NANO LETTERS
Large-Scale Fabrication of 4-nm-Channel Vertical Protein-Based Ambipolar Transistors
2009 Vol. 9, No. 4 1296-1300
Elad D. Mentovich,†,‡ Bogdan Belgorodsky,† Itsik Kalifa,‡ Hagai Cohen,§ and Shachar Richter*,†,‡ School of Chemistry and UniVersity Center for Nanoscience and Nanotechnology, Tel AViV UniVersity, Ramat AViV, Tel AViV 69978, Israel, and Department of Chemical Research Support, Weizmann Institute of Science, RehoVot, 76100, Israel Received September 4, 2008; Revised Manuscript Received January 13, 2009
ABSTRACT We suggest a universal method for the mass production of nanometer-sized molecular transistors. This vertical-type device was fabricated using conventional photolithography and self-assembly methods and was processed in parallel fashion. We used this transistor to investigate the transport properties of a single layer of bovine serum albumin protein. This 4-nm-channel device exhibits low operating voltages, ambipolar behavior, and high gate sensitivity. The operation mechanism of this new device is suggested, and the charge transfer through the protein layer was explored.
The increasing demand for smaller and faster complementary transistors (in which both n- and p-type devices exist on one wafer) arranged in dense arrays requires the development of new methods for the parallel fabrication of nanometersized transistors. However, because of limitations in current technology, the achievement of these goals is very challenging. Although, some isolated examples of such devices and architectures have been demonstrated, they exhibit only moderate or limited performance, or are constructed via sophisticated multistep methodologies.1 Here we suggest and demonstrate a universal method in which a new type of nanometer-sized, ambipolar, vertical molecular transistor is fabricated in parallel fashion. This central-gate molecular vertical transistor (C-Gate MolVeT) is fabricated by a combination of conventional microlithography techniques and self-assembly methods. The general fabrication methodology of the C-Gate MolVeT allows the process to be adapted for various materials and systems. To study an example of such a transistor, we have investigated the operation of an ultrathin, 4-nm-channel-length C-Gate MolVeT, composed of bovine serum albumin (BSA) protein. This transistor, approximately an order-of-magnitude smaller than the existing commercial devices, shows remarkable * Corresponding author,
[email protected]. † School of Chemistry, Tel Aviv University. ‡ University Center for Nanoscience and Nanotechnology, Tel Aviv University. § Department of Chemical Research Support, Weizmann Institute of Science. 10.1021/nl802694k CCC: $40.75 Published on Web 03/05/2009
2009 American Chemical Society
ambipolar properties, operates at extremely low voltages and very low leakage currents, and possess high gate sensitivity. The architecture of the device is based on the concept of the solid-state vertical transistor,2 in which a side gate is used to activate a vertically patterned semiconducting active-channel layer. For larger systems such as conjugated polymers and nanowires, vertical transistors have been demonstrated.3 For the molecular scale this task is very challenging due to geometrical and physical restrictions.4 In addition, none of the known existing methodologies is successful in the construction of an array of vertical molecular transistors on a wafer scale. In our design, the nanometer channel length is determined by a protein-based self-assembled monolayer, that is sandwiched between source and drain electrodes inside a microcavity, while a centered oxidized-metal-electrode column inside the cavity serves as the gate electrode (Figure 1). Note that while the channel length is determined by the thickness of the active layer (in our case ∼4 nm), the submicrometer transistor area is defined by the difference between the macrosized dimension of the cavity and the area of the gate electrode. This area is approximately π(RC2 -RG2), where RC and RG are the photolithography-defined microcavity and titanium-column radii, respectively. By this simple method, one can vary the lateral area of the transistor from several square micrometers down to the subnicrometer range. Furthermore, since RC and RG are micrometer-sized dimensions, the whole process can be performed in parallel by conventional lithographic methodologies, resulting in arrays
Figure 1. The C-Gate MolVet fabrication procedure. (a) A network of gold electrodes is defined on top of a highly doped silicon wafer covered with 100 nm thick thermal oxide, followed by the deposition of a 70 nm layer of Si3N4 dielectric. (b) Arrays of microcavities, ranging from 800 nm to 1.5 µm in diameter are created by drilling holes through the entire layer to the highly doped silicon substrate, followed by mild etching of several nanometers of the gold electrode. This undercut in the electrode provides space for oxide growth. (c) A titanium column is evaporated followed by the definition of a larger cavity and oxidation of the titanium column to form the gate electrode. (d) Adsorption of the proteinbased SAM on top of the exposed gold ring and definition of the upper electrode. (e) The final C-Gate MolVet structure is achieved by an indirect evaporation of palladium on top of the protein layer. (f, g) Tilted high-resolution scanning electron microscopy (HRSEM) images of a single device (f) and array (g) of transistors before molecular assembly. (h) Optical image of the transistor after stage (c). (i) HRSEM image featuring an array of C-Gate MolVet transistors.
