roll}$ & $(0.68 \pm 0.13) \times 10^{-9}$ & N & linear extrapolation from [D]\\
1109: Rolling-friction energy & $E_{\rm roll}$ & $(8.1 \pm 1.9) \times 10^{-16}$ & J & $E_{\rm roll} = F_{\rm roll} \frac{\pi}{2} s_0$\\
1110: Sticking threshold velocity & $v_{\rm stick}$ & 1.1 & $\rm m~s^{-1}$ & extrapolated from [E]\\
1111: Rolling-threshold velocity & $v_{\rm roll}$ & $1.5 \pm 0.3$ & $\rm
1112: m~s^{-1}$ & $v_{\rm roll} = \sqrt{(10 E_{\rm roll}) / m_0} $\\
1113: \hline
1114: {\underline{Agglomerates of spherical $\rm SiO_2$ particles}}\\
1115: Volume filling factor & $\phi$ & $0.15 \pm 0.01$ & & [I]\\
1116: Compressive strength & $C$ & 500 & $\rm N~m^{-2}$ &[I]\\
1117: Tensile strength & $T$ & 1,100 & $\rm N~m^{-2}$ &[I]\\
1118: \hline \hline
1119: \\[5cm]
1120: {\underline{Irregular diamond particles}}\\
1121: Material && C, diamond && [G]\\
1122: Morphology && irregular && \\
1123: Density & $\rho_0$ & 3,520 & $\mathrm{kg\,m^{-3}}$ & \\
1124: Size & $s_0$ & $0.75 \pm 0.25$ & $\rm \mu m$ & [E] \\
1125: \hline
1126: {\underline{Agglomerates of irregular diamond particles}}\\
1127: Volume filling factor & $\phi$ & $0.11 \pm 0.02$ && [I]\\
1128: Compressive strength & $C$ & 200 & $\rm N~m^{-2}$ &[I]\\
1129: Tensile strength & $T$ & 200 & $\rm N~m^{-2}$ & [I]\\
1130: \hline \hline
1131: {\underline{Irregular $\rm SiO_2$ particles}}\\
1132: Material && $\rm SiO_2$, non-porous && [H]\\
1133: Morphology && irregular && \\
1134: Density & $\rho_0$ & 2,600 & $\mathrm{kg\,m^{-3}}$ & [H] \\
1135: Size & $s_0$ & $\sim$ 0.05-5 & $\rm \mu m$ & 50\% of typical particle diameter \\
1136: \hline
1137: {\underline{Agglomerates of irregular $\rm SiO_2$ particles}}\\
1138: Volume filling factor & $\phi$ & $0.07 \pm 0.03$ &&[I]\\
1139: Compressive strength & $C$ & 200 & $\rm N~m^{-2}$ &[I]\\
1140: Tensile strength & $T$ & 300 & $\rm N~m^{-2}$ &[I]\\
1141: \enddata
1142: \end{deluxetable}
1143:
1144: \begin{figure}[htp]
1145: \plotone{f1.eps}
1146: \figcaption{\label{targetprojectiles}(a) Example of mm-sized
1147: projectiles consisting of spherical $\rm SiO_2$ particles. (b)
1148: Example of an unprocessed target with 2.5 cm diameter
1149: consisting of spherical $\rm SiO_2$ particles. (c) Example of a
1150: ``molded'' target with 2.5 cm diameter
1151: consisting of spherical $\rm SiO_2$ particles.}
1152: \end{figure}
1153:
1154: \begin{figure}[htp]
1155: \plotone{f2.eps}
1156: \figcaption{\label{setup}Schematics of the experimental
1157: setup used in this work.}
1158: \end{figure}
1159:
1160: \begin{figure}[ht]
1161: \plotone{f3.eps}
1162: \figcaption{\label{experimentalparametersa}The distribution of
1163: the experiment parameters impact angle, projectile
1164: mass and impact velocity for the
1165: three agglomerate types consisting of spherical, monodisperse
1166: $\rm SiO_2$ particles (top), irregular quasi-monodisperse
1167: diamond particles (middle) and polydisperse, irregular
1168: $\rm SiO_2$ particles (bottom). Open symbols denote collisions which
1169: resulted in sticking, filled symbols those in which the projectile
1170: bounced off the target after the impact.}
1171: \end{figure}
1172:
1173: \begin{figure}[htp]
1174: \plotone{f4.eps}
1175: \figcaption{\label{experimentalparametersc}Masses and impact
1176: velocities of the agglomerates consisting of spherical,
1177: monodisperse $\rm SiO_2$ particles with $\phi = 0.15$ for
1178: impacts into ``molded'' targets with $\phi = 0.15-0.20$. The open circles denote sticking, the
1179: full circles denote bouncing, the asterisks denote fragmentation
1180: without mass transfer to the target, and the diamonds denote
1181: fragmentation with mass transfer to the target.}
1182: \end{figure}
1183:
1184: \begin{figure}[htp]
1185: \plotone{f5.eps}
1186: \figcaption{\label{stickingprobability} Sticking probability in collisions
1187: between dust-aggregate projectiles and targets consisting of
1188: $1.5~\rm \mu m$ $\rm SiO_2$ spheres with a volume filling factor
1189: of $\phi =0.15$. The sticking probabilities were derived by
1190: averaging over 13 impacts.}
1191: \end{figure}
1192:
1193: \begin{figure}[ht]
1194: \plotone{f6.eps}
1195: \figcaption{\label{movie_examples} Examples for collisions of
1196: high-porosity dust-aggregate projectiles into targets of the same
1197: composition. The width of the full image is 27.5 mm.
