National Sun Yat-sen University Institutional Repository:Item 987654321/28475

行政院國家科學委員會專題研究計畫 成果報告

以溶膠凝膠法製備鈦酸鉛系列薄膜作為氫離子感測場效電

晶體特性之研究(2/2)

計畫類別： 個別型計畫

計畫編號： NSC92-2216-E-110-008-

執行期間： 92 年 08 月 01 日至 93 年 07 月 31 日

執行單位： 國立中山大學電機工程學系(所)

計畫主持人： 陳英忠

報告類型： 完整報告

處理方式： 本計畫可公開查詢

中 華 民 國 93 年 11 月 2 日

中文摘要

本計畫以溶膠-凝膠(Sol-gel)法製備(Pb,La)TiO

與(Pb,Mg)TiO

等鈦酸鉛系

列材料作為絕緣感測薄膜層於氫離子感測場效電晶體上，進行一系列感測特性

及其非理想因素之探討，以尋求摻雜與製程條件之最佳化參數，並完成一 pH

量測系統之設計與製作。根據結果顯示，Mg

_{及 La}

_{分別被摻雜於 PbTiO}

感

測膜中，以修飾(Modification)相關感測特性，發現受體(Acceptor) Mg 將有利

於 pH 響應度(58-59 mV/pH)、時漂(0.4 mV/h)、遲滯(1-3) mV 及衰減率(–0.2

µV/pH-day)之改善；反之，施體(Donor) La 將不利於感測特性之改善。

關鍵字：鈦酸鉛鑭、鈦酸鉛鎂、氫離子感測場效電晶體

英文摘要

In this project, sol-gel-derived (Pb,Mg)TiO

and (Pb,La)TiO

membranes are

applied as a novel pH-sensing layer of ion-sensitive field-effect transistor. Their

sensing responses and nonideal properties are examined. As a result, the former is

a great benefit to improve the pH-sensing characteristics, which exhibits the pH

response of 58-59 mV/pH, the drift of below 0.4 mV/h, the hysteresis of 1-3 mV

and the reduction rate of -0.2 µV/pH-day. Finally, a digital pH meter has been

successfully developed.

目 錄

中文摘要 ... I

英文摘要 ...II

圖表目錄 ... IV

一、 計畫緣由與目的 ... 1

二、結果與討論 ... 3

三、計畫成果自評 ... 6

四、參考文獻 ... 7

五、附錄... 17

圖 表 目 錄

圖一 (a)PMT 與(b)PLT EIS 結構之不同摻雜濃度的 pH 響應曲線(Firing

temperature of 400℃) ... 10

圖二 (a) PMT(4)與(b) PMT(7)薄膜之不同燒結溫度的 XRD 分析...11

圖三 數位酸鹼感測計基本架構圖 ... 12

圖四 系統設計流程 ... 13

表一 作用於非飽和區之 PT、 PMT(4)及 PLT(3) gate ISFET 之感測響應值與

pH 有效範圍 ... 14

表二 作用於飽和區之 PT、 PMT(4)及 PLT(3) gate ISFET 之感測響應值與線性

相關係數 ... 14

表三 PT、 PMT(4)及 PLT(3) gate ISFET 之非理想因素... 15

表四 不同感測材料之 pH 感測特性 ... 15

表五 自製及商用感測計之 pH 值量測結果 ... 16

一、 計畫緣由與目的

離 子 感 測 場 效 電 晶 體 (Ion-Sensitive Field Effect Transistor, ISFET) 是 P.

Bergveld 於 1970 年率先提出[1]。主要是應用金屬-氧化物-半導體場效應電晶

體(MOSFET)的原理，將其閘極金屬去除後，使感測絕緣層 SiO

直接與緩衝溶

液接觸，藉由緩衝溶液中的待測離子與絕緣感測膜吸附而產生感應通道

(Channel)，隨之發現 ISFET 與傳統的玻璃電極具有相同之酸鹼性感測特性。

一般而言，單層之 SiO

感測膜在感測度及穩定度上並未有良好的表現，因此，

近三十年來已有許多發展相繼提出，諸如：(1)以 Si

N

/SiO

[2-3]、Al

O

/SiO

[4-5]

及 Ta

O

/SiO

[2,6]等雙層結構作為離子感測閘極而具有較佳的感測響應；(2)

參考電極之微小化[7]；(3)結構之改善與吸附鍵結模型(Site-binding model)理論

上之研究[8]；(4)遲滯、時漂等非理想因素之探討[9-11]。

另外，在研究氫離子 ISFET 的過程中發現，利用單層的 SiO

或 Si

N

上披

覆相對的感測膜即可以感測出其它離子，如：血液中的鉀、鈣、鈉離子及葡萄

糖、尿素黴、青黴素等，對於醫學方面貢獻良多[12-14]。但由於各種離子感測

器仍存有諸多問題，因而限制了它的發展與應用；然究其根本，係因 ISFET

原有的非理想因素所引起，例如：(1)遲滯、時漂等非理想因素造成了離子感

測場效應電晶體之使用壽命減短，長期的穩定性亦不佳，導致商品化僅能用於

拋棄式；(2)溫度及光照所引起的輸出電壓不穩定，致使量測過程需在恆溫、

暗室中進行。

一般而言，由於不同感測材料具有不同離子活性與穩定性，因此材料之選

擇對於 ISFET 之感測特性將有重大之影響。鈦酸鉛(Lead Titanate, PbTiO

, PT)

薄膜已被我們發現對電解液中的酸鹼離子有較大的靈敏度及具有線性穩定化

的輸出，極適合作為 ISFET 的感測層[15-17]。然而無論如何非理想因素先天

的存在性是無可避免地，在過去僅有少數文獻針對氧化物感測層提出熱處理方

式 [18]尋求非理想因素改善與補償之道。因此，本計畫嘗試在 PT 薄膜內以補

償方式(Compensation)分別摻雜微量受體 Mg 元素及施體 La 元素使本體(Bulk)

得以修飾相關結構特性，以期改善與補償非理想因素。換言之，(Pb,La)TiO

與(Pb,Mg)TiO

等鈦酸鉛系列材料將作為絕緣感測薄膜層，並分別討論其作為

離 子 感 測 場 效 電 晶 體 之 特 性 ， 以 尋 求 其 最 佳 化 之 參 數 ， 並 了 解 遲 滯

(Hysteresis)、時漂(Drift)、溫度效應(Temperature effect)、生命週期(Lifetime)

