**5. Investigation of III-nitride devices by AFM**

The structure of a lateral blue or green InGaN LED is shown in Fig. 18. To observe the surface of each layer and know what happened after each device processing step, researchers use SEM and AFM to check the surface morphology of the patterned sapphire substrates, as-grown LED wafers, etched n-mesa, ITO and p/n metal.

Fig. 18. Structure of a typical lateral GaN based light emitting diode.

#### **5.1 Patterned sapphire substrates**

Most of the current commercial LEDs are grown on patterned sapphire substrates (PSS), not on flat sapphire substrates, because LEDs on PSS have demonstrated an enhanced light output power and external quantum efficiency (EQE) compared to conventional LEDs grown on planer sapphire substrates.

Fig. 17. AFM images of AlGaN superlattices on bulk AlN. The scan area of (a), (b), and (c) is 10 × 10 μm�, and (d), (e), and (f) is 1 × 1 μm�. (a) and (d), (b) and (e), and (e) and (f) are taken after the growth of superlattices 1, 2, and 3, respectively. Root-mean-square (rms)

The structure of a lateral blue or green InGaN LED is shown in Fig. 18. To observe the surface of each layer and know what happened after each device processing step, researchers use SEM and AFM to check the surface morphology of the patterned sapphire

Most of the current commercial LEDs are grown on patterned sapphire substrates (PSS), not on flat sapphire substrates, because LEDs on PSS have demonstrated an enhanced light output power and external quantum efficiency (EQE) compared to conventional LEDs

roughness from each scan is labeled.

**5.1 Patterned sapphire substrates** 

grown on planer sapphire substrates.

**5. Investigation of III-nitride devices by AFM** 

substrates, as-grown LED wafers, etched n-mesa, ITO and p/n metal.

Fig. 18. Structure of a typical lateral GaN based light emitting diode.

LED grown on PSS showed a higher internal quantum efficiency due to a dislocation density reduction by epitaxial lateral overgrowth technology (Tadatomo et al., 2001; Yamada et al., 2002 ; Hsu et al., 2004). Besides the elimination of threading dislocations due to the lateral growth of GaN on top of PSS, researchers believe that PSS improve the light extraction. It is well known that the large difference of refractive index between semiconductor and air leads to trap of large percentage of light emitting from LED (Lee et al., 2005). Thus, it is important to design and characterize the geometry of the PSS. AFM is the only available tool to characterize the bump shape accurately so far.

AFM images of PSS with different specs are shown in Figs. 19(a) and (b). The bump shapes, depths and widths could be characterized accurately. The average bump depth, pitch and gap of the patterns seen in Fig. 19(a) are 1.1 μm, 3 μm and 0.2 μm, respectively. While the average bump depth, pitch and gap of the patterns seen in Fig. 19(b) are 0.3 μm, 6 μm and 3 μm, respectively.

Fig. 19. (a) One patterned sapphire substrate with a cross sectional line to show the height, width and shape of the bumps. (b) 3-dimensional AFM image of another PSS with triangular pyramid cone shaped bumps.

#### **5.2 As grown LED surface and leakage characteristics**

Fig. 20 (a) is the surface morphology of an LED grown on a PSS. There is a p++ layer on the top of the LED surface. The bumps on the surface are caused by heavily doped Mg on the surface, usually rooted from a screw type dislocation. In the manufacturing of LEDs it is important to verify that the surface is free of pits, which implies the p-layer has coalesced well and sealed all the V-pits from the underneath layer of InGaN. Fig. 20(b) is an example of a LED with pits on the surface. LEDs with un-coalesced surfaces with pits usually have high leakage currents.

Screw type related pits on the GaN surface have been confirmed to be the source of reverse leakage in GaN films (Law et al., 2010). Fig. 21(a) shows an AFM topograph of a GaN sample grown by molecular beam epitaxy (MBE), in which the screw type TD related pits could be observed. Fig. 21(b)–(d) show conductive atomic force microscopy (CAFM) images obtained at dc biases of -14, -18, and -22 V of the area in Fig. 21(a). Fig. 21(b) shows several small, dark features that correspond to localized reverse-bias leakage paths observable

AFM Application in III-Nitride Materials and Devices 205

because it may produce high quality surface at low cost and fast material-removal rates. Fig. 22

Fig. 22. 25 × 25 μm�AFM scans of the surface topography of a polished sapphire substrate.

shown to have smaller particle and better uniformity.

magnetron sputtering.

