Litton C.W., . Reynolds D.C., Collins T.C. Zinc Oxide Materials for Electronic and Optoelectronic Device Applications

Подождите немного. Документ загружается.

CVT growth of ZnO is illustrated in Figure 7.1. In this conﬁguration, a quartz ampoule is

used to hold a polycrystalline charge of ZnO at one end with a seed crystal at the other end. By

placing the ampoule in a temperature gradient such that the polycrystalline charge is at a

higher temperature than the seed, the charge will be vaporized and transported to the cooler

end where it recrystallizes on the seed crystal. To form a single crystal, the seed would ideally

be of the same material as the growing crystal. However, in some cases, single crystal seeds

composed of other materials have been used.

[6]

“Self-seeding” (no seed is used; the ampoule

is usually conﬁgured to localize the initial growth to a few small single crystals) is also

possible but usually requires a tapered ampoule such that the ﬁrst deposited material will form

at the tip in such a small size that it will naturally form a very small single crystal. The rest of

the deposit then grows epitaxially on this seed and, as it gets thicker, expands in diameter as

the ampoule tapers out. This can require an extended growth time for the growing crystal to

expand to a usable diameter. A more economical method has been developed by ZN

Technology in which a full diameter seed is used with a straight wall ampoule. This method

has been shown to produce high quality, large area, single crystals of CdS, CdTe,

[7]

ZnTe,

ZnSe,

[8,9]

and many ternaries. These crystals are typically twin free without low angle grain

boundaries and have been grown at 2- and 3-in. diameters by the PVT method. The same

methodology was adapted for use in the CVTof ZnO and 2-in. diameter single crystals have

been grown. The addition of a full diameter seed to the PVT method results in the process

being named the Seeded Physical Vapor Transport (SPVT) method and for CVT, the Seeded

Chemical Vapor Transport (SCVT) method.

7.2 Transport Theory and Comparison with Growth Data

While PVT of ZnO is possible, the sublimation vapor pressures of zinc and oxygen over

ZnO are very low at practical growth temperatures: 3 10

6

atm at 1000



C and

1 10

4

atm at 1200



C for the Zn partial pressure.

[10]

Typical Zn vapor partial pressures

in the SCVT system are calculated

[10]

to be 1.5 10

2

atm at 1000



C and 1.9 10

2

atm at 1200



C. Thus the use of the chemical transporting agent should increase the

(g)

↔ Zn

(v)

ZnO(s) + H

→

Zn (v)

→

O (v)

(g)

←

Single

Crystal

Poly

Charge

Figure 7.1 Schematic illustration of the chemical vapor transport of ZnO in an enclosed

ampoule

172 Vapor Transport Growth of ZnO Substrates and Homoepitaxy of ZnO Device Layers

transport and growth rates by a factor of approximately 10

at 1000



C and 10

at 1200



Most vapor growth is done in quartz ampoules and the typical upper limit for use of quartz

is in the 1200–1300



C range. Experimental measurements by ZN Technology of the

transport rate of ZnO using PVT have typically resulted in rates lower than 1 10

6

mol

2

1

which is too low to be practical for crystal growth. In these experiments, a

buffer gas of helium at a pressure of 370 Torr was used with a temperature differential

between the charge and the seed of 50



C. Even this low rate may have been

enhanced by residua l moisture or other impurities in the growth charge and system that

may have acted as transporting agents and increased the transport rate beyond that of

sublimation only.

The SCVT method provides the potential for much higher transport rates of ZnO at

signiﬁcantly lower temperatures. In this method, the transporting agent reacts with the

source material causing a decrease in the partial pressure of the transporting agent and

production of high vapor pressure products. The vapor products formed by the reaction

then ﬂow towards the growing crystal where the reverse reaction takes place. At the growth

interface, the reverse reaction releases the transporting agent causing an increase in the

transporting agent’s partial pressure at that point. Th erefore, the pressure differentials of

the transporting agent and the crystal forming products force the net ﬂuxes of the zinc and

oxygen cont aining vapors to be from source to crystal while that of the transporting agent

is in the opposite direction. This counter ﬂow of gases does not normally exist in PVT and

results in a slightly more complex ﬂow system inside the ampoule.

Although not the only possibility, the transporting agent chosen for ZnO is hydrogen.