等非理想因素改善與補償情形，且嘗試建立合理之摻雜機制模型。最後，並與

無摻雜 PT 薄膜及傳統之 Si

N

、Al

O

等感測薄膜做一分析比較，更期能完成

二、結果與討論

圖一為摻雜不同含量 Mg 及 La 的 EIS 結構之 pH 響應曲線。當摻雜 La 時，

隨著摻雜量增加，感測度隨之下降。然而摻雜 Mg 時，含量於 2-6 mole%時其

感測度值皆在 55 mV/pH 之上，特別在 4 mole%時具有 59 mV/pH 之 Nernst 響

應，但含量達到 7 mole%時，感測度亦急速下降。此乃因摻雜雜質將使居理溫

度改變，由圖二之 XRD 可得知 PMT(4)於 450℃時為非晶形狀態，溫度增至

500 ℃時才呈現結晶狀態，但是當 Mg 含量增為 PMT(7)，於 450 ℃時則已呈

現結晶狀態。明顯地，Mg 含量增加導致 PMT 薄膜之居理溫度下降，導致薄

膜表面吸附鍵結減少，以致於待測離子與薄膜表面自由基不易吸附鍵結，感測

度受到影響而下降。

藉由電流-電壓(I-V)特性瞭解其感測度及線性度可區分為將工作點操作在

非飽和區之固定電流法與將工作點操作在飽和區之固定電壓法。表一與表二分

別指出 PT、PMT(4) 及 PLT(3) gate ISFET 作用於非飽和區及飽和區之感測響

應結果。當 Mg 及 La 分別被摻雜於 PbTiO

感測膜中，無論作用於非飽和區或

飽和區，由其結果得知受體 Mg 將有利於 pH 響應度，而施體 La 則反之。主

要係因 La

_{比 Pb}

_{多一正電荷，因此在薄膜中除了既有的鉛原子、氧原子的空}

位(Vacancy)外，還有施體原子 La 的帶電載子，這些缺陷(Defect)使得感測度變

低。當加入少量 La 時，將填補部份缺陷，並使表面趨於平整，亦減少因缺陷

而增加的表面電位勢；當 La 含量增多時，除了減少了薄膜中的缺陷載子數目

之外，相對的也使得待測離子與薄膜表面吸附鍵結變少，而使感測度下降。

表三為 PT、 PMT(4)及 PLT(3) gate ISFET 之非理想因素之結果。在時漂方

面，無論摻雜 Mg 或 La 元素均有助於時漂之改善，其主要係因時漂發生在幾

個原子厚的表面修正層(Modified surface layer)上，今微量之 Mg 或 La 元素摻

雜其中時，將中和(Neutralization) 表面修正層內之缺陷，使時漂值下降。同時，

在遲滯方面，Mg 元素摻雜其中時，將有助於被感測層捕陷於其中之 H

或 OH

離子的機率下降，如此遲滯量將獲得改善；反之，La 元素摻雜其中時，將提

供捕陷機制，使得被捕陷之 H

或 OH

離子大大增加。但無論如何，隨量測

時間延長將增加薄膜內所累積的電荷，進而導致整個遲滯的表現隨時間延長而

增加；由此可知遲滯量的大小是與量測時間的長短有相當大的直接關係。在元

件的老化現象方面，PMT 在酸鹼環境中其老化速率相當緩慢，不易被酸鹼溶

液所侵蝕而損壞，其使用壽命至少可以持續一年以上；然而， PLT 則有加速

元件老化現象的發生。此乃因摻雜 Mg 元素時，將有利於 PT 材料更為電中性，

形成更穩定之感測結構。

表四為不同感測材料之 pH 感測特性一覽表。比較 PT-series ISFET 與其它

ISFET 的感測度與 pH 有效範圍，發現 PT-series 與以 Al

O

、Ta

O

、SnO

為閘

極之 ISFET 相同，皆表現出相當高的酸鹼感測度，且與 Si

N

、SiO

相較之下，

表現更為突出。在 pH 7 之時漂量的比較上，發現 PT-series 薄膜與 Si

N

、Al

O

與 Ta

O

等薄膜相近，但相較於 SiO

、SnO

材料則有較佳表現。至於遲滯表

現上，雙層結構之 Ta

O

、PT-series 等氧化物感測材料之遲滯量相近似，其原

因可能此類型材料具有較穩定之鍵結，其缺陷所捕獲之 H

或 OH

離子數目

差異不大。此隱含著缺陷若能被適當中和將有助於此特性的改善，其中被修正

後之 PT 薄膜即為一典型例子。最後，溫度特性表現上，雙層結構之 ISFET 均

具有近似之結果，其乃因 ISFET 是以半導體 Si 為基礎之元件，其受溫度影響

主要是半導體 Si，而非絕緣感測材料。

酸鹼感測計之系統方塊電路如圖三所示，主要可分為：

(a) 類比前端訊號處理模組：訊號讀出電路。

(b) 類比/數位轉換模組：類比/數位轉換電路。

(c) 數位輸出訊號處理模組：AT89C51 及 LCD 顯示器。

系統中使用 AT89C51 單晶片微處理器作為資料處理單元，包含了資料的運算

處理和解碼的工作，其處理流程如圖四所示。在程式開始執行之後，8051 先

將系統及 LCD 予以初始化，並分別要求載入 pH 4 與 pH 7 之標準緩衝液，當

確定輸入後，會顯示 OK。等到 pH 4 與 pH 7 相對應之兩數位值輸入後，將進

行兩點校正，且經由 8051 之計算分別列出 pH 1 到 pH 13 相對應的數位值。其

後輸入待測溶液，程式將逐次比較，即可得到此溶液整數位數之 pH 值，再經

由細部之比較可精確地得出一位小數之 pH 數值。

最後，在系統測試上，以自製之酸鹼感測計與商用酸鹼感測計(SUNTEX

Digital pH/mV Meter：Model No.TS-1)來量測其 pH 值並記錄於表五中。由表五

可看出，在 pH 10 及 pH 12 其誤差值較大，誤差之原因主要為兩點校正，因為

ISFET 之感測度在每一 pH 值之間並非完全相同，故在數位之設定點上會造成

與實際 pH 值之數位點的誤差。此外，在系統中所使用的類比/數位轉換器為 8

位元，使得在量測上亦造成誤差。

三、計畫成果自評

本研究主要是討論以 PT-series 為絕緣層的 ISFET 的電特性，並將其結果分

述如下：

1. EIS 結構中，發現受體 Mg 含量為 4 mole％、膜厚 0.5 µm、燒結溫度 400 ℃

時，可得到最佳之感測度及穩定度，其在 pH 2-12 之最佳感測度約為

58-59mV/pH。

2. Mg

及 La

分別被摻雜於 PbTiO

感測膜中，以修飾相關感測特性，發現受

體 Mg

_{將有利於 pH 響應度(58-59 mV/pH)、時漂(0.4 mV/h)、遲滯(1-3) mV}

及衰減率(–0.2 µV/pH-day)之改善。

3. 一數位型 pH 量測儀已被設計與實現。

4. 有關本研究已發表之成果請見 Refs. 17, 23-24，詳見附錄。

綜合上述之結果，發現藉由補償方式摻雜適當微量元素，將能改善先天存

在的非理想因素，以期獲得一更精確與可靠之感測材料。

四、參考文獻

1. P. Bergveld, Development of an ion-sensitive solid-state device for

neurophysiological measurements, IEEE Transactions on Bio-Medical

Engineering 17 (1970) 70-71.

2. L. Bousse, S. Mostarshed, B. van der Schoot and N. F. de Rooij, Comparison of

the hysteresis of Ta

O

and Si

N

pH-sensing insulators, Sensors and Actuators B

17 (1994) 157-164.

3. M. N. Niu, X. F. Ding and Q. Y. Tong, Effect of two types of surface sites on the

characteristics of Si

N

-gate pH-ISFETs, Sensors and Actuators B 37 (1996)

13-17.

4. L. Bousse, H. H van der Vlekkert and N. F. de Rooij, Hysteresis in Al

O

-gate

ISFETs, Sensors and Actuators B 2 (1990) 103-110.

5. H. Van Den Vlekkert, L. Bousse and N. de Rooij, The Temperature dependence

of the surface potential at the Al

O

/electrolyte interface, J. Colloid Interface Sci.

122 (1988) 336-345.

6. S. Poghossian, The Super-Nernstian pH sensitivity of Ta

O

-gate ISFETs,

Sensors and Actuators B 7 (1992) 367-370.

7. S. D. Collins, Practical limits for solid-state reference electrodes, Sensors and

Actuators B 10 (1993) 169-178.

8. D. E. Yates, S. Levine and T. W. Healy, Site-binding model of the electrical

double layer at the oxide/water interface, J. Chem. Soc. Faraday Trans. I 70

(1974) 1807-1818.

9. Aine Garde, John Alderman and Willian Lane, Improving the drift and hysteresis

of the Si

N

pH response using RTP techniques, Sensors and Materials 9 (1997)

015-023.

10. Luc Bousse, Dean Hafeman and Nancy Tran, Time-dependence of the

chemical response of silicon nitride surface, Sensors and Actuators B 1 (1990)

361-367.

Dynamic behaviour of ISFET-based sensor-actuator systems, Sensors and

Actuators B 1 (1990) 416-420.

12. B. D. Liu, Y. K. Su, and S. C. Chen, Ion-sensitive field-effect transistor with

silicon nitride gate for pH Sensing, INT, J. Electronics 67 (1989) 59-63.

13. C. Cane, A. Gotz, A. Merlos, I. Gracia, A. Errachid, P. Losantos and E.

Lora-Tamayo, Multilayer ISFET membranes for microsystems applications,

Sensors and Actuators B 35-36 (1996) 136-140.

14. Wang Zheng-Xiao, Applications of penicillinase FET in penicillin-fermentation

engineering, Sensors and Actuators B 13-14 (1993) 568-569.

15. Shiun-Sheng Jan, Ying-Chung Chen, Jung-Chuan Chou, Chien-Chuan Cheng

and Chun-Te Lu, Preparation and Properties of Lead Titanate Gate ISFETs by

the Sol-Gel Method, Jpn. J. Appl. Phys. 41 (2002) 942–948.

16. Shiun-Sheng Jan, Ying-Chung Chen, Jung-Chuan Chou, Chien-Chuan Cheng

and Chun-Te Lu, Nonideal Factors of Ion-Sensitive Field-Effect Transistors with

Lead Titanate Gate, Jpn. J. Appl. Phys. 41 (2002) 6297-6301.

17. Shiun-Sheng Jan, Jung-Lung Chiang, Ying-Chung Chen, Jung-Chuan Chou

and Chien-Chuan Cheng, Characteristics of the hydrogen ion-sensitive field

effect transistors with sol-gel-derived lead titanate gate, Analytica Chimica Acta

469 (2002) 205-216.