Semi-transparent Ni/Au on Mg doped GaN was used as the p-contact material in the earlier LED devices. However, the transmittance of such semi-transparent Ni/Au contact is only around 60–75%. Although we could increase the transmittance by reducing Ni/Au metal layer thickness, the contact reliability could become an issue when the contact layer thickness becomes too small. Transparent indium in oxide (ITO) was popularized as the pcontact material because its high electrical conductivity and transparency to visible light. ITO could be deposited by electron beam evaporation or a magnetron sputtering method. It was found that the LED using sputtering ITO has 2-3% higher Light output (Lop) than that of E-beam ITO. Surface morphology and particle size of the ITO film deposited by two methods were measured by AFM as shown in Fig. 23. AFM analysis helped researchers to better understand the quality difference between the two ITO films. Sputtering ITO was

(a) (b)

Fig. 23. 5× 5 μm� AFM scans of the ITO deposited by (a) electron beam evaporation and (b)

is a typical 25× 25 μm� AFM image of a sapphire substrates polished by CMP.

Fig. 20. 5 × 5 μm�AFM scans of the (a) fully coalesced LED sample and (b) partially coalesced LED sample.

Fig. 21. (a) AFM topograph and (b)–(d) CAFM images obtained at tip dc bias voltages of -14, -18, and -22 V for MBE GaN sample.

at -14 V bias. In Fig. 21(c), the reverse-bias voltage magnitude was further increased and the density of observed conductive paths increased as well. This trend of increasing conductive path density as a function of increasing reverse-bias voltage magnitude continues in Fig. 21(d). According to these results, J. J. M. Law et al. suggested that changes in surface defects surrounding or impurities along screw-component threading dislocations are responsible for their conductive nature (Law et al., 2010).

#### **5.3 Other device processing characterized by AFM**

Chemical-mechanical polishing (CMP) is a polishing technique used to thin down the substrates for GaN devices, including Si, SiC, sapphire, AlN and GaN free-standing substrates,

(a) (b)

Fig. 21. (a) AFM topograph and (b)–(d) CAFM images obtained at tip dc bias voltages of -14,

at -14 V bias. In Fig. 21(c), the reverse-bias voltage magnitude was further increased and the density of observed conductive paths increased as well. This trend of increasing conductive path density as a function of increasing reverse-bias voltage magnitude continues in Fig. 21(d). According to these results, J. J. M. Law et al. suggested that changes in surface defects surrounding or impurities along screw-component threading dislocations are responsible

Chemical-mechanical polishing (CMP) is a polishing technique used to thin down the substrates for GaN devices, including Si, SiC, sapphire, AlN and GaN free-standing substrates,

Fig. 20. 5 × 5 μm�AFM scans of the (a) fully coalesced LED sample and (b) partially

coalesced LED sample.


for their conductive nature (Law et al., 2010).

**5.3 Other device processing characterized by AFM** 

because it may produce high quality surface at low cost and fast material-removal rates. Fig. 22 is a typical 25× 25 μm� AFM image of a sapphire substrates polished by CMP.

Fig. 22. 25 × 25 μm�AFM scans of the surface topography of a polished sapphire substrate.

Semi-transparent Ni/Au on Mg doped GaN was used as the p-contact material in the earlier LED devices. However, the transmittance of such semi-transparent Ni/Au contact is only around 60–75%. Although we could increase the transmittance by reducing Ni/Au metal layer thickness, the contact reliability could become an issue when the contact layer thickness becomes too small. Transparent indium in oxide (ITO) was popularized as the pcontact material because its high electrical conductivity and transparency to visible light. ITO could be deposited by electron beam evaporation or a magnetron sputtering method. It was found that the LED using sputtering ITO has 2-3% higher Light output (Lop) than that of E-beam ITO. Surface morphology and particle size of the ITO film deposited by two methods were measured by AFM as shown in Fig. 23. AFM analysis helped researchers to better understand the quality difference between the two ITO films. Sputtering ITO was shown to have smaller particle and better uniformity.

Fig. 23. 5× 5 μm� AFM scans of the ITO deposited by (a) electron beam evaporation and (b) magnetron sputtering.

AFM Application in III-Nitride Materials and Devices 207

Kobayashi J. T., Kobayashi N. P., & Dapkus P. D., (1998). Meeting of The American Physical

Koide Y., Itoh N., Itoh, K., Sawaki N., & Akasaki I. (1988). *Jpn. J. Appl. Phys*., Vol.27, pp.1156 . Law J. J. M., Yu E. T., Koblmüller G., Wu F., & Speck J. S., (2010). *Appl. Phys. Lett.* 96, 102111 LeeY.J., Hsu T.C., Kuo H.C., Wang S.C., Yang Y.L., Yen S.N., Chu Y.T., Shen Y.J., Hsieh

Li J., Nam K.B., Nakarmi M.L., Lin J.Y., Jiang H.X., Carrier P., & Wei S.H. (2003). *Appl. Phys.* 

Lin Y. S., Ma K. J., Hsu C., Feng S. W., Cheng Y. C., Liao C. C., Yang C. C., Chou C. C., Lee

Mayes K., Yasan A., McClintock R., Shiell D., Darvish S.R., Kung P., & Razeghi M. (2004).