The reversible chemical reactio n of hydrogen with ZnO proceeds as:

ZnOðsÞþH

ðgÞ$ZnðvÞþH

OðvÞð7:1Þ

At the source (polycrystalline charge) end, the reaction proceeds from left to right (see

Figure 7.1) resulting in vaporization of the solid ZnO through usage of the hydrogen. At

the growth end, the reaction proceeds in the opposite direction with solid ZnO formed and

molecular hydrogen released. Transport rates of 5 10

4

mol cm

2

1

were achieved by

ZN Technology using a 20% mixture of hydrogen in helium.

While it would be possible to use pure hydrogen to increase the transport rate, growth of

high quality crys tals often requires reduced growth rates. To control the growth rate, the

SCVT growth of ZnO has been conducted with a mixture of inert gas and hydrogen. This

not only reduces the hydrogen partial pressure, a controlling factor in the transport rate, but

also provides a medium through which the vapors must diffuse to reach the growing

crystal. If conditions are set such that the transport rate is limited by the diffusion of the

vapors from one end of the ampoule to the other, independent control of the growth rate

and growth temperature can be achieved resulting in improved reproducibility and long

term stability of the system.

Faktor et al . developed a one-dimensional model for vapor transport and applied it to

crystal growth of CdS by PVT.

[11]

In their analysis, they used a combination of drift and

diffusion to describe the ﬂux of a vapor per unit cross-sectional area given by:

vap



vap

ð7:2Þ

Transport Theory and Comparison with Growth Data 173

where the term

vap

represents the drift of the vapor and

vap

represents the

diffusion term. Boone

[12]

originally applied these equations to the SCVT system to

determine the transport and growth rates as functions of the growth parameters. This work

was extended by ZN Technology to more closely model the current growth system. The

results of these calculations along with experimental data are shown in Figure 7.2 where the

growth rate of ZnO is plotted vs furnace temperature. The experimental values are averages

over all growth runs conducted at each temperature. Though the ﬁt of the calculated curve is

not exact, it does exhibit the maximum in growth rate seen in the data.

In addition to growth rate, it is important to maintain an optimum stoichiometry in

the vapor at the crystal growth interface for vapor grow th of high quality II–VI

compound crystals. In the case of the SCVT ZnO growth process, this means

controlling the ratio of water, the oxygen bearing compound, to zinc vapor. A set

of experiments by Reisma n et al.

[13]

involving the gas phase reaction of Zn and H

vapors in an helium carrier gas indicated the conditions for producing smooth and

transparent layers involved a

ratio of 10–50. Similar observations have been made

for metal organic chem ical vapor deposition (MOCVD) growth processes

[14,15]

except the

ratios were in the 4–8 range. Toaccomplish vaporstoichiometry control, an external water

vaporsourcewas addedtotheSCVTZnOgrowthsystem.Thissourcep ermitscontrolofthe

water vapor pressure in the g rowth ampoule. The one-dimensional growth model utilized

for this growth system includes the external water vapor source. In conjunction with the

crystal growth temperature the inﬂuence of the external water vapor source can be used to

change the ratio of water vapor to zinc vapor parti al pressures a t the g rowing crystal

interface from 1.5athighgrowthtemperaturesandlow external water vapor pressure to

80 at low growth temperatures and high external water vapor pressure. This is illustrated

in Figure 7.3 by the two curves representing the lowest and highest temperatures used for

ZnO crystal growth.

0.00

0.10

0.20

0.30

-250 -200 -150 -100 -50

50 100

Furnace Temperature (Deviation from Standard Temperature) (ºC)

Growth Rate (a.u.)

Figure 7.2 Comparison of experimental growth rates (points) with the growth rate (solid line)

as calculated by the thermodynamic model described in the text. Each point is the average

growth rate for that growth temperature

174 Vapor Transport Growth of ZnO Substrates and Homoepitaxy of ZnO Device Layers

7.3 Characterization

7.3.1 Crystallinity

High resolution X-ray omega rocking curves were used to check the crystallinity of the

SCVT ZnO. In this procedure, the crystal is rocked along the omega axis through the

Bragg peak and the detector is unmasked such that all of the diffracted radiation is

collected. The values measured are the full width at half-maximum (FWHM), the peak

height relative to the incident beam intensity, and the omega position of the Bragg peak.