1198: Top left: irregular $\rm SiO_2$ particles; impact velocity
1199: $v \approx 0.8-1.0 ~\rm m~s^{-1}$. Top right: diamond particles;
1200: impact velocity $v \approx 1.7-2.3 ~\rm m~s^{-1}$. Bottom:
1201: spherical $\rm SiO_2$ particles; ``molded'' target; impact
1202: velocity $v \approx 0.38~\rm m~s^{-1}$. Movies are available
1203: in the online version of this paper.}
1204: \end{figure}
1205:
1206: \begin{figure}[ht]
1207: \plotone{f7.eps}
1208: \figcaption{\label{masstransferimage}Example of a non-sticking
1209: collision with mass transfer
1210: between the target and the projectile agglomerate (the movie
1211: of this impact can be found in the online material to Fig.
1212: \ref{movie_examples}). Shown is a
1213: sequence of four images recorded with the high-resolution
1214: high-speed camera. The inset in the bottom right frame is a copy of
1215: the projectile from the first frame. Projectiles and target consist of
1216: 1.5 $\rm \mu m$ $\rm SiO_2$ spheres and have volume filling factors
1217: of $\phi = 0.15$. The impact velocity is $v = 1.77~\rm m~s^{-1}$.}
1218: \end{figure}
1219:
1220: \begin{figure}[ht]
1221: \plotone{f8.eps}
1222: \figcaption{\label{masstransferhistogram}Histogram of the mass
1223: ratio $\mu$ defined in Eq. \ref{masstransfer}.}
1224: \end{figure}
1225:
1226: \begin{figure}[htp]
1227: \plotone{f9.eps}
1228: \figcaption{\label{coefficientofrestitutionfig}Coefficient of restitution
1229: as a function of impact velocity (top) and impactor mass
1230: (bottom) for the 18 non-sticking impacts of agglomerates consisting of
1231: $\rm SiO_2$ spheres. }
1232: \end{figure}
1233:
1234: \begin{figure}[htp]
1235: \plotone{f10.eps}
1236: \figcaption{\label{energyloss}The ratio of total kinetic energy after and before the
1237: collision as a function of the squared impact parameter (circles). The
1238: straight line gives the upper limit derived by \citet{blumue93},
1239: which is valid for perfectly plastic central and perfectly
1240: elastic and non-slipping grazing collisions.
1241: The energy loss for central collisions
1242: ($\sin^2{\theta}=0$) in this model
1243: is due to plasticity and for grazing collisions ($\sin^2{\theta}=1$)
1244: is due to an energy
1245: transfer from translation to rotation.}
1246: \end{figure}
1247:
1248: \begin{figure}[ht]
1249: \plotone{f11.eps}
1250: \figcaption{\label{semielasticrebound}An example of a
1251: partial rebound of a projectile agglomerate consisting of
1252: irregular $\rm SiO_2$ particles (the movie
1253: of this impact can be found in the online material to Fig.
1254: \ref{movie_examples}). On the top left
1255: image the projectile can be seen just before the impact. The impact
1256: angle is $\theta = 39.9^{\circ}$. The subsequent images -- taken 4.3, 8.6, 12.9,
1257: 21.5, 34.4, 47.3, and 137.6 ms after the first image -- show that the
1258: projectile penetrates into the target and then is slowly rebounding
1259: with decreasing velocity until it finally sticks. The impact velocity is
1260: $v = 1.02~\rm m~s^{-1}$.}
1261: \end{figure}
1262:
1263: \begin{figure}[htp]
1264: \plotone{f12.eps}
1265: \figcaption{\label{reboundingparticle}Displacement of the rebounding projectile
1266: in Fig. \ref{semielasticrebound} as a function of time. Also shown are
1267: two fit functions following Eqs. \ref{parabola} and
1268: \ref{exponential}}.
1269: \end{figure}
1270:
1271: \begin{figure}[ht]
1272: \plotone{f13.eps}
1273: \figcaption{\label{doienergy}Depth of intrusion for all sticking events as a function
1274: of the total kinetic energy of the projectiles
1275: for all three agglomerate types. The
1276: least-squares fits following Eq. \ref{intrusion2} are indicated by the
1277: three lines.}
1278: \end{figure}
1279:
1280: \begin{figure}[htp]
1281: \plotone{f14.eps}
1282: \figcaption{\label{hertzplot}Depth of intrusion of dust agglomerates consisting of spherical
1283: monodisperse $\rm SiO_2$ as a function of the parameter $\frac{m v_{\rm n}^2}{s^{1/2}}$.
1284: The indicated straight lines have slopes of 2/5 (see Eq. \ref{hertz4}).}
1285: \end{figure}
1286:
1287: \begin{figure}[ht]
1288: \plotone{f15.eps}
1289: \figcaption{\label{cvenergy}Crater volume as a function of the
1290: normal component of the impact energy for agglomerates
1291: consisting of spherical $\rm SiO_2$. A power-law least squares
1292: fit to the data points $E > E_{\rm kin,n}$
1293: with a slope of 0.75 is plotted as a solid line.}
1294: \end{figure}
1295:
1296: \begin{figure}[ht]
1297: \plotone{f16.eps}
1298: \figcaption{\label{pressureenergy}Dynamic pressure as a function of energy for agglomerates
1299: consisting of spherical $\rm SiO_2$.}
1300: \end{figure}
1301:
1302:
1303:
1304: \end{document}
1305: