1245321 九、發明說明： 【發明所屬之技術領域】 本發明係有關一種非晶矽退火的應用方法，特別是有關一種近紅外波 段飛秒雷射在非晶矽退火的應用方法。 【先前技術】 近年來液晶顯示器目前在市場上已經取代傳統陰極射線管監視器，再 加上後pc時代來臨及負訊豕電之需求，液晶顯示器所能應用之產品隨之增 加，所以液晶顯示器產業被譽為下一個台灣的半導體產業。其中，主動式 液晶顯示器（Active Matrix LCD)中關鍵元件—低溫複晶矽薄膜電晶體 (low-temperature poly-Si TFTs，LTPS-TFT)由於具有高移動率、自我對 準(Self-alignment)、以及與CMOS製程相容等優點，因此可以將元件體積 縮小、儲存電容減少、畫素(Pixel)精細度與開口率提高、並可將驅動電路 直接製作在顯示器基板上，是大面積TFT LCD產業發展的關鍵技術。 習知關於雷射退火結晶非晶矽的文獻皆屬於準分子雷射退火(ELA)，高 月匕脈衝綠光，或穩疋功率輸出綠光雷射退火的範_，其所需的退火能量密 度較咼，且光此吸收效率對雷射波長，基材特性極其敏感，能得到較大晶 粒所需的退火能量密度範圍也較小。 有鑑於此，本發明係針對上述之問題，提出一種近紅外波段飛秒雷射 在非晶矽退火的應用方法，其係利用紅外波段的超快鈦藍寶石雷射進行非 晶矽的飛秒雷射退火，可得到晶粒尺寸大的多晶矽，及較佳之表面平整度。 1245321 【發明内容】 本發明之主要目的， 的應用方法，其係紅外波段飛秒雷射退火 性吸收能量的方式達舰融再結晶形成多晶石力 的多晶矽，及較佳之表面平整度。 係在提供—種近紅精段飛秒料細晶概火 的製程’使得非晶梦以非線的過程，可得到晶粒尺寸大 的製程， 的應用方法，其係利用掃描的方式可得到均勻 用在低溫多晶石夕(Low Temp· P〇ly-Si)製程。 本發明之另-目的，係在提供—觀紅外波段飛秒雷射切晶梦退火 且大面積的退火區域，可應 本發明之再-目的’係在提供—種近紅外波段飛秒雷射在非晶石夕退火 的應用方法’其與壓縮脈衝寬度技術搭g⑶可進—步降低退火雷射能量， 增加產能。 本發明之又-目的，其係利用超短脈衝之飛秒雷射進行非晶石夕的退 火，透過非線性吸收的過程進行能量的轉換，對基材㈣較不靈敏，有利 於量產時之參數優化。 根據本發明’其係利用近紅外波段飛秒雷射以掃描方式退火使非晶矽 表面產生電漿，同時經由非線性吸收的過程引發非晶矽的相變，再藉由搭 配掃描退火可得到晶粒大多晶矽，透過此一超短脈衝退火的過程，所需的 退火能量密度遠小於線性吸收較強的準分子雷射退火所需能量密度，可以 得到一個均勻的大面積退火區域。 底下藉由具體實施例配合所附的圖式詳加說明，當更容易瞭解本發明 之目的、技術内容、特點及其所達成之功效。 1245321 【實施方式】 本發明在超短脈衝（50fs)鈦藍寶石雷射（amplified Ti:sapphiR laser)非熱退火系統應用近紅外光（人^ 8⑼咖）飛秒雷射退火（ρ[Α)薄 膜電晶體（TFTs)元件製程中之非晶石夕主動區結晶及佈植區活化步驟上。 請參閱第一圖為本發明雷射退火之步驟流程圖，步驟S1提供一基板， 基板為設有非晶矽模的透明基板，其中非晶矽模可為一傳導材料、一半導 體材料或一介電材料且摻雜有磷或硼離子；步驟S2近紅外波段飛秒雷射以 掃描方式退火，使用一超短脈衝，其短脈衝波長為8〇〇nm、脈衝寬5〇fs、 能量密度為45mJ/cm2，在該非晶石夕材料上經由非線性吸收的過程引發非晶 矽的相變產生高密度的電漿，使非晶矽材質相變到一多晶矽材質，多晶矽 材質表面平整度小於4.5nm，其晶粒包括大於〇·8ιμ之平均晶粒大小。 請參閱第二圖為本發明與其它雷射退火結晶比較表，本發明與準分子 雷射退火及其他綠光雷射退火所得到的結晶機制、最佳晶粒尺寸、相對應 能量密度的比較，在相同晶粒尺寸下飛秒雷射退火所需的能量密度遠小於 其他退火技術，所需的退火總能量及脈衝數量也最小。 又本發明利用超短脈衝(50fs)鈦藍寶石雷射(amplified Ti:sapphire laser)非熱退火系統，可使非晶矽以非線性多光子吸收方式（multiple photo absorption)吸收紅外光能(8〇〇nm)成為複晶或單晶矽，此反應是不 同於準分子雷射退火之線性熱吸收（吸收波峰〜3〇〇nm)因此為非熱過程，另 外其高重複率脈衝輸出（Ι-kHz)特性可增進製程效率及電晶體的載子遷移 率。且矽材料對紅外光波長為穿透性，因此對鈦藍寶石短脈衝雷射能量幾 無熱線性吸收，此可進一步降低退火熱效應及局部製程可能性，因飛秒雷 1245321 射本身為非熱之製程可以針對離子損傷處進行恢復而不經過再結晶之過 程’可以使通道的晶體結構與源/難部分之晶體結構減呈連續的晶體構 造，此外柯雜子健之觀分舞持相_料會錢航應產生， 減少擴散反麟於職要鱗縮小岐非常_之條件。 本發月疋從非03雜晶開始，請參閱第三圖為高倍率電子顯微鏡⑽） 下之晶糊，f射脈_柯_密度刊掃描與雜描方式退 火所得晶粒之SEM圖，在不進行掃描時所得到的平均晶粒尺寸皆小於5〇服， 這是因為雷射脈衝、结束後熔融區域快速冷卻時伴隨發生的隨機再結晶所造 成的，但是使崎描方式進行退火時，獨的能量密度下卻可以使得晶粒 大小產生·的變化。第四圖為雷射能量密度與平均晶粒大小_圖，從 第四’)得知，非晶雜職秒雷射敎後辭均晶減小是隨著雷射能 篁密度增加而先增賴。在脈衝寬度47fs、基板溫度40〇。(：、_重複掃描 B夺，有最大的晶粒約800nm，石夕表層方均根粗糙度（R〇〇t—_-square roughness，RMS roughness)為2.37nm。依照晶粒的大小，可以將晶粒成 長分成二個區域，能量由低至高分別為部分熔融區、幾乎全熔區與完全熔 解區。又從第四圖⑹可知，增加掃描的脈衝數目並不會增加結晶狀況，而 在此飛熱退火方式中，較短的脈衝和較大的能量密度可以得到較大的晶 粒。利用AFM量測到退火後樣品表面的方均根粗糙度一般皆小於5nm，如第 五圖為晶粒側面立體圖，遠優於準分子雷射退火樣品的表面平整度（約 15nm) ° 第六圖為脈衝寬度為50fs時的拉曼譜線(a)及XRD結果(b)，圖中所附 1245321 為多晶石夕之TEM影像，由拉曼譜線可知再結晶比例達98%，具有相當好的結 晶性’由XRD結果可知其結晶的優選方向（111)與其他雷射退火技術所得相 同。第七圖為不同溫度下所得到的最大晶粒尺寸及相對應的能量密度(3)及 不同脈衝寬度相對應的溫度變化所造成的晶格放大率(b)，基板溫度愈高所 得最佳晶粒尺寸也越大，但所需的最佳能量較低。隨脈衝長度增加，因為 溫度增加而得到的晶格尺寸放大率也逐漸減少。 由非晶矽結晶習知結果，也加以應用於活化摻雜。由第八圖.N—type(推 雜磷）與P-type (摻雜硼）的摻雜離子濃度對深度的圖形得知，一般在電 晶體製程的離子摻雜後都需要經過加熱活化使得晶格能夠修補，以得到較 高的載子遷移率與開/關電流比，傳統常見的活化方式為爐管加熱與紅外光 快速熱退火方式（Rapid Thermal Annealing, RTA)，其中又以RTA方式因 為不易造成摻雜離子擴散，因此較爐管加熱更適用於各種元件的活化製 程，由第八圖可知··相較於105(rc之下的RTA製程得到的摻雜離子擴散情 形，無淪在室溫、l〇〇°C與200°C環境下所做的飛秒雷射活化皆可得到較佳 的離子分佈情形，從（A)和（B)兩圖可知飛秒雷射活化使得離子並無明 顯的擴散發生，除了離子擴散情形較佳外，從第九圖可知在摻雜磷的試片 經過能量密度〜30mJ/cm2飛秒雷射活化後，片電阻約下降至1〇〇 ohm/sheet，較活化前降低許多（活化前約107〇hm/sheet);另從第十圖可 知摻雜硼的試片經過飛秒雷射活化亦可使得片電阻下降為1〇〇〜2〇〇 ohm/sheet 〇 上述之近紅外波段飛秒雷射在非晶矽退火方法所得之結晶矽及摻雜活 1245321 化區，整合於TFT元件，可得證其實效性。在TFT元件製程中，將非晶石夕 結晶成為多晶♦的步驟，若使用spc方式與使用FLA方式得到如第十一圖 汲極電流-難電壓圖（Drain eurrent—Gate _哪，Μ)，其載子遷 移率為分別為13 onVVS，及23 cmVVS。FLA方式得到的載子遷移率約為 SPC方式的兩倍，SPC方式中之FA結晶參數為，6〇〇°c、18hr ; FLA結晶的 雷射能量密度為45 mJ/αη2、脈衝寬度為5〇fs、掃描重複率為_，又由第 十二圖載子遷移率變化圖雌示，由爐管結晶石夕(spc Si⑴⑽及似活 化之源汲極所構造之TFTs元件，其載子遷移率達到8· 7cmVvs，驗證似 確可修補因雜而損壞晶格。