18. D. H. Kwon, B. W. Cho, C. S. Kim and B. K. Sohn, Effect of heat treatment on

Ta

O

sensing membrane for low drift and high sensitivity pH-ISFET, Sensors

and Actuators B 34 (1996) 441-445.

19. 鍾與采、趙守安、劉濤，pH-ISFET 輸出時漂特性的研究，半導體學報，

第 15 卷，第 12 期 (1994) 838-843。

20. Yu Dun, Wei Ya-dong and Wang Gui-Hua, Time-dependent mesponse

characteristics of pH-sensitive ISFET, Sensors and Actuators B 3 (1991)

279-285.

21. J. C. Chou and C. Y. Weng, Sensitivity and hysteresis effect in Al

O

gate

pH-ISFET, Material Chemistry and Physics 71 (2001) 120.

Temperature and optical characteristics of tin oxide membrane gate ISFET, IEEE

Transactions on Electron Devices 46 (1999) 2278.

23. Shiun-Sheng Jan, Ying-Chung Chen, Jung-Chuan Chou, Chien-Chuan Cheng

and Pei-Jane Jan, Preparation and Properties of the Hydrogen Ion-sensitive Field

Effect Transistors with Sol-gel-derived Mg-modified Lead Titanate Gate, Journal

of Noncrystalline Solids 323 (2003) 11.

24. Ying-Chung Chen, Shiun-Sheng Jan, Jung-Chuan Chou, Temperature effects

on the characteristics of the hydrogen ion-sensitive field-effect transistors with

sol-gel-derived lead titanate gate, Analytica Chimica Acta 516 (2004) 43.

0 2 4 6 30 40 50 60

(b)

(a)

Acceptor Mg

Donor La

Sensitiv

ity

(m

V/pH)

Doping Content (mol%)

圖一 (a)PMT 與(b)PLT EIS 結構之不同摻雜濃度的 pH 響應曲線

(Firing temperature of 400℃)

20 30 40 50 60 4000C (a) _PMT(4) 5000C 4500C In tensity (arb. units ) 2θ (deg) 20 30 40 50 60 4000C (b) _PMT(7) 5000C 4500C Inten s it y (arb. u n it s ) 2θ (deg)

圖二 (a) PMT(4)與(b) PMT(7)薄膜之不同燒結溫度的 XRD 分析

ISFET

Analog front-end

circuit

Analog-to-digital

converter

Single chip

microprocessor

AT89C51

LCD

圖三 數位酸鹼感測計基本架構圖

Set initial sate

to system

Load

pH 4

Testing and reading

digital values

for the solution

Check pH range

of the solution

No

Starting

Yes

Reading

pH 4 digital data

Yes

Reading

pH 7 digital data

No

LCD displays “OK”

Deriving

pH 1-13 digital data

Delivering to LCD

Calculating pH value

Load

pH 7

圖四 系統設計流程

表一 作用於非飽和區之 PT、 PMT(4)及 PLT(3) gate ISFET 之感測響應值

與 pH 有效範圍

PT-series sensing membrane*

PT

PMT(4)

PLT(3)

Sensitivity (mV/pH)

55-58

58-59

52-55

Available pH range

2-12

2-12

2-10

* The fabrication parameters is kept at the firing temperature of 400℃

and the thickness of 0.5 µm.

表二 作用於飽和區之 PT、 PMT(4)及 PLT(3) gate ISFET 之感測響應值

與線性相關係數

Gate-source

Voltage

(V)

pH response

(µA/pH)

Correlation

coefficient

γ

1 4.2

0.9491

3 24.8

0.9995

PT gate ISFET

5 31.3

0.9996

1 8.4

0.9907

3 24.3

0.9996

PMT(4) gate ISFET

5 32.7

0.9997

1

_{2.7 0.9607}

3 11.4

0.9784

PLT(3) gate ISFET

5 18.3

0.9976

表三 PT、 PMT(4)及 PLT(3) gate ISFET 之非理想因素

PT-series sensing membrane

PT

PMT(4)

PLT(3)

Drift (mV/h)

0.5-1

0.2-0.4

0.1-0.3

Hysteresis* (mV)

3-5

1-3

2-13

Reduction rate

(µV/pH-day)

10 0.18 79.2

*Measured at the short and long loops.

表四 不同感測材料之 pH 感測特性[19-22]

Structure

type

Sensing

membrane

Sensitivity

(mV/pH)

Available

pH range

Drift

(mV/h)

Hysteresis

(mV)

Temperature

coefficient

(mV/pH-℃)

Single

layer

SiO

25-48 4-10

Unstable Unstable

Unstable

Si

N

46-56

1-13

0.83

6.3-10.9

0.307

Al

O

53-57 1-13 0.1-0.2 3-6 0.315

Ta

O

57-58

2-12

0.72

3-4

0.134

SnO

55-58 2-10 1.65 ~2.5 0.166

Double

layer

PT-series

58-59 2-12 0.2-0.4 1-3 0.106

表五 自製及商用感測計之 pH 值量測結果

Testing pH value

Standard

pH value

commercial

home-made

Testing

digital value

Error

value

2 1.91 2.0 71

0.045

4 4.01 4.0 82

0.005

6 6.28 6.1 93

0.090

7 7.37 7.3

101

0.035

8 8.23 8.0

104

0.115

10 10.12 9.6 115

0.260

12 12.33 11.6 125

0.365

Analytica Chimica Acta 469 (2002) 205–216

Characteristics of the hydrogen ion-sensitive field effect

transistors with sol–gel-derived lead titanate gate

Shiun-Sheng Jan

, Jung-Lung Chiang

, Ying-Chung Chen

,

Jung-Chuan Chou

, Chien-Chuan Cheng

a_{Department of Electrical Engineering, National Sun Yat-Sen University, Kaohsiung 804, Taiwan, ROC} b_{Department of Electronic Engineering, Chung Chou Institute of Technology, Yuan-Lin, Chang-Hua 510, Taiwan, ROC}

c_{Institute of Electronic and Information Engineering, National Yunlin University of Science and Technology,} Touliu, Yunlin 640, Taiwan, ROC

d_{Department of Electronic Engineering, De-Lin Institute of Technology, Tucheng, Taipei 236, Taiwan, ROC}

Received 26 April 2002; received in revised form 22 July 2002; accepted 24 July 2002

Abstract

The sol–gel-derived lead titanate (PbTiO3) membrane has been successfully applied as a pH sensitive layer to form the

PbTiO3gate ion-sensitive field-effect transistor (ISFET). There exhibit the excellent quasi-Nernstian response of 56–59 mV

pH−1, good surface adsorption and anticorrosion characteristics via theC–V measurement of the EIS structure. At a specific pH concentration, the output and transfer characteristics are very similar to the behaviours of MOSFETs, and the ISFET model can be derived by the modified MOSFET model. As it operated in the nonsaturation region, there exhibits a linear pH response of about 56–59 mV pH−1. On the other hand, as it operated in the saturation region, the pH response and linearity can be controlled by adjusting theVGSvalues, e.g. the pH responses of−4.2, −24.8 and −31.3␮A pH−1and the correlation

coefficients of 0.9491, 0.9995 and 0.9996 atVGS = 1, 3 and 5 V can be obtained, respectively. Besides, in order to get the

best pH response and the minimized leakage current, the heat treatment temperature of the PbTiO3membrane must be limited

between 350 and 450◦C.

© 2002 Elsevier Science B.V. All rights reserved. Keywords: ISFET; pH response; Lead titanate; Sol–gel

1. Introduction

The first ion-sensitive field-effect transistor (ISFET)

was reported by Bergveld [1]. The ISFET can be

considered as a special type of the MOSFET without a metal gate, by which, the gate concerned with the insulator is directly exposed to the buffer solution. A change of the surface potential between electrolyte

∗_{Corresponding author. Tel.:+886-7-525-2000x4121;}

fax:+886-7-525-4199.

E-mail address: [email protected] (Y.-C. Chen).

and insulator will be resulted from a change of the pH concentration in the electrolyte, which will induce the alteration of the electric field in the insulator– semiconductor interface and the modulation of the channel conductance and current. In view of very small size, rapid response, low output impedance, compat-ible with standard MOS process and low cost, etc.

[2–4], it reveals a lot of advantages in comparison with a conventional ion-selecitive electrode particularly in the applications of the biomedical engineering[5,6].

In the past, many pH-sensitive materials, such as, SiO2[1], Si3N4[7], Al2O3[8], Ta2O5[9], WO3[10], 0003-2670/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.

