Permissions for Reuse

We demonstrate a technique for creating large area, electrically stable molecular junctions. We use atomic layer deposition to create nanometer thick passivating layers of aluminum oxide on top of self-assembled organic monolayers with hydrophilic terminal groups. This layer acts as a protective barrier and allows simple vapor deposition of the top electrode without short circuits or molecular damage. This method allows nonshorting molecular junctions of up to 9  mm² to be easily and reliably fabricated. The effect of passivation on molecular monolayers is studied with Auger and x-ray spectroscopy, while electronic transport measurements confirm molecular tunneling as the transport mechanism for these devices. ©2008 American Institute of Physics

Contents

• BODY OF ARTICLE
• REFERENCES
• FIGURES
• TABLES
• FOOTNOTES

The field of molecular electronics has generated excitement due to the promise of controlling electronic transport at molecular length scales. While there have been numerous advances in creating single-molecule electronic junctions,^1,2,3,4,5 practical molecular electronic devices will almost certainly require junctions with large numbers of molecules. However, fabricating large area molecular junctions has proven to be challenging. In particular, the molecular monolayers that make up the prototypical molecular electronic device are known to be susceptible to a wide range of defects, ranging from pinhole shorts that prevent the creation of large area junctions⁶ to damage caused by electrode deposition that can affect even the smallest devices.^7,8 There are several schemes for circumventing these limitations, such as creating nanopore junctions,⁹ indirect deposition of electrodes,¹⁰ using float-on electrodes,^11,12 or combining small area junctions (<100  µm²) with conjugated polymer spacer layers between the active molecular layer and the electrodes.¹³ Here, we present a complementary metal-oxide-semiconductor (CMOS)-compatible process that enables the creation of large area (9  mm²) nonshorting molecular junctions by using atomic layer deposition (ALD) to create a nanometer thick passivating layer between the molecular film and the top electrode. This process not only protects the molecular layer from damage during metal deposition, but also passivates defects in the monolayer. The process is successful for a range of terminal groups and could be adapted for a variety of organic electronic devices.

ALD has found widespread use in recent years in creating nanometer thick high permittivity coatings ^14,15 and there has been recent interest in low temperature deposition of oxides onto biological and organic systems.¹⁶ We demonstrate that depositing nanometer thick aluminum oxide (Al₂O₃) layers onto a variety of organic self-assembled monolayers (SAMs) is sufficient to eliminate metal penetration through the SAM during evaporation of top metal contacts.^7,17

Our process is schematically shown in Fig. 1. Typically, freshly evaporated Au films were cleaned with UV ozone for 5  min, then placed in solutions of alkanethiols functionalized with various tailgroups, with concentrations of 1  mg/ml in tetrahydrofuran (THF). After 24  h, the films were removed rinsed with THF and de-ionized water and placed into a commercial ALD reactor. Next, the samples were placed into a commercial ALD reactor (Cambridge Nanotech, Cambridge, MA), where the samples were alternately exposed to the precursors trimethyl aluminum (TMA) and water vapor, creating a Al₂O₃ monolayer with each cycle, where approximately 1  Å of Al₂O₃ is deposited per cycle. Due to the need for water absorption during the ALD process, Al₂O₃ can be deposited on SAMs with hydrophilic end groups, while hydrophobic molecules prevent the deposition of Al₂O₃ (Table I).¹⁸ To avoid desorption of the SAM at high temperatures,¹⁹ the ALD reactor temperature was kept at 60  °C. The samples were exposed to the precursors for between 5 and 30  s/cycle, and purging times varied from 20  to  50  s. Nitrogen was used as a purge gas, with a flow rate of 10  SCCM. Nonshorting devices were fabricated with Al₂O₃ thicknesses ranging from 1–4  nm. After Al₂O₃ deposition, a 30  nm thick top Au electrode was deposited onto the Al₂O₃ layer using electron beam evaporation with a shadowmask pattern, completing the device.

Figure 1.

