1267933 06810twf2.doc/006 95-7-12 九、發明說明: 本發明是與P型金氧半場效電晶體(Metal Oxide semiconductor Field Effect Transistors,MOSFET’s)可靠性 測試有關,且特別是有關於P型金氧半場效電晶體最大閘 極電流之預測方法。 半導體電晶體元件在經過一段時間後,元件之可靠性 和性能是非常的重要。影響元件壽命之主要現象是熱載子 導致的退化現象。根據功率定律函數(Power Law Function) 可發現P型金氧半場效電晶體之有效元件壽命(r)的極限 (此極限主要是由熱載子退化所造成)和閘極電流Ic有關。 由 Ong 等人於 1990 年發表於 IEEE Transactions on Electron Device期干[l(Vol.37,No·7,Jιιly)中所提出之「P型金氧半場 效電晶體的表面通道之熱載子電流模擬和元件退化 (Hot-Carrier Current Modeling and Device Degradation in Surface-Channel p-MOSFET’s)」中所記載。在數個較高汲 極(VDS)電壓下測量τ値,經由這些高應力壽命(high-stress lifetime),推斷在低操作電壓時的τ値。 一般在r値與Ic之間存在下列之關係: r =A!*(lG/W)m (l) 其中W爲元件的寬度,Ai爲常數,1(3爲對應於汲極電 壓所測得之最大閘極電流,m爲(log r )對(log Ic/W)作圖所 得之斜率。上述之方程式在此以方程式1來表示。典型之 m値大約m=-l·5,如Ong等人之文獻所記載。根據方程式 1267933 95-7-12 06810twf2.doc/006 1之形式’使用加速應力測試以測定元件壽命,r的測量値 與IG/W之間的關係常以數個VDS値,作成l〇g-l〇g關係圖。 然而,在較低之操作電壓時,很難量測到Ic之最大値。 一般而言,Ic之最大値小於ΙΟρΑ,通常超出傳統測試儀器 所能夠測量範圍之鑑別度及準確度。這使得由高電壓壽命 推斷在低操作電壓時的τ値之計算方法變得不可行。因此 需要一個模擬閘極電流IG之方法,能夠有效的測量元件壽 命。 本發明揭露一種對P型金氧半場效電晶體在不同汲極 電壓Vds下預測最大閘極電流I<3_MAX之方法。根據實驗數 據分析得到方程式IG_MAX =A2*exp(B/VDSAn),此方程式可 準確的預測最大閘極電流。其中η約爲4,A2和B爲在相 當高VDS電壓以實際數據利用趨勢線預測法所測出之係 數。在求出IG_MAX預測方程式之係數後,即可計算在標準 操作電壓(nominal operating voltage)之 IG_MAX 。之後以此 Ig_max値利用方程式計算P型金氧半場效 電晶體的有效壽命r。 圖式之簡單說明 爲使本發明之上述和其他目的、特徵、和優點能更明 顯易懂,並進一步提供發明專利範圍之解釋,下文特舉一 較佳實施例,並配合所附圖式,作詳細說明如下: 第1圖閘極電流最大値Ic_max與汲極電壓VDS的實 際數據關係圖。 第2圖有效壽命r與閘極電流最大値IC_MAX的實際 1267933 95-7-12 ' 06810twf2.doc/006 數據關係圖 實施例 本發明是有關於P型金氧半場效電晶體在不同汲極電 壓VDS下’預測較低操作電壓vDS之最大閘極電流Ic_MAX 之模式。藉由測出之最大閘極電流IG_MAX,可求出元件之 有效壽命r。 本發明在本質上是以實驗爲基礎的。是在不同電壓參 數下測量P型金氧半場效電晶體之最大閘極電流Ic_MAX, 並預測在較低操作電壓下所產生之最大閘極電流。 此種預測方法在1·8、2.5、3.3以及5伏特操作電壓下之P 型金氧半場效電晶體均得到良好之驗證。 以一具體的例子可得到下列實際數據: 表一1267933 06810twf2.doc/006 95-7-12 IX. Description of the invention: The present invention relates to reliability testing of P-type Metal Oxide Semiconductor Field Effect Transistors (MOSFET's), and in particular to P-type A method for predicting the maximum gate current of a gold-oxygen half-field effect transistor. After a period of time, the reliability and performance of the components of the semiconductor transistor are very important. The main phenomenon affecting the life of components is the degradation caused by hot carriers. According to the Power Law Function, the limit of the effective element lifetime (r) of the P-type MOS field-effect transistor (which is mainly caused by the degradation of the hot carrier) is related to the gate current Ic. The hot carrier current of the surface channel of a P-type gold-oxygen half-field effect transistor proposed by Ong et al. in 1990 published in IEEE Transactions on Electron Device [l(Vol.37, No.7, Jιιly) "Hot-Carrier Current Modeling and Device Degradation in Surface-Channel p-MOSFET's" is described. τ 测量 is measured at several higher dipole (VDS) voltages, and τ 在 at low operating voltages is inferred via these high-stress lifetimes. Generally, the following relationship exists between r値 and Ic: r = A!