206 S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 SnO2 [11,12], etc. have been investigated. The

sens-ing properties of the ISFETs were greatly dependent on various materials owing to the different reactivity of the electrolyte with these materials. However, these materials must be deposited in high vacuum cham-ber, which is not simple and moderate at all. Recently, an amorphous sol–gel-derived lead titanate (PbTiO3)

membrane, which exhibits the advantages of chemi-cal stability, mechanichemi-cal strength, high resistivity, high permittivity and greater ease of manufacturing [13], has been successfully prepared in our laboratory as a novel hydrogen ion-sensitive layer[14].

In this study, the electrolyte–insulator–semicon-ductor (EIS) structure will be adopted to examine the properties of the surface adsorption, pH response and chemical stability of the PbTiO3-sensitive

mate-rial. The electrical characteristics and pH response

of the PbTiO3 gate ISFET will be studied by the

current–voltage (I–V ) measurement. Simultaneously,

the annealing condition for the PbTiO3membrane to

obtain the best pH response and minimized leakage current will be also proposed. Finally, the pH

re-sponse of the PbTiO3gate ISFET will be practically

carried out by a read-out circuit.

2. pH–ISFET model

Assuming that the pH response belongs to Nerns-tian response or quasi-NernsNerns-tian response, the ISFET model can be derived by the modified MOSFET

model [15,16]. Fig. 1schematically shows the basic

structure and multi-phase diagram of ISFET. Ac-cording to the MOSFET model, the term of metal gate can be modified as the reference electrode, electrolyte and electrolyte/insulator interface. In

the nonsaturation region, the ideal I–V

character-istics for an n-channel enhancement mode ISFET

Fig. 1. Basic and multi-phase diagram of ISFET.

can be described as

IDS= Kn[2(VGS− VT)VDS− VDS2 ], (1)

where the conduction parameter,K_n = µ_nCiW/2L,

µn is the electron mobility in the inversion layer,

Ci the insulator capacitance per unit area, W/L the

width-to-length ratio, VT the threshold voltage of

ISFET, expressed as follows:

VT= VTM−ΦM

q + ERef+ χSol− ψ0, (2)

where VTM is the threshold voltage of MOSFET,

ΦM/q the work function of metal, ERef the

refer-ence electrode potential relative to vacuum,χSol the

surface dipole potential of the buffer solution, ψ0

the surface potential at electrolyte/insulator interface, described as[15] ψ0= 2.303 kT q β β + 1(pHPZC− pH), (3)

where k is the Boltzmann constant, T the absolute

temperature, q the electron charge, β the sensitive

parameter (0< β < 1), and pH_PZCis the pH value at the point of zero charge.

By mergingEqs. (2) and (3)intoEq. (1), theI–V

characteristics can be rewritten as

IDS= Kn
2  VGS− 2.303 kT q β β + 1pH− VT∗  × VDS− VDS2  , (4)

whereV_T∗is the modified threshold voltage andVT−

2.303(kT/q)[β/(β + 1)]pH, which is a constant at a

specific reference electrode potential. When VDS is

a constant voltage operated in nonsaturation region, the pH concentration can be described by a linear equation

IDS= αVGS− S1× pH − Θ, (5)

whereα = 2K_nVDS,S1= 2.303(kT/q)[β/(β + 1)]α

andΘ = α(V_T∗+ 0.5VDS).

On the other hand, in the saturation region, the

ideal I–V characteristics for an n-channel

enhance-ment mode ISFET can be described as

S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 207

Similarly, theI–V characteristics can also be rewritten by the equation IDS= Kn  (VGS− VT∗) 2_{− 4.606(V} GS− VT∗) kT q × β β + 1pH−  2.303kT q β β + 1 2 pH2  . (7)

When the condition of VGS  2.303(kT/q)[β(β +

1)]pH+V_T∗is satisfied, the term of{2.303(kT/q)[β(β+ 1)]}2pH2 can be neglected, and Eq. (7) can be de-scribed by a linear equation

IDS= A − S2× pH, (8)

whereA = K_n(VGS−VT∗)2andS2= 4.606Kn(VGS−

V∗

T)(kT/q)[β/(β + 1)] are constant at a specific VGS.

3. Experimental

3.1. Sample preparation

The schematic diagram of the PbTiO3 gate ISFET

is shown inFig. 2. The sensitive layer of the PbTiO3

membrane is deposited onto the SiO2 gate ISFET

prepared by the standard MOS technique to form

the double-layer PbTiO3gate ISFET. The fabrication

parameters are listed in Table 1, and the fabrication procedures are listed as follows.

(1) Thermal growth of silicon dioxide (0.1␮m).

(2) First photomask and oxide etching. (3) Phosphorus ion implantation (1015cm−2). (4) Second photomask and oxide etching.

Fig. 2. Schematic of the PbTiO3 gate ISFET.

Table 1

Fabrication parameters of the double-layer PbTiO3gate ISFET

Fabrication parameter Type or value

Wafer p-type (100) silicon

Dose (cm−2) 1015_(phosphorus) Width-to-length ratio 20 SiO2 thickness (␮m) 0.1 PbTiO3thickness (␮m) ∼0.5 Pyrolysis temperature (◦C) 350 Pyrolysis time (h) 1 (5) Gate oxide (0.1␮m).

(6) Third photomask and oxide etching. (7) Al sputtering (0.5␮m).

(8) Fourth photomask and Al etching.

(9) Deposition of sol–gel-derived PbTiO3membrane

(0.5␮m).

(10) Bounding and encapsulation using epoxy resin. Comparing with the ISFET structure, the source and the drain are removed in the EIS structure, that is, only the gate and the bottom electrodes are included.

3.2. Packaging and measurement

Epoxy resins are used for encapsulation of samples, which are generally used in sensor technology because

208 S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216

Fig. 4. Schematic of the read-out circuit.

of their low permeability[17], and the sensitive region must retain to be directly exposed to the standard pH buffer solution (phosphate). The Ag/AgCl reference electrode is simultaneously immersed in the standard pH buffer solution. A family of capacitance–voltage (C–V ) curves can be obtained by applying a se-ries of step voltages at the reference electrode using an HP4284A LCR parameter analyzer, as shown in

Fig. 3. The Keithley 236 semiconductor parameter

analyzer is used to measure the I–V characteristics

of the PbTiO3gate ISFET. In order to avoid

temper-ature and photoelectric effects, all the measurement processes are carried out at the room temperature of

25◦C, set by a temperature control system, and in

the dark.

3.3. Read-out circuit

The measurement circuit used for reading out the pH response of ISFET is shown inFig. 4. In order to obtain the true chemical response of the ion sensitive layer and the linearity of the output signal, a specific circuit is designed to keep both drain-source current

IDS and drain-source voltageVDS constant to obtain

the output voltage as a function of pH concentration

[18,19]. The desired constant drain-source current and drain-source voltage can be adjusted individually by

R2andR1, respectively.

4. Results and discussion

4.1. Electrolyte–insulator–semiconductor structure

In order to examine the properties of the surface adsorption, pH response and corrosion of the sensitive material, the EIS structure device is often employed. TheC–V measurement is well-known for the charac-terization of MOS capacitances, which is also suitable for characterizing an EIS structure[20].Fig. 5shows

a family of C–V curves of the PbTiO3 EIS

struc-ture with the membrane thickness of 0.5␮m in the

standard buffer solution in the range of pH 2–12 at

the frequency of 10 kHz. A C–V curve consists of

the accumulation, flat-band, depletion and inversion regions. Because the space charge is dominated by minority carriers in the accumulation and inversion regions, these regions are not suitable for the study of pH response. However, at the flat band condition, the charge density is low and not dominant, so the pH response of the EIS structure with different pH

concentration can be carried out. TheC–V curves are

shifted in parallel with the pH concentration of the buffer solution, which is ascribed to the flat-band volt-age shift towards positive value with increasing pH concentration. The quasi-Nernstian response of about

56.7 mV pH−1, which is near the Nernstian response

S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 209

Fig. 5. A family of theC–V curves of the PbTiO3EIS structure with the PbTiO3membrane thickness of 0.5␮m at the frequency of 10 kHz.

be obtained from the flat-band shift variation ofC–V curves, as shown in the inset of Fig. 5. This result reveals that the electrolyte/PbTiO3 interface has an

excellent absorption mechanism and less interference effect for the electrolyte ions due to the sufficiently

large surface site density [18]. Furthermore, based

on the experimental observation, the sensitive layer

of the PbTiO3 membrane is difficult to be corroded

in the strong acid and base regions. As the results

mentioned earlier, the PbTiO3 membrane is suitable

to act as a pH sensitive layer.