Auger electron spectroscopy confirmed that the ALD processing did not displace the SAM, as can be seen from Table I. Individual Auger spectra are shown in the supplemental information (EPAPS).²⁰ All samples with SAMs show strong sulfur and carbon signals, indicating that the monolayers survive the passivation process. SAMs made of 1-dodecanethiol (DDT), which contains a hydrophobic methyl tailgroup, act as an ALD resist, with no measurable aluminum signal in the Auger spectra, agreeing well with previous work studying SAM ALD resists on silicon.¹⁸ In contrast, SAMs with hydrophilic tailgroups such as 1-8 octanedithiol (ODT), 11-mercaptio-1-undecanol (MUD), 11-mercaptoundecanoic acid (MUA), and 4-mercaptobenzoic acid (MBA), readily support the growth of ALD Al₂O₃, as seen from the strong aluminum signal, with the more hydrophilic −COOH and −COH groups showing faster growth of Al₂O₃ than the −SH group. Additionally, the elemental analysis of the spectra for all SAMs correlates with the known elemental composition of the constituent molecules. For example, the SAM of 1-8 octanedithiol (ODT), which has a sulfur tailgroup, exhibited a significantly stronger sulfur signal than the SAMs with tailgroups not containing sulfur.

X-ray photoelectron spectroscopy (XPS) demonstrated that the chemical functionality and oxidation state of the molecular tailgroups was not significantly changed by the ALD passivation. High resolution scans of the carbon 1s (C1s) peak of several different SAMs both before and after passivation are shown in Fig. 2. The chemical shifts of the C1s peak of the thiol, alcohol, and carboxyl terminal groups before passivation are in agreement with published values.²¹ After passivation with Al₂O₃, all samples show slight broadening of the primary C1s peaks. In addition, the carbon peak for the carboxyl group in the 11-mercaptoundecanoic (MUA) acid sample shifts 0.8  eV toward lower binding energy. This change in chemical shift is consistent with the ALD process causing the oxygen atoms on the carboxyl groups to become less electronegative. There was no significant change in chemical shift for the hydroxyl tailgroup in the 11-mercapto-1-undecanol (MUD) SAM. The chemical shift due to the sulfur tailgroup in the ODT sample is known to be very small²¹ and is not resolved in our measurements.

Figure 2.

To demonstrate the viability of the ALD processing technique for molecular electronics, we have created large area tunnel junctions with device areas ranging from 0.0025  to  9  mm² and measured their I-V characteristics. Figure 3(a) illustrates six typical I-V traces taken between 166 and 293  K, showing that the current is independent of temperature, indicating tunneling as the transport mechanism.⁹ The current shows linear behavior at low bias (<0.2  V), and an exponential dependence on bias at higher voltages, which also supports tunneling as the transport mechanism.

Figure 3.

To confirm that tunneling occurred through the molecular monolayer and not through Al₂O₃ present in the SAM defects, the current density was measured for devices with several different alkane chain lengths with the same carboxyl tailgroup (8-mercaptooctanoic acid, MUA acid, and 16-mercaptohexadecanoic acid), as shown in Fig. 3(b). The samples had a 100% success rate (n>30) and all junctions had very similar and reproducible electrical behavior for biases up to 1  V. It should be noted that junctions with different device areas gave the same current density, with device areas in this sample set ranging from .03  to  .05  mm². The current density through the devices exponentially depends on the molecular chain length, as can be seen in Fig. 3(c), confirming that the transport through the device is through the molecular layer. Assuming a current density dependence of the form J  exp^−d, where is the decay coefficient and d is the molecular length, we measure a value of 0.50±0.04  Å⁻¹ at a bias of 0.5  V. This value is slightly lower than some previous results,^9,12,13 possibly due the effects of the additional Al₂O₃ tunnel barrier. It is worth noting that the ability to control the thickness of the Al₂O₃ to within an atomic layer should also enable precise tuning of the molecule-electrode coupling in these devices.

As an additional confirmation of molecular tunneling as the transport mechanism for the devices, tunneling spectroscopy was performed on the junctions at cryogenic temperatures. It is well known that inelastic electron tunneling spectroscopy (IETS) can measure vibrational modes of molecular junctions.²² In Fig. 4 we show both the differential conductance, dI/dV, and the IETS spectrum, d²I/dV², of one of the 8-mercaptooctanoic acid tunnel junctions from Fig. 3. Measurements were performed in a Desert Cryogenics cryogenic probe station at 28  K. A digital lock-in amplifier was used to directly measure harmonics of the modulated tunnel current as a function of dc bias voltage, with an ac modulation amplitude of 10  mV and a frequency of 141.3  Hz. Vibrational modes of the molecular monolayer are apparent as steps in the differential conductance, but are more clearly seen as peaks in the IETS spectrum. The IETS spectrum shows the characteristic vibration modes of carboxylic acids, and the peak widths of ~20  meV agree well with the peak broadening expected from the temperature and ac excitation amplitude.

