The quality of crystals is to a large extent determined by the surface structure and

dynamics during growth. The growth of many crystals is strongly influenced by

the presence of impurities, defects, surface reconstructions or adsorption layers.

This thesis describes X­ray scattering experiments in which the structure of crystal

interfaces is determined in order to get an atomic­scale understanding of the kinetic

and thermodynamic processes involved in growth. The first part of this thesis is

concerned with crystals and their growth in an ultra­high vacuum (UHV) environ­

ment, where the conditions of the surface can be controlled very well. In the last

two chapters we describe experiments on a crystal in its growth solution.

In homoepitaxial growth of Ag(111) it is known that by adding a surfactant

like Sb the growth mode can be changed from three dimensional (rough) to layer­

by­layer (smooth). The equilibrium surface structure of an Sb­covered Ag(111)

surface depends on the Sb coverage. For coverages below 1/3 monolayer, the Sb

atoms substitute for Ag atoms at normal fcc positions in the top surface layer.

There is no lateral ordering of the Sb atoms. At a coverage of 1/3 monolayer a

(Ö 3 x Ö 3)R30° reconstruction is formed. We have determined the atomic structure

of this reconstruction for the Ag(111)­Sb as well as for the similar Cu(111)­Sb

surfaces (chapter 2). Contrary to previous reports we found that all top layer

atoms reside at stacking fault positions. Each (Ö 3 x Ö 3)R30° surface unit cell

contains one substitutional Sb atom. We determined the out­of­plane relaxations

of the top layer atoms and the in­plane distortions in the second layer. When Ag

is deposited on this surface at 100 ° C, the Sb segregates and the Ag atoms return

to the correct fcc stacking, while the new Ag atoms in the top layer again have the

hcp stacking. This thus effectively leads to a floating stacking fault. Because of

kinetic limitations, the same effect occurs for Sb coverages below 1/3 monolayer.

For growth above 100 ° C, all lower lying Ag layers return to the correct stacking,

and no twin crystallites are formed.

In chapter 4 we study a model solid­liquid interface. We present a structural

analysis of the b ­Ge(111) (Ö 3 x Ö 3)R30° -Pb® 1 x 1 phase transition at ~180 ° C

for a Pb coverage of 1.25 monolayer. Below the phase transition the b phase

has a saturation coverage of 4/3 monolayer. Our atomic structure model for this

phase, consisting of three Pb atoms on off­centered T1 sites and one on a H3

site in the unit cell, is consistent with other studies reported for this system. We

find that above the phase transition the single layer of Pb gives rise to a ring

of diffuse scattering indicative of a two­dimensional liquid. However, of all the

Pb geometries considered, an ordered layer with large in­plane thermal vibration

amplitude is found to provide the best agreement between calculated and measured

structure factors. The Pb atoms appear to rapidly diffuse over the surface, but spend

a significant fraction at the lattice sites that are occupied at the low temperature

b ­phase. The Pb layer has thus both liquid and solid properties.

Although most crystals are grown from the liquid phase, the atomic structure

of the growing interface is hardly studied because of a lack of suitable techniques.

Most surface science techniques need a UHV environment and cannot be applied to

surfaces in a fluid. X­ray diffraction using the latest synchrotron radiation facilities

make these studies feasible for the first time. We have studied the interface atomic

structure of the inorganic crystal KDP. KDP crystals are grown from an aqueous

solution. Ex situ measurements were performed in vacuum and in air. In order

to be able to do in situ measurements, where the crystal is in contact with its

growth solution, we have designed and built a crystal growth chamber which

is compatible with X­ray diffraction experiments (chapter 6). The surface atomic

structure has been determined of the two natural existing faces, the prismatic {100}

and pyramidal {101} faces. We found that the {101} faces are terminated by a

layer of ions and not by groups, resolving a long-standing issue that could

not be predicted by theory (chapter 5). From our measurements we cannot find clear

differences between the surface structure in air, vacuum or in solution. However, the

quality of the surface, also as function of time, is better controlled in situ.

It is known that when trivalent metal ion impurities like Fe3+ or Cr3+ are present

in the growth solution the macroscopic crystal habit is elongated in the direction

of the pyramidal faces. From the atomic structure of the two different faces in

solution, we can explain this phenomenon. With only K+ ions on the {101} face

of the crystal, impurity ions will experience a large barrier for adsorption onto the

positively charged face. The {100} face has both the positive K+ ions and the

negative at the interface. On these faces cations can adsorb easily, and

small amounts of these ions will already block the growth. When Fe impurities are

added to the saturated KDP solution, no evidence was found for an ordered Fe layer

on the prismatic face. However, the surface becomes significantly more

rough. The impurities locally pin the moving steps, which causes an increased

meandering of the steps leading to a rougher surface.