Un-Cooled Microbolometers with Amorphous Germanium-Silicon (a-Ge x Si y :H) Thermo-Sensing Films

Thermo-Sensing


Introduction
Silicon integrated circuits (IC) in conjunction with the micro-machining technology for thin films, have opened new ways for the development of low cost and reliable night vision systems based on thermal detectors.Among the thermal detectors used as pixels on IR focal plane arrays, the microbolometer appears as one of them.A microbolometer is a device in which the IR transduction is performed through a change in the resistivity of its thermosensing material, due to the heating effect caused by the absorbed radiation.Among the requirements for the materials used as thermo-sensing layer in microbolometers it can be mentioned a high activation energy (E a ), high temperature coefficient of resistance (TCR), low noise, and compatibility with standard CMOS fabrication processes.A variety of materials have been used as thermo-sensing elements in microbolometers, as vanadium oxide (VO x ) (B. E. Cole, 1998Cole, , 2000)), metals (A.Tanaka, 1996), polycrystalline (S.Sedky, 1998) and amorphous semiconductors (A.J. Syllaios, 2000).
Those materials have shown good characteristics but also some disadvantages.VO x has a moderated value of TCR (0.021 K -1 ) and low resistivty, however it is not a standard material in the IC technology.Metals as titanium are compatible with the standard IC technology, have low resistivity but also have very low TCR values.Polycrystalline semiconductors have high TCR values (0.05 K -1 ) and moderated resistivity, however they are deposited a relatively high temperatures (700 -900 °C), which results in an incompatibility with a microbolometer fabrication post-process on a silicon wafer surface, containing an readout integrated circuit (ROIC).
as thermo-sensing material, since intrinsic a-Si:H has a very large activation energy (E a ) of above 1 eV, and therefore, provide a very large thermal coefficient of resistance (TCR) of 0.13 K -1 .However intrinsic a-Si:H has a very low room temperature conductivity ( RT ≤ 1x10 -9 (Ωcm) -1 ), resulting in a very high pixel resistance when is used as thermo sensing material in microbolometers (R pixel ≥ 10 9 Ω).Such high pixel resistance causes a mismatch with the input impedance of the CMOS ROIC.For commercial microbolometers, boron doping is commonly used in order to decrease the undesirable resistivity of intrinsic a-Si:H (A.J. Syllaios, 2000), to values of pixel resistance of around 30x10 6 Ω, however it also results on a reduction on the activation energy (E a ≈ 0.22 eV) and on the TCR (-0.028K -1 ), and therefore in a decrement on the pixel performance.
In this chapter we present a summary on the study of a-Ge x Si y :H and a-Ge x Si y B z :H thin films and their application as thermo-sensing element in microbolometers.We have fabricated, characterized and studied two devices configurations labeled as planar (the standard configuration used in commercial microbolometer arrays) and sandwich structures.The later shows several advantages when intrinsic materials are used as thermo-sensing element.Finally we studied the performance characteristics of the different device configurations and compared them with commercial devices and those reported on literature.

Principle of operation of un-cooled microbolometers
The operation of a microbolometer is based on the temperature rise of the thermo-sensing material by the absorption of the incident IR radiation.The change in temperature causes a change on its electrical resistance, which is measured by an external circuit.Microbolometers based on amorphous semiconductors have advantages over other types of thermal detectors, including microbolometers that use other kind of thermo-sensing materials.The advantages are mainly technological, since these microbolometers are fully compatible with silicon CMOS fabrication technology, there is no need of additional fabrication equipment in a IC production line.Are relatively of simple fabrication and can be processed at relatively low temperature by PECVD.The above make them ideal for a post-process fabrication over a CMOS read-out circuit.
Fig. 1 shows a scheme of one microbolometer (B.E. Cole, 1998); it is built on a membrane usually made of SiN x .Over the membrane is deposited the thermo-sensing material and the IR absorber material.The membrane provides thermal isolation to the thermo-sensing film.Germanium-Silicon (a-Ge x Si y :H) Thermo-Sensing Films 25 Fig. 1.Microbolometer scheme.

Thermal insulation
There are three mechanism of heat transfer that occurs in a thermal detector, they are conduction, convection and radiation.Conduction mechanisms occur when the heat flows from the thermo-sensing area along the supporting legs to the substrate.Conduction is critical when the pixels are very close, since the heat can flow from one pixel to a neighbor pixel.Convection occurs when the heat flows in the presence of a surrounding atmosphere, this mechanism is not very important if the detector is encapsulated in a vacuum package.Radiation mechanism is presented by the fact that the detector radiates to its surroundings and the surroundings radiate to it.
When the microbolometers are encapsulated in an evacuated package, with an IR transmitting window, convection and radiation mechanism are minimized.Thus the main loss of heat mechanism is conduction from the thermo-sensing material to the substrate through the supporting structure.
The supporting structure is a very important part of thermal detectors, it provides three functions, mechanical support, electrical conducting path and thermal conducting path.In order to avoid heat losses in microbolometers, it is necessary to improve the thermal insulation.In microbolometers there are two main thermal insulation configurations: singlelevel and two-level configurations.
Single level configuration consist in deposit a membrane over the silicon (Si) substrate and after that, open a hole in the Si substrate, employing bulk micromachining techniques.Bulk micromachining consumes area, since the Si substrate is etched with a side wall angel of 54.9 degrees.The electronic circuit (which forms part of the read out circuit) is fabricated next to the pixel, consuming area also.That result in a 20% fill factor.
The two-level configuration allows the fabrication of the electronics circuit in the substrate and after that, the fabrication of the microbolometer in a low temperature post process over the electronics, by using surface micromachining techniques.With this configuration is saved substrate area, achieving a fill factor of above 70%.
In order to fabricate thermal sensors in a post process, over a wafer surface, containing an IC circuit; it is necessary to use low temperatures during the fabrication process.By employing Plasma Enhanced Chemical Vapor deposition (PECVD) it is possible to deposit thin films at relatively low temperatures (150 -350 o C).

