Wafer Characterization and Properties Measurement MSE-630

4-point probe r = 1/q(mnn+mpp) W-cm r= 2ps V/I MSE-630 Used to measure sheet resistivity Outer probe forces current through wafer; inner probes measure voltage drop r = 1/q(mnn+mpp) W-cm us. n>>p or p>>n, so only one term is of interest r= 2ps V/I Typically, 0.5-mm< s <1.5-mm MSE-630

2-point probe MSE-630 Useful to determine material type (n- or p-type) Apply two probes, one 25 – 100 oC hotter than other Thermally excited electrons flow away from hot probe, leaving holes and build up around cold electrode Measure Seebeck voltage using high impedance volt meter If material is p-type polarity will be reversed We can measure either short circuit current or open circuit voltage. Current for an n-type material is: Jn = qmnnPndT/dx Pn is thermoelectric power, either (-) for e- or (+) for h+ MSE-630

Hall effect MSE-630 Using Hall effect, we can determine material: Type Carrier concentration Carrier mobility Current, Ix, is forced through sample Results in measurable voltage drop, Vx r = wt/s Vx/Ix Applying a magnetic field, B, deflects electrons: F = q(x + v x B) electrons will be forced in –y direction MSE-630

Hall effect MSE-630 mH = ׀RH׀/r = ׀RH׀ s xy = Bz Ix/qwtn, Since no current flows in the y-direction, an electric field e must build to offset magnetic force: Fy = q(xy + vxBz) = 0 → xy = -vxBz Since vx = Ix/wtn, xy = Bz Ix/qwtn, or Vy = Bz Ix/qtn Define the Hall Coefficient: RH = tVy/BzIx = 1/qn → n = ± 1/qRH The “Hall mobility”, mH is mH = ׀RH׀/r = ׀RH׀ s Hall mobility is typically ~2 x e- or h+ mobility Consistent units for calculating Hall effect: V = volts A = Amps length = meters B = Tesla (1T = 104 Gauss = 1 V-s/m2) RH = m3/C MSE-630

K2Cr2O7 (1.5M):HF or Cr2O3 (.15M):HF Imaging Defects Dislocations and stacking faults introduced during processing and nucleated by oxygen, thermal stress and oxidation processes Chemical etchants reveal density, size and location of defects. Etchants attack areas with high chemical or strain energy and are visible with a microscope Etch Composition Sirtl Cr2O3 (5M): HF 1:1 Secco K2Cr2O7 (1.5M):HF or Cr2O3 (.15M):HF 1:2 Dash HF:HNO3:acetic acid 1:3:10 MSE-630

FTIR: Fourier Transform Infrared Spectroscopy Used to measure concentrations of O and C (down to ~1015/cm3) Molecules absorb energy at characteristic wavelengths E = hn = hc/l Si-O-Si absorbs at wave number 1106/cm C absorbs at wave number 607/cm MSE-630

IR beam split & follows two separate paths to sample and detector Moving mirror causes two beams to interfere constructively or destructively in a sinusoidal manner The Fourier transform of the signal will be a delta function proportional to incident beam intensity If the frequency of the source is swept, the FT of the resulting transform will produce an intensity spectrum If we insert a sample, the intensity spectrum will change because of absorption of specific wavelengths Scan of sample is compared to a baseline scan to identify absorbed frequencies MSE-630

Clean Room and Contamination Reduction Particles on wafers must be detected to <10 per wafer Blank wafers are scanned with a laser – particles will cause light to be reflected Detects particles ~ 0.2 mm Computer enhancement of scanned image produces a “map” of particles on wafers MSE-630

Test Structures Test structures, as shown below, can identify shorts and opens. A short between pads 1 and 2 indicate bridging due to a particle or etching lines An open between pads 1 and 4 is a break in a continuous line, probably due to a particle In a MOS capacitor, a thin insulator is grown and metal gates are patterned. Voltage applied to each capacitor structure and ramped up until dielectric breakdown occurs. Results are plotted on a histogram Premature failure usually due to contamination MSE-630

Surface Analysis Techniques Incoming e- collide inelastically with target e-, ejecting it from solid. These “secondary e-” have lowest energy and are used to generate an SEM image Highest energy e- “backscatter” elastically Energy released in an EL1→EK transition may b given to a 3rd e-, aka “Auger” electron (pronounced oh-jay) X-rays or emitted Auger e- carry a unique signature that is element dependent Lighter elements produce Auger e-, heavier elements produce x-rays Intermediate energy e- release an e- from inner bands of atom (L or K level). An incident e- causes an EK e- to be ejected. Then, an EL1 e- falls into the vacant EK level, producing an x-ray. This is XES or microprobe MSE-630