The rocking curves resulting from this procedure are similar to those in Figure 7.4. The

100

0.0450.0350.0250.015

0.005

External H

O partial pressure (atm)

O/Zn vapor pressure ratio

Figure 7.3 The ratio of water vapor to zinc vapor partial pressures at the growth interface vs

the applied external water vapor pressure for two different growth temperatures. Upper curve:

240



C; lower curve: þ60



Omega

0.2

0.4

0.6

0.8

I/Io

24"

23"24"

23"

26"24"27"24"24"

22"

25"

Figure 7.4 X-ray rocking curves (omega) for one of the best SCVT ZnO crystals. The

measurements shown are for different spots along the diameter of the wafer and cover a

length of approximately 40 mm

Characterization 175

FWHM (shown in Figure 7.4 as the number above each peak in arc-sec) is measured as the

width of the peak at a vertical position half the total height of the peak. The rocking curves

shown in Figure 7.4 are from several positions across a single 40 mm diameter SCVT ZnO

wafer. The uniformity of the FWHM values and the peak heights are indicative of both the

high crystalline quality and excellent uniformity of the materia l. Although lower (better)

FWHM values have been measured in individual areas on a wafer, the crystalline quality

indicated by the FWHM values in Figure 7.4 should be sufﬁcient for excellent homo-

epitaxial results and subsequent device fabrication.

7.3.2 Purity

One characteristic of the SCVT ZnO growth process is the extremely high purity of the

resulting ZnO crystals. This can be very important for homoepitaxial growth of ZnO since

any impurities in the substrate may diffuse into the homoepitaxial ﬁlm and alter its

electrical characteristics. Glow discharge mass spectrometry (GDMS) was used to

characterize the impurities in the SCVT ZnO crystals. Table 7.1 shows the results for

the impurities detected in three samples from different growth runs. All other elements

were either not detectable down to a few parts per billion atomic (ppba) or were a part of

the sample matrix or sample mount such as zinc, oxygen, and tantalum. The majority of

the impurity content is silicon and nitrogen. Silicon is expected since the crystals are

grown in quartz ampoules. There has been no indication this level of silicon doping

changes the electrical properties of SCVT ZnO. The amount of nitrogen present in the

three samples shows signiﬁcant variation, but is consistently present. This nitrogen

presence may be related to imperfect sealing of the growth apparatus resulting in a minor

component of nitrogen in the growth atmosphere. Of the remaining detected impurities,

only aluminum and boron are known to be n-type dopants and their total concentration of

20–50 ppba is not likely to impact the doping of any epitaxial ﬁlms grown on these

substrates by outdiffusion from the substrate to the ﬁlm. This high purity is also evidenced

in the optical data discussed in a later section.

Table 7.1 Results of GDMS of one of the SCVT ZnO crystals. Only detected elements are

shown. Note that the data are shown in parts per million atomic (ppma)

Sample A Sample B Sample C

(ppma) (ppma) (ppma)

B 0.012 0.012 0.028

C 0.040 0.004 0.091

N 0.028 0.180 1.200

Na 0.015 ND ND

Al 0.009 0.007 0.020

Si 0.330 0.280 0.660

Ti 0.001 ND ND

Sn ND 0.077 ND

Pb 0.002 ND ND

ND, not detected.

176 Vapor Transport Growth of ZnO Substrates and Homoepitaxy of ZnO Device Layers

7.3.3 Electrical

All undoped ZnO grown by the SCVT method has been n-type conductive. Typical room

temperature electrical characteristics are 0.5 ohm cm with a mobility of 200 cm

1

and a net donor concentration of 6 10

3

. Growth parameters have been found

to cause the net donor concentration to change, though not drastically. Figure 7.5

illustrates the decrease observed in the net donor concentration as the growth temperature

decreases. Although the direct cause of this has not been investigated, as was shown in an

earlier section, the thermodynamic model describing the transport process predicts that the

O/Zn ratio at the growth interface will increase with decreasing growth temperature.

Thus, changes in the stoichiometry at the growth interface for different growth tempera-

tures may be the cause of this trend. Changes in mobility are also observed, but they appear

to be associated with the decrease in the net donor concentration. Figure 7.6 plots the

mobilities and net donor concentrations for all samples measured showing the apparent

correlation.

The conductivity of the bulk SCVT ZnO has been attributed to three donors by Look

et al.:

[16]

one with a 30 meV activation energy proposed to be due to a native defect

complex donor, the second at 44 meV activation energy associated with hydrogen

interstitials, and the third at 75 meV that is probably associated with a group III

impurity (most likely Ga or Al) on a Zn site. From Hall vs temperature measurements,

the density of each of these was estimated to be 4.5 10

3

for the 30 meV donor

level, 3.0 10

3

for the 44 meV donor level, and 2.0 10

3

for the 75 meV

level. The acceptor density was calcul ated to be 1.3 10

3

. Therefore, the room

temperature n-type conductivity of the SCVT ZnO is likely due to a combinatio n of the

three sources and the compensation level is low with acceptor densities more than a factor

of 20 lower than the donor densities. In addition, the peak electron mobility has been

measured at H2000 cm

1

at low temperatures after a thermal anneal that is thought

to remove hydrogen.

