Sensitive gas sensor embedded in a vertical polymer space-charge-limited transistor
Hsiao-Wen Zan, Chang-Hung Li, Chih-Kuan Yu, and Hsin-Fei Meng
Citation: Applied Physics Letters 101, 023303 (2012); doi: 10.1063/1.4734498
View online: http://dx.doi.org/10.1063/1.4734498
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/101/2?ver=pdfcov Published by the AIP Publishing
Articles you may be interested in
CSA doped polypyrrole-zinc oxide thin film sensor AIP Conf. Proc. 1512, 500 (2013); 10.1063/1.4791130
High output current in vertical polymer space-charge-limited transistor induced by self-assembled monolayer Appl. Phys. Lett. 101, 093307 (2012); 10.1063/1.4748284
High-performance space-charge-limited transistors with well-ordered nanoporous aluminum base electrode Appl. Phys. Lett. 99, 093306 (2011); 10.1063/1.3632045
Polymer space-charge-limited transistor as a solid-state vacuum tube triode Appl. Phys. Lett. 97, 223307 (2010); 10.1063/1.3513334
Low voltage active pressure sensor based on polymer space-charge-limited transistor Appl. Phys. Lett. 95, 253306 (2009); 10.1063/1.3266847
Sensitive gas sensor embedded in a vertical polymer space-charge-limited
transistor
Hsiao-Wen Zan,1,a)Chang-Hung Li,1Chih-Kuan Yu,1and Hsin-Fei Meng2
1
Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, Taiwan
2
Institite of Physics, National Chiao Tung University, Taiwan
(Received 23 May 2012; accepted 17 June 2012; published online 11 July 2012)
We report a very sensitive gas sensor embedded in a vertical polymer space-charge-limited transistor. The oxidizing and reducing gases act as electron dedoping and electron doping agents on the transistor active layer to change the potential distribution in the vertical channel and hence to change the output current density. With a 30-ppb detection limit to ammonia, the sensor can be used for non-invasive breath monitor in point-of-care applications. The integration of a sensitive gas sensor and a low-operation-voltage transistor in one single device also facilitates the development of low-cost and low-power-consumption sensor array.VC 2012 American Institute of Physics.
[http://dx.doi.org/10.1063/1.4734498]
Organic semiconductor materials (OSMs) have been investigated and applied to thin-film transistors because of the low-cost and large-area fabrication on flexible substrates.1The gas-sensing ability of OSMs is an unique property that allows the integration of vapor sensors with organic thin-film transis-tors (OTFTs), which has been extensively studied recently.2,3 However, high operation voltage (>10 V) is usually required in these OTFT-based gas sensor. Sensitivity may be limited because the gas molecules mostly contact with bulk area (the exposing area) rather than channel area (buried under bulk region).4,5In this work, we propose a gas sensor based on a vertical polymer transistor. The current flows in the bulk region of the vertical channel. Exposing the vertical channel to gas molecules creates a significant interaction between chan-nel and gas molecules and hence a high gas sensitivity.
The vertical polymer transistor is called a space-charge-limited transistor (SCLT). In our previous works, we had demonstrated that SCLT exhibits a high on/off current ratio (>10 000) at a low operation voltage (<2 V). The opera-tional mechanism of the vertical polymer SCLT is similar to the solid-state vacuum tube triode.6 The hole current from the bottom (emitter) to the top (collector) of the bottom-injection SCLT is modulated by the metal-grid, which forms the base terminal of the vertical channel. The bias of the metal-grid controls the potential profile in the vertical chan-nel and hence switches the on and off states of SCLT. It has been demonstrated that SCLT exhibits a very low switching swing, indicating that a slight change of the potential barrier results in a significant current variation when SCLT is biased in the switching transition region.7In this work, we demon-strate that the low switching swing of SCLT is critical to pro-vide high gas sensitivity. The lowest detectable ammonia concentration is 30 ppb for a poly(3-hexylthiophene) (P3HT) SCLT. On the other hand, for P3HT OTFT with a conven-tional bottom-gate structure, the detection limit to ammonia gas is larger than 1 ppm.2 In literatures, it is known that ammonia molecules (NH3) act like electron donors to
pro-vide additional electrons to recombine holes in P3HT and hence lower down the conductivity in P3HT.8–10 In this work, by using computer-aided simulator, we demonstrate that SCLT is sensitive to electron doping in channel region. The simulated response of SCLT to channel electron doping agrees well with the experimental response of SCLT to am-monia molecules. The high sensitivity to amam-monia achieved in this work enables the development of non-invasive breath ammonia analysis for monitoring dysfunction of the human body.11 For such applications, a portable and real-time am-monia sensor with a detection limit of 50 ppb is critical but is still challenging.11 Our results may facilitate the develop-ment of low-cost point-of-care technology.