composed of thousands of nanometer-sized molecular transistors. Our device was operated using a BSA-based self-assembled monolayer (SAM) as the active-channel layer. Discovered more than 150 years ago, BSA (with 583 aminoacid residues and a molecular mass of 66.43 kDa) is one of the most studied proteins.5 The properties and biological functions of BSA are well-known, it is widely used in research in numerous diagnostic applications and as a nutrient in cell and microbial cultures.5 Although high-purity ligandfree BSA is manufactured in multiton quantities, its electronic properties are not well understood.6 The X-ray crystallography based structure of BSA is shown in Figure 2a. It is a prolate ellipsoid with major and minor axes of 14 and 4 nm, respectively. The Cys34 residue of BSA is responsible for the formation of the dense, selfaligned monolayer on the surface of the gold source Nano Lett., Vol. 9, No. 4, 2009
electrode. In this monolayer, each individual protein assumes “side-on” or “end-on” arrangements, having minor or major axes oriented perpendicular to the surface.7 By controlling the pH and concentration of the BSA solution, we obtained a closed packed, 4 nm height, SAM film that was evaluated by X-ray photoelectron spectroscopy and spectroscopic ellipsometry (see Supporting Information). Figures 2 and 3 show the results of transfer, sensitivity, leakage, and amplification measurements taken for the C-Gate MolVeT. Several unique features of the transistor can be deduced from the data. (i) The source-drain transport curves (ISD-VSD) do not resemble those of conventional metal-oxide-semiconductor field-effect transistors (Figure 2). Whereas in the standard metal-oxide-semiconductor field-effect transistor (MOSFET), saturation current is observed at a certain voltage, this is not the case here. The lack of saturation current is also known for very short channel length transistors.8 (ii) ISD is extremely sensitive to the gate voltage, VG; we have successfully measured ISD changes with VG intervals of 50 mV (Figure 3 top) and 15 mV at 77 K. This sensitivity is much better than that previously reported for planar SAMbased transistors.9 (iii) The operating voltages of the C-Gate MolVeTs do not exceed 1 V (Figure 2a-c). This voltage range is at least 1 order of magnitude smaller than that of conventional organic transistors, thus making the C-Gate MolVeT a good candidate for low-power-circuit applications. (iv) Despite the narrow gate oxide (8 nm), the gate-source leakage currents are 3 orders of magnitude lower than those of the drain-source currents (Figure 3 middle). Similar leakage measurements performed between the drain and the gate terminals showed even lower currents. The leakage current can therefore be neglected in this device in the range of approximately VG e 0.45 V. At higher values of VG, the leakage current cannot be neglected, and the transistor action resembles a triode-type transistor.10 We attribute this property to the high dielectric constant of the titanium dioxide-gate material.11 The effect of the gate voltage in the case of the nanometer-sized transistor is a fundamental issue that is still not fully understood.12 Kagan et al. claimed that the gate effect on thin layers is limited because of material and geometric constraints.13 It was suggested that, above a certain channel length-to-oxide width ratio, a strong gate effect cannot be produced. The maximum width of a silicon oxide layer, t, can be estimated as toxide e L/1.5, where L is the channel length. However, because of the high k of the titanium oxide gate material in our case, the effective thickness of the gate layer can be calculated by tTiO2 )
KTiO2 KSiO2
tSiO2
hence this rule of thumb still holds.13 The important consequence of using high-k gate materials in nanometersized devices is the ability to use a relatively thick (8 nm) gate oxide and still achieve a strong gate effect.13 It should be noted that although devices comprised of thicker gate 1297
Figure 2. The BSA-based C-Gate MolVet. (a) The BSA molecule. (The free cysteine group is highlighted in yellow.) (b-d) The transfer characteristics of the C-Gate MolVeT taken at room temperature. (b) ISD-VSD taken for different gate voltages. (c, d) Zoom of the negative (c) and of the positive (d) VSD regions demonstrating the ambipolar characteristics of the device.