*(lG/W)m (l) where W is the width of the component, Ai is a constant, and 1 (3 is measured corresponding to the drain voltage) The maximum gate current, m is the slope obtained by plotting (log r ) versus (log Ic/W). The above equation is expressed here by Equation 1. Typically m値 is approximately m=-l·5, such as Ong According to the literature of et al., the accelerated stress test is used to determine the life of the component according to the equation of the equation 1267933 95-7-12 06810twf2.doc/006 1 . The relationship between the measured 値 and IG/W of r is often several VDS.値, create a graph of l〇gl〇g. However, at lower operating voltages, it is difficult to measure the maximum I of Ic. In general, the maximum 値 of Ic is less than ΙΟρΑ, which is usually beyond the reach of traditional test instruments. The degree of discrimination and accuracy of the range makes it impossible to infer the calculation method of τ値 at low operating voltage from high voltage life. Therefore, a method of simulating the gate current IG is required to effectively measure the life of the component. The invention discloses a prediction of P-type gold-oxygen half field effect transistor under different gate voltage Vds The method of large gate current I<3_MAX. According to the experimental data analysis, the equation IG_MAX = A2*exp(B/VDSAn) is obtained, which can accurately predict the maximum gate current. Among them, η is about 4, and A2 and B are equivalent. The high VDS voltage is measured by the trend line prediction method based on the actual data. After the coefficient of the IG_MAX prediction equation is obtained, the IG_MAX at the nominal operating voltage can be calculated, and then the equation is calculated by using Ig_max値. The useful life of the P-type MOS field-effect transistor is the following description of the invention and other objects, features, and advantages of the present invention, and further explanation of the scope of the invention patent, A preferred embodiment, together with the drawings, is described in detail as follows: Figure 1 shows the relationship between the maximum gate current Ic_max and the actual voltage of the drain voltage VDS. Figure 2 The effective life r and the maximum gate current Actual 126783 of IC_MAX 95-7-12 '06810twf2.doc/006 Data Diagram Embodiment The present invention relates to P-type gold-oxygen half field effect transistors under different gate voltages VDS' The mode of measuring the maximum gate current Ic_MAX of the lower operating voltage vDS. The effective lifetime r of the component can be found by measuring the maximum gate current IG_MAX. The invention is essentially experimental based. The maximum gate current Ic_MAX of the P-type MOS field-effect transistor is measured under voltage parameters, and the maximum gate current generated at a lower operating voltage is predicted. This prediction method is at 1.8, 2.5, 3.3, and 5 volts. The P-type MOS half-field effect transistor under operating voltage has been well verified. The following actual data can be obtained with a specific example: Table 1