Three-dimensional and two-dimensional SCLT sche-matic diagrams are shown in Fig. 1(a); a scanning electron microscope (SEM) image of SCLT is shown in Fig. 1(b). We prepared an indium tin oxide (ITO) glass substrate as the emitter (E). Cross-linkable poly(4-vinyl phenol) (PVP) (8 wt. %) (Mw approxiamately 20 000, Aldrich) and cross-linking agent poly(melamine-co-formaldehyde) (PMF) were dissolved in propylene glycol monomethyl ether acetate (PGMEA) with a PVP:PMF mass ratio of 11:4. The solution was then spun onto ITO at 1600 rpm for 40 s and annealed at 200C for 1 h to form a 200 nm-thick organic dielectric layer. A 1.5 wt. % P3HT (RR > 98.5%, Rieke Metals Inc.) solution dissolved in chlorobenzene was then spun onto PVP to form a 20 nm thick layer. The substrate was spin-rinsed with p-xylene to increase the P3HT surface polarity. The substrate was then immersed into a solution of 100 nm-diameter negatively charged polystyrene (PS) balls. After the PS balls had adhered to the P3HT surface, 40 nm-thick Al metal was deposited onto the prepared PVP substrate by thermal evaporation, to serve as the base (B). We used Scotch tape (3M) to remove the PS balls and reveal the metal-grid base. O2plasma at 100 W was applied to etch the
bare PVP for 8 min, to form vertical channels. A 400 nm-thick P3HT active layer was spun onto the substrate and was annealed at 200C for 10 min. The substrate was immersed into a solution of 100 nm-diameter PS balls again. Once the balls had adhered to the P3HT surface, a 40 nm-thick Al a)Author to whom correspondence should be addressed. Electronic mail:
hsiaowen@mail.nctu.edu.tw.
0003-6951/2012/101(2)/023303/4/$30.00 101, 023303-1 VC2012 American Institute of Physics
layer was deposited by thermal evaporation. The PS balls were then removed by tape to form a collector (C) electrode with high-density nano-meter pores. The test gas easily inter-acted with the active layer via the nano-meter pores. The transfer characteristics, the collector current density (JCE) as
a function of the base voltage (VBE), of the SCLT with
porous collector (porous SCLT) are shown in the inset of Fig. 1(b). With a collector bias (VCE) as 2.4 V, porous
SCLT exhibits an on/off current ratio as 4750 and a switch-ing swswitch-ing as 140 mV/dec. When porous SCLT is biased at VCE¼ 1.2 V, on/off ratio and switching swing are 890 and
122 mV/dec, respectively.