Infrared absorber films
An absorber element is a very important part in un-cooled IR microbolometers, its role is based in the absorption of IR radiation and the transfer of heat to the thermo-sensing material.The main requirements of absorbing materials for un-cooled microbolometers are: A high absorbance coefficient in the range = 8 -12 m, simple fabrication and compatibility with the silicon CMOS technology.
The IR absorption can be improved employing a resonant micro-cavity, where the thermosensing film is separated from the substrate by a gap equivalent to one quarter of the wavelength at which it will be operating.A mirror (Al or Ti) is deposited over the substrate surface, under the thermo-sensing material.In this configuration the radiation that was not absorbed by the thermo-sensing film will resound inside the cavity and will be re-absorbed by the thermo-sensing element.
Terrestrial objects have temperatures around of 300K, with IR emission centered in 10 m.Thus un-cooled microbolometers employed for detection of objects at room temperature, should have a gap from the substrate of 2.5 m, for the fabrication of the resonant microcavity.
Several materials have been employed as absorbing films in microbolometers, which are deposited over the thermo-sensing film.Among the most employed are some metals, as black gold film (M.Hirota, 1998), which has a very high absorption coefficient of IR radiation (more than 90 %), however it is not a standard material in CMOS technology.SiN x films are employed commonly as absorber films in microbolometers (A.Schaufelbühl, 2001, S. Sedky, 1998), since its absorption coefficient can be tuned by the deposition parameters and it is a standard material in CMOS technology.

Thermo-sensing films
The thermo-sensing material is perhaps the most important element in a microbolometer.The increment in temperature in the sensing material causes a change in some temperaturedependent parameter.In the case of a microbolometer that parameter is the resistance.
The thermo-sensing material should have a large temperature coefficient of resistance, TCR ( (T)), which is defined by Eq. 1, where E a is the activation energy, K is the Boltzman constant and T is temperature.
A large TCR means that a small change in temperature in the sensing material will result in a large change in resistance.Eq. 1 shows that the TCR and E a are directly related, thus a high E a in the material is needed.
For un-cooled microbolometers vanadium oxide, VO x , was the first thermo-sensing element employed (B.E. Cole, 1998), since it has a moderated TCR, (T) ≈ 0.021 K -1 , however it is not a standard material in silicon CMOS technology.Some metals have been employed also, Hydrogenated amorphous silicon (a-Si:H) prepared by PECVD is very attractive to be used as thermo-sensing film in microbolometers, for room temperature operation (A.J. Syllaios, 2000).It is compatible with the IC technology, has a high activation energy, E a ≈ 0.8 -1 eV and high value of TCR, (T) ≈ 0.1 -0.13 K -1 , however it also has a very high undesirable resistivity, which often cause a mismatch with the input impedance of the read-out circuits.
In order to reduce the a-Si:H high resistance, boron doping has been employed.The B doped a-Si:H films have a significant reduction in its resistivity, however a reduction in E a and TCR is obtained also, Ea ≈ 0.22 eV and TCR ≈ 0.028 K -1 (A.J. Syllaios, 2000).In our work (R. Ambrosio, 2004;A. Kosarev, 2006;A. Torres, 2008;M. Moreno, 2007M. Moreno, , 2008M. Moreno, , 2010)), amorphous germanium-silicon, a-Ge x Si y :H, deposited by PECVD has been studied as thermo-sensing films in un-cooled microbolometer, obtaining high activation energy, E a = 0.34 eV, consequently a high value of TCR = 0.043 K -1 and improved but still high resistivity.

Material
Table 1 shows the most common materials employed as thermo-sensing films in microbolometers.As can be seen in the table, there are available several materials which can be used as thermo-sensing films.Intrinsic amorphous silicon, a-Si:H and a-Ge x Si y :H, show the largest TCR values and are fully compatible with the silicon CMOS technology, however they have also the smallest values of room temperature conductivity, RT .

Main figures of merit of un-cooled microbolometers
In this section the different figures of merit of a microbolometer, as thermal characteristics, responsivity and detectivity are presented.The different types of noise in microbolometer are described also.