Figure 4.

In summary, we have developed a method to easily create large area, electrically stable molecular junctions using atomic layer deposition of Al₂O₃ on SAMs with hydrophilic tailgroups. Auger spectroscopy shows that the monolayer is preserved during the passivation process, while XPS indicates that the functional tailgroups of the molecules also survive processing. Finally, transport measurements show that this process can create large area, electrically robust molecular junctions. This passivation process is applicable to molecules with a variety of hydrophilic endgroups, and enables facile CMOS processing to integrate molecules into robust electronic circuits.

We would like to thank J. Sulpizio and H. Chou for assistance with the ALD processing and L. Jimison for help with I-V measurements. ALD of alumina dielectric was performed using equipment purchased under contracts AFOSR (F49620-02-1-0383) and the ONR (YIP, N00014-01-1-0569). This work was supported by the NSF Career Award DMR 0449385 and DOE Project No. DE-AC02-76SF00515. M.J.P. acknowledges the support of the NSF Graduate Fellowship and the NDSEG Graduate Fellowship.

Auxiliary Material (EPAPS)

1. M. A. Reed, C. Zhou, C. J. Muller, T. P. Burgin, and J. M. Tour, Science 278, 252 (1997). first citation in article
   
2. H. Park, J. Park, A. K. L. Lim, E. H. Anderson, A. P. Alivisatos, and P. L. McEuen, Nature (London) 407, 57 (2000). [MEDLINE] first citation in article
   
3. X. D. Cui, A. Primak, X. Zarate, J. Tomfohr, O. F. Sankey, A. L. Moore, T. A. Moore, D. Gust, G. Harris, and S. M. Lindsay, Science 294, 571 (2001). [MEDLINE] first citation in article
   
4. B. Xu and N. J. Tao, Science 301, 1221 (2003). [MEDLINE] first citation in article
   
5. R. McCreery, Chem. Mater. 16, 4477 (2004). first citation in article
   
6. X.-M. Zhao, J. L. Wilbur, and G. M. Whitesides, Langmuir 12, 3257 (1993). [ISI] first citation in article
   
7. A. Walker, T. Tighe, J. Stapleton, B. Haynie, S. Upilli, D. Allara, and N. Winograd, Appl. Phys. Lett. 84, 4008 (2004). first citation in article
   
8. S. Chang, Z. Li, C. Lau, B. Larade, and R. Williams, Appl. Phys. Lett. 83, 3198 (2003). first citation in article
   
9. W. Wang, T. Lee, and M. Reed, Phys. Rev. B 68, 035416 (2003). first citation in article
   
10. J. Deng, W. Hofbauer, N. Chandrasekhar, and S. O'Shea, Nanotechnology 18, 155202 (2007). first citation in article
    
11. A. Vilan and D. Cahen, Adv. Funct. Mater. 12, 795 (2002). [ISI] first citation in article
    
12. K. Shimizu, J. Fabbri, J. Jelincic, and N. Melosh, Adv. Mater. (Weinheim, Ger.) 18, 1499 (2006). first citation in article
    
13. H. Akkerman, P. Blom, D. de Leeuw, and B. de Boer, Nature (London) 441, 69 (2006). [MEDLINE] first citation in article
    
14. A. Javey, H. Kim, M. Brink, Q. Wang, A. Ural, J. Guo, P. McIntyre, P. McEuen, M. Lundstrom, and H. Dai, Nat. Mater. 1, 241 (2002). [MEDLINE] first citation in article
    
15. M. Leskela and M. Ritala, Thin Solid Films 409, 138 (2002). first citation in article
    
16. M. Groner, F. Fabreguette, J. Elam, and S. George, Chem. Mater. 16, 639 (2004). first citation in article
    
17. B. de Boer, M. M. Frank, Y. Chabal, W. Jiang, E. Garfunkel, and Z. Bao, Langmuir 20, 1539 (2004). [ISI] [MEDLINE] first citation in article
    
