The Influence of Thermal-Mechanical Effectson Resistance Changes During and After Electromigration Experiments
- PDF / 546,975 Bytes
- 7 Pages / 414.72 x 648 pts Page_size
- 79 Downloads / 148 Views
al. [13,14].
513 Mat. Res. Soc. Symp. Proc. Vol. 391 01995 Materials Research Society
(a)
S S
700- linear phase
160(b)
non-linear d
600. U=2.i0cm
1600 1400. 81200
e
500300-
•,
200100"
non-linear Cicrease
•
U=2MA/cm2]
0
0
10
20
400 20•0 0-
06 1'0 io 30 4O0 5
30 40
time (h) Figure 1: a) Ratio of stripes at 220°C with was switched off. b) j = 2 MA/cm 2 . After
600-
6O
time (h)
resistances (ROR, see after for a definition) observed in 9 A1-1%Si/TiN/Ti a current density j = 2 MA/cm 2 . After about 21 hours the stressing current Ratio of resistances of two lines stressed at 240' with a current density about 29 hours the stressing current was switched off.
In this paper step-temperature measurements without stress current are compared to EM tests at a stress current inducing the same temperature steps by means of Joule heating. The experimental results are obtained with different HRRMs. The goal is to demonstrate that, when typical stress conditions are used and even if stabilization thermal treatments are performed prior to an EM test, pure thermal effects (including mechanical stress variations due to temperature changes) may be an important component, or perhaps can dominate the non-linear resistance changes observed in the very early stages of EM and after EM. EXPERIMENTAL Experiments were performed on passivated (0.7pm thick PSG, deposited at 400'C), A1-1%Si (900nm)/TiN(60nm)/Ti(50nm) lines. Stripes of two different geometries and dimensions were patterned, namely a straight (ASTM) 0.9 pm wide and 8 0 0 aym long line, and a meandered 4.0 pm wide and 2460 pm long line. The two lines run very closely apart, in order to share thermal fluctuations. Average grain size was > 1.4 pm. EM measurements were performed by means of the wafer-level "ratio of resistances" (ROR) technique (for a detailed description see [2]). A high stressing current j = 2 MA/cm 2 was introduced in the narrow lines only, while wide lines were used as reference resistors monitoring temperature fluctuations. During the very rapid (•< 5 s) measurement cycles, the stressing current is switched off and a low, alternating current is introduced in both resistors. This assures that during the measurement cycles Joule heating is absent. Dices containing the narrow and wide lines were also mounted in ceramic packages. Two oil baths, characterized by a temperature stability better than 0.01 0 C, were prepared in order to perform in-situ step-temperature experiments [3]. Each sample underwent the temperature cycle reported in the inserts of Fig. 2. These measurements were also devoted to the investigation of possible differences in the resistance changes between wide (4.0,4m) and narrow (0.9pm) lines. ELECTROMIGRATION VS. STEP-TEMPERATURE MEASUREMENTS In this section resistance changes collected during EM tests are compared to step-temperature
514
•b)
(a)
200 -240
-
-200
0-200
li
2Mo -
0-200 -20
2W0
-400 -600 -
220 -11
1
IS15
180
T=232 C
T'232*C
-400
(
time (a.u.)
-600
-800 -0T=2I0C
T=2100 C T=2000C
-800 -
Data Loading...