Before analyzing the experimental gas sensing response, we usedTCAD SILVACO ATLASsoftware to simulate the potential distribution in the vertical channel and the ideal transfer char-acteristics of SCLT. Particularly, to reflect the ammonia sens-ing response, the influences of electron dopsens-ing (e-dopsens-ing) on SCLT are investigated. In literatures, ammonia molecules dif-fuse into P3HT layer to serve as e-doping agents.9,10To dis-cuss equilibrium condition, we assume uniform e-doping distribution in the whole P3HT layer. Material parameters of TCAD simulation were defined in Ref.12. Figures 2(a)and 2(b) show 2-dimensional potential profiles of the vertical SCLT channel for e-doping concentrations of 1015and 1016, respectively. SCLT is biased in off state with VCE¼ 1.2 V
and VBE¼ 1.5 V. It is observed that increasing e-doping
con-centration in P3HT results in the increase of the potential bar-rier, particularly in the central region of the vertical channel. The potential distributions along the central vertical channel from top Al (C) to bottom ITO (E) with various e-doping con-centrations are plotted in Fig.2(c). As e-doping concentration
increases, the potential barrier increases. The corresponding ideal JCE VBEcurves are shown in Fig.2(d). With a fixed
VCE as 1.2 V, JCE VBE curves shifts to the left when
e-doping concentration increases, indicating that a more nega-tive base potential is required to lower down the channel potential barrier and to turn on the transistor. To further verify the influence of e-doping on channel potential barrier, the cur-rent density variation ratios (DJ/J0) as a function of VBE
for various doping concentrations are plotted in the inset of Fig. 2(d). The current density variation ratio (DJ/J0),
repre-senting the response sensitivity, is defined as (JCE JCE0)/
JCE0, where JCE0is the collector current density at a electron
doping concentration of 1015cm3. We found that DJ/J0 is
strongly dependent on VBE and the maximum DJ/J0 is
FIG. 1. (a) The three-dimensional and two-dimensional porous SCLT sche-matic diagrams; (b) The SEM cross-section image of the porous SCLT. Inset of (b) shows the transfer characteristics of the porous SCLT.
FIG. 2. The two-dimensional potential distribution of P3HT SCLT with e-doping concentration as (a) 1015and (b) 1016cm3at VCE¼ 1.2 V and
VBE¼ 1.5 V. (c) A simulated potential distribution at the central vertical
channel from bottom to top. (d) The corresponding simulated JCE VBE
curves with various e-doping concentrations. Inset of (d) shows the current density variation ratio (DJ/J0) as a function of VBE.
obtained in the switching region (e.g., 0.5 V < VBE< 2 V in
this case). As aforementioned, when SCLT is biased in the switching region, a slight change of base potential causes a significant current variation. The maximum DJ/J0 in the
switching region indicates that porous SCLT biased in the switching region exhibits largest response to e-doping as well as the gas molecule interaction. When VCE is2.4 V,
simu-lated DJ/J0also has a peak value in the switching region (not
shown).
The experimental gas sensing response is then investi-gated. Two kinds of gas molecules, ammonia and nitric ox-ide, are used to serve as acceptor (i.e., e-doping) and donor (i.e., hole doping) to P3HT, respectively.10 The porous SCLT was placed in a micro-fluid sensing chamber contain-ing an atmosphere of high purity (99.9999%) nitrogen gas. We used an electrical syringe pump system to inject the test gas (NH3, 99.9999% pure) into a tube to mix with the N2.
The gas mixture then entered the micro-fluid system. The amount of N2was controlled by a mass-flow controller, and
specific concentrations of test gas were obtained by adjusting the injection speed of the syringe pump. The concentration of nitric oxide (NO) gas was controlled by a mass-flow controller.
Figure 3(a) shows a plot of JCE VBE, representing
the porous SCLT’s sensing response to NH3. VCEwas fixed
as 1.2 V and the NH3 concentrations were ranged from
30 ppb to 1000 ppb. The response of the switching region
(VBE¼ 0.4 V to 0 V) of JCE VBE plot is shown in the
inset of Fig.3(a). Increasing NH3concentration makes JCE
VBE curves shift to the left, indicating an increase of the
potential barrier in the vertical channel. This sensing response agrees well with the simulation results for JCE
VBEdescribed in Fig.2(d). The sensing sensitivities DJ/J0as
a function of VBEfor various NH3concentrations are plotted
in Fig. 3(b). We found that the sensitivity was strongly de-pendent on VBE, and that the maximum sensitivity occurred
in the switching zone (0.5 V < VBE< 0 V). As anticipated,
the experimental results agree well with the simulated results in the inset of Fig. 2(d). For NH3concentrations of 30 ppb,
100 ppb, and 1000 ppb, the maximum sensitivities measured at VBE¼ 0.2 V and VCE¼ 1.2 V are 0.09, 0.23, and
0.56, respectively. As shown in the inset of Fig. 3(b), a power law relationship is found between the maximum sen-sitivity and NH3concentration, indicating that the proposed
NH3sensor is particularly sensitive in low-concentration
re-gime (i.e., 30 ppb to 1000 ppb). Note that when VCE is
2.4 V, the maximum sensitivity to 100-ppb NH3is 0.19
at VBE¼ 0.2 V. Compared to porous SCLT biased at
VCE¼ 1.2 V, porous SCLT biased at VCE¼ 2.4 V
exhib-its a larger initial on/off current ratio as 4750 but a slightly degraded NH3 sensitivity. The large on/off ratio is due to
the large on current. Since the sensing response is defined as the current variation ratio, the large on current also serves as a large background signal to suppress the NH3 sensing
response.