4.2. Output and transfer characteristics

The gate of ISFET consists of three parts of the ion-sensitive layer, the pH buffer solution and an Ag/AgCl reference electrode, which is different from the metal gate of MOSFET. Therefore, two param-eters of the gate must be concurrently considered; one is the gate-source voltage (VGS) between the

ref-erence electrod and the source, and the other is the pH concentration in the pH buffer solution. There-fore, the two parameters (VGS,VDS) of the operation

point with a specific threshold voltage (VT) must be

considered to carry out the detecting of pH concen-tration.

Fig. 6 shows typical sets of output (IDS–VDS) and

transfer (IDS–VGS) characteristics of the PbTiO3gate

ISFET with different VGS and pH concentration,

re-spectively. There exhibits the thresold voltage VT of

about 0.8–0.9 V at pH = 7, and the calculated

mod-ified thresold voltage V_T∗ of about 0.41–0.51 V can

be obtained. When VGS is greater than VT and the

drain-source voltage (VDS) is applied, the drain-source

current (IDS) is generated. So, at a specific pH

con-centration, the output and transfer characteristics are in good agreement with those of similarly fabricated

MOSFET and can be represented asEqs. (1) and (6)in

the nonsaturation and saturation regions, respectively

[15,16,21–23]. However, when the ISFET operated in the saturation region is used to detect the pH con-centration, bothVGSandVDSmust be well-chosen in

the consideration of linearity, minimization and lower

power dissipation. So, the PbTiO3gate ISFET can be

operated atVGSin the range of 3–4 V andVDSin the

range of 3–4 V. When the ISFET operated in the non-saturation region is used to detect the pH concentra-tion,VDSin the range of 0.2–0.4 V will be suitable.

210 S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216

Fig. 6. (a)IDS–VDScharacteristics of the PbTiO3gate ISFET as a function of gate-source voltage, (b) transfer characteristics of the PbTiO3

gate ISFET as a function of pH concentration.

4.3. pH response

In order to detect the pH concentration, two types of pH response can be obtained by the ISFET operated in the nonsaturation and saturation region, respec-tively. In the nonsaturation region withVDS= 0.3 V,

a typical set of IDS, Gm versus VGS curves of the

PbTiO3 gate ISFET with different pH concentration

is obtained, as shown inFig. 7. TheIDS–VGS curves

are shifted in parallel with the pH concentration of the buffer solution, which is ascribed to the threshold volt-age shift towards positive value with increasing pH

S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 211

Fig. 7.IDS,Gm vs.VGS curves of the PbTiO3 gate ISFET operated in the nonsaturation region withVDS= 0.3 V, measured in pH 2, 4,

6, 8, 10, and 12 buffer solutions, respectively.

concentration. A linear pH response of about 56.6 mV

pH−1 is obtained by calculating the shifts in the

threshold voltage of the ISFET for different pH con-centration, as shown inFig. 8(a). Simultaneously, as a substantial application, the ISFET is always operated at a constant drain-source current, e.g.IDS= 37 ␮A,

Fig. 8. pH response of the PbTiO3 gate ISFET operated in the nonsaturation region: (a) measured, (b) simulated byEq. (5).

and the pH response of about 56.6 mV pH−1can also

be obtained by the gate-source voltage shift. So, the pH response operated at a well-chosen drain-source current is in good agreement with that calculated from the threshold voltage shift, and it can be directly read out and is a practical and simple method.

212 S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216

Table 2

Relative parameters adopted for the simulation of the PbTiO3gate

ISFET Parameter Value Electron mobility,µn(cm2 V−1s−1) 580 Insulator capacitance (nF cm−2) 16 Width-to-length ratio,W/L 20 Boltzmann constant,k (J K−1) 1.38× 10−23 Temperature,T (K) 300 Sensitive factor,β 15.7

Modified threshold voltage, V∗_T (V) 0.41–0.51

Furthermore, the theoretical relationship of IDS,

VGS and pH concentration in the nonsaturation

re-gion has been described in Eq. (5). There exhibits

the pH response of 56.1 mV pH−1by simulating VT

or VGSwith different pH concentration, as shown in

Fig. 8(b), where the relative parameters adopted are listed inTable 2. This result is in good agreement with

that measured in Fig. 8(a). However, there exists a

parallel-shift deviation of about 0.26–0.44 V between

Fig. 8(a) and (b). This may be resulted from the po-tential within the insulator and the interface, in which, lots of charges exist and form a gate-source voltage

Fig. 9.IDS–VDScharacteristics of the PbTiO3 gate ISFET as a function of pH concentration.

shiftVGS. This phenomenon can be evidenced by

means of the shift in the C–V measurement of the

MIS structure[14].

In the saturation region, as an example of the ISFET held atVGS= 3 V, a typical set of IDS–VDScurves of

the PbTiO3 gate ISFET with different pH

concentra-tion is shown inFig. 9. TheIDS–VDScurves are shifted

in parallel with the pH concentration of the buffer so-lution, which is ascribed to the transconductance shift towards negative value with increasing pH concentra-tion in view of the transfer characteristics inFig. 6(b). Simultaneously, three families of the pH response of

the PbTiO3 gate ISFET were obtained with different

VGS, as shown inFig. 10. When these data were

sub-jected to the linear curve-fitting processes, there

ex-hibit the pH responses of−4.2, −24.8 and −31.3 ␮A

pH−1and the correlation coefficients, the description for linear regression, of 0.9491, 0.9995 and 0.9996 at

VGS= 1, 3 and 5 V, respectively. These results show

that the pH response can be controlled by adjusting

VGS and increases with the increased VGS. The

lin-earity of pH response also increases asVGSincreases,

which is in good agreement with the linear condition inEq. (7). In addition, as the data inFig. 10(a) were subjected to the two-order polynomial curve-fitting

S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 213

Fig. 10. Three families of pH responses of PbTiO3gate ISFET operated in the saturation region with (a)VGS= 1 V, (b) VGS= 3 V and

(c)VGS= 5 V, respectively.

processes, the results dovetail neatly with the two-order terms in Eq. (7). So, in the saturation region, there exhibits a linear pH response as described in

Eq. (8).

On the other hand, the theoretical relationship of

IDSand pH concentration in the saturation region has

been described in Eq. (8) and the simulated results

are shown inFig. 10(b) and (c). There exhibit the pH

responses of−26.4 and −47.3 ␮A pH−1 inVGS =

3 and 5 V, respectively, which is similar to the mea-sured results. However, compared with the meamea-sured results, there exists an exiguous difference in the shift and magnitude between the two curves. This can be

ascribed to the considerable variations in theA and

S2terms inEq. (8)generated by a gate-source voltage

shiftVGS.

4.4. Annealing temperature

In order to obtain the best pH response and avoid the damage of the device, the annealing temperature of ion-sensitive layer must be considered. The PbTiO3

gate ISFETs fired at 350◦C are annealed at various

temperatures from 400 to 550◦C with step 50◦C.

Be-low the annealing temperature of 400◦C, the linear

quasi-Nernstian response of about 57 mV pH−1can be

obtained with a negligible leakage current, which is the drain-source current atVDS= 0 V, as shown inFig. 11.

214 S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216

Fig. 11. pH response and leakage current of the PbTiO3gate ISFET as a function of annealing temperature.

When the annealing temperature goes up over 450◦C,

the pH response and leakage current behave a drastic decrease and increase, respectively. This phenomenon is resulted from the increase of crystallization and the

Fig. 12. XRD patterns of the sol–gel-derived PbTiO3 membrane fired at 350◦C for 1 h: (a) unannealed, (b) annealed at 450◦C and (c)

annealed at 550◦C for 1 h, respectively.

decrease of the surface site density of PbTiO3

mem-brane. As shown in Fig. 12, the PbTiO3 membrane

appears an amorphous-crystal transition between 450

S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 215

Fig. 13. Output voltage vs. the different pH concentration by the read-out circuit.

suggest a decrease in the surface site density[24–26]. Besides, many researchers had demonstrated that ions (e.g. Na+, Li+, K+, etc.) could move through crys-talline SiO2 at temperatures below 250◦C[27], and

result in an increased leakage current by the current penetrability of the gate.This effect can also occur

in the PbTiO3 gate ISFET owing to the diffusion of

lead ions from PbTiO3 membrane to Si/SiO2

inter-face. Consequently, higher annealing temperature will cause a decrease in the pH response and an increase in the leakage current, so the annealing temperature

should be limited between 350 and 450◦C.

4.5. Read-out circuit

The PbTiO3 gate ISFET is practically tested by a

read-out circuit in Fig. 4. The output voltageVo

be-haves as a linear function of pH concentration, as

shown inFig. 13. There exhibits the pH response of

58.1 mV pH−1 in the range of pH 2–12, which is in

good agreement with that obtained byIDS–VGS

char-acteristics inFig. 7.