汲極電壓VDS Ι〇_ΜΑΧ測量値 閘極電壓vc I(}_MAX預測値 -3.3 3.34 ρΑ -0.65 V 3.08 pA -3.6 88.32 ρΑ -0.75 V 107 pA -3.9 1344.2 ρΑ -0.8 V 1110 pA -4.2 5087.9 ρΑ -0.75 V 5435 pA -4.5 13504 ρΑ -0.8 V 16493 pA -4.8 44125 ρΑ -0.8 V 36514 pA 產生上述數據之元件爲3.3 V操作電壓之Ρ型金氧半場 效電晶體,此Ρ型金氧半場效電晶體之元件閘極寬度爲20 微米,閘極長度爲0.205微米。其他不同尺寸之電晶體(此 尺寸是指元件之閘極寬度與閘極長度)及不同製程也可得 1267933 06810twf2. doc/006 95-7-12 到其他數據,但是爲簡述其重要性,在此只以一組數據作 說明。並且要注意閘極電壓之調整,直到發現最大閘極電 流 Ig_max 0 發明者也發現其他PMOS電晶體元件之數據也符合下 列之預測模式。經由數據測試下列之數學模式可以精確的 預測 Ig_max :The drain voltage VDS Ι〇 ΜΑΧ ΜΑΧ measurement 値 gate voltage vc I (} _MAX prediction 値 -3.3 3.34 ρ Α -0.65 V 3.08 pA -3.6 88.32 ρ Α -0.75 V 107 pA -3.9 1344.2 ρ Α -0.8 V 1110 pA -4.2 5087.9 Α 0.7 -0.75 V 5435 pA -4.5 13504 ρΑ -0.8 V 16493 pA -4.8 44125 ρΑ -0.8 V 36514 pA The 产生-type MOS field device with the above-mentioned data is 3.3 V operating voltage, this 金 type gold oxide half field The gate width of the effect transistor is 20 micron and the gate length is 0.205 micron. Other different sizes of transistors (this size refers to the gate width and gate length of the device) and different processes can also be obtained 1267933 06810twf2. doc /006 95-7-12 to other data, but to briefly describe its importance, only a set of data is used here. Also pay attention to the adjustment of the gate voltage until the maximum gate current Ig_max 0 is found. The inventor also found The data of other PMOS transistor components also conforms to the following prediction modes. The following mathematical modes can be used to accurately predict Ig_max:
Ig_max -A2*exp(B/VDsAn) (2) 其中VDS爲元件汲極電壓,人2爲常數,B爲ln(IG)對 l/VDSn作圖所得之斜率。上述之方程式在此以方程式2表 示。根據實驗的測量,若η爲4,使R2趨近爲1。元件在 電壓1.5和5伏特之間操作所歸納出之η値約爲4。然而可 由此領會對於其他未測試之不同元件,η値可能會不同。由 實驗上所得之實際收集數據以ln(Ic)對l/VDSn作圖可測得 常數人2與B之値。請參照第1圖,相關之數據列在表一。 使用傳統之數學線性趨勢線預測法技術’測定此p型金氧 半場效電晶體試驗結果,可得到常數A2爲542382以及常 數B爲-1432.4。因此使用相同參數於P型金氧半場效電晶 體之預測模式上,此預測模式成爲:Ig_max -A2*exp(B/VDsAn) (2) where VDS is the component's drain voltage, human 2 is a constant, and B is the slope obtained by plotting ln(IG) versus l/VDSn. The above equation is expressed here by Equation 2. According to the experimental measurement, if η is 4, R2 approaches 1 . The component is operated at a voltage between 1.5 and 5 volts and the sum of η 値 is about 4. However, it can be appreciated that for other untested different components, η値 may be different. The actual collected data obtained from the experiment can be measured by ln(Ic) versus l/VDSn to determine the constant between humans 2 and B. Please refer to Figure 1, the relevant data is listed in Table 1. Using the conventional mathematical linear trend line prediction technique to determine the p-type gold oxide half field effect transistor test results, the constant A2 was 542382 and the constant B was -1432.4. Therefore, using the same parameters on the prediction mode of the P-type MOS field-effect transistor, this prediction mode becomes:
