厉建龙

论文题目：低微纳米结构的分子束外延生长和变温扫描隧道显微镜研究

作者简介：厉建龙，男，1970年01月出生，1995年09月师从于中国科学院物理研究所薛其坤教授，于2001年08月获博士学位。

摘 要

本论文的工作主要是利用分子束外延和扫描隧道显微镜研究了两类低维纳米结构的生长过程及其原子结构。这两类低维结构是： 1) III族金属In、Ga、合金Ag/In以及磁性金属Mn在Si(111)-7x7模板衬底上形成的全同量子点阵；2)金属In在Si(001)-2xn模板衬底上形成的有序量子线阵列。

I) 在制备全同的金属量子点阵的研究中，我们首先发现具有幻数尺寸的III族金属元素In、Ga的团簇可以在Si(111)-7x7衬底上形成有序的量子点阵。其中，In的团簇会优先占据Si(111)-7x7单胞中有层错的一半(Faulted Half Unit Cell)，直到它们占满整个表面上所有的FHUC从而形成有序的三角形晶格超点阵(superlattice)，其晶格周期与7x7衬底相同。在更高覆盖度情况下，7x7单胞的两半都会被In的团簇占据而形成周期性的honeycomb点阵结构。这种结构依然与7x7表面具有相同的对称性，但其每个单胞中具有两个In团簇，是一种复式晶格。扫描隧道显微镜研究表明，这样获得的每个In的团簇由六个In原子构成，在空态STM图像中呈现正三角形结构。第一原理计算表明，a)六个In原子在Si(111)-7x7表面形成一种扭曲的六边形结构，每个In原子与一个衬底Si原子及两个相邻的In原子形成共价键，这些共价键具有类sp²构型；b)单个In原子吸附在Si(111)-7x7单胞的FHUC时的吸附能较吸附在没有层错的一半(Unfaulted Half Unit Cell)的吸附能低0.2eV/原子，故而In团簇优先在FHUC成核生长并进而形成有序的点阵结构。

在Ga幻数团簇的生长研究中我们发现，Ga原子对7x7单胞的两半不存在优先占据的问题。低覆盖度下Ga团簇是随机分布的，不具有明显的有序性。计算表明单个Ga原子吸附在7x7单胞的两半具有几乎相同的吸附能(差别仅为5meV)，这意味着Ga原子会随机地占据7x7单胞的任意一半成核生长，从而Ga团簇在低覆盖度下是随机分布的。高分辨的STM图像显示，每个Ga团簇由三个Ga原子构成。当Ga的团簇覆盖整个7x7表面时就会形成尺寸全同、空间分布均匀的量子点阵。具体地，Ga团簇晶格的对称性与高覆盖度下In团簇晶格的对称性是相同的Honeycomb结构。

总之，通过对In、Ga团簇的生长研究，我们发现某些金属原子具有形成尺寸全同的团簇的趋势，并进而在特定的周期性模板衬底上形成有序的周期结构。利用金属的这种趋于形成具有特定尺寸团簇的属性极大地压缩了所生长的量子点的尺寸分散性，加上周期模板的作用，就形成了一种有效的生长尺寸、空间分布均匀的量子点阵的方法。

我们进一步把这种设想在具有催化性能的贵金属Ag上进行了尝试。结果发现，Ag在7x7的半个单胞内会形成几种不同尺寸的团簇，而不是象In、Ga那样只形成单一大小的团簇。令人惊异的是我们发现在全同的In团簇点阵(In团簇只占据FHUC情况)上生长Ag时，Ag原子会占据In团簇的位置(按第一原理计算的结果，In团簇的位置已经不再有悬挂键)，而不占据Si的UFHU位置，从而形成周期良好、尺寸均一的Ag/In合金团簇点阵。这种全同的合金点阵结构为研究纳米尺度下的催化反应提供了一种可能的候选材料。