18. R. Chen and S. Bent, Chem. Mater. 18, 3733 (2006). first citation in article
    
19. D. Lavrich, S. Wetterer, S. Bernasek, and G. Scoles, J. Phys. Chem. B 102, 3456 (1998). first citation in article
    
20. See EPAPS Document No. E-APPLAB-92-090817 for individual Auger spectra. For more information on EPAPS, see http://www.aip.org/pubservs/epaps.html. [EPAPS] first citation in article
    
21. J. Moulder, W. Stickle, P. Sobol, and K. Bomben, Handbook of X-Ray Photoelectron Spectroscopy (Physical Electronics, Chanhassen, MN, 1995). first citation in article
    
22. J. Lambe and R. Jaklevic, Phys. Rev. 165, 821 (1968). [ISI] first citation in article
    
23. N. M. D. Brown, R. B. Floyd, and D. G. Walmsley, J. Chem. Soc., Faraday Trans.75, 261 (1979). first citation in article
    

Full figure

Fig. 1. (Color) Schematic of ALD passivation process. (a) A thiolated molecular layer with a hydrophilic tailgroup is self-assembled onto a gold film. (b) TMA reacts with the hydrophilic endgroup until the surface is completely passivated. Unreacted reagents are pumped away. (c) Water vapor is pulsed onto the system, reacting with the surface, forming aluminum-oxygen bridges. Steps (b) and (c) are then repeated to create an Al₂O₃ layer of the desired thickness. (d) After ALD passivation, the top electrode is evaporated onto the device. First citation in article

Full figure

Fig. 2. (Color) High resolution XPS scans of the carbon 1s peak for several SAMs before and after passivation for ODT (a), MUD (b), and MUA (c). The chemical shifts due to the functional endgroup are clearly visible for MUD and MUA. After ALD passivation, the carboxyl peak for MUA shifts toward lower binding energy, indicating higher local electron density on the tailgroup carbon after passivation. First citation in article

Full figure

Fig. 3. (Color) (a) I-V measurements of a single MUA tunnel junction as a function of temperature. The I-V curves are temperature independent, indicating tunneling as the dominant transport process. (b) Current density vs applied bias for 8-mercaptooctanoic acid, MUA and 16-mercaptohexadecanoic acid junctions. All samples were covered with 25  cycles of ALD deposited Al₂O₃, and ellipsometery measurements indicated an Al₂O₃ thickness of 2  nm. The devices had areas ranging from .03  to  .05  mm². Each curve shown is the average of eight different devices fabricated under identical conditions, and the error bars indicate the standard deviation of the measurements. The current density exponentially depends on chain length, as can be seen in (c). The straight lines are the results of exponential fits (see text). First citation in article

Full figure

Fig. 4. (Color) Differential conductance and IETS of a 8–mercaptobenzoic acid tunnel junction. Steps in the differential conductance dI/dV indicate vibrational modes of the molecular monolayer. This can be more clearly seen in the IETS spectrum d²I/dV², where the vibrational modes now appear as peaks (Ref. 22). Following Ref. 23, we assign peak (1) at 3620  cm⁻¹ as the (O–H) mode, peak (2) at 2960  cm⁻¹ as the (C–H) mode, and peak (4) at 1450  cm⁻¹ as either the (CO₂) or the (CH₃) mode. We have one unassigned peak (3) at 2050  cm⁻¹. First citation in article

Table I. Measured atomic percentages from Auger spectra on SAMs covered by Al2O3 deposited via ALD (10  cycles), individual spectra are shown in Ref. 20. Spectra from samples without a SAM show no sulfur and only a weak carbon signal. ODT, MUD, MUA, and 4-mercaptobenzoic acid (MBA) all have strong sulfur and carbon peaks. All spectra were taken with a 20  kV, 10  nA beam on a PHI 700 scanning Auger nanoprobe.
	Sulfur	Carbon	Oxygen	Aluminum	Gold
DDT	2.3	31.2	0.5	2.6	63.4
No SAM	<0.1	3.9	35.7	27.6	32.7
ODT	1.8	20.0	21.6	18.5	38.0
MUD	0.2	7.0	35.6	25.5	31.7
MUA	0.3	8.0	34.6	25.2	31.9
MBA	0.4	8.4	34.3	25.6	31.4

First citation in article

^aElectronic mail: nmelosh@stanford.edu.

Up: Issue Table of Contents
Go to: Previous Article | Next Article
Other formats: HTML (smaller files) | PDF ( kB)