以上所述係藉由實_說明本發明之特點， 其目的在使熟習該技術者能暸解本發明之内容並據以實施，轉限定本發 明之專利範圍，故，凡其他未脫離本發明所揭示之精神所完成之等效修飾 或修改，仍應包含在以下所述之申請專利範圍中。 1245321 【圖式簡單說明】 第一圖為本發明雷射退火之步驟流程圖。 第二圖為本發明與其它雷射退火結晶比較表。 第三圖為高倍率電子顯微鏡下之晶粒圖。 第四圖為雷射能量密度與平均晶粒大小關係圖。 第五圖為晶粒側面立體圖。 第六圖為脈衝寬度為50fs時的拉曼譜線及XRD結果。 第七圖為不同溫度下所得到的最大晶粒尺寸及相對應的能量密度及不同脈 衝寬度相對應的溫度變化所造成的晶格放大率。 第八圖為使用本發明後N-type與Ρ-type的摻雜離子濃度對深度的圖形。 第九圖為試片摻雜鱗的片電阻。 第十圖為試片摻雜侧的片電阻。 第十一圖為汲極電流_閘極電壓圖。 第十一圖為載子遷移率變化圖。 【主要元件符號說明】1245321 IX. Description of the invention: [Technical field to which the invention belongs] The present invention relates to an application method for annealing amorphous silicon, and more particularly to an application method for annealing near-infrared band femtosecond lasers on amorphous silicon. [Previous technology] In recent years, liquid crystal displays have replaced traditional cathode ray tube monitors in the market. Coupled with the advent of the post-PC era and the demand for negative signal electricity, the number of products that can be applied to liquid crystal displays has increased, so liquid crystal displays. The industry is known as the next semiconductor industry in Taiwan. Among them, the low-temperature poly-Si TFTs (LTPS-TFT), a key element in Active Matrix LCDs, have high mobility, self-alignment, And compatibility with CMOS process, so it can reduce the size of components, reduce storage capacitance, improve pixel fineness and aperture ratio, and directly drive circuits on the display substrate. It is a large-area TFT LCD industry. Key technologies for development. The literatures about laser annealing of crystalline amorphous silicon are all examples of excimer laser annealing (ELA), high moon pulse green light, or stable power output green laser annealing. The required annealing energy is The density is relatively high, and the absorption efficiency of light is extremely sensitive to the laser wavelength and substrate characteristics. The range of annealing energy density required to obtain larger crystal grains is also small. In view of this, the present invention is directed to the above-mentioned problem, and proposes a method for applying near-infrared band femtosecond laser to annealing amorphous silicon, which uses ultra-fast titanium sapphire laser in infrared band to perform femtosecond laser of amorphous silicon. Radiation annealing can obtain polycrystalline silicon with large grain size and better surface flatness. 1245321 [Summary of the invention] The main purpose of the present invention is an application method, which is a method of absorbing energy in the infrared band femtosecond laser annealing to achieve polycrystalline silicon with polycrystalline silicon power by melting and recrystallization, and a better surface flatness. The application method is to provide a kind of near red fine segment femtosecond material fine-grained rough fire process, which enables the amorphous dream to take a non-linear process to obtain a process with a large grain size. The application method is obtained by scanning. Uniformly used in the process of Low Temp. Poly-Si. Another object of the present invention is to provide an infrared annealing region of femtosecond laser cut crystals and a large area of annealed area. The purpose of the present invention is to provide a near infrared region femtosecond laser. The application method of annealing in amorphous stone, which can be combined with the compression pulse width technology to further reduce the annealing laser energy and increase the production capacity. Another purpose of the present invention is to use an ultra-short pulsed femtosecond laser to anneal amorphous stones, and to perform energy conversion through a process of non-linear absorption, which is less sensitive to substrates and is