Ig_max =542382*e(-1432.4/VDsAn) 使用上述方程式所得到之1C-MAX預測値列在表一之最 右邊一欄。如第1圖,模擬方程式”擬合(fit)”2 r2値非常 接近1,表示預測値對於測量數據相當吻合(r2=0·9974)。使 用此模式可在非常低之VDS電壓下預測1C-MAX (此低VDS 電壓在常態所引起之閘極電流是非常小而且可能無法測 1267933 06810twf2.doc/006 95-7-12 量)。這些低VDS電壓一般是小於3·3伏特’並且在次世代 半導體元件上降的更低,大約只有丨·8伏特。 因此藉由預測在低操作電壓之1C_MAX。可在經由加速 應力測試,可以方程式1精確的推斷元件壽命^: r =A!*(lG/W)m (i) 請參照第2圖,預測元件寬度爲20微米,閘極長度爲 0.205微米之P型金氧半場效電晶體元件之有效壽命r。首 先,實際的測量有效壽命r。有效壽命τ是由不同的IC_MAX 値所推算出來(而1CLMAX則由不同的VDS値所推算出來)。之 後即可根據實驗的計算得到常數Αι以及常數m。例如繪於 第2圖中之具體的例子,測量常數値爲5.9812以及常數 m 値爲-1.2792。 接著對標準操作電壓VDS使用上述之閘極電流IC_MAX 預測模式,可計算有效壽命τ。此技術圖解的顯示在第2 圖中,其中3.3伏特之標準操作電壓產生大約1Ε21小時之 有效壽命。 上述之模擬技術,可很容易的運用在製程可靠性分析 中。尤其是在特別設計之Ρ型金氧半場效電晶體上預測有 效壽命r。例如,假設一個寬度設計爲20微米,閘極長度 設計爲0.18微米之P型金氧半場效電晶體。進一步的假設 此P型金氧半場效電晶體之標準操作電壓爲1.8伏特。以 目前之技術中,因爲在低電壓測量是很困難,所以 預測有效壽命τ也很困難。使用本發明之方法,對於大量 95-7-12 1267933 06810twf2.doc/006 製造P型金氧半場效電晶體之生產製程’可有效且快速得 到P型金氧半場效電晶體熱載子測試’正常操作電壓下之 有效壽命r。 首先,對於P型金氧半場效電晶體提供較標準操作電 壓1.8伏特爲高之不同VDS電壓値’進行加速應力測試。例 如,將P型金氧半場效電晶體加壓在等於或高於3伏特v〇s 電壓下操作。接著’對每一個vds電壓値測量最大閘極電 流IG_max。更進一步的由這些不同vds電壓値可得到相對 應之有效壽命r。這些數據可得到下列形式: VDS Ig max r 3.0伏特 500 pA 100小時 3.5伏特 3000 pA 20小時 4.0伏特 10000 pA 3.0小時 4.5伏特 60000 pA 0.5小時 此數據可用來測量閘極電流模擬方程式(方程式2)之 常數A2以及B的値。更進一步的也可以用來測量方程式1 之係數Α:以及m的値。 當測出方程式1與方程式2之係數値後,方程式2以 標準操作電壓爲1.8伏特,測出最大閘極電流IC_MAX値。 之後,以方程式1 (連同測出之係數Ai以及m的値)利用 方程式2所預測之最大閘極電流IC_MAX値計算有效壽命 r。使用上述之方法可減少相當多的循環測試時間。另外 1267933 06810twf2.doc/006 95-7-12 可建立軟體指示電腦設備執行趨勢線預測,計算步驟如上 述。 雖然本發明已以一較佳實施例揭露如上,然其並非用 以限定本發明,任何熟習此技藝者,在不脫離本發明之精 神和範圍內,當可作各種之更動與潤飾,因此本發明之保 護範圍當視後附之申請專利範圍所界定者爲準。Ig_max = 542382 * e (-1432.4 / VDsAn) The 1C-MAX prediction obtained using the above equation is listed in the rightmost column of Table 1. As shown in Fig. 1, the simulation equation "fit" 2 r2 値 is very close to 1, indicating that the prediction 相当 is quite consistent with the measured data (r2 = 0.9974). Use this mode to predict 1C-MAX at very low VDS voltages (the gate current caused by this low VDS voltage in the normal state is very small and may not be able to measure 1267933 06810twf2.doc/006 95-7-12). These low VDS voltages are typically less than 3.3 volts and are lower on the next generation of semiconductor components, which is only about 丨8 volts. Therefore, by predicting 1C_MAX at a low operating voltage. The component life can be accurately estimated by Equation 1 through the accelerated stress test ^: r =A!*(lG/W)m (i) Please refer to Figure 2 to predict the component width of 20 μm and the gate length of 0.205 μm. The effective life of the P-type gold-oxygen half-field effect transistor component. First, the actual measured effective life r. The effective lifetime τ is derived from different IC_MAX ( (and 1CLMAX is derived from different VDS値). The constant Αι and the constant m can then be obtained from the experimental calculations. For example, in the specific example depicted in Figure 2, the measurement constant 値 is 5.9812 and the constant m 値 is -1.2792. The effective lifetime τ can then be calculated using the