Since the proposed sensor is embedded in a vertical tran-sistor, it is therefore important to evaluate the switching properties of the transistor under NH3sensing. The switching
function of the porous SCLT under NH3sensing is shown in
Fig.4(a). The porous SCLT is biased at VCE¼ 1.7 V to
ex-hibit an initial on/off current ratio as 3300 and a good
FIG. 3. (a) A plot of JCE VBEof porous SCLT under various NH3
concen-trations. Inset of (a) shows the response to NH3 in the switching region
(VBE¼ 0.4 V to 0 V) of JCE VBE. (b) The sensing sensitivities DJ/J0as a
function of VBEfor various NH3concentrations. Inset of (b) shows the
maxi-mum sensitivity as a function of NH3concentration.
FIG. 4. The real-time sensing switching function of porous SCLT under (a) NH3and (b) NO.
enough sensitivity to NH3. The porous SCLT exhibits a
sig-nificant current drop under NH3sensing (the shaded areas).
During NH3sensing, a good on/off switching property is still
obtained when switching VBEbetween0.9 V (on state) and
0.9 V (off state). This result suggests that the proposed po-rous SCLT integrates a NH3sensor and a switching transistor
in one single device. In sensor array technology, to save the operational power and to improve the signal-to-noise ratio, the pixel circuit is composed of one sensor and one switching transistor.13 Our proposed device can serve as a pixel cir-cuitry by itself, facilitating the development of low-power sensor array technology. After removing NH3, the recovery
behavior is found to be dependent on VBE(not shown). A
0.9 -V VBEbias (on state) causes a poor recovery while a
0 -V or þ0.9 -V VBE bias (off state) for 600 s leads to an
almost complete recovery of the current level. The negative VBEbias may attract NH4þions, which are generated by the
decomposition of NH3, and hence impede the desorption of
NH3.
14
In addition to study the sensor response to e-doping agents like NH3, the response to hole doping agents is also
investigated. Fig.4(b)shows the real-time sensing response of JCE to NO. To observe the influences of NO on both on
and off state of the transistor, porous SCLT is biased at VCE¼ 2.5 V to exhibit a large initial on/off current ratio as
5000. In opposite to the current-drop response to NH3, the
sensor exhibits a significant current increase when NO is injected into the sensing chamber. Since NO is known to serve as the hole doping agents on P3HT,15it is suggested that hole concentrations in P3HT is increased and the hole current is also increases during NO sensing. The increased hole concentration in P3HT, however, deteriorates the switching function of SCLT. As shown in the shaded area (NO sensing area), off-state current increases significantly when SCLT is biased at VBE¼ 0.9 V (off state). The poor
base control over the channel potential may be due to the base-field shielding effect (or the screening effect) when P3HT is doped by hole-doping agents. Similar base-field shielding effects in SCLT have been observed in our previ-ous reports when we added the tetrafluoro-tetracyano-quino-dimethane (F4-TCNQ) to dope P3HT or when we irradiated light to create electron-hole pairs in P3HT.16,17 A poor re-covery is also found in NO sensing as shown in Fig. 4(b). Changing VBEno longer improves the recovery. The poor
re-covery when exposing conducting polymers to strong oxidiz-ing gases was also observed in other reports.18 The mechanism is still not clear.
