Study on dopant activation of phosphorous implanted polycrystalline silicon thin films by KrF excimer laser annealing

(1)

Study on dopant activation of phosphorous implanted

polycrystalline silicon thin ﬁlms by KrF

excimer laser annealing

Chang-Ho Tseng, Ching-Wei Lin, Teh-Hung Teng, Ting-Kuo Chang,

Huang-Chung Cheng

*

_{, A. Chin}

Department of Electronics Engineering, Institute of Electronics, Building IV, National Chiao Tung University, 1001 Ta Hsueh Rd., Hsinchu 300, Taiwan, ROC

Received 19 August 2001; received in revised form 2 November 2001; accepted 4 December 2001

Abstract

The KrF excimer laser annealing of phosphorous implanted polycrystalline silicon (poly-Si) films had been inves-tigated completely. Resistors were fabricated to measure sheet resistance of poly-Si film. The interference effect, heat absorption of capping oxide as well as transformation of poly-Si grain size during laser annealing were reported. De-pending on the carrier concentrations, poly-Si exhibits different sheet resistance behavior when the excimer laser fluence is higher than the full-melt threshold fluence. The sheet resistance of poly-Si film has an abnormal increase from 5 104

to 4 106 _X=_{when the excimer laser energy is higher than full-melt threshold energy.} _{Ó 2002 Elsevier Science Ltd.}

Keywords: Excimer laser annealing; Dopant activation; Polycrystalline silicon thin ﬁlm transistor; Active-matrix liquid crystal display

1. Introduction

Excimer laser annealing (ELA) has been studied ex-tensively on high-performance polycrystalline silicon (poly-Si) thin-film transistors (TFTs) for active-matrix liquid crystal display (AMLCD) applications. These studies contain not only the crystallization of amorphous silicon (a-Si) thin film but also the dopant activation of source/drain regions. For a-Si thin film crystallization, many researches have been made completely. Those of them include effects of laser energy density, number of laser shot [1], annealing ambient, and laser beam shape [2]. For example, large grain size of poly-Si can be formed with special orientation, (1 1 1) orientation, by utilizing multi-shot process with exact laser energy of flat

top excimer laser beam [1]. Such kind of grain structure could also be obtained from the irradiation of multi-scan with Gaussian shape excimer laser beam [2]. By using such large gain and excellent crystallinity of poly-Si thin ﬁlm, the performance of poly-Si TFTs can be improved drastically.

Besides the crystallization of a-Si thin ﬁlm, dopant activation is another application of ELA in low-tem-perature poly-Si TFTs fabrication. Low-sheet resistances (Rs) of source/drain regions and contacts are necessary

for high-performance poly-Si TFTs fabricating on low-cost glass substrate. Comparing with conventional fur-nace annealing, laser annealing can efficiently activate the doping atoms in poly-Si film at room temperature. Several studies have been reported about the dopant activation by ELA process. Those of them include spin-on-dopant layer annealed by KrF excimer laser, gas immersion laser doping (GILD) process, and projection-GILD [3–5]. But the dopant activation of poly-Si thin film capped with silicon oxide by ELA is less studied. In addition, it is well known that the conducting behavior

www.elsevier.com/locate/sse

*

Corresponding author. Tel.: +886-3-5712121x54218; fax: +886-3-573-8343/8403.

E-mail address:[email protected](H.-C. Cheng).

(2)

of poly-Si ﬁlm varies with the poly-Si grain size especially when the carrier concentrations are closed to the bulk trap concentrations. However, study related with this topic was also seldom reported.

This paper investigates the dopant activation of phosphorus implanted poly-Si thin film by KrF ELA process. The effects of laser fluence, capping oxide thickness and the grain structure transformation after KrF ELA on the sheet resistance of poly-Si films were studied systematically.