Thermal capacitance, C th , thermal conductance, G th and thermal response time, τ th
A simple representation of a microbolometer is shown in Fig. 2, the detector has a thermal capacitance, C th , and it is coupled to the substrate which is a heat sink, by a thermal conductance, G th .
When the detector receives modulated IR radiation, the rise in temperature is found by solving the balance equation, Eq. 2; where C th (expressed in JK -1 ) is the thermal capacitance of the supporting membrane containing the thermo sensing film, while G th (expressed in WK -1 ) is the thermal conductance of the legs, which is considered the main heat loss mechanism.ΔT is the temperature difference of the hot and reference junctions.A cell is the detector area, β is the fill factor, which is the radio of the thermo-sensing film area to the total cell area, η is the optical absorption coefficient, defined as the fraction of the radiant power falling on the thermo-sensing area, which is absorbed by that area.P o is the intensity of the IR modulated radiation, ω is the angular modulation frequency and The solution of the balance equation is shown in Eq. 3: Where, th (expressed in seconds) is the thermal response time of the microbolometer, it is defined by Eq. 4, which establishes a relation between th , C th and G th .Typical values of thermal time constant are in the range of milliseconds, which are much longer than the typical time of photon detectors.
For unmodulated radiation Eq. 3 can be reduced to: Eq. 5 shows that the increment of temperature, ΔT, in the detector is inversely proportional to the thermal conductance G th of its legs.In order to achieve a high performance microbolometers ΔT should be as high as possible and therefore G th as small as possible, which can be done by making very thin the detector legs.

Responsivity
Responsivity, R, is defined as the ratio of the pixel output signal to the incident radiant power (in Watts) falling on the pixel (P.W. Kruse, 2001).The output signal is an electrical signal that can be voltage or current, thus R can be expressed in Volts/Watts (voltage responsivity, R u ) or Amps/Watts (current responsivity, R I ).In order to obtain R, we can use the simplest model, where it is assumed that there is no heating due the electrical bias in the detector (Joulean heating), and also it is assumed a constant electrical bias to the detector.
When the microbolometer is current biased, the output signal is voltage, V s , given by Eq. 6, where I b is the bias current, R cell is the electrical resistance of the microbolometer, is the TCR, described by Eq. 1 and ΔT is the increment of temperature in the detector, obtained in Eq. 5. T Voltage responsivity, R v , is obtained by combining equations 3 and 6, and dividing by P o A cell , which is the incident radiant power, the result is shown in Eq. 7.
When the microbolometer is voltage biased equations 7 and 8 are transformed to Eq. 9 and Eq. 10 respectively, where R I is current responsivity.
 

Noise in microbolometers
There are four main sources of noise in microbolometers (P.W. Kruse, 2001), which are Johnson noise, 1/f noise, temperature fluctuation noise and background fluctuation noise, these noise types are uncorrelated and are described in the following subsections.

Johnson noise
The Johnson noise component, V j , is described by Eq. 11, where k is the Boltzmann constant, T cell is the bolometer temperature, R cell is the bolometer resistance and Δf is the bandwidth of the integration time.

1/f noise
The 1/f noise is characterized by a spectrum that depends inversely on frequency and is described by Eq. 12, where V is the product of the bias current -I b and the electrical resistance of the microbolometer -R cell , f is the frequency at which the noise is measured and n is the 1/f noise parameter, which depend on the material detector.
  1/f noise is the dominant noise at low frequencies and falls below the Johnson noise at higher frequencies, the transition is commonly called the "knee".

Temperature fluctuation noise
A thermal detector which is in contact with its environment (by conduction and radiation), exhibits random fluctuations in temperature, since the interchange of heat with its surrounding has a statistical nature; this is known as temperature fluctuation noise.The mean square temperature fluctuation noise voltage is given by Eq. 13 (P.W. Kruse, 2001). (13)

Background fluctuation noise
When the heat exchange by conduction between the detector and its surroundings is negligible, in comparison with the radiation exchange, the temperature fluctuation noise will be identified as background fluctuation noise.
The mean square background fluctuation noise is given by Eq. 14, where T cell is the detector temperature and T B is the background temperature.
The total noise voltage is obtained by adding the 4 noise contributions as is shown in Eq. 15.

Detectivity
Detectivity, D * (expressed in cmHz 1/2 Watt -1 ), is a figure of merit for all types of detectors, it is defined as the pixel output signal to noise ratio per unit of incident radiant power falling on the detector, measured in a 1 Hz bandwidth.In other words, D * is the normalized signal to noise ratio in the detector and is shown in Eq. 16.
In Eq. 16 R v is the voltage responsivity, A cell is the detector area, Δf is the frequency bandwidth and V N is the contribution of the four noises.It is clear that in order to achieve a high D * the responsivity should be as high as possible and the noise as small as possible.
The fundamental limit to sensitivity of any thermal detector is set by random fluctuations in the temperature of the detector due to fluctuations in the radiant power exchange between the detector and its surroundings.The highest possible value of D * of a thermal detector operated at room temperature is D* = 1.98 x10 10 cmHz 1/2 W -1 (A.Rogalski, 2003).
Intrinsic amorphous silicon (a-Si:H) prepared by PECVD is a very attractive material to be used in microbolometers as thermo-sensing film.It has a high activation energy, E a ≈ 0.8 -1 eV and high value of temperature coefficient of resistance, TCR, (T) ≈0.1 -0.13 K -1 , however it also has a high undesirable resistivity.
Amorphous germanium-silicon (a-Ge x Si y :H) films deposited by PECVD have been studied as thermo-sensing film in microbolometers (R. Ambrosio, 2004; A. Kosarev, 2006;A. Torres, 2008;M. Moreno, 2007M. Moreno, , 2008M. Moreno, , 2010)), due its high activation energy and consequently high TCR, and its relatively high room temperature conductivity, RT, in comparison with a-Si:H films.In this section is presented a description of the deposition by PECVD of intrinsic amorphous germanium-silicon (a-Ge x Si y :H) and amorphous germanium-silicon-boron (a-Ge x Si y B z :H) thin films, and its electrical and compositional characterization.