In summary, we proposed a low-power sensitive gas sensor embedded in a vertical polymer transistor, SCLT. In
SCLT, current flows in the bulk region of the vertical chan-nel. Exposing the vertical channel to gas molecules creates a significant interaction between channel and gas molecules and hence a high gas sensitivity. Moreover, the potential pro-file in the vertical channel is sensitive to the channel doping concentration as well as the base bias. When P3HT-based po-rous SCLT is biased in the switching region, a maximum sen-sitivity to NH3 can be obtained with a detection limit as
30 ppb. The sensitivity is much higher than P3HT-based OTFT, which has a NH3detection limit higher than 1 ppm.
The high NH3sensitivity achieved in this work facilitates the
development of non-invasive breath ammonia analysis for point-of-care applications. The proposed sensor integrating a gas sensor and a switching transistor in one single device also can be used as pixel circuitry in sensor array applications.
This work was supported in part by the National Science Council under Grant 2628-E-009-018-MY3 and 100-2628-M-009-002.
1
B. Crone, A. Dodabalapur, Y. Y. Lin, R. W. Filas, Z. Bao, A. LaDuca, R. Sarpeshkar, H. E. Katz, and W. Li,Nature (London)403, 521 (2000).
2
J. W. Jeonga, Y. D. Leea, Y. M. Kima, Y. W. Parka, J. H. Choia, T. H. Parka, C. D. Soob, S. M. Wonb, I. K. Hanc, and B. K. Jua,Sens. Actuator B146, 40 (2010).
3
R. S. Dudhe, S. P. Tiwari, H. N. Raval, M. A. Khaderbad, R. Singh, J. Sinha, M. Yedukondalu, M. Ravikanth, A. Kumar, and V. R. Rao,Appl. Phys. Lett.93, 263306 (2008).
4
B. Li and D. N. Lambeth,Nano Lett.8, 3563 (2008).
5
A. N. Sokolov, M. E. Roberts, and Z. Bao,Mater. Today12, 12 (2009).
6
Y. C. Chao, M. C. Ku, W. W. Tsai, H. W. Zan, H. F. Meng, H. K. Tsai, and S. Fu Horng,Appl. Phys. Lett.97, 223307 (2010).
7
Y. C. Chao, H. K. Tsai, H. W. Zan, Y. H. Hsu, H. F. Meng, and S. F. Horn,Appl. Phys. Lett.98, 223303 (2010).
8
G. Bischoff and W. F. Schmidt,Macromol. Mater. Eng.208, 151 (1993).
9F. Mohammad,J. Phys. D: Appl. Phys.31, 951 (1998). 10
H. Bai and G. Shi,Sensors7, 267 (2007).
11
T. Hibbard and A. J. Killard,Crit. Rev. Anal. Chem.41, 21 (2011).
12
The SCLT characteristics simulation is made based on the device structure shown in Fig.2. The thicknesses of PVP, Al grid, Al2O3, and P3HT are
200, 40, 10, and 400 nm, respectively. The opening diameter is 100 nm. The highest occupied molecular orbital and lowest unoccupied molecular orbital levels of P3HT are 5.2 and 3.0 eV. The work functions of emitter and collector are 5.2 and 4.3 eV. The hole mobility and electron mobility in P3HT are 105and 106cm2/V s.
13
B. Guo, A. Bermak, P. C. H. Chan, and G. Z. Yan,Solid-State Electron.
51, 69 (2007).
14P. Foot, T. Ritchie, and F. Mohammad,J. Chem. Soc., Chem. Commun.
1536–1537 (1988).
15
V. Saxena, D. K. Aswal, M. Kaur, S. P. Koiry, S. K. Gupta, J. V. Yakhmi, R. J. Kshirsagar, and S. K. Deshpande, Appl. Phys. Lett. 90, 043516 (2007).
16
Y. C. Chao, C. Y. Chen, H. W. Zan, and H. F. Meng,J. Phys. D: Appl. Phys.43, 205101 (2010).
17
H. W. Zan, W. W. Tsai, and H. F. Meng,Appl. Phys. Lett.98, 053305 (2011).
18
M. K. Ram, O. Yavuz, and M. Aldissi,Synth. Met.151, 77 (2005).