[17]

1 x10

100500-50-100-150

-200

Furnace Temperature

(Deviation from Standard Temperature) in (ºC)

Net Carrier Concentration (cm

−3

)

Figure 7.5 Plot of the net carrier concentrations for SCVT ZnO samples grown over a range of

furnace set point temperatures

Characterization 177

The donor level associated with hydrogen has been eliminated by annealing the SCVT

ZnO in air at temperatures aroun d 800



C. Look et al.

[16]

report elimination of this donor

level using a 715



C anneal. This resulted in a reduction of the room temperature net donor

concentration by a factor of 1.7. The results of annealing in air at 1000



C have been found

to reduce the room temperature net donor concentration by a factor of two to three.

[10]

7.3.4 Optical

SCVT grown ZnO has very good optical properties as measured by photoluminescence

and cathodoluminescence. At room temperature, the typical photoluminescence response

shows a strong UV peak at 378.6 nm along with a very low intensity visible (green)

luminescence.

[10]

Cathodoluminescence also shows the extremely low value of the visible

as compared with the UV luminescence. This is illustrated in Figure 7.7

[18]

where, in this

experiment, the cathodoluminescence visible peak height is less than 1% of the UV peak

height. Figure 7.7 also includes a comparison of the cathodoluminescence of SCVT ZnO,

hydrothermal ZnO, and melt grown ZnO showing that the SCVT ZnO ratio of visible/UV

luminescence is much better than the samples grown by the other two techniques.

Using low temperature cathodoluminescence, the large number of observable phonon

replicas in Figure 7.8

[18]

also indicates the extremely high optical quality of the SCVT

ZnO sample. The uniformity of the optical properties was also tested on this sample,

showing less than 0.1% variation in the cathodoluminescence when map ped across the

face of the wafer.

[18]

Similar luminescence spectra exhibiting phonon replicas are

observed using low temperature photoluminescence.

[19]

Room temperature, optically

pumped stimulated emission has also been observed in SCVT ZnO samples.

[10]

Figure 7.9

shows the spectra of the stimulated emission. The luminescence spectra are shown for

various pump densities from below to above the stimulated emission threshold. This

observation of room temperature stimulated emission in an SCVT ZnO substrate with no

special optical cavity preparati on demonstrates the exceptional optical quality of the

SCVT ZnO and the efﬁciency of ZnO’s radiative processes.

1 x 10

250230210190170150

Hall Mobility (cm

–1

)

Net Carrier Concentration (cm

–3

)

Figure 7.6 Plot of average net carrier concentrations of SCVT ZnO grown over a range of

furnace set point temperatures

178 Vapor Transport Growth of ZnO Substrates and Homoepitaxy of ZnO Device Layers

0 5 10 15 20

0.01

0.1

ZnO, SCVT, Zn

= 2.0 μW

ZnO, Hydrothermal, Zn

= 2.0 μW

ZnO, Melt Grown, Zn

= 2.0 μW

Green/NBE

Voltage (kv)

Figure 7.7 Ratio of green luminescence to near band edge (NBE) luminescence as a function

of electron beam energy for SCVT grown bulk ZnO (lower plot) and those of two other methods

showing that the SCVT ZnO has very low green luminescence. Reprinted from L. J. Brillson

et al., Nanoscale depth-resolved cathodoluminescence spectroscopy of ZnO surfaces and

metal interfaces, Superlattices and Microstructures 45, 206–213. Copyright (2009) with

permission from Elsevier

3.02.4

1.8

100

1000

10000

100000

1000000

3.3605 eV

3.3111 eV

4,6,8,9

Phonon Replicas

(0.071eV)

2.48 eV

Constant Power SEM CL Intensity at 12K

CL Intensity (a.u.)

Photon Energy (eV)

Electron Beam Energy

2keV

5keV

10keV

20keV

Figure 7.8 Cathodoluminescence (conducted at Ohio State University) of an (0001) SCVT

ZnO substrate. The sample shows no discernible dependence on electron beam energy

indicating that surface damage was not detectable. Reprinted from H. L. Mosbacker, et al.,

Role of subsurface defects in metal-ZnO(000) Schottky barrier formation, J. Vac. Sci. Technol.

