IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 16, NO.

8, AUGUST 2015 1927

Thin Film Silicon Nanoparticle UV Photodetector

O. M. Nayfeh, S. Rao, A. Smith, J. Therrien, and M. H. Nayfeh

AbstractWe constructed ultraviolet (UV) photodetectors by

room-temperature deposition of Si nanoparticle films on Si p-
type substrates. Silicon nanoparticles of 1-nm diameter are
dispersed from Si wafers using electrochemical etching. The
currentvoltage characteristics indicate a photoconductor in
series with a diode-like junction with a large enhancement in the
forward current under UV illumination. With increasing
wavelength, the response drops rapidly, dropping to a few
percent at 560 nm. These results point to a sensitive UV detector
with good visible blindness where the particle films effectively
constitutes a wide-bandgap material.
Index TermsNanotechnology, photoconducting devices, pho- Fig. 1. Schematic of the silicon nanoparticles UV photodiode.
toconductivity, photodetector, photodiode, semiconductor devices, silicon, thin films.

T HERE HAS been great interest in emitters, sensors, or

filters that are efficient and compatible with the existing
silicon-based integrated circuit technology [1][8]. Silicon,
the most common semiconductor, has an indirect bandgap of
1.1 eV, corresponding to an infrared wavelength of 1.1 m,
making it not useful for ultraviolet (UV) detection. Thus, none
of the material available for UV detection so far is based on
elemental silicon. We present here UV photodetectors with
superior efficiency created by deposition of Si nanoparticle
films on Si substrates. Contrary to bulk Si, Si nanoparticles
exhibit behavior of direct wide-bandgap materials. This makes
it very sensitive to UV light and transparent to visible light,
with good rejection ratio. Fig. 2. IV characteristics of a typical diode (a) taken under dark conditions,
(b) taken with 365-nm light from a mercury lamp directed onto the sample, and
Recent research has shown that reducing the size of a Si crystal (c) the difference of the responses (a) and (b).
to a few tens of atoms ( 1 nm) effectively creates a di-rect
wide-bandgap material [9], [10]. UV light couples to the particles
efficiently to produce electron-hole pairs. Contrary to bulk Si or We electrochemically dispersed crystalline Si in the mixture
large particles, the electron hole pairs do not appre-ciably HFH O [12][16]. We prepared blue luminescent 1-nm
recombine via nonradiative processes, allowing charge separation particles, which were characterized by electron microscopy and
and collection. Nonradiative recombination is not strong in ultra- fluorescent spectroscopy [17], [18]. We start with a p-type
small Si nanoparticles due to strong quantum confinement [11]. device-quality (100) Si wafer of 48 cm conductivity
This is evidenced by the strong and fast ra-diative recombination ( dopant/cm ). An oxide layer of 500 nm is grown on the
they exhibit. For instance, the emission proceeds over 110-ns substrate. A single pattern or an array of patterns is etched using
time scale typical of a direct-gap mate-rial with brightness that HFNH F. We immerse the wafer, positively biased relative to a
exceeds that of common dyes [12][15], platinum electrode, in a suspension of particles in alcohol. A
[6]. In this letter, we describe the fabrication of UV current is established, depositing nanoparticles in the hole,
photodetec-tors using the new Si nanoparticle material. producing a film of 500 nm thick with an active region of 5 mm
in diameter. Semitransparent gold of 4-nm thickness is deposited
on the particle film. The 300-nm-thick gold pads are deposited on
Manuscript received December 12, 2003; revised April 21, 2004. This work
was supported by the U.S. NSF under Grant BES-0118053, by the State of the gold layer and on the back of the substrate, providing means
Illi-nois under Grant IDCCA 00-49106, by the U.S. DOE under Grant to bias the device, as shown in Fig. 1.
DEFG02-ER9645439, by the NIH under Grant RR03155, and by the We used a Hg lamp of a fluorescence microscope at powers of
University of Illi-nois at Urbana-Champaign.
O. M. Nayfeh is with the Electrical Engineering Department, 13 mW focused to a spot 2 mm in diameter. Fig. 2(a) shows the
Massachusetts Institute of Technology, Cambridge, MA 02139-4307 USA. currentvoltage ( ) curves under dark conditions. They show
S. Rao, A. Smith, J. Therrien, and M. H. Nayfeh are with the Department of somewhat rectifying behavior. Similar behavior was observed in
Physics, University of Illinois at Urbana-Champaign, Urbana, IL 61801 USA
(e-mail: [email protected]). silicon nanoparticle-based devices of equivalent configuration
Digital Object Identifier 10.1109/LPT.2004.831271 constructed to conduct electroluminescence studies [19]. The