Surface Analysis Techniques Incident x-ray beam can cause an e- to be emitted (XPS) or fluoresce (XRF) when x-ray is emitted Rutherford backscattering (RBS) detects ions which are elastically scattered by atomic nuclei in the substrate when bombarded by an incident ion beam (usually He) SIMS (secondary ion mass spectroscopy) uses O+ or Cs+ as bombarding ions. This is the dominate means of determining the doping profile MSE-630

Method Attributes AES/XPS XES/XRF SIMS RBS MSE-630 Good depth resolution and surface sensitivity; limited depth XES/XRF Not surface sensitive, not good depth resolution, easy to implement on SEM SIMS Very sensitive, excellent depth resolution, destructive RBS Good depth resolution, sensitivity ~ 0.1 at%, poor lateral resolution MSE-630

Interface characterization MSE-630 Interface characterization Need to measure Thickness Dielectric constant Index of refraction Dielectric strength Defect density Three groups of measurements: Physical (destructive): etching and measuring using AFM or cross section and view on SEM or TEM Optical Electrical

Interface characterization – Optical Methods MSE-630 Interface characterization – Optical Methods For a constant wavelength, l, incident and reflected waves will add constructively for certain values of l, destructively for others lmin,max = 2n1cocos(b)/m m=1,2,3… for maxima m=1/2,3/2,5/2… for minima Conversely, b = sin-1(nosin(f1)/n1) For a fixed f, vary l to find lmax and lmin. Solve for co – dielectric thickness n1 must be known

Optical Characterization Methods Elipsometry Measures thicknesses <10nm Works similar to reflectance Uses polarized light and measures degree of change in polarization between surface and substrate n must be known, and substrate must be transparent to l Change in polarization only depends on film thickness and n Color Charts If white light is used to illuminate a surface, destructive interference causes a particular color to emerge in the reflected light Color of layer corresponds to thickness Resolution ~10-20 nm Good only for thicker samples MSE-630

Electrical Characterization Methods 4 types of charge at interface: Qm: mobile charge in oxide, usually metals (Na+ and K+) Qot: oxide trapped charge between Si-O-Si bonds, usually from ion implantation Qf: fixed oxide charge – incompletely oxidized Si with a (+) charge Qit: interface trapped charge. Incomplete O bonds like Qf, but may be (+), neutral or (-). Energy levels of trapped charge can be anywhere in the forbidden band gap, but usually close to band edges Charges associated with SiO2-Si system Density of Qf, Qit ~ 109 – 1011 cm-2 eV-1 Qot usually repaired by high temperature anneal MSE-630

Capacitance-Voltage Technique (CV) Electrical Characterization Methods Capacitance-Voltage Technique (CV) In (a), + VG attracts e- to Si surface. Apply small AC signal (10 kHz – 1MHz) and measure Cox (increase impedance due to oxide capacitance even though Si acts as a resistor in series with Cox. (b), continued Capacitance of depletion region: CD = xs/xD where xs = permittivity of Si, 11.9 Measured capacitance now CD in series with Cox (c) for larger values of VG, surface layer inverts from n-type to p-type. Negative VG attracts minority carrier holes in substrate to the surface to form an inversion layer of p-type carriers. Occurs at Vth, xD stops expanding, becomes xDmax In (b), -VG repels e- from surface creating a depleted region. Donor atoms have net + charge after losing e- and act to balance negative VG ׀QD׀=׀QG ׀ = NDxD where ND = doping in substrate, xD is depletion region depth, QD, QG: charge on gate fro donor atoms, #/cm2 accumulation depletion The gate charge must always be balanced by the substrate charge: QG = NDxD+QI Once QI forms, QD stops expanding MSE-630 inversion

Interface Capacitance, Continued Oxide prevents current flow (+) voltage in accumulation region places EF closer to Ec, at the surface. e- population is higher at surface than in bulk Negative gate voltage in depletion region causes bands to bend upward. This creates depletion region with depth xD. EF is far from EC and EV, and n and p are small, i.e., mobile carrier concentration << ND. In the inversion layer, EV is close to EF, resulting in holes building up in EC. Hole population is significant at surface – inversion has occurred and material is p-type. Now, increasing VG tries to move EV closer to EF, but since #p’s increase exponentially, QI offsets QG and QD = constant at xDmax MSE-630

Interface Capacitance, Continued In inversion region, we superimpose high frequency voltage on VG, causing QG to fluctuate. To balance charge, QI or QD must change. But, if we modulate faster than QI can respond, only QD changes. Charge balancing occurs between gate and bottom of depletion region, DQD = DQG. For any –VG in the inversion region xD = XDmax and CD = CDmin At low frequency (<1Hz) QI can follow changes in QG and measured capacitance is just Cox because the effective tops and bottoms of the capacitor are at the interface. Deep depletion occurs when rate of change in VG exceeds ability of QI to respond MSE-630