Orenberg N. C., Martin M. G., Oliver R. A., Castell M. R., & Briggs G. A. D., (2002). *J. Phys.* 

Ren Z., Sun Q., Kwon S.-Y., Han J., Davitt K., Song Y. K., Nurmikko A. V., Cho H.-K., Liu W., Smart,J. A. & Schowalter L. J., (2007). *Appl. Phys. Lett.,* Vol.91, pp.051116.

Scholz F., Off J., Fehrenbacher E., Gfrorer O., & Brockt B., (2000). *Phys. Status Solidi A*

Sun, W.H., Zhang, J.P., Adivarahan, V., Chitnis, A., Shatalov, M., Wu, S., Mandavilli, V.,

Tadatomo K., H Okagawa., Ohuchi Y., Tsunekawa T., Imada Y., Kato M., Taguchi T., (2001)

Takeuchi Y., T., Amano H., Akasaki I., Yamada N., Kaneko Y., & Wang S. Y., *Appl. Phys. Lett*.

Wang X.L., Zhao D.G., Jahn U., Ploog K., Jiang D.S., Yang H. & Liang J.W. (2007). *J. Phys. D:* 

Weimann N.G., Eastman L.F., Doppalapudi D., Ng H.M., & Moustakas T.D. *J. Appl. Phys.,* 

Weyher J.L., Macht L., KamleG. r, Borysiuk J., Grzegory I., (2003). *Phys. Stat. Sol. (c),* Vol..3

Wickenden, D.K., Bargeron, C.B., Bryden, W.A., Miragliotta, J., & Kistenmacher, T. J. (1994).

Yang, J.W., & Khan, M.A. (2004). *Appl. Phys. Lett.*, Vol.85, pp.531.

Liliental-Weber Z., Kisielowski C., Smith M., Lin J. Y., & Jiang H. X., (1997). *Appl.* 

Sharma N., P Thomas., Tricker D., & Humphreys C., (2000). *Appl. Phys. Lett. Vol.*77, 1274. Shatalov, M., Chen, Z., Gaevski, M., Adivarahan, V., Yang, J., & Khan, A., (2006). *International Workshop on Nitride Semiconductors*, Kyoto, Japan., pp.101. Sun C. J., Anwar M. Z., Chen Q., Yang J. W., Khan M. A., Shur M. S., Bykhovski A. D.,

Northrup J. E., Romano L. T., & Neugebauer J., (1999). *Appl. Phys. Lett.* Vol.74, pp.2319. O'Leary S. K., Foutz B. E., Shur M. S., Bhapkar U. V., L Eastman. F., J., (1998). *Appl. Phys.*

C. M., & Chyi J. I., (2000). *Appl. Phys. Lett.* Vol.77, 2988.

Polyakov V. M. & Schwierz F., (2006). *Appl. Phys. Lett.* Vol.88, pp.032101.

Sangwal K., (1987). *Etching of Crystals*, North-Holland, Amsterdam , pp. 87-160.

M.H., Jou M.J., Lee B.J., (2005) *Materials Science and Engineering B* Vol.122 pp.184–

Society, Los Angeles, pp.16–20.

*Lett.*, Vol.83, pp.5163.

Vol.83, pp.826.

Vol.180, 315.

*Phys. Lett.* 70, 2978.

(1998). Vol.72, pp.710

Vol.83, 3656. (1998).

pp.21.

*Appl. Phys.,* Vol.40, pp.1113.

*Appl. Phys. Lett.*, Vol.65, pp.2024.

*Jpn. J. Appl. Phys*. Vol.40 pp.L583–L585.

Taniyasu Y., Kasu M., & Makimot, T. (2006). *Nature*, Vol.441, pp.325.

*D: Appl. Phys.*pp.35 615.

*Appl. Phys. Lett.*, Vol.84, 1046.

187

### **6. Conclusion**

In summary, the application of AFM in GaN, In(Ga)N and Al(Ga)N materials research and device fabrication have been reviewed. In regard to the GaN materials, the threading dislocations, including edge, screw and mixed types dislocations, as well as surface features are investigated by AFM. The study of V-shaped defects and topography in InGaN, InN films and InGaN/GaN multiple quantum wells by AFM has been reviewed. For AlN and AlGaN materials, how to utilize AFM to characterize the films and optimize the growth condition are demonstrated. Results also show that AFM is a powerful tool for device characterization and can shed light on device processing optimization.