Films preparation for characterization
An intrinsic film (a-Ge x Si y :H) was deposited in a capacitive discharge low frequency (LF) PECVD reactor at frequency f = 110 KHz, substrate temperature T s = 300 o C, pressure P = 0.6 Torr and RF power W = 350 W, with a gas mixture of SiH 4 , GeH 4 and H 2 and gas flow rates of Q SiH4 =25 sccm, Q GeH4 =25 sccm and Q H2 =1000 sccm respectively.This result in a Ge gas content X g = 0.5.The film was labeled as process A.
The a-Ge x Si y B z :H films were also deposited in a capacitive discharge low frequency (LF) PECVD reactor at frequency f = 110 KHz, substrate temperature T s = 300 o C, pressure P = 0.6 Torr and RF power W = 350 W. Three sets of films were deposited from SiH 4 (100%), GeH 4 (100%) and B 2 H 6 (1% on H 2 ) gas mixture, with a fixed SiH 4 and B 2 H 6 gas flow rates: Q SiH4 =50sccm and Q B2H6 =500 sccm, respectively, while the GeH 4 gas flow was set at the following values: Q GeH4 =25, 50 and 75 sccm.
The late resulted in a Ge gas content X g = 0.3, 0.45, 0.55 and a B gas content Z g = 0.11, 0.09, 0.07 in the samples labeled as process number B, C and D, respectively.Since those films are studied for applications as thermo-sensing films for microbolometers, we measured the film electrical properties after patterning them with photolithography in one cell of dimensions 70 x 66 m 2 .
Assuming that stress arisen in the film deposited over a SiN x micro-bridge could have an effect on the film conductivity, we also studied the films deposited on a micro-bridge.For that purpose, we prepared three different kinds of samples for each type of the four thermosensing films (three boron alloys with different Ge content and the intrinsic film).The films were prepared as is shown in Fig. 3.We performed measurements of temperature dependence of conductivity (T) in the a-Ge x Si y :H and a-Ge x Si y B z :H thermo-sensing films in the range of T= 300-400 K.The measurements were performed in a vacuum chamber at a pressure P≈20 mTorrs.A temperature controller (model K-20, MMR Inst.) for the temperature measurement control and an electrometer (model 6517-A, Keithley Inst.) for the current measurements were employed.These measurements allowed us to obtain the (T) temperature dependence and then to determine the E a , the TCR and the room temperature conductivity, RT .
The conductivity temperature dependence can be well described by (T)= 0 exp(-E a /kT), where 0 is the prefactor, E a is the activation energy, k is the Boltzmann constant and T is the temperature.Fig. 4 shows (T) curves for four different thermo-sensing films (three boron alloys with different Ge gas content, Ge x = 0.3, 0.45, 0.55 and the intrinsic film with Ge x = 0.5), fabricated in three different sample configurations (stripes, patterns and micro-bridges).
From (T) measurements with temperature in the thermo-sensing films, we found that the boron alloys (a-Ge x Si y B z :H) have a significantly larger conductivity (by about 2-3 orders of magnitude) in comparison with that of the intrinsic film (a-Ge x Si y :H).We observed that an increment in the Ge content in gas phase in the boron alloys results in an increase of the room temperature conductivity, from RT = 2.8 x10 -3 (Ωcm) -1 (for Ge x = 0.3) to RT = 1 x10 -2 (Ωcm) -1 (for Ge x = 0.45) and RT = 2.5 x10 -2 (Ωcm) -1 (for Ge x = 0.55), while for the intrinsic film the room temperature conductivity is RT = 6 x10 -5 (Ωcm) -1 (for Ge x = 0.5).The increment in the is accompanied with a reduction in the E a .We obtained an E a = 0.22 eV (for Ge x = 0.3), E a = 0.21 eV (for Ge x = 0.45) and E a = 0.18 eV (for Ge x = 0.55), while in the intrinsic film is E a = 0.345 eV (for Ge x = 0.5).E a as a function of Ge x is shown in Fig. 5 A).
The reduction in the thermo-sensing films dimensions, from the stripes samples (10x1.5 mm 2 ) to the patterned samples (70 x 66 m 2 ), has no significant effect on E a , however it has on the RT .We observed a reduction of above 50-80 % of the RT value in the patterned samples in comparison with that of the stripes samples.
Practically no change in E a of the thermo-sensing films deposited over a SiN x micro-bridge was observed, in comparison with that of the stripes and patterned samples; however the micro-bridge samples showed a larger reduction in the RT values, of 60-90 %.The dependence of RT with the Ge x content and the sample structure are shown in Fig. 5 B), while the deposition rate dependence of Ge x content in the thermo-sensing films is shown in Fig. 5 C).Table 3 show a comparison of E a , TCR, RT and 0 in stripes, patterned and microbridges samples for the different thermo-sensing films.
The micro-bridges samples have the largest reduction of conductivity, and it could be explained by the stress arisen in the SiN x micro-bridge, affecting the thermo-sensing film electrical conductivity.The deposition rate in the boron alloys is around 2 -3 times larger than that of the intrinsic film.Boron incorporation during the thermo-sensing deposition, enhance the deposition rate as is shown in Fig. 5 C).It is important to point out that, doping on amorphous semiconductors reduce E a and increases the films conductivity ( RT ).In Fig 5A) for reference, is shown an intrinsic a-Ge x Si y :H film produced with a gas content of Ge x =50% and Si y =50%, which has a E a of 0.34 eV.This is the largest value for a-Ge x Si y :H films (doped or un-doped, using a gas content of Ge x =50% and Si y =50%).
When boron is introduced in the film deposition, E a is reduced and the conductivity is increased in the films.In fig. 5 A) is shown that E a is reduced to values in the range of 0.18 -0.22 eV, while the conductivity is increased in more than one order of magnitude.Also it is important to notice that a-Ge x Si y :H films have an intermediate E a value, between a-Si:H and a-Ge:H.Intrinsic a-Si:H has E a values close to 1 eV, while a-Ge:H have Ea values of above 0.3 eV.Thus, varying the Ge (and Si) gas contents in the a-Ge x Si y :H films, it is possible to modify E a (an also the conductivity) on intrinsic films.
Larger Ge x content in the films will reduce the value of E a .In fig 5 A), we observe a decrement on E a of the a-Ge x Si y B z :H films, not just because the B z gas content (which in fact decreases), but because the Ge content (which increases).In fig 5 A) is shown that for a-Ge x Si y B z :H films, the Ge x gas content vary between 0.3 and 0.55, while the B z gas content vary just between 0.07 and 0.11.Thus the effect of the variation of the Ge gas content on E a is dominant in the a-Ge x Si y B z :H films .