beneficial to mass production. Parameter optimization. According to the present invention, it uses a near-infrared band femtosecond laser to perform scanning annealing to generate plasma on the surface of amorphous silicon, and at the same time initiates a phase change of amorphous silicon through a nonlinear absorption process, which can be obtained by matching scanning annealing. Most of the grains are crystalline silicon. Through this ultra-short pulse annealing process, the required annealing energy density is much smaller than the energy density required for linearly-absorbed excimer laser annealing, and a uniform large-area annealing region can be obtained. In the following, detailed descriptions are provided by specific embodiments in conjunction with the accompanying drawings to make it easier to understand the purpose, technical content, features and effects of the present invention. 1245321 [Embodiment] The present invention applies near-infrared light (human ^ 8⑼ca) femtosecond laser annealing (ρ [Α) film to an ultra-short pulse (50fs) titanium sapphire laser (amplified Ti: sapphiR laser) non-thermal annealing system In the process of crystal (TFTs) element manufacturing, the steps of crystallization and activation of the active region of the amorphous stone are performed. Please refer to the first figure for the flowchart of the laser annealing process of the present invention. Step S1 provides a substrate. The substrate is a transparent substrate provided with an amorphous silicon mold. The amorphous silicon mold may be a conductive material, a semiconductor material, or a substrate. The dielectric material is doped with phosphorus or boron ions; step S2 the near-infrared band femtosecond laser is annealed in a scanning manner using an ultrashort pulse with a short pulse wavelength of 800 nm, a pulse width of 50 fs, and an energy density It is 45mJ / cm2. The amorphous silicon material causes a phase transition of amorphous silicon to generate a high-density plasma through a non-linear absorption process. The amorphous silicon material is transformed into a polycrystalline silicon material, and the surface flatness of the polycrystalline silicon material is less than At 4.5 nm, its grains include an average grain size greater than 0.8 μm. Please refer to the second figure for a comparison table of the present invention and other laser annealing crystals. A comparison of the crystallization mechanism, optimal grain size, and corresponding energy density obtained by the present invention with excimer laser annealing and other green laser annealing. In the same grain size, the energy density required for femtosecond laser annealing is much smaller than other annealing techniques, and the total annealing energy and pulse number required are also the smallest. In addition, the present invention uses an ultra-short pulse (50fs) titanium sapphire laser (amplified Ti: sapphire laser) non-thermal annealing system to enable amorphous silicon to absorb infrared light energy in a non-linear multiple photo absorption (80 °). 〇nm) becomes complex or single crystal silicon. This reaction is different from the linear thermal absorption (absorption peak ~ 300nm) of excimer laser annealing, so it is a non-thermal process, and its high repetition rate pulse output (I- (kHz) characteristics can improve process efficiency and transistor carrier mobility. And the silicon material is transparent to the wavelength of infrared light, so it has almost no thermal linear absorption of short-pulse laser energy of titanium sapphire, which can further reduce the thermal effect of annealing and the possibility of local process, because the femtosecond lightning 1245321 is itself non-thermal. The process can recover the ion damage without undergoing the process of recrystallization. 