2. Experimental

Resistors were used to investigate the sheet resistance of poly-Si film. The fabrication process of resistor is described as follows. a-Si film (100 nm) was deposited on the top of oxidized (0.5 lm) Si substrate by LPCVD at 550°C. Sample A, B, C, and G were crystallized by solid phase crystallization (SPC) at 600°C for 24 h in order to investigate the interference effect resulted from the cap-ping oxide and excimer laser. Sample D, E, and F were crystallized by ELA with optimum annealing conditions at substrate temperature of 400 °C and capped with silicon oxide films of 76, 100, and 117.5 nm, respectively. Phosphorous ion implantation was performed at accel-eration voltage of 40 keV and with dosage of 5 1015

cm2_{for all samples. After ion implantation, sample A,}

B, and C were capped with 76, 100, and 117.5 nm thick silicon oxide films by plasma enhanced chemical vapor deposition (PECVD) system. Scanning type KrF (248 nm) excimer laser with Gaussian shape of 1.8 mm width and 23 mm length were applied on these samples. The overlapping of two sequent shots and the laser frequency were fixed at 95% and 20 Hz. After ELA, the contact holes of resistors were formed. Al (500 nm) and Au/Al (200=500 nm) films were deposited on sample A, B, C, G and sample D, E, F respectively and patterned for metal pads. Al sintering process was carried out at 400°C for 30 min in a N2 gas ambient furnace. NH3 plasma

treatment was done at 300°C for 3 h. Finally, HP4156 was used to characterize the resistance of resistors. In order to investigate resistor-to-resistor variation, 25 re-sistors were probed for each laser irradiation condition.

3. Results and discussion

The reflectance of poly-Si film capped with various thicknesses of silicon oxide film is shown in Fig. 1. On the other hand, inset shows the simulation result of poly-Si film reflectance which varies with the thickness of silicon oxide under KrF excimer laser (248 nm) irradiation. It is known as the interference effect. The measured reflec-tance is slightly lower than that of simulation. It should be caused from the light scattering by rough surface of poly-Si film [6].

According to the analysis from Fig. 1, the sheet re-sistances of sample A, B, and C have a good agreement with the interference effect as shown in Fig. 2. To reduce the sheet resistance of poly-Si thin film, sample C needs just the two-thirds amount of energy what sample A does. Differing from the results published in reference [1], the sheet resistance does not increase again due to Nomenclature

Symbols

Rs sheet resistance, X=

Ea activation energy, eV

Ed energy density, mJ/cm2

Vg voltage across grain boundary, volt (V)

Va applied voltage across resistor, volt (V)

I current, ampere (A)

J current density, A/cm2

A area perpendicular to the ﬂow, cm2

A _{constant, A/cm}2_K2

k Boltzmann constant, J/K

T temperature, K

q elementary charge, 1:60218 1019_C

Fig. 1. Reflectance of poly silicon film capped with 76, 100, and 117.5 nm silicon oxide films. Inset shows the calculation result of poly-Si film reflectance varying with silicon oxide film thickness under KrF (248 nm) excimer laser exposing.

(3)

the amorphization of poly-Si film when high-laser flu-ence is applied. Due to the benefit of the scanning type laser beam with Gaussian shape, the silicon film can be retained in polycrystalline status during higher laser flu-ence annealing in our case. Consequently, the resistivity of poly-Si film does not increase again.

The sheet resistance of poly-Si film in sample G and A are shown in Fig. 3. From the optical analysis in Fig. 1, sample A absorbs the same or even higher laser energy than blank poly-Si film does under KrF ELA. However, the sheet resistance of poly-Si film in sample G is less than that of sample A. A reasonable explanation is that the capping oxide film acted as a heat absorber when poly-Si film was annealed with excimer laser. It de-creased the activation ability of excimer laser particu-larly at low-laser fluence annealing. Although the silicon oxide consumes some heat from the molten poly-Si film, the sheet resistances of poly-Si thin film in sample A, B, and C actually meet the requirements of low-tempera-ture poly-Si TFTs applications.

Figs. 4, 5, and 7 show the sheet resistance of sample D, E and F, respectively. After phosphorous ion im-plantation, the dopant concentrations of each sample are different due to the various thickness of capping oxide. Hall measurement was performed to take the carrier concentrations and Hall mobility of poly-Si film. Unfortunately, rough surface of poly-Si annealed by ELA, fluctuation of excimer laser energy as well as laser beam with Gaussian shape result in non-uniform grain structure of poly-Si film. They increase the difficulties to take exact data in Hall measurement. The approximate carrier concentrations of sample D, E, and F are 1:9 1020_{, 2:2}₁₀19_{, and 2:5}₁₀18_cm3_{, separately. It}

is well known that the carrier transport mechanism of

poly-Si ﬁlm greatly diﬀers from that of c-Si [7]. At low-level dopant concentrations, the potential barrier re-sulted from the charges around the grain boundary is low, and depletion regions of grain boundary will be through the entire grain. Carriers can easily transport across the grain boundary by thermionic emission. As the dopant concentration increases, more carriers are trapped at the grain boundary; the height of potential barrier of the grain boundary increases and the deple-tion region within the grain does not change until all

Fig. 2. Sheet resistance of poly-Si ﬁlm in sample A, B, and C.