Charges at interfaces cause shifts in the CV curves DC are “charge traps”. Because HF traps cannot charge and discharge, but at LF they can, DC is proportional to trap density DC are “charge traps”. Because HF traps cannot charge and discharge, but at LF they can, DC is proportional to trap density As VG shifts, EF goes from EC to EV, and QIT traps fill and empty as EF moves through their energy levels, distorting CV curve. QF is the fixed positive charge in the oxide. This induces a mirror negative charge in the Si, making Si more n-type at the surface and harder to invert to p-type. Result: lateral shift by the amount qQF/Cox Additional lateral shift from work function, fMS. fMS is known and known for any experiment. DIT: interface state density # traps/cm2 eV MSE-630

Bias Temperature Stressing (BTS) Purpose: to characterize Qm – the mobile charge density of the insulator Mobile charges typically due to Na+ or K+ Make initial HF measurement at room temperature Heat to 200oC with VG applied. Na+ and K+ are highly mobile at this temperature and migrate up and down Keep under bias and cool to room temperature. Take second CV measurement Curve will shift laterally as it did with QF. The extent of shift is due to QM Two tests with opposite bias can be conducted to get total QM MSE-630

Breakdown Voltage Test Ramp V until dielectric breakdown. Thick oxides usually break down at ~15MV/cm in high quality SiO2. Same process can be used to measure tunneling currents in thin oxides (<10 nm) If DC currents are forced through a MOS capacitor for a period of time and CV measurements are taken before and after, shift in CV curve are due to QOT. A series of experiments with varying time will produce a plot of QOT vs time Since xD depends on doping, CV can be used to extract NA or NP vs. depth MSE-630

MSE-630 Measure Concentration vs. Depth to get: Channel doping profile Source-drain profile Well profiles field region profiles Use SIMS to measure primary doping profile and chemical concentration of dopants Usually large squares, 200 x 200-mm, are incorporated in layout and SIMS is performed on these Sputtered ions are accelerated, mass analyzed and counted to get depth profile Dopant profile resolutions are 1016-1017/cm3 n-type dopants (As, P, Sb) use Cs beam, p-type dopants (B, In) use O May be unreliable in thin regions. Primary ion beam energies typically 200eV to 5 keV. To do surface profiles to identify contaminants over first 50-nm, use lower beam energy and adjust the beam angle MSE-630

Spreading Resistance MSE-630 Purpose: to obtain doping profiles Technique: measure resistivity on beveled sample and compare to standard Measurements made with 2 probes on beveled samples at 2-10-mm increments MSE-630

Mechanical Measurements Differential stresses occur from: Thermal expansion Intrinsic stress from Ion bombardment Lattice constant mismatch Gas or inclusions on film s in interconnects lead to failures from: Cracking Loss of Adhesion Void and hillock formation MSE-630

Residual thermal stress Residual thermal stress can be calculated from: sf = (as-af)DT Yf/(1-nf) Thermal stresses can also be measured using x-ray diffraction to determine the strain in the lattice: sf = -Yf/2nf·(ao-as)/as Film may lose adhesion due to stresses. Adhesion often measured using the tape test. A more quantitative method to measure adhesion is to epoxy a pin onto the film, then pull on the pin with a calibrated weight or at a calibrated rate. A third method is to grow a 3rd layer on top of the film/substrate until adhesion is lost. In this method you must know top layer’s stress as a function of thickness. A third way to measure residual stress is from the curvature of the film: sf = 1/6R·Ysxs2/(1-ns)xf MSE-630

Electrical Characterization Methods Electrical measurements are used to measure: Sheet resistance Contact resistance on interconnects Dielectric breakdown voltage Contact reliability Contact resistance, RK, measured from V/I using the configuration at right. RK = V/I (W) then calculate contact resistance: RK = rc/l2 Note: Average resistance may also be measured in a series of interconnected contacts on test sheets. MSE-630

MTTF = A·J-nexp(EA/kT) Hillocks, Valleys and Median Time to Failure Hillocks and valleys lead to shorts and open circuits. They are due to residual thermal stresses, formation of second phases, and electromigration. Electromigration is current-induced hillock and void formation. H&V formation is measured using high T’s (~200oC) and I’s (1-2 MA/cm2) Voltage and resistance are monitored; a 20% change in resistance indicates failure Here, A = constant J = current density n ~ 2 typical (1-3) EA = Activation energy for migration, us. 0.5-0.8 eV Median time to failure, t50, is the time at which 50% cumulative failure occurs. MTTF = A·J-nexp(EA/kT) Extrapolate to room temperature MSE-630 Goal: 0.1% fail in not less than 10-20 years

MSE-630