1041-1135/04$20.00 2015 IEEE

1928 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 16, NO. 8, AUGUST 2015

Fig. 3. Response (in arbitrary units) versus wavelength.

forward current, beginning to rise at 0.75 V, increases nonlin-

Fig. 4. IV characteristics of a single 1-nm particles on p-type substrate
early and saturates locally to 0.075 mA at 2 V. Beyond 2 V, the recorded by a scanning tunnelling microscope whose tip is above the particle.
current rises linearly to 0.4 mA at 9 V, with an impedance of 21 k
. At 2 V, the ratio of forward to reverse current is seven. Fig.
2(b) was taken with 2 mW at 365 nm. It shows very strong hence, there are no unoccupied levels. The high impedence also
rectification. The forward current is enhanced by a factor of 20, indicates a film with negligible cracking, and consistent with the
while the reverse current is essentially unaffected. The ratio of measured low leakage current ( 0.075 mA). The film, therefore,
forward to reverse current at 2 V is now 160. Fig. 2(c) gives the acts as a good insulator in the dark, but becomes capable of
difference of the responses taken with and without 365-nm light. conducting electric current when exposed to light. We believe
The spectral responsitivity is the product of the quantum that conduction is based on hole states generated by the UV
efficiency for carrier generation and that of charge collection. The radiation. We examined the mechanism using scanning tunnelling
nonlinearity in the response reflect space charge ef-fects, spectroscopy. We placed a low coverage of 1-nm nanoparticles on
including multiplication (gain) and saturation, under high charge a Si substrate. The room-temperature spectrum, taken using
densities and fields ( 40 kV/cm). At 1 V, the sensitivity is 150 electron injection from the microscope tip, shows that UV
mA/W. With voltage, it rises to a local maximum of enhances the transport at a period of 1.0 eV, as shown in Fig. 4.
750 mA/W at 2 V. We use , where is the The period is consistent with carrier confinement energy levels
wavelength of the incident radiation, is Plancks constant, is [16]. If a hole is generated in each particle, and for a closely
the electronic charge, is the light speed, is the quantum effi- packed film, this would correspond to a very high electronhole
ciency, and is the gain. For A/W, this gives a gain density, corresponding
of 2.6 at 365 nm, assuming . Other devices tested showed to an effective doping of /cm .
wider saturation regions (plateau), with softer rises. Responsiv- In the device, the nanoparticle film/p-Si substrate represents
ities ranging from 80 to 180 mA/W at 1 V, and 350750 at 2 V the diode-like behavior [19], [20]. The marginal rectification in
have been measured for different depositions. The device has the dark (a factor of seven) is related to the fact that the effec-tive
good visible blindness. The responsitivity, measured using Hg or doping of the particle film is very low, with very little con-
Xe lamps (shown in Fig. 3), drops to 0.036 at 560 nm com-pared duction electrons present. The enhancement in the rectification
to 22 at 250 nm. Due to propagation problems of deep UV, the upon irradiation (to a factor of 160) is caused by the creation of
gain at 200 is underestimated. the above high electronhole or effective doping density.
The dark current is due to leakage from imperfections and The gain (2.6) and spectral responsitivity (750 mA/W) are
cracks in the film. If the device operates as a shot-limited de- related among other factors to the uniformity of the particles (1
tector with response time of 0.3 s ( MHz device band- nm in diameter). Size uniformity provides uniform confine-ment
width), and a leakage current of mA, the signal-to- energies and bandgaps (3.6 eV). Uniform bandgaps allow
noise ratio of the shot noise SNR , where is the resonant tunnelling between particles, which is much faster than
electronic charge is ensured at a level of ( 69 dB). We nonresonant processes. Resonant tunnelling-limited trans-port
should note that no temperature studies have been conducted to provides higher carrier mobility , hence, shorter transit
determine the contribution of the thermal (Johnson) noise to the time according to (assuming a single electron-hole
dark current. Moreover, we made no measurements to control the mobility), where is the applied bias, is the film thickness, and
dark current using mesoscopic patterning. is the effective mobility [20]. This is to be contrasted with the
We now discuss the response in terms of the films electronic laser ablation or chemical vapor deposition (CVD), which
and physical properties. First, the open circuit impedance is a produce a wide size distribution, hence, a wide bandgap distri-
few . The high impedance of the film is a direct result of bution. In this case, the conduction proceeds via complicated
the fact that the individual nanoparticles, under low doping, paths in search of the same size particles, resulting in longer
are essentially nonconducting. Under doping of /cm paths, hence lower mobility. For instance, the effective mobility
(48- cm conductivity) as in this sample, less than one in a across films of nanoparticles prepared by pulsed laser ablation
million of the 1-nm particles contain one dopant atom, was measured to be of the order of cm /Vs [20]. On the
NAYFEH et al.: THIN FILM SILICON NANOPARTICLE UV PHOTODETECTOR 1929