Fig. 3. Sheet resistance of poly-Si ﬁlm in sample A (capped with 76 nm oxide ﬁlm) and sample G (without capping oxide).

(4)

traps are charged entirely. Carriers here are difficult to pass the potential barriers of grain boundaries. At high-dopant concentrations, all traps are charged and the added carriers, which are not trapped, cause the neutral regions within the grains. From the charge neutrality, the width of depletion regions decrease with increasing dopant concentration and so does the potential barrier heights. Carriers can transport across the grain bound-aries easily by thermionic emission or even by tunneling. The thickness of capping oxide of sample D is 76 nm. From the analysis of ion implantation simulation, we find that almost half amount of phosphorous ions pass through the oxide layer and stop in the poly-Si film. A high-carrier concentration of 1:9 1020_cm3_suppresses

the potential barrier of grain boundary and results in low-sheet resistance of sample D, as shown in Fig. 4. We also observe a peak around the laser energy density of 384 mJ/cm2_{. This energy density is similar with the}

op-timum crystallization energy of a-Si thin film with 100 nm thickness at room temperature. In addition, the poly-Si films of sample D, E, and F prepared by ELA with optimum laser energy at 400°C substrate temper-ature have average grain size about 600 nm, which is examined by SEM. The grain size of poly-Si film under room temperature laser annealing will not reduce until that the laser fluence is higher than full-melt threshold (FMT) fluence. The average grain size decreases to about 150 nm after room temperature laser annealing with laser energy which is higher than FMT fluence. So the peak in Fig. 4 could be realized as the grain size shrinkage, and the decrease of sheet resistance at higher laser energy should be the increase of poly-Si crystal-linity.

The sheet resistance of sample E exhibits an abnor-mal increase from 5 104 _{to 4}₁₀6 _X=_{around the}

laser energy density of 290 mJ/cm2 _{as shown in Fig. 5.}

From the analysis of Hall measurement, the approxi-mate carrier concentration of sample E is 2:2 1019

cm3_{. However, the Hall mobility of poly-Si ﬁlm}

de-creases from 29 to 8 cm2_{/V s when increasing laser}

energy. Besides the increase of grain boundaries, the increase of grain boundary potential barrier height should be the other reason resulting in abnormal in-crease of sheet resistance of poly-Si ﬁlm in sample E. In order to clarify this assumption, the I–V characteristics of poly-Si ﬁlm resistors irradiated with laser energy density of 250 and 336 mJ/cm2 _{were measured over a}

range of temperatures.

Assuming the potential barriers at each grain boundary of poly-Si ﬁlm are too wide for appreciable tunneling, the predominant transport mechanism is thermionic emission of majority carriers over the barri-ers. We suppose the applied voltage Va, which is across

the contacts of resistor, is dropped equally across the depletion zones at each of the grain boundaries and the voltage divides equally on both sides of each grain boundary. Then the voltage drop, Vg, across an

indi-vidual grain boundary is Vg¼

Va

N ð1Þ

where N is equal to the number of grain boundaries between the contacts.

By considering thermionic emission in the two di-rections, forward and reverse current, the net current density can be found [8].

J¼ JF JR ¼ A_T2_exp qVB kT exp qVg 2kT exp qVg 2kT ð2Þ or I¼ AJ ¼ 2K sinh qVa 2kTN ð3Þ where VB is the grain boundary barrier height at zero

bias, A is the area perpendicular to the ﬂow and K¼ AA_T2_exp qVB

kT

ð4Þ which were derived from Eqs. (2), and (3).