5. Conclusions

In this study, the amorphous sol–gel-derived

PbTiO3 membrane which possesses the excellent

characteristics in the surface adsorption and anticor-rosion is first proposed as a novel type pH-sensitive layer. The sensing properties are examined by the

C–V measurement of the EIS structure in the different

buffer solutions, which reveals that the pH response

of approximately 56–59 mV pH−1can be obtained.

The PbTiO3membrane is deposited onto the SiO2

gate ISFET prepared by the standard MOS technique to form the double-layer PbTiO3gate ISFET. The

out-put and transfer characteristics are similar to those of a MOSFET at a specific pH concentration. Two types of pH responses have been obtained by the ISFET op-erated in the nonsaturation or saturation region, one is

about 56–59 mV pH−1, and the other is about−4.2,

−24.8 and −31.3 ␮A pH−1_{with the correlation}

co-efficients of 0.9491, 0.9995 and 0.9996 atVGS = 1,

3 and 5 V, respectively. The latter can be controlled

by adjusting VGS, by which, both pH response and

the linearity can be increased with the increasedVGS.

Besides, in order to get the best pH response and the minimized leakage current, the annealing temperature

of the PbTiO3membrane must be limited between 350

and 450◦C. Finally, according to the ISFET model,

modified from the MOSFET model, there exhibits the

pH response of 56.1 mV pH−1 in the nonsaturation

region, and the pH response of−26.4 and −47.3 ␮A

pH−1 in VGS = 3 and 5 V, respectively, in the

216 S.-S. Jan et al. / Analytica Chimica Acta 469 (2002) 205–216 experimental results. Incidentally, the pH response

of the ISFET operated in the nonsaturation region is mainly dominated by the behaviors of the sensitive material, but that operated in the saturation region is largely affected by the fabrication parameters of the device and is not easy to be controlled.

Acknowledgements

This study is partly supported by the National Science Council, ROC under contract No. NSC 91-2216-E-110-013. We are grateful to Mr. Chun-Te Lu, Mr. Yii-Fang Wang and Mr. Hsjian-Ming Tsai for their assistance in the experiments.

References

[1] P. Bergveld, IEEE Trans. Bio-Med. Eng. 17 (1970) 70. [2] A. Merlos, E. Cabruja, J. Esteve, Sens. Actuatators B 24/25

(1995) 228.

[3] T. Matsuo, M. Esashi, Sens. Actuatators 1 (1981) 77. [4] J. Bausells, J. Carrabina, A. Errachid, A. Merlos, Sens.

Actuatators B 57 (1999) 56.

[5] J.R. Sandifer, J.J. Voycheck, Mikrochim. Acta 131 (1999) 91. [6] T.C. Yeow, M.R. Haskard, D.E. Mulcahy, H.I. Seo, Sens.

Actuatators B 44 (1997) 434.

[7] B.D. Liu, Y.K. Su, S.C. Chen, Int. J. Electr. 67 (1989) 59. [8] L. Bousse, H.H. Van Den Vlekkert, N.F. De Rooij, Sens.

Actuatators B 2 (1990) 103.

[9] S. Poghossian, Sens. Actuatators B 7 (1992) 367.

[10] J.C. Chou, J.L Chiang, Sens. Actuatators B 62 (2000) 81. [11] L.L. Chi, J.C. Chou, W.Y. Chung, T.P. Sun, S.K. Hsiung,

Mater. Chem. Phys. 63 (2000) 19.

[12] H.K. Liao, L.L. Chi, J.C. Chou, W.Y. Chung, T.P. Sun, S.K. Hsiung, Mater. Chem. Phys. 59 (1999) 6.

[13] A.J. Moulson, J.M. Herbert, Electroceramics, 1st ed., Chapman and Hall, New York, 1990, Chapters 1 and 6, pp. 1 and 263.

[14] S.S. Jan, Y.C. Chen, J.C. Chou, C.C. Cheng, C.T. Lu, Jpn. J. Appl. Phys. 41 (2002) 942.

[15] P. Bergveld, Sens. Actuatators 8 (1985) 109.

[16] S.D. Moss, C.C. Johnson, J. Janata, IEEE Trans. Bio-Med. Eng. 25 (1978) 49.

[17] R.E.G. van Hal, P. Bergveld, J.F.J. Engbersen, D.N. Reinhoudt, Sens. Actuatators B 23 (1995) 17.

[18] C.D. Fung, P.W. Cheung, W.H. Ko, IEEE Trans. Electr. Devices 33 (1986) 8.

[19] R.L. Smith, D.C. Scott, IEEE Trans. Bio-Med. Eng. 33 (1986) 83.

[20] R.E.G. van Hal, P. Bergveld, J.F.J. Engbersen, D.N. Reinhoudt, Sens. Actuatators B 23 (1995) 17.

[21] M. Grattarola, G. Massobrio, S. Martinoia, IEEE Trans. Electr. Devices 39 (1992) 1700.

[22] H.K. Liao, J.C. Chou, W.Y. Chung, T.P. Sun, S.K. Hsiung, Sens. Actuatators B 50 (1998) 104.

[23] L.L. Chi, J.C. Chou, W.Y. Chung, T.P. Sun, S.K. Hsiung, Mater. Chem. Phys. 63 (2000) 19.

[24] C.M. Wang, Y.C. Chen, M.S. Lee, J.W. Wu, C.C. Chiou, Jpn. J. Appl. Phys. 37 (1998) 951.

[25] W.M. Siu, R.S.C. Cobbold, IEEE Trans. Electr. Devices 26 (1979) 1805.

[26] D.H. Kwon, B.W. Cho, C.S. Kim, B.K. Sohn, Sens. Actuatators B 34 (1996) 441.

[27] R.F. Pierret, Semiconductor Device Fundamentals, 1st ed., Addison-Wesley, New York, 1996 (Chapters 18, 654).

Preparation and properties of hydrogen ion-sensitive field

effect transistors with sol–gel-derived Mg-modified lead

titanate gate

Shiun-Sheng Jan

, Ying-Chung Chen

, Jung-Chuan Chou

, Pei-Jane Jan

,

Chien-Chuan Cheng

a_{Department of Electrical Engineering, National Sun Yat-Sen University, Kaohsiung 804, Taiwan, ROC} b_{Institute of Electronic and Information Engineering, National Yunlin University of Science and Technology,}

Touliu, Yunlin 640, Taiwan, ROC

c

Department of Electronic Engineering, De-Lin Institute of Technology, Tucheng, Taipei 236, Taiwan, ROC Received 29 May 2002; received in revised form 29 July 2003

Abstract

A sol–gel-derived Mg-modified lead titanate (PMT) membrane has been applied successfully as a novel hydrogen ion-sensitive layer. The fabrication parameters and characteristics of the amorphous PMT membrane are determined at a firing temperature of 350–450°C and Mg content below 7 mol% via differential thermal, thermogravimetric and X-ray diffraction analyses. The optimized fabrication parameters of the PMT membrane for hydrogen ion-sensitive field effect transistors (ISFETs) include a Mg content of 4–5 mol%, thickness of 0.5 lm and firing temperature of 400°C. These parameters were determined via capacitance–voltage measurements of the electrolyte–insulator–semiconductor struc-ture. These membranes exhibit a quasi-Nernstian response of 58–59 mV/pH, good surface adsorption and corrosion resistant characteristics. The output characteristics of the PMT gate ISFET are very similar to those of metal oxide field effect transistors at a specific pH concentration. In the saturation region, the pH response and linearity can be con-trolled by adjusting the gate–source voltage (VGS), e.g. the pH responses of)8.4, )24.3 and )32.7 lA/pH and the correlation coefficients of 0.99067, 0.99959 and 0.99963 at VGS¼ 1, 3 and 5 V can be obtained, respectively. In the non-saturation region, a linear pH response exists at about 59 mV/pH.

Ó 2003 Elsevier B.V. All rights reserved.

PACS: 77.84.Dy; 82.65.)I; 83.80.Lz; 84.60.Dn

1. Introduction

An ion-sensitive field effect transistor (ISFET) based on the metal oxide field effect transistor (MOSFET) operation theory has been developed. This ISFET can be considered a special type of MOSFET without a metal gate. The gate and *_{Corresponding author. Tel.: +886-7 525 2000x4121; fax:}

+886-7 525 4199.

E-mail address:[email protected](Y.-C. Chen).

0022-3093/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved.