biasing voltage. The same architecture may be used as a light

emitter, or a UV filter for monitoring enzyme reactions. High
efficiency allows the study of extremely small samples, assays
in small volumes, and sensor array for high-speed screening.

REFERENCES
[1] P. Ball, Let there be light, Nature, vol. 409, pp. 974976, Feb. 2001.
[2] L. Fletcher, In brief, Nature Biotechnol., vol. 20, p. 351, Apr. 2002.
[3] M. Green, J. Zhao, A. Wang, P. Reece, G. Weedon, and S. Harris,
Effi-cient silicon light-emitting diodes, Nature, vol. 412, pp. 805
808, Aug. 2001.
[4] W. Ng, M. A. Lourenco, R. M. Gwilliam, S. Ledain, G. Shao, and K.
P. Homewood, An efficient room-temperature silicon-based light-
emit-ting diode, Nature, vol. 410, pp. 192194, Mar. 2001.
[5] European commision technology roadmap-optoelectronics intercon-
nects for integrated circuits, Office for Official Publications of the
European Communities, Luxemburg, 1998.
Fig. 5. Bandgap versus the lattice constant of materials used in UV detection. [6] D. M. Brown, E. T. Downey, M. Ghezzo, J. W. Kretchmer, R. J. Saia,
Y. S. Liu, J. A. Edmond, G. Gati, J. M. Pimbley, and W. E. Schneider,
other hand, under the resonant conditions in the present devices, Silicon carbide UV photodiodes, IEEE Trans. Electron Devices, vol.
40, pp. 325343, Feb. 1993.
we expect an effective mobility as large as cm /Vs. [7] M. Krause, M. Topic, H. Stiebig, and H. Wagner, Thin-film UV de-
The enhanced responsitivity and gain are also related to the tectors based on hydrogenated amorphous silicon and its alloys,
Phys. Stat. Sol A, vol. 185, pp. 121127, May 2001.
quality of particle packing. Unlike laser ablation or CVD de- [8] J. M. Pimbley and G. J. Michon, Charge detection modeling in solid-
position, which provides no direct control on the size and the state image sensors, IEEE Trans. Electron Devices, vol. 34, pp. 294
degree of packing, we utilize uniform particles in close packing 300, Feb. 1987.
[9] G. Allan, C. Delerue, and M. Lannoo, Nature of luminescent surface
arrangement. In this case, there is minimum non-Si material be- states of semiconductor nanocrystallites, Phys. Rev. Lett., vol. 76, pp.
tween particles that may lead to strong barriers. In our samples, 29612964, Apr. 1996.
we expect barriers of 1 nm, thus tunnelling is the dominant [10] M. H. Nayfeh, N. Rigakis, and Z. Yamani, Photoexcitation of Si-Si
surface states in nanocrystallites, Phys. Rev. B, vol. 56, pp. 2079
factor. Since the gain is inversely proportional to the transit time, 2084, July 1997.
the reduction of potential barriers enhances the gain and respon- [11] M. Lannoo, C. Delerue, and G. Allan, Theory of radiative and nonra-
sitivity [19]. In addition to packing effects, the sensitivity is de- diative transitions for semiconductor nanocrystals, J. Lumin., vol. 70,
pp. 170184, Oct. 1996.
pendent on the thickness of the film. It must be thin enough to [12] O. Akcakir, J. Therrien, G. Belomoin, N. Barry, J. D. Muller, E.
allow penetration of the radiation, such that a path for carriers is Gratton, and M. H. Nayfeh, Detection of luminescent single