Composition of the a-Ge x Si y :H and a-Ge x Si y B z :H films
In Figure 5 the showed results are related to the Ge x , Si y , and B z gas contents, not to solid contents.There exists a significant difference between the gas content used for the films deposition, and the solid content in the films produced.The composition in solid phase of the different films (three boron alloys with different Ge gas content and the intrinsic film) was characterized by secondary ion mass spectroscopy (SIMS).The samples used for SIMS characterization were the stripes samples described in section 4.1.Fig. 6 shows the SIMS profiles obtained.
From SIMS profiles we calculated the solid composition in the thermo-sensing films.For the film with gas content: Gex=0.3 and B z =0.11 (process B), we observed an increase in the solid content: Ge x =0.59 and B z =0.32 respectively.For the film with Ge x =0.45 and B z =0.09 (process C), we observed Ge x =0.67 and B z =0.26, respectively.For the film with Ge x =0.55 and Bz=0.07 (process D), we observed Ge x =0.71 and B z =0.23, respectively.These results suggested a strong preferential B and Ge incorporation from gas phase during the film deposition process.The B z solid content demonstrated values about 3 times larger than the content in gas phase B z , while the Ge x solid content increased by a factor of 1.3-2 from the Ge x gas content.Those results are shown in Table 4.  TCR (K -1 ) -0.047 -0.028 -0.025 σRT (Ωcm) -1 2.2x10 -5 1.2x10 -3 7x10 -3 σ0 (Ωcm) -1 32.8 5.94 15.58

Thermo-sensing films
Table 3.Comparison of E a , TCR, RT and 0 in stripes, patterned and micro-bridges samples for the different thermo-sensing films.
In Fig. 6 the intrinsic film has a Boron content of 10 18 cm -3 (which represents a B solid content of 2x10 -3 %) as is shown in table 4. The reason of the above is the fact that all the films were deposited on the same chamber.Even though, the chamber was extensively cleaned and coated with a SiN x film before the intrinsic film deposition, Boron impurities remained in the chamber walls, which were re-deposited in the intrinsic films.