25, 1405. Copyright (2007) with permission from American Vacuum Society

Characterization 179

7.4 In-situ Doping

Even though the SCVT ZnO is electrically conductive as grown, a higher conductivity

would be beneﬁcial for its use as a substrate for light-emitting diodes (LEDs) and lasers.

By adding n-type dopants to the growth system, it is possible to incorporate them during

growth. The key is to ﬁnd dopants or dopant compounds that have appropriate volatility in

the growth system with normal growth parameters. In addition, a method of control of the

doping rate must be devised to keep its vapor concentration constant for the entire growth

run to assure uniform doping. Group VII dopants such as ﬂuorine, chlorine and bromine

are potential candidates for use. However, they may also act as transporting agents,

complicating the growth system; not to mention that they are hazar dous and can severely

deteriorate the growth equipment during an accidental leak. Possible group III dopants are

boron, aluminum, gallium and indium. These are more benign but elemental forms are not

volatile at the growth temperatures typically employed in the SCVT syst em. Most reports

of highly doped, high conductivity ZnO utilize gallium as the dopant.

[20–22]

However,

calculation of the probable dopant compound vapor pressures in the SCVT growth system

showed that those most suited to maintaining a desirable and controllable vapor pressure

were compounds of boron and indium. Both have previously been reported for use in thin

ﬁlm deposition methods to increas e the conductivity of ZnO ﬁlms.

[23–25]

Figure 7.10

shows plots of the calculated ratios of In/Zn and B/Zn in the vapor for a range of growth

temperatures. As the ratio affects the degree of doping, there will be an optimum dopant to

Zn vapor pressure ratio. This is usually determined experimentally but one could estimate

that the optimum ratio should be in the 10

5

–10

3

range as a starting point. The optimum

value will also depend upon how effectively the dopant is incorporated into the growing

interface. The In/Zn ratio is higher than wha t is considered optimum but can be lowered

somewhat by maintaining a high growth temperature. The B/Zn vapor ratio is much lower

than the In/Zn ratio for the sam e temperatures. This ratio is also temperature dependent

with the lowest ratio being at the highest growth temperatures.

200

400

600

800

1000

1200

460

440

420

400

380

360

Wavelength (nm)

Intensity (a.u.)

Figure 7.9 Room temperature, optically pumped, stimulated emission from a centimeter

square, standard chemo-mechanically polished SCVT ZnO wafer. At low pumping power, the

emission peaks are broad

180 Vapor Transport Growth of ZnO Substrates and Homoepitaxy of ZnO Device Layers

ZnO crystals were doped in situ by this method with indium and boron. Both dopants

were successfully incorporated into ZnO ingots and the room temperature Hall measure-

ment data for the resulting substrates are shown in Table 7.2. While the indium seems to be

the more effective dopant of the two, giving a resistivity of the order of 10

3

ohm cm with a

mobility of the order of 40 cm

1

, the indium-doped ZnO was completely opaque for

this dopin g level. The boron-doped ZnO crystals showed a much larger decrease in the

room temperature Hall mobility (6–10 cm

1

) than did the indium-doped crystals but

the transparency of the ZnO was much better. Further work is needed to ﬁnd the correct

dopant and concentr ation to give the increased conductivity without having a dramatic

impact on the transparency of the ZnO.

7.5 ZnO Homoepitaxy

The primary use of SCVT ZnO is projected to be substrates for homoepitaxy of ZnO-based

optoelectronic devices, particularly LEDs and laser diode s (LDs). Therefore, the ultimate

test of the quality of the substrates is their ability to be used to produce high quality

epitaxial layers. The results of the homo epitaxy (deﬁned here as growth of ZnO-based

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1200-120-240

-360

Furnace Temperature

(Deviation from Standard Temperature) (ºC)

Dopant/Zn Ratio

In/Zn Ratio in the Vapor B/Zn Ratio in the Vapor

Figure 7.10 Calculated ratios of dopant atoms to Zn atoms in the vapor of the SCVT growth

system for boron and indium

Table 7.2 Room temperature Hall measurement results for boron- and indium-doped SCVT

ZnO. A typical undoped result is included for comparison

Resistivity

(ohm-cm)

Mobility

(cm

/Vs)

Net carrier

concentration (cm

3

)

In Doped 7.9 10

3

41.5 1.9 10

B Doped

1.2 10

1

9.5 5.4 10

2.6 10

1

5.9 4.1 10

Typical Undoped 5.0 10

1

200 6.0 10

ZnO Homoepitaxy 181