Journal of Non-Crystalline Solids 332 (2003) 11–19

insulator are directly exposed to the buffer solu-tion. A change in the surface potential between the electrolyte and the insulator will result from a change in the pH concentration in the electrolyte. This will induce an alteration in the electric field in the insulator–semiconductor interface and channel conductance and current modulation. Since Berg-veld first employed the field effect transistor for neurophysiological measurements in 1970 [1], the ISFET has been developed into a new type of chemical-sensing electrode. In view of its very small size, rapid response, low output impedance, compatibility with standard MOS processes and low cost, etc. [2–4], this device has advantages in comparison with the conventional ion selective electrode. This ISFET is particularly advanta-geous in biomedical engineering applications [5,6]. In the past, various dielectrics, e.g. SiO2 [1], Si3N4[7], Al2O3[8], Ta2O5[9], WO3[10], and SnO2 [11,12], were investigated as pH-sensitive gate in-sulator materials. The ISFET sensing properties are greatly dependent upon various materials owing to the different electrolyte reactivity with these materials. These materials must be deposited in a high vacuum chamber. Generally, a series of ceramic lead titanate materials exhibits chemical stability, mechanical strength, high resistivity, high permittivity and greater ease of manufacture ad-vantages [13]. Earlier we developed an ISFET with a lead titanate (PbTiO3) membrane and obtained excellent performance [14–16]. We found that many negatively charged defects exist within the amorphous PbTiO3 membrane. This was proven by the negative shift in capacitance-voltage (C–V ) curve on the Al/PbTiO3 MIS structure [14]. Con-sequently, this revealed that the ISFET with PbTiO3 membrane can be further improved using acceptor ions such as Mg2þ _{to substitute the Ti}4þ sites [13]. This would produce reduced localized states and slow response. In this study, sol–gel-derived Mg-modified lead titanate (PMT(x), where x is in mol%) materials were used as a new pH-sensitive layer. To prepare a precise stoichiometric and homogeneous lead titanate based membrane composition distribution, chemical solution methods such as sol–gel technology can be used to provide flexible and precise control over the stoi-chiometry. The sol–gel method has drawn a

con-siderable amount of attention in the scientific and technological fields because of its generally lower temperature processing condition, easier compo-sition control and homogeneity, non-vacuum technique and low cost [17,18].

In this study, the electrolyte–insulator–semi-conductor (EIS) structure will be adopted to ex-amine the pH response, surface adsorption and chemical stability properties of the PMT sensitive material. The pH response will be estimated via the C–V measurement. The various fabrication parameters, i.e. firing temperature, thickness and Mg content influence on the sensing properties will be also discussed. Finally, the electrical charac-teristics and pH responses of PMT gate ISFET will be studied by the current–voltage (I–V ) measure-ment.

2. Experimental

To investigate the PMT pH-sensitive material, the electrolyte/PMT/SiO2/Si/Al electrolyte–insula-tor–semiconductor structure (PMT EIS structure) and PMT/SiO2 gate ISFET (PMT gate ISFET), shown in Fig. 1, was fabricated using the sol–gel method. The PMT membrane was deposited onto silicon dioxide (SiO2) with a thickness of 1000 AA on p-type (1 0 0) Si with drain and source elec-trodes prepared using the standard MOS tech-nique, to form the double-layer PMT/SiO2 gate ISFET. Compared with the ISFET device, the source and the drain were removed in the EIS structure, that is, only the gate and the bottom aluminum electrode were included.

Fig. 1. Schematic of PMT gate ISFET. 12 S.-S. Jan et al. / Journal of Non-Crystalline Solids 332 (2003) 11–19

Fig. 2 shows a flow chart of the PMT(x) mem-brane fabrication process with a general chemical formula prepared using a sol–gel method. Lead acetate trihydrate Pb(CH3COO)2Æ3H2O, magne-sium acetylacetonate hydrate Mg(CH3 COCHC-OCH3)3Æ2H2O and titanium diisoprop-oxide bis (acetylacetonate) (TIAA) Ti(OC3H7)2(CH3 CO-CHCOCH3)2, produced by Aldrich Chemical Co., Inc., USA, were used as the precursors. 1,3-pro-panediol HO(CH3)2OH was used as the solvent. The gravimetrically assayed Pb(CH3COO)2Æ3H2O and Mg(CH3COCHCOCH3)3Æ2H2O reagents were dissolved in 1,3-propanediol at a 1:5 molar ratio of (Pb + Mg) to diol, to obtain various PMT(x) sol compositions. Membranes with Mg contents of x¼ 2, 3, 4, 5, 6 and 7 mol% are

ab-breviated as PMT(2), PMT(3), PMT(4), PMT(5), PMT(6) and PMT(7), respectively. The solutions were refluxed at 140°C for 0.5 h in an atmosphere and then cooled to 80°C. After adding TIAA, the solutions were further refluxed at 120 °C for 1 h and 80 °C for 2 h to promote solution homoge-neity. Stock solutions of about 1 M concentration were then obtained. Inductively coupled plasma mass spectrometry was used to confirm that the deviation from stoichiometry was within about ±1%. After stock solution deposition onto SiO2/Si substrate (or SiO2 gate ISFET) and firing at 350 °C, a crack-free monolayer of about 0.25 lm thickness was obtained [14]. A layer with a thick-ness of about 0.5 and 0.75 lm can be prepared by repeatedly coating and firing. The thermal de-composition property of a gel system with tem-perature was examined using differential thermal analysis (DTA) and thermogravimetric analysis (TGA). The phase transition of the PMT mem-brane was observed using X-ray diffraction (XRD) analysis.

Fig. 1 shows a schematic diagram of the PMT gate ISFET. Epoxy resins, which are generally used in sensor technology owing to their low per-meability [19], were used for sample encapsulation. The sensitive region must be directly exposed to the standard pH buffer solution (phosphate). The Ag/AgCl reference electrode was simultaneously immersed in the standard pH buffer solution. A family of capacitance–voltage (C–V ) curves was obtained by applying a series of step voltages at the reference electrode using a HP4284A LCR parameter analyzer. The Keithley 236 semicon-ductor parameter analyzer was used to measure the I–V characteristics of the PMT gate ISFET. To avoid the photoelectric and temperature effects, measurements were carried out in the dark at room temperature.

3. Results

3.1. Membrane characteristics

To examine the thermal gel decomposition properties, the gels were pre-dried at 100°C for 12 h and analyzed via TGA and DTA at a heating A: Lead acetate trihydrate

B: Magnesium acetylacetonate hydrate C: 1, 3-propanediol Reflux at 140o_{C for 0.5 h} Molar ratio (A+B) : C = 1 : 5 Cool to 80o_C Add TIAA Reflux at 120o_{C for 1 h} Cool to 80o_{C, reflux 2 h} Stock solution

Deposited onto SiO2/Si

substrate or SiO2 gate ISFET

Fired at 350o_{C for 1 h} Repeated coating Amorphous PMT membrane Heat-treated at 400, 450 or 500 o_{C for 1 h} Annealed PMT membrane

Fig. 2. Flow chart for the sol–gel-derived PMT membrane.

rate of 10 °C/min. The thermolytical behavior of the PMT(2) gel is shown in Fig. 3. The TGA data indicated a major weight loss between 100 and 350 °C and a final weight loss near 500 °C. The DTA data revealed a series of exothermic peaks. The peaks at temperatures between 100 and 350°C are associated with the former weight loss. The final peak near 500°C is associated with the weight loss

concerning a phase transition of the PMT(2) membrane from amorphous to crystalline.

The amorphous–crystalline transition of PMT membranes can be obtained by heat treatment. Fig. 4 shows the XRD patterns of the PMT(4) and PMT(7) membranes as a function of the heat-treatment temperature. The amorphous–crystal-line transition of the PMT(4) and PMT(7) membranes occurs at 450–500 and 400–450 °C, respectively.

3.2. Electrolyte–insulator–semiconductor structure The EIS structure is often employed to examine the surface adsorption, pH response and corrosion properties of the sensing material. The C–V mea-surement is well-known for MOS capacitance characterization, which is also suitable for char-acterizing an EIS structure [19]. Fig. 5 shows a typical set of C–V curves for the PMT(4) EIS structure with a membrane thickness of 0.5 lm in the standard buffer solution of pH 2–12 at a fre-quency of 10 KHz. A response of about 59 mV/pH can be obtained using the flat-band shift variation in C–V curves, as shown in the inset in Fig. 5.

0 200 400 600 800 40 30 20 10 0 TGA Endo EXo DTA Weight Loss (%) Temperature (C)o

Fig. 3. TGA and DTA data for the gel of PMT(2).

20 30 40 50 60

4000C

(a) _PMT(4)

5000C

4500C

Intensity (arb. units)

2θ (deg) 20 30 40 50 60 4000C (b) _PMT(7) 5000C 4500C

Intensity (arb. units)

2θ (deg)

Fig. 4. XRD patterns of the PMT membranes with Mg content at (a) 4 mol% and (b) 7 mol% as a function of heat-treatment temperature.