'The crystal structure of the channel and the crystal structure of the source / difficult part can be reduced to a continuous crystal structure. In addition, the perspective of Ke Zazi's point of view is consistent. The meeting of Qian Hang should be generated, and the conditions for reducing proliferation should be reduced. This issue begins with non-03 heterocrystals. Please refer to the third picture for high-magnification electron microscopy.) The SEM image of the crystal paste under f-pulse _ke_density scan and anodized annealing. The average grain size obtained when no scanning is performed is less than 50 μm. This is caused by laser pulses and random recrystallization that occurs during rapid cooling of the molten region after the end. Under the unique energy density, the grain size can be changed. The fourth graph is the laser energy density and the average grain size. From the fourth '), it is known that the amorphous heterogeneous second laser is reduced after the average homogeneity is increased as the laser energy density increases. Lai. The pulse width was 47 fs and the substrate temperature was 40 °. (:, _ Repeated scanning B, there is a maximum grain size of about 800nm, the surface root mean square roughness (RMS) of Shi Xi is 2.37nm. According to the size of the grain, the crystal can be The grain growth is divided into two regions, and the energy from low to high is a partial melting zone, an almost full melting zone and a complete melting zone. It can be seen from the fourth figure that increasing the number of scanning pulses does not increase the crystallization state, and flying here In the thermal annealing method, larger grains can be obtained with shorter pulses and larger energy densities. The root mean square roughness of the surface of the annealed samples measured by AFM is generally less than 5 nm. , Which is much better than the surface flatness of the excimer laser annealing sample (about 15nm) ° The sixth figure shows the Raman spectrum (a) and XRD result (b) when the pulse width is 50fs. The attached 1245321 in the figure is mostly From the TEM image of Spar Stone, it can be seen from the Raman spectrum that the recrystallization ratio is 98%, which has quite good crystallinity. From the XRD results, it is known that the preferred direction of the crystal (111) is the same as that obtained by other laser annealing techniques. The picture is obtained at different temperatures The maximum grain size and corresponding energy density (3) and the lattice magnification (b) caused by temperature changes corresponding to different pulse widths, the higher the substrate temperature, the greater the optimal grain size, but the The required optimal energy is lower. As the pulse length increases, the lattice size magnification obtained due to the increase in temperature also gradually decreases. The known results from amorphous silicon crystals are also applied to active doping. From the eighth figure. The pattern of doped ion concentration versus depth for N-type (doped phosphorus) and P-type (doped boron) is known. Generally, after ion doping in the transistor process, it needs to be activated by heating to enable the lattice to be repaired. In order to obtain a higher carrier mobility and on / off current ratio, the traditional and common activation methods are furnace tube heating and infrared light rapid thermal annealing (RTA), and the RTA method is not easy to cause doping. Ion diffusion, so it is more suitable for the activation process of various elements than furnace