oxides electrodes are expected to give lower leakage currents, the Gate-induced electric field in the channel should decrease dramatically, thus reducing the ability to modulate the current with VG.13 A systematic study aiming for the optimization of these parameters is currently under investigation. (v) The shape of the transport data is asymmetric. We (and others) have attributed the source of this phenomenon to the difference in the nature of the electrode contacts and their interface with the proteins.15 While strong chemical coupling between BSA and gold is achieved by a chemical bond, the coupling to the palladium electrode is weak since no chemical bond is created.16 As shown previously, this phenomenon can lead to strong asymmetry in the transport behavior. (vi) The transistor shows ambipolar characteristics (Figure 2c,d), namely, both n- and p-type transistor properties can be measured on the same transistor by tuning the appropriate gate voltages. In general, ambipolarity of transistors is highly desirable since it reduces the number of steps needed to fabricate complementary circuits.17 Furthermore, the low operating voltages of the ambipolar transistors compared to ambipolar organic FETs make them far superior to conventional organic transistors. (vii) The transistor shows amplification properties. The static amplification properties of the transistor are shown in Figure 3, bottom, where it can be seen that the transistor shows moderate amplification with transconductance of 1-2 mS. From these observations, a general operational mechanism of the transistor can be deduced. At zero bias, the potential landscape is determined by the work functions of the metal 1298
electrodes, which are 5.1 eV for the gold source electrode, 5.12 eV for the palladium drain electrode, and 3.84 eV for the titanium gate electrode. If the dielectric constant of the titanium oxide is taken into account as well, the formation of a Schottky barrier between the gate and the source/drain electrodes and penetration of an inhomogeneous static electric field into the monolayer are expected. Thus, due to the ultrasmall dimensions of the C-Gate MolVeT, a small variation in the gate voltages should change the electricfield distribution in the SAM layer dramatically for both VSD polarities. Similar behavior, including the lack of a saturation current, has been observed in both larger-scale staticinduction transistors18 and polymer triode devices.19 To explore the charge-transfer mechanism of BSA, transport measurements were performed at 77 K (Figure 4). The data can be explained by the two known dominant mechanisms for conventional organic vertical transistors: Fowler-Nordheim tunneling (FN)20 and trap-charge-limiting (TCL) behavior.21 In the case of FN, the current density, J, is related to the electric field, F, by J ∝ F2 exp(-k/F) where k is proportional to the charge effective mass at the protein and to the barrier height for carrier tunneling.22 In the TCL regime, transport is governed mainly by the electronic properties of the BSA moiety and less by the electrode properties. Thus, an increment in the bias increases the charge flux, thereby filling the limited number of traps. As a result, the overall currents increase and a power dependence of voltage on current is expected.10 Figure 4 presents typical FN plots of the transistor. The linear correlation obtained at low electric fields suggests that Nano Lett., Vol. 9, No. 4, 2009
Figure 3. Sensitivity, leakage, and amplification characteristics of the C-Gate MolVet. (Top) Sensitivity characteristics of the transistor taken for different VG with 50 mV steps. (Middle) Leakage-current characteristics of the C-Gate MolVet (triangle) compared to the source-drain currents (solid line) taken at VSD ) -0.3V. The value of the leakage current is at most 4 orders of magnitude smaller than the ISD under working VG conditions (and 2 orders of magnitude at higher gate voltages). (Bottom) Transfer characteristics of the C-Gate MolVeT (taken at VSD ) -0.4V) featuring its amplification properties.
the FN mechanism dominates in this region. At a certain defined turning point, there is an apparent change in the transport mechanism, with FN no longer effective (Figure 4). McElvain23 et al. showed that in the case of organic triode transistors, this turning point is related to a critical electric field, in which the TCL mechanism starts to dominate. In the TCL region, it is expected that I ∝ Vm.24 Thus, log-log plots should indicate the existence of this phenomenon. Figure 4 shows TCL plots of the transistor for different values of VG. The data show that the slope is highly gatedependent, thus indicating the existence of unstable charges in the BSA medium and the gate effect on it. Nano Lett., Vol. 9, No. 4, 2009
Figure 4. Transistor characteristics at 77 K. (a) Typical transistor action data. (b) FN plot of the negative drain voltage region. (c) log-log plots (TCL) of the I-V curves shown in a, taken at the high-field region. The power order, m, is highly field-dependent, ranging between 2 and 4.