我们还利用覆盖了全同In、Ga量子点阵的Si(111)-7x7表面为模板，在其基础上尝试了磁性金属Mn、Co、Fe量子点阵的生长。我们发现，与在7x7衬底上直接生长磁性金属的情况不同(磁性金属与Si之间的反应活性较强，通常会很快形成硅化物，从而不能形成有序的点阵结构)，In、Ga团簇的存在减缓甚至抑制了硅化物的形成，使磁性金属原子得以形成团簇结构，同时模板结构又没有完全被破坏。尤其有趣的是，我们发现Mn在In团簇形成的模板衬底(In团簇只占据7x7单胞的FHUC情况)上生长时，Mn形成了尺寸均一的团簇，且只占据7x7单胞的UHUC，从而形成了具有完好周期性的磁性金属点阵结构。这是目前我们所知的空间密度最大的周期性磁性金属团簇点阵，其密度约为1.6x10¹⁵/cm²(目前实际使用的磁存储材料的存储密度量级为10⁷/cm²)。

II) 在Si(001)-2xn模板衬底上生长金属In纳米线阵列的研究中，我们发现In原子优先占据具有悬挂键的Si二聚体链位置，而不占据空位缺陷位置。当In覆盖整个2xn表面的2x1台面(terrace)时，由于空位线的存在，就会形成有序的In量子线阵列。这种In量子线阵列形成局域的Si(001)-2x2-In再构。在这种再构结构中，每个In原子与两个衬底Si原子及一个相邻的In原子形成共价键。In在2xn衬底上的生长过程由一种称为表面聚合反应(surface-polymerization-reaction)的生长机制所控制。这种特别的生长机制、加上In在Si(001)表面迁移的高度各向异性以及2xn模板衬底的作用，三者共同导致了有序的In纳米线阵列的形成。

In纳米线的宽度是由2xn衬底上的2x1台面的宽度决定的，而2x1台面的宽度是可以通过精细控制2xn衬底的制备过程(例如控制导致2xn结构的杂质浓度)进行调节的，所以这种方法实际上可以制备出不同尺寸的纳米线。另外我们指出，因为III族金属Al、In、Ga及IV族金属Pb、Sn在Si(001)上具有相同的生长机制，那么在2xn衬底上同样可以制备出Al、Ga、Pb、Sn等金属的纳米线。

总之，我们的工作发现了两类制备低维有序结构的新方法，尤其是第一次给出了晶体表面生长的小尺寸团簇的原子结构以及构成团簇的原子与衬底原子间的结合关系。其实验结果对理解其他种类团簇的物理、化学性质都有重要帮助。同时这些大面积有序的低维结构也为制备全新的催化材料、磁存储材料、电子器件等提供了新的可能性。

关键词：低维结构、纳米结构、量子点、量子线、幻数团簇、半导体、分子束外延生长、扫描隧道显微镜。

Abstract

In this dissertation, fabrication, formation mechanism and structural properties of two kinds of low dimensional quantum structures were studied by using molecular beam epitaxy (MBE), scanning tunneling microscopy (STM) and first-principles total energy calculations. The two low-dimensional nanostructures are: (1) Ordered identical-sized metal quantum dots array on the Si(111)-7x7 surface, and (2) ordered metal quantum wire array on the Si(001)-2xN Surface.

I) In the study of identical quantum dots of metal atoms, we find that In, Ga atoms form identical clusters on the Si(111)-7x7 surface. In the case of In clusters, they prefer to occupy the faulted half unit cells (FHUCs) of the Si(111)-7x7, and can develop into a well-ordered quantum dots array with all the In clusters occupying the FHUC. Further In deposition will lead to a honeycomb structure in which In clusters occupy both FHUCs and unfaulted half unit cell (UFHUCs) of the 7x7 surface. High-resolution STM images show that each In cluster is made up of six In atoms. First principle calculation shows that, a) the six In atoms develop a six-member-ring structure or a distorted-hexagon. Each In atom is bonded with one Si atoms and two neighboring In atoms; b) The adsorption energy of single In atom on the FHUC of the Si(111)-7x7 is 0.2eV higher than that on the UHUC. As a result, In atoms prefer to enucleate at the UHUC and develop into ordered dots array accordingly.

