Formation mechanisms of needles in an aqueous HF solution in the transition region can be described through pore formation models, such as the current burst model [176], the Zahng pore formation model (distribution of the electric field at different regions of a pore) [27], and the Lehmann macro-pore formation model [186]
with some hypothesis. In this way, nucleation of pores and formation of irregular islets can be respectively explained by the current burst model and the Zhang pore formation model. Formation of needles through further etching of side walls of irregular islets can be then described by the Lehmann macro-pore model.
In general, a certain amount of Si-dissolution may occur prior to and during initiation of pores [27]. Such a dissolution or etch layer before pores’ initiation is involved in all types of porous silicon since an etching which causes roughening of the surface, is required for initiation of pores [232, 233]. Under conditions where current exceeds JPSi and the direct dissolution prevails, current bursts begin to correlate positively in time [178]. This means that the likelihood of nucleation of a new current burst in the place where an old one had occurred, is increased, and current bursts begin to cluster.
Clustered current bursts are then resulting in formation of macro-pores (see Figure 6c) due to their correlations in space [190]. In a condition where the oxidation reaction is low (e.g., in a water free electrolyte), current bursts result in macro-pores covered with a micro-or meso-pores PSi layer (transition layer) with a strong preferential growth of both pore types in <100> direction since no oxidation reaction removes the left-over
77 Si between the current lines (see Figure 24) [189]. However, in a condition where the oxidation reaction is dominant (e.g., in an aqueous electrolyte), oxidation takes over current bursts and smooths the surface. This results in macro-pores with no meso-pores coverage (complete dissolution of the transition layer).
Figure 24. Cross-sectional view of two formed PSi layers, a transition micro- or meso-PSi layer on top of a macro-PSi layer (retrieved from [234]).
A transition layer is almost formed in all n-types silicon (with different doping concentrations), and its thickness layer is related to size of pores; the larger the pores, the thicker the transition layer surface [207]. However, for p-type substrates, it is found to be only formed on lowly doped substrates [207]. In a lowly doped p-type substrate, coverage of macropores with micropores is due to the large difference between potential drops at the SCR and the substrate; voltage drop at the SCR is associated with formation of micropores, whereas the non-linear effect of the substrate resistivity is responsible for formation of macropores [234]. In the transition region, the surface is not completely covered by oxide; therefore, both inhomogeneous and homogenous dissolutions may occur on different parts of the surface [27]. This may result in inhomogeneous dissolution of the transition layer and formation of small islets (remaining bulk Si in the transition layer) above macro-pores at certain times during the etching. With increasing time, the transition layer dissolves entirely and gives access to underneath macro-pores, which have already widened and overlapped.
Widening and overlapping of the macro-pores occur because of comparable dissolution rates occurring at edge of pores due to 𝐽𝑏 and pores tips due to 𝐽𝑡
according to the Zhang pore formation model (see Eq. 13) [27]. This causes a substantial dissolution at the edge of pore bottoms before further propagation of pore tips, and results in complete dissolution of walls between the pores. The remaining bulk Si between widened and overlapped macro-pores are then resulted in large islets.
p-type Si
Macro-pores
Micro- or Meso-pores
78
Assuming formed large islets (with diameters or widths of ~ 4.5 ± 1.4 µm measured from Figure 13d) as remaining and surrounding walls of widened and overlapped macro-pores in which their thicknesses are almost 3 to 4 times larger than the 𝑆𝐶𝑅𝐿 (1 - 1.2 µm for 12-17 Ωcm p-type Si, calculated using Eq. 10), the Lehmann macro-pore model [186] can be used to describe growth of vicinity of macro-pores and reshaping of islets (by etching their side walls) to needles. In this way, with increasing etch time, when diameters or widths of remaining bulk Si between widened pores (islets) reach to 2 - 2.4 µm (twice the 𝑆𝐶𝑅𝐿) due to further etching of their side walls, islets which have narrowed and reshaped to needles, become passivated. At this stage, further etching does not affect passivated needles; it develops only tips of their vicinity pores, and consequently increases height of needles. These characteristics can be easily observed from Figure 25 where a single needle with diameter or width of about 2.4 µm (almost equal to twice the 𝑆𝐶𝑅𝐿, 1.87 - 2.6 µm for 10 - 20 Ωcm p-type Si) and length of about 25 µm is bent, stretched, and attached to other needles from top.