above-described gate current IC_MAX prediction mode for the standard operating voltage VDS. The diagram of this technical diagram is shown in Figure 2, where a standard operating voltage of 3.3 volts yields an effective lifetime of approximately 1 Ε 21 hours. The above simulation techniques can be easily applied to process reliability analysis. In particular, the effective lifetime r is predicted on a specially designed 金-type MOS field effect transistor. For example, suppose a P-type MOSFET is designed to have a width of 20 microns and a gate length of 0.18 microns. Further hypothesis The standard operating voltage of this P-type MOS field-effect transistor is 1.8 volts. In the current technology, since it is difficult to measure at low voltage, it is difficult to predict the effective lifetime τ. Using the method of the present invention, a large number of 95-7-12 1267933 06810twf2.doc/006 manufacturing process for producing a P-type gold-oxygen half-field effect transistor can effectively and quickly obtain a P-type gold-oxygen half-field effect transistor thermal carrier test. Effective life r at normal operating voltage. First, an accelerated stress test was performed for a P-type MOS field-effect transistor with a different VDS voltage 値' higher than a standard operating voltage of 1.8 volts. For example, a P-type MOS field effect transistor is operated at a voltage equal to or higher than 3 volts v 〇s. Next, the maximum gate current IG_max is measured for each vds voltage 値. Further, the corresponding effective lifetime r can be obtained from these different vds voltages. These data can be obtained in the following form: VDS Ig max r 3.0 volts 500 pA 100 hours 3.5 volts 3000 pA 20 hours 4.0 volts 10000 pA 3.0 hours 4.5 volts 60000 pA 0.5 hours This data can be used to measure the gate current simulation equation (Equation 2) The constant A2 and the B of B. Further, it can also be used to measure the coefficient 方程 of Equation 1 and the 値 of m. When the coefficients of Equation 1 and Equation 2 are measured, Equation 2 measures the maximum gate current IC_MAX値 with a standard operating voltage of 1.8 volts. Thereafter, the effective lifetime r is calculated by Equation 1 (along with the measured coefficients Ai and m) using the maximum gate current IC_MAX値 predicted by Equation 2. Using the above method can reduce a considerable amount of cycle test time. In addition, 1267933 06810twf2.doc/006 95-7-12 software can be established to instruct the computer equipment to perform trend line prediction. The calculation steps are as described above. Although the present invention has been described above in terms of a preferred embodiment, it is not intended to limit the invention, and it is obvious to those skilled in the art that various changes and modifications can be made without departing from the spirit and scope of the invention. The scope of the invention is defined by the scope of the appended claims.
1010