To find suitable Mg content for the PMT membrane, PMT EIS structure samples with the Mg content between 2 and 7 mol% were prepared. Fig. 6 shows the pH responses of PMT EIS structures with various Mg contents. A pH re-sponse above 54 mV/pH can be obtained as the Mg content falls below 7 mol%. The optimized pH response of 58–59 mV/pH exists at a Mg content of about 4–5 mol%.

To find a suitable PMT membrane thickness, PMT EIS structure samples with thicknesses of

0.25, 0.5 and 0.75 lm were prepared. The pH re-sponses of PMT(4) membranes are 57.8, 59 and 33.16 mV/pH for membrane thicknesses of 0.25, 0.5 and 0.75 lm, respectively, as shown in Fig. 7. A response of 59 mV/pH in the range of pH 2–12 occurs at a thickness of about 0.5 lm.

To find a suitable firing temperature for the PMT membrane, PMT(4) EIS structure samples with annealing temperatures of 350, 400 and 450 °C for 1 h were prepared. As shown in Fig. 8, pH responses of 54.3, 59 and 48.9 mV/pH were ex-hibited for annealing temperatures of 350, 400 and 450 °C, respectively. A response of 59 mV/pH in -6 -4-2 0 0.7 0.8 0.9 1.0 pH= 2, 4, 6, 7, 8 ,10 ,12 pH 2 pH 12 Normalized Capacitance, C/C 0 Gate Voltage, VG(V) 2 46 8 10 12 -1.6 -1.4 -1.2 -1.0 59 mV/pH Gate Voltage, V G (V) pH value

Fig. 5. A typical set of C–V curves of the PMT(4) EIS structure with the PMT(4) thickness of 0.5 lm at a frequency of 10 KHz.

2 3 4 5 6 7 54 56 58 60 pH Response (mV/pH) Mg Content (mol%)

Fig. 6. pH response of PMT EIS structures with various Mg content at a firing temperature of 400°C.

0.25 0.50 0.75 20 30 40 50 60 pH Response (mV/pH) Thickness (µm)

Fig. 7. pH responses of PMT(4) EIS structures with various PMT(4) membrane thicknesses. 350 400 450 40 50 60 pH Response (mV/pH) Annealing Temperature (ΟC)

Fig. 8. pH responses of PMT(4) EIS structures with various firing temperatures.

the range of pH 2–12 occurs at an annealing temperature of 400°C.

3.3. Ion-sensitive field effect transistor

From the results obtained above, the optimized fabrication parameters for the PMT membrane include a thickness of about 0.5 lm, a fring tem-perature of 400 °C and a Mg content of about 4 mol% in relation to the pH response. The PMT gate ISFET was prepared using the optimum fabrication parameters. Fig. 9 shows a typical set of IDS–VDS characteristics for the PMT(4) gate ISFET at pH 7 as a function of VGS. A threshold voltage VT of about 0.6–0.7 V was exhibited at pH 7. The drain–source current (IDS) is generated when VGS is greater than VT and the drain–source voltage (VDS) is applied. At a specific pH concen-tration, the output (IDS–VDS) characteristics are in good agreement with those of similarly fabricated MOSFET [19–22].

Two types of pH response can be obtained by the ISFET operated in the saturation and non-saturation region to detect the pH concentration. In the saturation region, as the ISFET is held at VGS¼ 3 V, a typical set of IDS–VDS curves for the PMT(4) gate ISFET with different pH concentra-tions is shown in Fig. 10. Three families of pH responses for the PMT(4) gate ISFET were ob-tained for different VGS, as shown in Fig. 11. When

0 1 2 3 4 5 6 0.0 0.5 1.0 1.5 2.0 2.5 V_DS(sat) = V_GS - V_T 1 V 2 V 3 V 4 V Drain-Source Current, I DS (mA) Drain-Source Voltage, V_DS (V) pH= 7 VGS= 5 V

Fig. 9. IDS–VDScharacteristics of the PMT(4) gate ISFET as a

function of the gate–source voltage.

0 1 2 3 4 5 6 0.0 0.3 0.6 0.9 1.2 Drain-Source Current, I DS ( µ A) Drain-Source Voltage, V_DS (V) VGS = 3 V, pH = 2, 4, 6, 7, 8, 10, 12 pH 12 pH 2

Fig. 10. A typical set of IDS–VDScurves of PMT(4) gate ISFET

as a function of pH concentration. 1.7 1.8 1.9 2.0 2 10 12 0.03 0.06 0.09 0.12 0.64 0.72 0.80 0.88 (c) V GS= 5 V - 3 2 . 7µA / p H (a) V GS= 1 V - 8 . 4µA / p H pH Value VGS= 3 V (b) Drain-Source Current, I DS (mA) - 2 4 . 3µA / p H 4 6 8

Fig. 11. Three families of pH responses of PMT(4) gate ISFET operated in the saturation region with (a) VGS¼ 1 V, (b)

VGS¼ 3 V and (c) VGS¼ 5 V, respectively.

these data are subjected to the linear curve-fitting processes, there exhibit the pH responses of)8.4, )24.3 and )32.7 lA/pH and the correlation coef-ficients, describing linear regression, of 0.99067, 0.99959 and 0.99963 at VGS ¼ 1, 3 and 5 V, spectively. These results show that the pH re-sponse can be controlled by adjusting VGS and increases with increased VGS. The pH response linearity also increases as VGS increases.

In the non-saturation region with VDS ¼ 0:3 V, a typical set of IDS–VGS curves of PMT(4) gate ISFET with different pH concentrations is shown in Fig. 10. A pH response of about 59.3 mV/pH can be obtained by calculating the average shift in the gate–source voltage at IDS¼ 20 lA in the range of pH 2–12, as shown in the inset in Fig. 12.

4. Discussion

According to the DTA and TGA analyses, the gel exhibits about 40% weight loss in the 100–500 °C temperature range owing to adsorbed water and solvent (b.p. of diol is 212°C) elimination and the decomposition of organic by-products. The pyrolysis procedure was carried out at a temper-ature of about 350 °C. The XRD analyses show that the temperature of the amorphous–crystalline transition decreased with the increase in Mg

con-tent. To maintain the ion binding sites at the electrolyte–insulator interface, it was necessary to maintain an amorphous PMT membrane micro-structure. Therefore, the fabrication temperature was chosen between 350 and 450 °C for the PMT membranes with various Mg content below 7 mol%.

In the EIS structure, the C–V curves consist of the accumulation, flat-band, depletion and inver-sion regions. Because the space charges are domi-nated by minority carriers in the accumulation and inversion regions, these regions are not suitable for a pH response study. However, at the flat-band condition, the charge density is low and not dominant; therefore the pH response of the EIS structure with different pH concentrations can be developed. The C–V curves shift in parallel with the buffer solution pH concentration, which is as-cribed to the flat-band voltage shift towards a positive value with the increased pH concentra-tion. This response (59 mV/pH) is a Nernstian response. This result reveals that the electrolyte/ PMT(4) interface has an excellent absorption mechanism and less interference effect for the electrolyte ions due to the sufficiently large surface site density [20]. Furthermore, based on the ex-perimental observations, the PMT(4) sensitive layer membrane is difficult to corrode in strong acid and base regions. This PMT(4) membrane is suitable to act as a pH-sensitive layer.

The Mg contents affect the microstructures and properties of PMT membranes and result in considerable variations in pH response. However, as the Mg content increases above 7 mol%, the pH response decreases drastically. This might re-sult from the decrease in surface site density owing to the increased crystallization with the increased Mg content via the XRD analysis in Fig. 4.

The thickness of a pH-sensitive layer will affect the capacitance capability and the surface site density of the membrane and result in considerable variations in pH response [23]. A Nernstian re-sponse in the range of pH 2–12 occurs at a thick-ness of about 0.5 lm. This phenomenon can be ascribed to the pH response being directly pro-portional to the insulator thickness. However, a great many charges inside the sensitive layer as a

0 1 2 3 0 20 40 60 80 Drain-Source Current, I DS ( µ A) Gate-Source Voltage, V_GS (V) pH 2 _{pH 12} 2 46 8 10 12 0.6 0.8 1.0 1.2 59.3 mV/pH pH Value VGS (V)

Fig. 12. A typical set of IDS–VGS curves for the PMT(4) gate

ISFET operated in the non-saturation region with VDS¼ 0:3 V,

measured in pH¼ 2, 4, 6, 7, 8, 10, and 12 buffer solutions, re-spectively.