Polytypism in SiC crystals
Polytypism in SiC crystals
复制标题
SiC晶体的多型性
DOI：
10.1107/s0365110x54000837
复制
发表时间：
1954
期刊：
影响因子：
--
通讯作者：
H. Jagodzinski
中科院分区：
文献类型：
--
作者：
H. Jagodzinski
最近Frank（1951）提出了SiC晶体复杂多型性的一种解释，他指出不同结构的存在可能是由于环生长过程中螺旋位错产生的同构期。Verma（1951）等人在SiC（0001）晶面上观察到的生长螺旋证实了这一观点。有理由证明，尽管有这些实验上的修正，弗兰克的观点并不能正确地解释所观察到的多类型现象。SiC晶体的x射线照片通常显示一维无序的特征（Jagodzinski, 1949 a, b, c）。这种类型的无序可能是由边缘位错引起的，在晶体的（0001）平面上延伸，并产生SiC层位移到另一个可能的位置（a， B， C）。这种错位的强度是frac，+。（t，]或t， t）”显然，产生螺旋位错所做的功要比形成边位错所做的功大得多。因此，不能理解为什么在SiC晶体中不应该发生边缘位错并破坏由缠绕生长螺旋产生的秩序。另一方面，SiC晶体的片状习惯表明，除了在晶体生长的最后阶段，面（0001）不是最大生长的面。用x射线法研究了随机选取的150个SiC晶体的晶体学性质；这些晶体中有62个属于一维无序型。我们开发了一种特殊的方法来评估无序度A （o~= Nd/N， Na是相对于晶体中达到的特殊有序的位移SiC层数，N是层总数）。所有晶体中平均故障的概率分布有两个最大值，一个在a= 0处（完全有序晶体），第二个在----0.12处（部分无序晶体）。这种复杂的行为可以通过引入振动熵来解释。定性估计给出的最大值在有序状态下，具有较低的恒等周期。无序状态在所有情况下都具有较低的振动熵。因此，振动熵有利于有序态，而构型熵不利于有序态。这样就会出现上述两个最大值，第一个最大值是由振动entrb~ 5~ tl~ e引起的，第二个最大值是由构型熵引起的。这里将不充分说明这一假设，但很明显，SiC晶体中的有序仅仅是由振动熵稳定的，通常观察到的长程有序是可以理解的。我们可以通过假设（参见Frank， 1951）三次修正是稳定的修正来推进对SiC结构复杂性的解释。这一观点得到了观察结果的支持，即大量正常合成的SiC属于立方结构（Ott, 1926），但它通常是如此精细的颗粒，如果用单晶照片进行研究，它很容易被忽视。因此可以得出结论，生长障碍会阻止这些立方晶体的生长。立方结构中能量最小的晶面是两个四面体（111）和（ilT），其中一个（四个）共价键指向开放空间。由于这些面形成一个封闭体，晶体生长将在相当高的过饱和时停止。如果任何位移是由边缘位错造成的，只有两个面（例如111和llT）将保持其稳定性——所有的面都将被“阶梯式”；这意味着这些面部更容易生长。这样晶体就会变成板状，特别是当有相当数量的位移存在时。在立方结构中，两个SiC层必须被置换。
An explanation of the complex polytypism in SiC crystals was recently advanced by Frank (1951), who pointed out that the existence of different structures might be due to the creation of identity periods by screw dislocations du-ring growth. This idea was confirmed by growth spirals observed on the (0001) crystal faces of SiC by Verma (1951) and others. There are reasons which prove that, in spite of these experimental remfits, Frank's idea does not give the correct explanation of the observed poly-typism. X-ray photographs of SiC crystals often show the features of one-dimensional disorder (Jagodzinski, 1949 a, b, c). This type of disorder may be caused by edge dislocations, extending in the (0001) plane of the crystal and producing a displacement of the SiC layer into another possible position (A, B, C). The strength of such dislocations is a frac,+. ion of the unit-cell length, or a multiple of it (t,] or t, t)" Obviously, the work done to generate a screw dislocation is much larger than that to form an edge dislocation. Therefore it cannot be under-stood why edge dislocations should not occur in SiC crystals and destroy the order produced by the winding up growth spiral. On the other hand the plate-like habit of SiC crystals shows that the faces (0001) are not those of maximal growth, excepting, perhaps, in the last stage of crystal growth. A total of 150 SiC crystalS, randomly chosen with respect to crystallographic properties, were investigated by X-ray methods; of these crystals 62 belonged to the one-dimensionally disordered type. A special method was developed for evaluating the degree of disorder a (o~= Nd/N, Na being the number of displaced SiC layers with respect to the special order achieved in the crystal, and N the total number of layers). The prob-ability distribution of the average faults in all crystals shows two maxima, one at a= 0 (completely ordered crystalS), the second at a----0.12 (partially disordered crystalS). This complex behaviour may be explained by introducing the vibration entropy. A qualitative estimate gives largest values for it in ordered states, which have a low identity period. Disordered states will in all cases have a lower vibration entropy. Thus ordered states are favoured by vibration entropy, but disfavoured by con-figuration entropy. In this way the two maxima described above may occur, the first being caused by vibration entrb~ 5~ tl~ e second by configuration entropy. No full account of this assumption will be given here, but it becomes clear that order in SiC crystals is merely stabilized by vibration entropy, and the long-range order commonly observed is understandable. We can advance the explanation of the complexity of SiC structures by assuming (see Frank, 1951) that the cubic modification is the stable one. This idea is supported by the observation that a great deal of normally syn-thesized SiC belongs to the cubic structure (Ott, 1926),but it is generally so finely grained that it may easily be overlooked if investigated by single-crystal photo-graphs. It may therefore be concluded that growth hindrance will prevent these cubic crystals from growing. The crystal faces of least energy in the cubic structure are the two tetrahedra (111) and (ilT), where one (of four) covalent bond is directed into the open space. As these faces form a closed body, crystal growth will stop at a fairly high supersaturation. If any displacement is created by an edge dislocation, only two faces (eg 111 and llT) will preserve their stability--all o~ hers will be'stepped'; that means that growth is favoured on these faces. Thus the crystals will become plate-like, especially if a fair number of displacements is present. In the cubic structure two SiC layers have to be displaced …