To summarize, we demonstrate a new type of ambipolar nanometer-sized, molecular-based complementary vertical transistor which is fabricated by conventional lithographic methods. The transistor characteristics showed unsaturated I/V curves with high gate sensitivity and low operating voltages. Using this structure, we explored the transport properties of self-assembled monolayers of proteins. Both FN and CL mechanisms were detected in a single layer of BSA. Further investigation of this phenomenon was carried out. We believe that this universal device can be used as a platform for various applications, and that the C-GateMolVet architecture can pave the way for further investigation of three-terminal devices on the molecular scale. Acknowledgment. The authors thank Mrs. Netta Hendler, Mr. Noam Sidelman, and Mr. Gregory Avushenko for technical support, Dr. Michael Gozin for the BSA supply, and Professor Abraham Nitzan for fruitful discussions. This work was supported by the Israel Science Foundation (Project 1299
Number 604/06), USAF (project No. 073003), and the James Frank Foundation. Supporting Information Available: Monolayer preparation procedure and detailed XPS analysis of the SAM. This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Lu, W.; Lieber, C. M. Nat. Mater. 2007, 6, 841–850. (2) Stutzmann, N.; Friend, R. H.; Sirringhaus, H. Science 2003, 299, 1881– 1884. (3) Schmidt, V.; Riel, H.; Senz, S.; Karg, S.; Riess, W.; Gosele, U. Small 2006, 2, 85–88. (4) Lee, J. O.; Lientschnig, G.; Wiertz, F.; Struijk, M.; Janssen, R. A. J.; Egberink, R.; Reinhoudt, D. N.; Hadley, P.; Dekker, C. Nano Lett. 2003, 3, 113–117. (5) Peters, T. J. All About Albumin: Biochemistry, Genetics, and Medical Applications; Academic Press: San Diego, 1996. (6) Pompa, P. P.; Della Torre, A.; del Mercato, L. L.; Chiuri, R.; Bramanti, A.; Calabi, F.; Maruccio, G.; Cingolani, R.; Rinaldi, R. J. Chem. Phys. 2006, 125, 021103/1-021103/4. (7) Ying, P. Q.; Viana, A. S.; Abrantes, L. M.; Jin, G. J. Colloid Interface Sci. 2004, 279, 95–99. (8) Oyamada, T.; Uchiuzou, H.; Akiyama, S.; Oku, Y.; Shimoji, N.; Matsushige, K.; Sasabe, H.; Adachi, C. J. Appl. Phys. 2005, 98, 074506/1–074506/7.
1300
(9) Maruccio, G.; Biasco, A.; Visconti, P.; Bramanti, A.; Pompa, P. P.; Calabi, F.; Cingolani, R.; Rinaldi, R.; Corni, S.; Di Felice, R.; Molinari, E.; Verbeet, M. R.; Canters, G. W. AdV. Mater. 2005, 17, 816.822-. (10) McElvain, J.; Keshavarz, M.; Wang, H.; Wudl, F.; Heeger, A. J. J. Appl. Phys. 1997, 81, 6468–6472. (11) Majewski, L. A.; Schroeder, R.; Grell, M. AdV. Mater. 2005, 17, 192, 1535–1539. (12) Tao, N. J. Nat. Nanotechnol. 2006, 1, 173–181. (13) Kagan, C. R.; Afzali, A.; Martel, R.; Gignac, L. M.; Solomon, P. M.; Schrott, A. G.; Ek, B. Nano Lett. 2003, 3, 119–124. (14) Tinoco, J. C.; Estrada, M.; Romero, G. Microelectron. Reliab. 2, 003, 43, 895–903. (15) Mentovich, E. D.; Kalifa, I.; Tsukernik, A.; Caster, A.; RosenbergShraga, N.; Marom, H.; Gozin, M.; Richter, S. Small 2008, 4, 55–58. (16) Haick, H.; Cahen, D. Prog. Surf. Sci. 2008, 83, 217–261. (17) Meijer, E. J.; De Leeuw, D. M.; Setayesh, S.; Van Veenendaal, E.; Huisman, B. H.; Blom, P. W. M.; Hummelen, J. C.; Scherf, U.; Klapwijk, T. M. Nat. Mater. 2003, 2, 678–682. (18) Fujimoto, K.; Hiroi, T.; Kudo, K.; Nakamura, M. AdV. Mater. 2007, 19, 525–530. (19) Yang, Y.; Heeger, A. J. Nature 1994, 372, 344–346. (20) Fang, X. Y.; Tan, O. K.; Wei, Q.; Yao, M. W.; Tjin, S. C. Sens. Actuators, B 2006, 119, 78–83. (21) Chao, Y. C.; Meng, H. F.; Horng, S. F. Appl. Phys. Lett. 2006, 88, 223510/1–223510/3. (22) Fowler, R. H.; Nordheim, L. Proc. R. Soc. London, Ser. A 1928, 119, 173–181. (23) McElvain, J.; Heeger, A. J. J. Appl. Phys. 1996, 80, 4755–4766. (24) Burrows, P. E.; Shen, Z.; Bulovic, V.; McCarty, D. M.; Forrest, S. R.; Cronin, J. A.; Thompson, M. E. J. Appl. Phys. 1996, 79, 7991–8006.
NL802694K
Nano Lett., Vol. 9, No. 4, 2009