created across the film. Thicknesses as low 100 nm can be used. ultrasmall silicon nanoparticles using fluctuation correlation
spectroscopy, Appl. Phys. Lett., vol. 76, pp. 18571859, Apr. 2000.
Although more susceptible to nonuniformity and cracks, which [13] M. H. Nayfeh, O. Akcakir, J. Therrien, Z. Yamani, N. Barry, W. Yu, and
cause leakage, thin films allow usage of larger active areas, and E. Gratton, Highly nonlinear photoluminescence threshold in porous
provide higher gain. For weak radiation or/and thick films, fo- silicon, Appl. Phys. Lett., vol. 75, pp. 41124114, Dec. 1999.
[14] M. H. Nayfeh, N. Barry, J. Therrien, O. Akcakir, E. Gratton, and G.
cusing the radiation may lead to better sensitivity. Belomoin, Stimulated blue emission in reconstituted films of
Current UV technology utilizes wide-bandgap materials, as ultrasmall silicon nanoparticles, Appl. Phys. Lett., vol. 78, pp. 1131
shown in Fig. 5 [5]. These include metal zinc oxide and mag- 1133, Feb. 2001.
[15] M. H. Nayfeh, S. Rao, N. Barry, J. Therrien, G. Belomoin, A. Smith, and
nesium zinc oxide, IIIV materials as well as Schottky-type TiO S. Chaieb, Observation of laser oscillation in aggregates of ultrasmall
UV photodiodes. The distinctive aspect of IIInitrides is that their silicon nanoparticles, Appl. Phys. Lett., vol. 80, pp. 121123, Jan. 2002.
bandgap are tuneable within the ultraviolet range. Other novel [16] J. Therrien, G. Belomoin, and M. H. Nayfeh, Light-induced conduc-
tance resonance in ultrasmall Si nanoparticles, Appl. Phys. Lett., vol.
approaches include organics and phosphors. Sil-icon-based 77, pp. 16681670, Sept. 2000.
detection utilizes silicon carbide material [6] and amorphous [17] G. Belomoin, J. Therrien, A. Smith, S. Rao, R. Twesten, S. Chaieb, M.
silicon alloys [7], [8]. The many UV detectors that revolve H. Nayfeh, L. Wagner, and L. Mitas, Observation of a magic discrete
family of ultrabright Si nanoparticles, Appl. Phys. Lett., vol. 80, pp.
around nitride-based heterostructures are grown on incompatible 841843, Feb. 2002.
substances. Si nanoparticle-based devices may offer an [18] L. Mitas, J. Therrien, R. Twesten, G. Belomoin, and M. H. Nayfeh,
opportunity for integration on Si. Moreover, the measured Effect of surface reconstruction on the structural prototypes of ultra-
small ultrabright Si nanoparticles, Appl. Phys. Lett., vol. 78, pp.
responsitivity is nearly an order of magnitude improvement over 19181920, Mar. 2001.
some commercial thin film detectors, and is an improvement over [19] T. A. Burr, A. A. Seraphin, E. Werwa, and K. D. Kolenbrander,
the sensitivities of 175 and 90 mA/W recently measured in SiC Carrier transport in thin films of silicon nanoparticles, Phys. Rev. B,
vol. 56, pp. 48184824, Aug. 1997.
[6] and in amorphous Si, respectively. [20] T. Yoshida, Y. Yamada, and T. Orii, Electroluminescence of silicon
In conclusion, we demonstrated UV photodetectors based on nanocrystallites prepared by pulsed laser ablation in reduced pressure
Si nanoparticle thin film. It provides high sensitivity under low inert gas, J. Appl. Phys., vol. 83, pp. 54275432, May 1998.