Elucidation of the mechanism of thermal-evaporated PAK2

In previous section, we studied the electrical characteristics of BPhen doped PAK2 layer from I-V curves and AS measurements and demonstrate the incorporation of PAK2 can reduce the electron injection barrier from Al cathode.

Similar result was reported for sodium acetate (CH3COONa) used as an efficient EIL in OLED devices [24]. Furthermore, higher sodium (Na) content in overlaid Al cathodes was observed for Alq3/CH3COONa/Al than Alq3/NaF/Al by x-ray photoelectron spectroscopy of detached Al cathodes [23]. It was also found by XPS that Na metal with a small amount of CH3COONa was deposited during vapor deposition of CH3COONa [30]. However, it is still a difficult task to evaluate the possibility of alkali metals formation by thermal decomposition.

For this purpose, the QCM is a simple and extremely sensitive method to measure the negative frequency shift caused by increasing mass during deposition of alkali metal salts. By using QCM, Ganzorig and Fujihira demonstrated the thermal decomposition of the Na salts of acetate and benzoate during vacuum vapor deposition [31]; Qiu also investigated the electron injection mechanism of evaporated Cs2CO3 [32]. To estimate the possible evaporation behavior of PAK2, we use QCM method to measure the negative frequency shift caused by the vacuum thermal evaporation of PAK2. In order to reduce the possible experimental errors, two thermally stable materials (NPB and MADN) were selected for comparison, and their slopes of the linear relationship (frequency shift vs mass loss) are compared to that of PAK2.

The frequency shift of the QCM (∆f) can be converted to the mass loss load on the surface of the quartz crystal wafer using the standard Sauerbrey formula

142 change in the mass and adsorbed onto the crystal, and A (cm²) is the deposition area on the quartz crystal.

If the source material is assumed to be thermally stable during the evaporation, the mass adsorbed onto the crystal surface (∆m) would be in direct ratio the mass loss in the boat (∆M): Then, according to equation 6.9, the following equation can be obtained:

A

where K is a positive constant that is independent of the materials used. It can be seen that ∆f is in direct ratio to ∆M. On the contrary, if the source material decomposes and only a fraction (d%) is deposited, ∆m would be in direct ratio to d% × ∆M: initial value if the source material decomposes.

143

Approximately 10, 20, 30, and 40 mg of NPB, MADN, PAK2 were introduced into the tungsten boat in the vacuum chamber, respectively. The resistive heating was controlled by adjusting the source current of the power supply, which was increased by 5 A every 2 min. The resistive heating of the bare tungsten boat could have led to a positive (∆f) value when the current was increased, but the shift of the bare quartz crystal proved to be smaller than 25 Hz, which was negligible.

The negative frequency shift–current (∆f vs I) characteristics of these materials are shown in Figure 6-6. For NPB, MADN, and PAK2, the materials in the boat were not used up until the applied current became 50, 45, and 70 A (critical currents), respectively. The ∆f values are therefore calculated from the frequencies at the critical currents and the frequencies at 0 A.

The ∆f vs ∆M curves are plotted in Figure 6-7, reveal that the experimental data of each material can be fitted to a straight line (∆f = K∆M +B). The values of the slopes (K), intercepts (B), and linear correlation coefficients (r) are summarized in Table 6-1, show that ∆f and ∆M have a good linear relationship for all the three materials. According to Equations 6.10 and 6.12, the intercept (B) should be zero, we attribute the experimental value of B is owing to the system errors. Note that the values of K∆M are far form than those of B, therefore, the value of B can be omitted.

144

Figure 6-6 ∆f vs I characteristics of (a) NPB, (b) MADN, and (c) PAK2.

Figure 6-7 ∆f vs ∆M characteristics of NPB, MADN, and PAK2.

(a) (b)

(c)

145

Table 6-1 Parameters of the linear fitting(∆f = K∆M +B) from Figure 6-7 (r is the linear correlation coefficient).

In Table 6-1, it is found that the K values of NPB and MADN are almost identical when the measurement errors are considered, indicating that the behaviors of NPB and MADN fit Equation 6.10 because they do not decompose during evaporation. In the case of PAK2, it is apparent that the negative ∆f value is less than those of NPB and MADN, indicating that PAK2 is decomposed during thermal evaporation and the behavior of PAK2 also fits the Equation 6.12.

Then, we consider the NPB as the standard material, and from the slopes (K) of NPB and PAK2, the d% of PAK2 can be determined to be 28.3%.

The weight percentage of potassium (K) atom in PAK2 molecule is 32.3%.

Furthermore, it can be observed that the chamber pressure will suddenly rise to the order of 10^-5 torr and then drop back to 10^-6 torr when PAK2 is evaporated.

According to the d% value of PAK2 from OCM measurements and the variation of chamber pressure, we suggest that PAK2 was partially decomposed during the thermal evaporation process, and some kind of gaseous byproducts are also generated, which are the reasons for the pressure variation. The result agrees with some reported observations [31].

In order to further confirm the mechanism of PAK2 deposition, we studied the depth profiling of PAK2 thin film (32 nm, deposited on silicon wafer) by

146

scanning auger spectroscopy. From Figure 6-8, which plots the characteristics of atomic percentage and etch time, shows the potassium (K) atoms are indeed deposited on the sample along with carbon (C) and oxygen (O) atoms. Based on the results of QCM measurements and depth profiling, we can conclude that the some species contain highly reactive potassium atom are generated during the evaporation process of PAK2. As a result, in the case of our n-doped layer composed of BPhen and PAK2, the those generated species with high dipole moments can reduce the electron injection barrier and further improve the electron injection from Al cathode.

We also tested the stability of PAK2 and compared the common used n-type material Cs2CO3, which absorbs easily the moisture in the ambient. The absorbing characteristics of PAK2 (23.6 mg, 9.7 × 10^-5 mole) and Cs2CO3 (23 mg, 7.06 × 10^-5 mole) were studied at a relative humidity of 60% and a temperature of 21 ⁰C by a analytical balance in our clean room. The relationship between mass change (∆m) and the time of exposing the samples to the air is plotted in Figure 6-9, in which the smaller mass change can be observed for PAK2, assuring the advantages of using PAK2 into OLED devices.

Figure 6-8 Depth profiling of PAK2-deposited film.

147

Figure 6-9 Relationship of mass change (∆m) vs time.

Figure 6-10 AFM topographic images of BPhen and 5% PAK2-doped BPhen thin films deposited on silicon wafers (50 nm). (a) BPhen before heating; (b) BPhen after heating; (c) 5% PAK2-doped BPhen before heating; (d) 5%

PAK2-doped BPhen after heating (scanned area: 5 m × 5 m).

(a) (b)

(c) (d)

148

On the other hand, the thin film of BPhen is known to be morphologically unstable and has been found to affect device degradation [36]. Figure 6-10 pictures the results of AFM measurements, in which the root-mean-square surface roughnesses (Rms) of unheated/heated BPhen films were 0.45 nm/48.8 nm, respectively, indicating the crystallization is evident in pristine BPhen film after heated at 80 °C for 30 min. However, the Rms of unheated 5% PAK2-doped BPhen film was 0.38 nm, and the degradation of surface morphology is clearly suppressed even after heated (Rms was 0.45 nm). Consequently, it is expected the