Microbolometer configurations and fabrication process flow
In this section we show a comparative study of the performance characteristics of three configurations of un-cooled microbolometers based on amorphous germanium thin films: a) Planar structure with intrinsic amorphous germanium-silicon a-Ge x Si y :H thermo-sensing film.In this configuration the metal electrodes are placed under the thermo-sensing film (Fig. 7 A); b) Planar structure with amorphous germanium-boron-silicon alloy a-Ge x B y Si z :H thermo-sensing film (Fig. 7 B) and c) Sandwich structure with intrinsic a-Ge x Si y :H thermosensing film, this configuration consists of metal electrodes which sandwich the thermosensing film (Fig. 7 C).The planar structure microbolometer with the boron alloy (a-Ge x B y Si y :H) thermo-sensing film is fabricated as the previous one, with difference in the thermo-sensing film deposition parameters.The boron alloy film is deposited from a SiH 4 + GeH 4 + B 2 H 6 + H 2 mixture with the following gas flows: Q SiH4 =50sccm, Q GeH4 =50 sccm, Q B2H6 =5 sccm and Q H2 =500 sccm.This results in a Ge content in solid phase Ge x =0.67 and B content in solid phase B y =0.26.Those values were obtained by SIMS measurements.The sandwich structure microbolometer with the a-Ge x Si y :H film is fabricated in the same way as the planar microbolometer with some differences, due to the placing of metals as bottom and top electrodes.In this structure the electrodes sandwich the thermo-sensing film.The bottom Ti electrode is 0.2 m-thick and is deposited before the thermo-sensing film.Then the a-Ge x Si y :H film is deposited and it is covered with a top thin electrode (10 nm) forming a sandwich structure.The active area of the thermo-sensing layer in the three configurations studied is A b =70x66m 2 .Fig. 8 shows the fabrication process of the microbolometer structures.www.intechopen.com Un-Cooled Microbolometers with Amorphous Germanium-Silicon (a-Ge x Si y :H) Thermo-Sensing Films 39

Microbolometers electrical characterization
A microbolometer is a resistor sensitive to temperature change, its operation is based on the temperature increase of the thermo-sensing film by the absorption of the incident IR radiation.The change in temperature causes a change on its electrical resistance, which is measured by an external circuit.
In this section we present a comparative study of 3 configurations of un-cooled microbolometers based on amorphous silicon-germanium thin films deposited by plasma.

I(U) measurements in dark and under Infrared Radiation (IR)
In this section is described the procedure performed in order to obtain the current voltage I(U) characteristics of the microbolometers, from this measurement it is possible to determine the microbolometer electrical resistance and responsivity.
The current-voltage characteristics I(U) and current noise spectral density (NSD) have been measured in the devices in order to compare the performance characteristics, such as responsivity and detectivity in the 3 configurations of microbolometers: a. Planar structure with an intrinsic germanium-silicon (a-Ge x Si y :H, Ge x =0.5) thermosensing film (process A of section 4).b.Planar structure with a germanium-silicon-boron alloy (a-Ge x Si y B z :H, Ge x =0.45, B z =0.09) thermo-sensing film (process C of section 4).c.Sandwich structure with an intrinsic (a-Ge x Si y :H, Ge x =0.5) thermo-sensing film (process A of section 4).
The samples were placed in a vacuum chamber at pressure P20 mTorr, at room temperature and illuminated through a zinc selenide window (ZnSe).The window has a 70% transmission in the range of =0.6 -20 µm.The source of IR light is a SiC globar source, which provides intensity I 0 =5.3x10 -2 W/cm 2 in the range of =1 -20 µm.The current was measured with an electrometer ("Keithley"-6517-A) controlled by a PC in dark and under IR illumination.
Fig. 9 A) shows the current-voltage I(U) characteristics in dark and under IR illumination for the planar configuration with a-Ge x Si y :H thermo-sensing film (process A, section 4); Fig. 9 B) shows these characteristics for the planar configuration with a-Ge x Si y B z :H thermo-sensing film (process C, section 4); and Fig. 9 C) shows the same characteristics for the sandwich configuration with a-Ge x Si y :H thermo-sensing film (process A, section 4).
In those figures we can see the increment in current due to IR illumination, ∆I=I IR -I Dark , where I IR is the current under IR radiation and I Dark is the current in dark.The planar configuration with the a-Ge x Si y :H (Ge x =0.5) film has a ∆I = 5.4 nA (at bias voltage U=7 V); the planar configuration with the a-Ge x Si y B z :H (Ge x =0.45, B z =0.09) film has a ∆I = 65 nA (at bias voltage U=7 V); and the sandwich configuration with the a-Ge x Si y :H (Ge x =0.5) film has a ∆I = 35 A (at bias voltage U=4 V).The inset in those figures show the Log I(Log U) characteristics, where we can see their linear behavior.The gamma ( ) constant indicates the slope of the curves.

Current and voltage responsivity
The current responsivity, R I , is described by Eq. 17, where ∆I is the increment in current (∆I=I IR -I Dark ) and P incident is the IR incident power in the device surface.P incident is described by Eq. 18 and is the product of the cell area, A cell and the IR source intensity, I 0 .
Ii n c i d e n t R I / P  (17) The intensity of the IR source is I 0 = 0.053 Wcm -2 , while the cell area is A cell = (70x10 -4 )(66x10 - 4 ) cm 2 = 4.6x10 -5 cm 2 .Therefore the IR incident power in the device surface is P incident = 2.475x10 -6 W. The planar microbolometer with a-Ge x Si y :H (Ge x =0.5) film has a R I =2x10 -3 A/W (at U=7 V); the planar microbolometer with a-Ge x Si y B z :H (Ge x =0.45, B z =0.09) film has a R I = 3x10 -2 A/W (at U=7 V); and the sandwich microbolometer with a-Ge x Si y :H (Ge x =0.5) film has a R I = 14 A/W (at U=4 V).
Fig. 11.Extraction of ∆U from I(U) characteristics: A) planar with a-Ge x Si y :H (Ge x =0.5).B) planar with a-Ge x Si y B Z :H (Ge x =0.45, B z =0.09).C) sandwich with a-Ge x Si y :H (Ge x =0.5).