heating. As shown in the eighth figure, compared with the doped ion diffusion obtained by the RTA process under 105 (rc), it is not reduced to room temperature, l 〇〇 ° C 和 20 Femtosecond laser activation performed at 0 ° C can obtain better ion distribution. From the two graphs (A) and (B), we can know that femtosecond laser activation does not cause significant diffusion of ions, except for ions. In addition to the better diffusion situation, it can be seen from the ninth figure that after the phosphor-doped test piece is activated by a femtosecond laser with an energy density of ~ 30 mJ / cm2, the sheet resistance is reduced to about 100 ohm / sheet, which is much lower than before activation ( (About 107 ohm / sheet before activation); from the tenth figure, it can be seen that the boron-doped test strip can also reduce the sheet resistance to 100-200 ohm / sheet by femtosecond laser activation. The wave band femtosecond laser is integrated in the crystalline silicon and the doped active 1245321 region obtained by the amorphous silicon annealing method, and is integrated into the TFT element, which can prove its effectiveness. In the TFT element manufacturing process, the step of crystallizing amorphous stone into polycrystalline silicon. If the spc method and the FLA method are used, the drain current-difficult voltage diagram (Figure 11) is obtained (Figure 11). , Its carrier mobility is 13 onVVS and 23 cmVVS, respectively. The carrier mobility obtained by the FLA method is about twice that of the SPC method. The FA crystallization parameter in the SPC method is 600 ° C, 18hr; the laser energy density of the FLA crystal is 45 mJ / αη2, and the pulse width is 5 〇fs, the scanning repetition rate is _, and it is shown in the twelfth figure of the carrier mobility change chart. The carrier migration of TFTs elements constructed by the furnace tube crystal stone evening (spc Si⑴⑽ and the activated source drain) The rate reaches 8.7cmVvs, and it seems that it can indeed repair the damaged lattice due to impurities. The above description explains the characteristics of the present invention by realizing its purpose, so that those skilled in the art can understand the content of the present invention and implement it accordingly. , To limit the scope of patents of the present invention, so all other equivalent modifications or modifications made without departing from the spirit of the present invention should still be included in the scope of patent application described below. 1245321 [Schematic description of the diagram] The first picture is a flowchart of the laser annealing process of the present invention. The second picture is a comparison table of the invention and other laser annealing crystals. The third picture is a grain diagram under a high magnification electron microscope. The fourth picture is the laser energy Density vs. average grain size The fifth figure is the side perspective view of the crystal grains. The sixth figure is the Raman spectrum and XRD results when the pulse width is 50 fs. The seventh figure is the maximum grain size and corresponding energy density obtained at different temperatures and Lattice magnification caused by temperature changes corresponding to different pulse widths. The eighth figure is a graph of the doping ion concentration versus depth of the N-type and P-type after using the present invention. The ninth figure is the doped scale of the test piece The tenth figure is the sheet resistance of the doped side of the test strip. The eleventh figure is the drain current_gate voltage chart. The eleventh figure is the carrier mobility change chart. [Description of the main component symbols]