In the study of growing Ga clusters, we find that they equally occupy both the FHUCs and UHUCs of the 7x7 unit cell. First principle calculation reveals that the adsorption energy for single Ga atom on the FHUC and the UHUC is almost the same, which accounts for why Ga clusters occupy both the half unit cells of the 7x7 when they start to grow. High-resolution STM images show that each Ga dot contains three Ga atoms. At high coverage, these identical-sized Ga dots develop into an ordered honeycomb structure on whole substrate surface.

After successful preparation of the identical Ga and In quantum dots arrays on the Si (111)-7x7 surface, we realize that metal atoms do have a tendency to develop into clusters with magic sizes. This characteristic together with the ordered template can narrow greatly the size distribution of quantum dots array (they are identical indeed). This provides a new way to fabricate uniformly-sized and –distributed quantum dots arrays.

We apply this strategy further onto noble metal Ag, since nanometer Ag clusters may be a promising candidate to study nanocatalytics. It turns out that Ag do not form identical clusters, but rather, several differently sized clusters on the Si(111)-7x7. However, when we deposit Ag onto the In cluster covered Si(111) (when In clusters occupy merely the FHUCs), to our surprise, Ag clusters nucleate on the In clusters only, but not on the bare UHUCs of the 7x7 substrate. As a result, they develop into a well-ordered Ag/In alloy clusters array, which suggests that the nanostructured In cluster array can be an ideal template to grow other nanostructures.

Then, we take advantage of In or Ga cluster occupied Si(111)-7x7 surface as template to grow magnetic clusters, e.g. Mn, Co clusters and so on. Usually Mn, Co atoms react with Si atoms, and form silicides on the Si(111) surface, thus it is daunting to grow well ordered magnetic quantum dots array. Our experiment reveals that on the In occupied 7x7 surface (In clusters merely occupy the FHUCs), Mn develop into very uniform clusters and they only occupy the UFUCs of 7x7. Therefore, they form well-ordered quantum dots array too. In the case of Co, we find that Co break the In/Ga template, and cannot form uniform quantum dots array.

II) In the fabrication of ordered In nanowire array on the Si(001)-2xn surface, we find that In atoms prefer to occupy the 2x1 terrace, rather than the vacancy lines. They form well-ordered nanowire array when all 2x1 terraces are covered. High resolution STM images show that such nanowires develop into a local Si(001)-2x2-In structure. In such structure, each In atom is bonded with two Si substrate atoms and one neighboring In atom, thus all three valency of In are saturated. Such growth process is controlled by a mechanism so called as “surface polymerization reaction”. This special growth mechanism, anisotropic migration of In atoms on the Si(001) and the 2xn template, account for the well ordered In nanowire array formed.

It is found that the width of such In nanowires is determined by the width of the 2x1 terrace, which can be adjusted by careful control of the preparation procedure of the 2xn template (such as the impurities concentration and annealing temperature). Thus, nanowires of different sizes can be fabricated with this approach. Furthermore, since group III metals Al, Ga, group IV metal Pb, Sn have similar growth behavior on Si(001), this strategy can be used to fabricate nanowires of Al, Ga, Pb and Sn as well.

In summary, two novel ways to fabricate low dimensional quantum structures with atomic precision are developed. For the first time, the atomic structure of surface-supported nanoclusters has been determined. The results provide useful information to understand the physical or chemical properties of other clusters. At the same time, such low dimensional structures, which are well ordered in large area, may be promising systems for the next generation microelectronics, high-density magnetic storage and novel catalytic reaction.

Key words: Low-dimensional Nanostructures, Quantum Dots, Quantum Wires, Magic Cluster, Semiconductor, Molecular Beam Epitaxy, Semiconductor, STM.

　回主页