Noise spectral density measurements and detectivity calculations
Noise measurements in the microbolometers were performed with a lock-in amplifier ("Stanford Research Systems" -SR530).The noise of the system and the total noise (system + cell noise) were measured separately, and a subtraction of the system noise allowed us to obtain the noise of the device.The detectivity was calculated from the responsivity values and noise measurements.The current noise spectral density (NSD), I cell noise (f), of the fabricated devices with the different thermo-sensing films are shown in Fig. 12.
The NSD in the cell is obtained as (I cell noise (f)) 2 = (I system + cell noise (f)) 2 -(I system noise (f)) 2 , where I cell + system noise (f) is the NSD measured at the microbolometer with the measuring system and the I system noise (f) is the NSD measured in the system without the microbolometer.In noise curves we observed different slopes at different frequencies and different cone frequencies.That data are shown in Table 6, where fc1 is the cone frequency 1, fc2 is the cone frequency 2, is the slope of the curve in region 1 and is the slope of the curve in region 2.
The procedure for the detectivity calculation is shown in Eq. 19, where R I is the current responsivity, A cell is the detector area, I noise is the cell NSD and ∆f = 1 is the bandwidth of the measurement system.We calculated the detectivity values D * in the 3 structures.For the planar structure with the a-Ge x Si y :H (Ge x =0.5) film we obtained D * = 7x10 9 cmHz 1/2 W -1 ; for the planar structure with the a-Ge x Si y B z :H (Ge x =0.45, B z =0.09) film it is D * = 5.9x10 9 cmHz 1/2 W -1 ; and for the sandwich structure microbolometer with the a-Ge x Si y :H (Ge x =0.5) film it is D * = 4x10 9 cmHz 1/2 W -1 .

Microbolometers thermal characterization and calibration curve
In order to estimate the temperature dependence of the thermal resistance of the microbolometers, I(U) measurements were performed in the range from 260 K to 360 K, as is shown in Fig. 13 A), where the bias is plotted as a function of current.
Fig. 13.A) U(I) curves of a planar structure microbolometer with a-Ge x Si y :H (Ge x =0.5).B) Calibration curve of a planar structure microbolometer with a-Ge x Si y :H (Ge x =0.5).
The slope of the linear part of each curve showed in Fig. 13 A) corresponds to the electrical resistance of the microbolometer for each temperature value (in the range of 260 K -360 K).
Once that is obtained the value of the electrical resistance for each value of temperature, it is possible to graph the electrical resistance of the microbolometer as a function of the temperature, also called the calibration curve, as is shown in Fig. 13 B).
The calibration curve is very important because from this curve is possible to calculate an increment in temperature in the microbolometer by measuring a change in its resistance.
The voltage-current curves in Fig. 13 A) have different resistance values as the current increases.The value of the resistance is affected by the temperature.Thus the resistance is calculated for each point of the Voltage-Current curves (obtained at different temperatures, 260, 270, etc.).If the resistance values obtained are compared with the calibration curve, it is possible to extract the increment of temperature (ΔT) for each point.

a-Ge x Si y :H and a-Ge x Si y B z :H microbolometers compared with literature
The results obtained from the study of fabrication and characterization of different microbolometer structures, containing intrinsic a-Ge x Si y :H films and boron alloys a-Ge x Si y B z :H, are discussed in the present section and compared with data reported in literature.
The sandwich structure microbolometer with the intrinsic a-Ge x Si y :H film, presents the shortest cell resistance of the devices reported in literature, R cell ≈ 1 x10 5 Ω, which is 3 orders of magnitude less than that of the planar devices with the same intrinsic film; one order of magnitude shorter than that of the boron alloy devices; 2 orders shorter than that of the a-Si:H,B devices; and near to 1 order of magnitude shorter than that of the a-Ge x Si 1-x O y microbolometers.The TCR in sandwich structures is very high, around =0.043 K -1 , the current responsivity is in the range of R I = (0.3 -14) AW -1 , which is around 2 -3 orders of magnitude larger than that of the boron alloys (a-Ge x Si y B z :H) planar structure microbolometers and around 3 -4 orders of magnitude larger than the intrinsic a-Ge x Si y :H film planar structure devices.However the sandwich structure presents a larger current NSD, I noise ≈ 10 -11 AHz -1/2 , which results in a detectivity D * = 4 x10 9 cmHz 1/2 W -1 .