Figure 25. a single needle with a diameter or width of ~ 2.4 µm (≈ 2 × 𝑆𝐶𝑅𝐿, 1.87 - 2.6 µm for 10-20 Ωcm p-type Si)) and a length of ~ 25 µm bent, stretched, and attached to other needles from top. Anodization parameters: current density = 70 mA/cm², substrate = 10 - 20 Ωcm
<100> CZ p-type Si, electrolyte = 7.2 wt.% aqueous HF, and anodization time = 40 minutes.
4.4 Summary
Generation of needle-like surfaces in the transition region was experimentally confirmed by anodizing lowly doped p-type Si wafers in a 7.2 wt.% aqueous HF solution at a constant temperature of 21 °C. Impacts of anodization parameters, substrate resistivity range, substrate fabrication method, and drying process on morphology of needle-likes surfaces and geometrical properties of clustered needles were experimentally investigated. Table 2 gives a prompt view over used materials and applied conditions for every employed experimental investigation.
79 For a particular substrate resistivity range (10 - 15 Ωcm or 10 - 20 Ωcm), increasing current density in the transition region was resulted in an increase in clustered needle density and height of clustered needles, and increasing anodization time was resulted in an increase in height of clustered needles. However, no considerable influence on diameter of clustered needles was observed in respect to current density and anodization time.
Addition of a surfactant agent (ethanol) to 7.2 wt.% aqueous HF electrolyte changed the electrolyte properties (e.g., conductivity or viscosity) and the semiconductor/electrolyte interface. This resulted in a change in the dissolution kinetics through improved or worsened passivation kinetics and prohibited the needle formation.
Silicon wafer resistivity and its fabrication method showed a significant impact on morphology of needle-like surfaces and geometrical properties of clustered needles.
For a specific anodization condition (j = 50 mA/cm², T = 40 minutes, and electrolyte = 7.2 wt.% aqueous HF), needle-like surfaces were only obtained from
10 - 15 Ωcm and 15 - 20 Ωcm p-type Si substrates. However, 10 - 15 Ωcm wafers resulted in longer clustered needles than 15 - 20 Ωcm wafers. For a specific anodization condition (j = 68 mA/cm², T = 70 minutes, 10 - 15 Ωcm p-type Si, and electrolyte = 7.2 wt.% aqueous HF), the CZ wafer was resulted in a needle-like surface with higher number of needles and with longer needles compared to the FZ wafer.
However, the needle-like surface obtained from the FZ wafer showed more homogeneity in respect to height and diameter of clustered needles and their distributions on the surface compared to the CZ one.
Pentane drying showed a slight impact on clustering behavior of needles. 1 - 2 % increase in clustered needle density and 2 - 5 % decrease in diameter of clustered needles were observed using pentane drying in comparison to natural drying.
However, no considerable difference was realized between surfaces dried by ethanol and by ethanol-pentane.
Formation mechanisms of needles in an aqueous HF solution in the transition region were monitored by surface SEM images taken after various anodization times.
Formation of needles were then described through the current burst model, the Zhang pore formation model, and the Lehmann macro-pore formation model with some
80
assumptions. In this way, nucleation of pores and formation of irregular islets (remaining bulk Si between widened and overlapped macro-pores) were respectively explained by the current burst model and the Zhang model. Formation of needles by further etching of side walls of irregular islets were then described by the Lehmann macro-pore model.
81 Table 2. Used materials and applied conditions for employed experimental investigations.
Experimental solution + the additives at 21 °C
82
83
5 | Room temperature Si-Si bonding technique using silicon needle-like surfaces
This chapter demonstrates the room temperature Si-Si bonding technique using silicon needle-like surfaces. Four different needle-like surfaces with various needles properties are prepared and employed to demonstrate the bonding method.
Geometrical properties of needles, their bond strengths, and bond interface widths are measured and presented.