The I–V characteristics of poly-Si ﬁlm irradiated with laser energy density of 250 and 336 mJ/cm2 _{at various}

substrate temperatures are shown in Fig. 6. Empty scatter symbol indicates the measured results and solid line shows the fitted I–V curves of poly-Si films ac-cording to Eq. (3). Actually, the fitted I–V curves have a

(5)

good agreement with measured ones. The value of K can be extracted from each one of the curves. The inset in Fig. 6 shows the plot of lnðK=T2_{Þ vs. 1=kT . Again, the}

potential barrier height of grain boundary can be ex-tracted depending on Eq. (4). The potential barrier height of poly-Si ﬁlm annealed with laser energy density of 336 mJ/cm2 _{is 0.13 eV, which is larger than that of}

poly-Si ﬁlm annealed with laser energy density of 250 mJ/cm2 _{(0.04 eV). Therefore, the increase of grain}

boundary potential barrier height and shrinkage of grain size cause the abnormal increase in sheet resistance. In addition, the sheet resistance of poly-Si ﬁlm in sample E is similar with that of poly-Si ﬁlm in the lightly doped drain region of poly-Si TFT. These results can also be applied to the low-temperature lightly doped drain poly-Si TFTs fabrication.

The sheet resistance of sample F is not similar to that of sample E as shown in Fig. 7. Besides the deviation of sheet resistance near the FMT fluence, unobvious rela-tionship between sheet resistance and laser energy den-sity is observed, even if high-laser energy annealing decreases the grain size of poly-Si film. The approximate carrier concentration of poly-Si film in sample F is as low as 2:5 1018_cm3_{. It results in high resistivity within}

poly-Si grain. Hence, we cannot obviously observe the transformation of sheet resistance depending on the excimer laser energy density.

4. Conclusions

Dopant activation experiments have been accom-plished in phosphorus ion implanted poly-Si film by KrF ELA. The consequences point out that the poly-Si film capped with silicon oxide film will cause the interference effect during KrF ELA. And the sheet resistance of poly-Si film annealed with sufficient laser energy can meet the requirements of low-temperature poly-Si TFTs applica-tions. Moreover, we also clarify the mechanism of ab-normal increase of sheet resistance when the carrier concentration of poly-Si film approximates to 2:2 1019

cm3_{. The increase of potential barrier height from 0.043}

to 0.13 eV and the shrinkage of poly-Si grain size from 600 to 150 nm result in the abnormal increase of sheet resistance.

Acknowledgements

The authors would like to acknowledge the ﬁnancial support by the National Science Council, Taiwan, ROC, under the Contract NSC-89-2215-E-009-108 and tech-nical aid by Semiconductor Research Center (SRC) and SPRAD LAB, National Chiao Tung University, Tai-wan, ROC.

References

[1] Giust GK, Sigmon TW. High-performance thin-ﬁlm tran-sistors fabricated using excimer laser processing and grain engineering. IEEE Trans Electron Dev 1998;45(4):925–32. [2] Aaron M, Apostolos TV, Raj S. A systematic study and

optimization of parameters aﬀecting grain size and surface roughnes in excimer laser annealed polysilicon thin ﬁlms. J Appl Phys 1997;82(9):4303–9.

Fig. 6. Fitted and measured I–V characteristics of resistors in sample E over a range of temperature. Inset shows the ﬁtting of grain boundary potential barrier height.

(6)

[3] Al-Nuaimy EA, Marshall JM. Excimer laser crystallization and doping of source and drain regions in high quality amorphous silicon thin ﬁlm transistors. Appl Phys Lett 1996;69:3857–9.

[4] Weiner KH, Carey PG, McCarthy AM, Sigmon TW. An excimer-laser-based nanosecond thermal diﬀusion technique for ultra-shallow pn junction fabrication. Microelectron Eng 1993;20:107–30.

[5] Kramer KJ, Somit T, Isabella TL, John ED, Kenneth AW, Keith AB, Kurt HW. Resistless, area-selective ultrashallow

P+/N junction fabrication using projection gas immersion laser doping. Appl Phys Lett 1996;68:2320–2.

[6] Chiang KL, Dell’Oca CJ, Schwettmann FN. Optical eval-uation of polycrystalline silicon surface roughness. J Elect-rochem Soc 1979;126:2267–9.

[7] Ted Kamins. Polycrystalline silicon for integrated circuits and displays. 2nd ed. Dordrecht: Kluwer Academic; 1998. [8] George JK, Richard SM. Conduction properties of lightly

doped, polycrystalline silicon. Solid-State Electron 1978; 21:1045–51.