Conclusion
Uncooled microbolometers are reaching performance levels which previously only were possible with cooled infrared photon detectors.For uncooled infrared bolometer arrays based on amorphous silicon films the efforts have been conducted to increase the number of pixels included in the arrays, rather than improve the performance characteristics of the microbolometers.Plasma deposited amorphous germanium-silicon (a-Ge x Si y :H) and amorphous germanium-silicon-boron (a-Ge x Si y B z :H) used as thermo-sensing films provided a high TCR and, as a consequence, a high responsivity and high detectivity with a improved conductivity.Thus a-Ge x Si y :H and a-Ge x Si y B z :H are very promising materials for its integration on IR detector arrays, and its circuitry in the same chip, avoiding the problems of matching with the input impedance of the electronic circuits.Moreover the manufacture of those devices is aligned with standard CMOS and MEMS foundry processes.

Fig. 4 .
Fig. 4. Conductivity dependence with temperature for the different thermo-sensing films (process: A, B, C and D).

Fig. 5 .
Fig. 5. Characterization of a-Ge x Si y :H and a-Ge x Si y B z :H films.A) E a as function of Ge gas content (Ge x ).B) Conductivity as a function of Ge gas content.C) Deposition rate as a function of Ge gas content.

Fig. 6 .
Fig. 6.SIMS profiles of a-Ge x Si y :H and a-Ge x Si y B z :H thermo-sensing films.
Fig. 7 D)  shows a picture of one device fabricated.The fabrication process of the planar structure microbolometer with the a-Ge x Si y :H thermosensing film is as follows.A 0.2 m-thick SiO 2 layer is deposited by CVD on a c-Si wafer and a 2.5 m-thick sacrificial aluminum layer is deposited by e-beam evaporation and patterned.A 0.8 m-thick SiN x film is then deposited at low temperature (350 o C) by low frequency PECVD over the aluminum sacrificial film.The SiN x film is patterned by reactive ion etching (RIE) in order to form a SiN x bridge.A 0.2 m-thick titanium contacts are deposited by ebeam evaporation over the SiN x bridge and a 0.5 m-thick thermo-sensing a-Ge x Si y :H film is deposited over the Ti contacts by low frequency LF PECVD technique at a rf frequency f=110 kHz, temperature T=300 o C, power W=350 W and pressure P=0.6 Torr.The a-Ge x Si y :H film is deposited from a SiH 4 + GeH 4 + H 2 mixture with gas flows: Q SiH4 =25sccm, Q GeH4 =25 sccm, Q H2 =1000 sccm.This results in a Ge content in solid phase Y=0.88 and a Si content in solid phase Y=0.11.The thermo-sensing film is covered with a 0.2 m-thick absorbing SiN x film deposited by PECVD and finally the aluminum sacrificial layer is removed with wet etching.

Fig. 7 .
Fig. 7. Microbolometers: A) Planar wit intrinsic film a-Ge x Si y :H, B) Planar with boron doped film a-Ge x B y Si z :H, C) Sandwich with intrinsic film a-Ge x Si y :H, D) A device fabricated.
Fig. 9. I(U) characteristics of 3 microbolometers: A) planar with a-Ge x Si y :H (Ge x =0.5).B) planar with a-Ge x Si y B Z :H (Ge x =0.45, B z =0.09).C) sandwich with a-Ge x Si y :H (Ge x =0.5).The inset in those figures show the Log I(Log U) characteristics, where we can see their linear behavior.The gamma ( ) constant indicates the slope of the curves.

Fig. 11 B
) shows a ∆U= 0.3 V extracted from a fixed current I = 1.35x10 -6 A, in the planar structure microbolometer with the a-Ge x Si y B Z :H (Ge x =0.45, B z =0.09) film.Fig.11 C) shows a ∆U= 0.54 V extracted from a fixed current I = 1.16x10 -4 A, in the sandwich structure microbolometer with the a-Ge x Si y :H (Ge x =0.5-I dark(A)

Fig. 14 A
Fig.14 A) shows the increment of temperature (ΔT) as a function of the power (P=U*I) applied to the microbolometer, for each temperature.

Fig. 15 .
Fig. 15.Microbolometer with a-Ge x Si y B z :H (Ge x =0.45, B z =0.09).A) Calibration curve.B) Thermal resistance.The thermal resistance of the microbolometer, R th , is then obtained as the slope of the increment of temperature in the microbolometer (ΔT) as a function of the power applied to it, for each temperature value.Fig.14 B) shows the temperature dependence of the thermal resistance (R th ).Fig.15 A) shows the calibration curve and Fig.15 B) shows the temperature dependence of the thermal resistance (R th ) of the planar structure microbolometer with a-Ge x Si y B z :H (Ge x =0.45, B z =0.09).

Table 1 .
Common materials employed as thermo-sensing films in microbolometers.

Table 2 .
Table 2 shows the deposition parameters for the 4 thermo-sensing films.Deposition parameters of a-Ge x Si y :H and a-Ge x Si y B z :H films.www.intechopen.com

Table 4 .
Gas content and solid content obtained by SIMS for the thermo-sensing films.

Table 5 .
Current and voltage responsivity values for 3 microbolometer configurations.

Table 6 .
NSD at different frequency regions in the different microbolometers structures.